final report daad13 02 c 0015 part5 app a f
DESCRIPTION
TRANSCRIPT
ldquosfrdquo RTA P200105 DOD SBIR Phase II Contract Number DAAD13-02-C-0015 Final Report December 2005
74
Appendices A Lee YH and S Farquharson Rapid chemical agent identification by surface-enhanced Raman spectroscopy
SPIE 4378 21-26 (2001) B Farquharson S WW Smith YH Lee S Elliott and J F Sperry Detection of bioagent signatures A
comparison of electrolytic and metal-doped sol-gel surface-enhanced Raman media SPIE 4575 62-72 (2002)
C Farquharson S P Maksymiuk K Ong and SD Christesen Chemical agent identification by surface-enhanced Raman spectroscopy SPIE 4577 166-173 (2002)
D Farquharson S A Gift P Maksymiuk and F Inscore ldquoRapid dipicolinic acid extraction from Bacillus spores detected by surface-enhanced Raman spectroscopyrdquo Applied Spectroscopy 58 351- 354 (2004)
E Farquharson S A Gift P Maksymiuk F Inscore and W Smith ldquopH dependence of methyl phosphonic acid dipicolinic acid and cyanide by surface-enhanced Raman spectroscopyrdquo SPIE 5269 117-125 (2004)
F Farquharson S A Gift P Maksymiuk F Inscore W Smith K Morrisey and SD Christesen ldquoChemical agent detection by surface-enhanced Raman spectroscopyrdquo SPIE 526916-22 (2004)
G Inscore F A Gift P Maksymiuk and S Farquharson ldquoCharacterization of chemical warfare G-agent hydrolysis products by surface-enhanced Raman spectroscopyrdquo SPIE 5585 46-52 (2005)
H Farquharson S A Gift P Maksymiuk and F Inscore ldquoSurface-enhanced Raman spectra of VX and its hydrolysis productsrdquo Applied Spectroscopy 59 654-660 (2005)
I Inscore FE AD Gift Stuart Farquharson ldquoDetect-to-treat development of analysis of Bacilli spores in nasal mucus by surface-enhanced Raman spectroscopyrdquo SPIE 5585 53-57 (2005)
J Farquharson S W Smith C Brouillette and F Inscore ldquoDetecting Bacillus spores by Raman and surface-enhanced Raman (SERS) spectroscopyrdquo Spectroscopy June supplement 8-15 (2005)
K Inscore F A Gift P Maksymiuk JF Sperry and S Farquharson ldquoIdentifying surfaces contaminated with Bacillus spores using surface-enhanced Raman spectroscopy to detect dipicolinic acidrdquo in Applications of Surface-Enhanced Raman Spectroscopy Ed S Farquharson CRC Press Boca Raton FL accepted
L Christesen S K Spencer S Farquharson F Inscore K Gonser J Guicheteau ldquoSurface-enhanced Raman detection of chemical agents in waterrdquo in Applications of Surface-Enhanced Raman Spectroscopy Ed S Farquharson CRC Press Boca Raton FL accepted
M Farquharson S F Inscore S Christesen ldquoDetecting chemical agents and their hydrolysis products in waterrdquo in Surface-Enhanced Raman Scattering ndash Physics and Applications Eds K Kneipp M Moskovitz and H Kneipp Springer accepted
N Inscore F S Farquharson ldquoDetecting hydrolysis products of blister agents in water by surface-enhanced Raman spectroscopyrdquo SPIE 5993 19-23 (2005)
O Inscore F P Maksymiuk S Farquharson ldquoSurface-enhanced Raman spectroscopic characterization of the chemical warfare agent vesicant HD and related mono-sulfidesrdquo JRS in preparation
P ROC curve data from measurements of CN HD and VX at the US Armyrsquos Edgewood ChemBio Center
SPIE-4378-2001
21
Rapid chemical agent identification by surface-enhanced Raman spectroscopy
Yuan-Hsiang Lee and Stuart Farquharson
Real-Time Analyzers 87 Church Street East Hartford CT 06108
ABSTRACT Although the Chemical Weapons Convention prohibits the development production stockpiling and use of chemical warfare agents (CWAs) the use of these agents persists due to their low cost simplicity in manufacturing and ease of deployment These attributes make these weapons especially attractive to low technology countries and terrorists The military and the public at large require portable fast sensitive and accurate analyzers to provide early warning of the use of chemical weapons Traditional laboratory analyzers such as the combination of gas chromatography and mass spectrometry although sensitive and accurate are large and require up to an hour per analysis New chemical specific analyzers such as immunoassays and molecular recognition sensors are portable fast and sensitive but are plagued by false-positives (response to interferents) To overcome these limitations we have been investigating the potential of surface-enhanced Raman spectroscopy (SERS) to identify and quantify chemical warfare agents in either the gas or liquid phase The approach is based on the extreme sensitivity of SERS demonstrated by single molecule detection a new SERS material that we have developed to allow reproducible and reversible measurements and the molecular specific information provided by Raman spectroscopy Here we present SER spectra of chemical agent simulants in both the liquid and gas phase as well as CWA hydrolysis products Keywords Chemical warfare agent simulant hydrolysis product SERS Raman spectroscopy sol-gels vapor
1 INTRODUCTION Chemical warfare has been banned since the 1925 Geneva Protocol yet the use of chemical agents has persisted1 This can be attributed to the simplicity in manufacturing ease of deployment and the relatively low cost of chemical warfare agents (CWAs) These attributes make these weapons especially attractive to low technology countries and terrorists Well known examples include the large-scale use of tabun (GA) during the Iran-Iraq war (1984-1948)2 and the release of sarin (GB) in the Tokyo subway in 1995 The latter is the first documented terrorist use of a chemical weapon34 This ever-present threat was again substantiated by the United Nations Special Commissions report that described Iraqrsquos facilities for nerve agents anthrax and small pox production5-7 These uses of chemical weapons have motivated the development of fast and accurate analytical techniques to warn soldiers and the public at large The development of these analytical techniques is challenging in that these techniques must not only measure extremely low concentrations quickly (microgramliter in lt 1minute) but must also be capable of measuring both gas phase and liquid phase to be effective The latter is required since chemical agents can also be used to poison water supplies89 The earliest technologies developed for CWA detection were based on electrochemical ionization or colorimetric analysis (eg phosgene tape) Although these analyzers were easy to use they were not generally agent specific and suffered from false-positives1 More traditional laboratory methods have also been investigated and in particular combined gas chromatography and mass spectrometry (GCMS) has been very successful at eliminating false-positives1011 However GCMS requires extraction repeated calibration and long analysis times (typically 20 to 60 minutes)11 making it labor intensive and less than desirable for field use More rapid analysis of agents in the solid liquid and gas phase has been demonstrated by vibrational spectroscopy12-15 Hoffland et al12 reported infrared absorbance spectra and absolute Raman cross sections for several chemical agents while Christesen measured Raman cross sections for sarin tabun mustard gas and VX (ethyl S-2-diisopropylamino ethyl methylphosphonothioate)16 Again however these techniques also have limitations Raman spectroscopy is simply not a very sensitive technique and detection limits are typically 01 (1000 ppm) While infrared spectroscopy would have limited value in analyzing poisoned water since the very strong infrared To whom correspondence should be addressed emailfarqureal-time-analyzerscom
SPIE-4378-2001
22
absorption of water would obscure most other chemicals present Nevertheless efforts to overcome these limitations have been demonstrated Braue and Pannella13 quantified the G-series nerve agents (tabun sarin and soman) in terms of infrared attenuated total reflectance using a circle-cell And Alak and Vo-Dinh demonstrated the possibility of surface-enhanced Raman spectroscopy (SERS) to identify CWAs by measuring several organophosphonates that simulate the nerve agents17 However quantitative measurements have not been demonstrated for the SER-active material used (silver coated on alumina particles) or other SER-active media18 Recently we developed silver-doped sol-gels to promote the SER effect19-22 The porous silica network of the sol-gel matrix offers a unique environment for stabilizing SER-active metal particles and the sol-gel provides a high surface area that effectively increases the number of molecules observed within the Raman scattering volume The silver-doped sol-gels have been coated on the internal walls of standard glass vials to allow rapid SER analysis of small sample quantities (lt 01 mL) without preparation We have used p-aminobenzoic acid (PABA) as a test chemical to demonstrate surface enhancements greater than 107 reversible measurements in a flowing system reproducible measurements from vial-to-vial and batch-to-batch and measurements in multiple solvents including water19-22 Here we present preliminary measurements of chemical agent simulants in both the liquid and gas phases as well as chemical agent hydrolysis products using our SER-active vials
2 EXPERIMENTAL The chemical agent simulants employed were obtained at their purest commercially available grade from Aldrich (Milwaukee WI) and were dissolved in water or methanol for analysis All chemicals used to prepare the silver-doped sol-gels were spectroscopic grade and also purchased from Aldrich The sol-gel vials were coated in a manner similar to that previously reported by adding ammonium hydroxide to a solution of silver nitrate tetramethyl orthosilicate and methanol22 After mixing 02 mL of the sol-gel solution was transferred into a glass vial (2 mL) dried and heated The incorporated silver ions were then reduced using dilute sodium borohydride The vials were washed and dried prior to the addition of a sample solution The patent pending SER-active vials are commercially available from Real-Time Analyzers (Simple SERS Sample Vials RTA East Hartford CT) Dimethyl metylphosphonate (DMMP) pinacolyl methylphosphonate (PMP) and methylphosphonic acid (MPA) were prepared in aqueous solution while 2-chloroethyl ethyl sulfide (CEES) was prepared in methanol at 1 mM for SERS measurements Neat samples were employed for normal Raman measurements All samples were prepared in a chemical hood and transferred into plain or SER-active vials for analysis Special precaution was followed for CEES since it is a severe blistering agent23 Once prepared the vial was placed into the sample compartment of a Raman spectrometer for analysis A prototype Fourier transform Raman spectrometer (RTA) was used for these measurements24 The system consisted of a NdYAG laser (Brimrose) for excitation at 1064 nm an interferometer built by On-Line Technologies (OLT East Hartford CT) for frequency separation an uncooled InGaAs detector for signal detection (RTA) and an Intel 400 MHz Pentium II based laptop computer (Dell Round Rock TX) for interferometric control data acquisition (OLT) and analysis (LabVIEW by National Instruments Austin TX) Additional components included a Notch filter (Kaiser Ann Arbor MI) and interferometer entrance and exit optics (Edmund Scientific Barrington NJ) Fiber optics were used to deliver the excitation beam to the sample and the scattered radiation to the interferometer (1 meter lengths of 200 and 365 micron core diameter respectively Spectran Avon CT) A second Notch filter (Kaiser) was used as a beam splitter to direct the excitation beam along the same axis as the collected radiation A microscope object (20x04 Newport Irvine CA) was used to focus the beam into the sample and to collect the scattered radiation back along the same axis In this co-axial backscattering arrangement the excitation beam was passed through the outside of a glass vial and focused onto the silver-doped sol-gel film (01 mm thickness) containing the sample
3 RESULTS AND DISCUSSION As a prelude to chemical agent measurements in water we evaluated the quantitative performance of the SER-active vials by measuring PABA over the concentration range from 10-7 M to 10-2 M Figure 1 shows the spectra for 7 35 and 70 micromolar concentrations while Figure 2 shows a plot of the 1450 cm-1 band intensity as a function of concentration The SER response is linear over nearly three orders of magnitude to just over 10-4M at which point the band intensity suggests that the silver surface is becoming saturated
SPIE-4378-2001
23
In an effort to demonstrate the broad capabilities of the SER-active vials to measure chemical agents spectra of a nerve agent simulant dimethyl methylphosphonate a mustard gas simulant 2-chloroethyl ethyl sulfide and hydrolysis products pinacolyl methylphosphonate and methylphosphonic acid were collected DMMP is widely used by the US Army as a chemical warfare simulant because its chemical structure volatility and water solubility are similar to those of nerve agents25 DMMP is completely miscible and stable in water at room temperature26 Figure 3 compares the SER spectrum to the normal Raman spectrum of DMMP A number of the normal Raman bands are SER-active such as the P-C stretching mode which shifts from 715 to 735 cm-1 and the C-H stretching modes at 2855 2930 2960 and 3000 cm-1 which shift slightly Surprisingly the P=O stretching mode at 1250 cm-1 virtually disappears However the most dramatic change is the appearance of an intense triplet in the SER spectrum near 1000 cm-1 The bands at 1000 cm-1 1030 cm-1 and 1075 cm-1 likely involved the P-O-C bond This is supported by the nearly identical triplets observed for the SER spectra of fonofos and fonofoxon1719 It is also worth noting that a band appears at 425 cm-1 in the SER spectrum that may be unique to DMMP and useful for identification The enhancement factor is estimated at 120000 based on the normal Raman and SER P-C band intensity taking into account the difference in sample concentrations and spectral acquisition conditions A detection limit based on a signal-to-noise ratio of 3 can be estimated at 16 ppm
10-7 10-6 10-5 10-4 10-3 10-2 10-110-2
10-1
100
101
102
Figure 1 SER spectra of A) 70 B) 35 and C) 7 micromolar p-amino benzoic acid in water Conditions 80 mW of 1064 nm laser excitation 100 averaged scans (15 min) at 8 cm-1 resolution
Figure 2 SER spectral intensity for p-aminobenzoic acid as a function of concentration using RTA SER-active vials
Concentration (M)
Figure 3 A) SER and B) normal Raman spectra of dimethyl methylphosphonate Conditions SERS as in Figure 1 normal Raman 500 mW and 200 scans
Figure 4 A) SER and B) normal Raman spectra of 2-chloroethyl ethyl sulfide Conditions as in Figure 3
Wavenumbers (∆cm-1)
CH3O-P-OCH3
=
O
_
CH3
CH3O-P-OCH3
=
O
_
CH3
Cl-CH2-CH2-S-CH2-CH3 A
B
A
B
A
B
C
Wavenumbers (∆cm-1)
Wavenumbers (∆cm-1)
500 1000 1500 2000
SPIE-4378-2001
24
2-Chloroethyl ethyl sulfide a blister agent simulant has a chemical structure similar to the mustard gas (Cl-CH2-CH2-S-CH2-CH2-Cl) with only one terminal chlorine Due to its low solubility in water CEES was dissolved in methanol for the SER measurement Again the prominent Raman modes are SER-active and even maintain relative intensity (Figure 4) The primary difference is that the SER bands appear to broaden such that the triplet near 700 cm-1 becomes a doublet and the shoulders at 2875 and 2970 cm-1 become less defined Again the latter bands are assigned to C-H stretching modes A single band at 700 cm-1 which is attributed to the C-S-C asymmetric stretch dominates the reported infrared spectrum of mustard gas12 A corresponding symmetric stretch is reported at 705 cm-1 in the Raman spectrum of mustard gas27 Here a corresponding symmetric stretch appears but as a doublet at 700 and 755 cm-1 presumably due to the loss in symmetry for CEES The band at 655cm-1 can also be confidently assigned to a C-Cl stretch The SER spectral bands at 620 and 730 cm-1 are probably due to the same modes ie C-Cl and C-S-C stretches respectively The enhancement factor for CEES was somewhat less than DMMP at approximately 62000 as is the estimated detection limit of 22 ppm The ability to rapidly detect trace quantities of chemical agents in the gas phase would be invaluable as an early warning system Although the Raman scattering cross-sections for the nerve agents suggest that remote detection by Raman-based LIDAR is unlikely16 a SER-based system for perimeter monitoring could prove successful As a preliminary measurement we prepared a 10 by volume solution of CEES in methanol exposed a SER-active vial to the equilibrium vapor phase in a sealed jar and monitored the SER spectrum as a function of time Initially the vial was removed through a transfer chamber every hour to record the SER spectrum After ten hours spectra were recorded only every ten hours As illustrated by Figure 5 the sol-gel performed as a dosimeter in that the spectra increased as a function of exposure time The most intense SER bands at 620 and 2930 cm-1 are discernable in the first few hours The spectrum after 40 hours is nearly identical to the solution phase spectrum except for a diminished intensity of the 730 cm-1 band This may be due to methanol solvation effects or surface-orientation effects Based on the relative concentrations of methanol and CEES and their partial pressures we estimate the equilibrium concentration of CEES to between 1 and 2 micromolar Although not shown this concentration could be detected in one hour As previously stated the analysis of chemical agents in water is important in identifying poisoned water It is also important to decommissioning activities in which agents are destroyed by hydrolysis (acid or base) Furthermore any analytical technique used must be capable of distinguishing between parent CWA and hydrolysis products to assess safety or effectiveness of decommissioning For example soman has a hydrolysis half-life of ~23 hours at ambient temperatures and neutral pH28 and forms hydrofluoric acid (somewhat toxic) and pinacolyl methylphosphonate (relatively non-toxic)2930 PMP further hydrolyzes to form methyl phosphonic acid and 33-dimethyl-2-butanol (both non-toxic) The structural similarities between soman PMP and MPA are expected to produce similar Raman as well as SER spectra Figure 6 compares PMP and MPA but not the highly toxic parent CWA soman As with DMMP the P-C stretch the P-O-C mode and C-H stretches are readily apparent Yet it is worth noting that the band positions are reasonably different The former two bands appear at 764 and 1042 cm-1 for MPA while they are at 788 and 1032 cm-1 for PMP More importantly a unique band at 546 cm-1 as yet unassigned appears in the PMP spectrum
Figure 6 SER spectra of A) methyl phosphonic acid and B) pinacolyl methylphosphanate (note unique band at 546 cm-1) Spectral conditions as in Figure 1
Figure 5 SER spectra of 2-chloroethyl ethyl sulfide vapor as a function of time (10 hour increments to top which is 40 hours) Bottom trace is a blank Spectral conditions as in Figure 1
A
B
Wavenumbers (∆cm-1) Wavenumbers (∆cm-1)
HO-P-OH
=
O
CH3
_
HO-P-OH
=
O
=
O
CH3
_
CH3
_HO-P-O-CH-C-CH3
=O
CH3
_
CH3
_CH3_CH3
_
HO-P-O-CH-C-CH3
=O
=O
CH3
_
CH3
_
CH3
_
CH3
_CH3_CH3
_CH3
_
SPIE-4378-2001
25
4 CONCLUSIONS We have successfully measured the SER spectra of chemical agent simulants dimethyl metylphosphonate and 2-chloroethyl ethyl sulfide and chemical agent hydrolysis products pinacolyl methylphosphonate and methylphosphonic acid using silver-doped sol-gel coated sample vials Measurements were obtained in both aqueous and gas phase The P-C stretching mode was SER-active for all four chemicals allowing identification by class Within this group each chemical contained at least one unique spectral band that could be used for identification (Table 1) Furthermore these bands do not appear to coincide with SER spectra reported for organophosphorus pesticides the most likely source of false-positives Although surface enhancement factors appear to be an order of magnitude better than those previously presented in the literature for similar chemicals17 measurement sensitivity needs to be improved by 1 to 2 orders of magnitude to provide adequate warning of chemical agent use Current research efforts to increase surface-enhancement optical collection efficiency and instrument design are being pursued to achieve the required sensitivity
Table 1 Enhancement factors detection limits and unique SER bands fro chemicals studied Agent Simulant Enhancement Detection limit Unique bands (cm-1)
Dimethyl methylphosphonate 120000 90 microM (16 ppm) 425 2-Chloroethyl ethyl sulfide 62000 60 microM (22 ppm) 620
Methylphosphonic acid 110000 3 microM (60 ppb) 764 1042 Pinacolyl methylphosphonate 150000 70 microM (14 ppm) 546 788 1032
5 ACKNOWLEDGEMENTS The authors would like to thank Drs Janet Jensen and Steven Christesen of Aberdeen Proving Ground for encouraging this work They would also like to thank Advanced Fuel Research for making their laboratory facilities available
6 REFERENCES 1 ldquoThe Chemical Weapons Convention ndash A Guided Tour the Organization for the Prohibition of Chemical Weaponsrdquo at
httpwwwopcwnlguidehtm 2 Robinson JP and J Goldblat Chemical Warfare In The Iraq-Iran War Stockholm International Peace Research
Institute Fact Sheet at httpprojectssiprisecbwresearchfactsheet-1984html (1984) 3 ldquoChemistry of GB (Sarin)rdquo at httpwwwmitretekorgmissionenvenechemicalagentssarinhtml 4 Tu Anthony ldquoOverview of Sarin Terrorist Incidents in Japan in 1994 and 1995rdquo 6th CBW Protection Symposium
Stockholm Sweden 10-15 May 1998 5 Staff Reporter ldquoGoing out with a bangrdquo Newsweek June 28 1999 6 See UNSCOM reports in httpwwwunorgdeptsunscom (1999) 7 Treven T Saddamrsquos Secrets Harper Collins (1999) 8 ldquoDecaying Sarin-filled Rockets Spark Fearsrdquo Janersquos Defense Weekly 25(20)3 (1996) 9 ldquoThe Chemical Weapons Convention Redefines Analytical Challengerdquo Analytical Chemistry News amp Features June 1
397A (1998) 10 Johnston RL Hoefler CM Fargo JC and Moberley B ldquoThe Defense Nuclear Agencyrsquos ChemicalBiochemical
Weapons Agreements Verification Technology Research Development Test and Evaluation Program and its Requirements for On-Site Analysisrdquo AT-ONSITE 5-8 (1994)
11 Black RM Clarke RJ Read RW and Reid MT ldquoApplication of gas chromatography-mass spectrometry and gas chromatography-tandem mass spectrometry to the analysis of chemical warfare samples found to contain residues of the nerve agent sarin sulphur mustard and their degradation productsrdquo J Chromatography 662 301-321 (1994)
12 Hoffland LD Piffath RJ Bouck JBrdquoSpectral signatures of chemical agents and simulantsrdquo Optical Engineering 24 982-984 (1985)
13 Braue EHJ Pannella MGrdquoCIRCLE CELL FT-IR Analysis of Chemical Warfare Agents in Aqueous Solutionsrdquo Applied Spectroscopy 44 1513-1520 (1990)
SPIE-4378-2001
26
14 Tseng C-H Mann CK and Vickers TJ ldquoDetermination of Organics on Metal Surfaces by Raman Spectroscopyrdquo
Applied Spectroscopy 47 1767-1771 (1993) 15 Staff Reporter ldquoUS Army Tests Lidar to Detect Biological Toxinsrdquo Photonic Spectra pg 50 December 1998 16 Christesen SD Raman cross sections of chemical agents and simulants Applied Spectroscopy 42 318-321 (1988) 17 Alak AM and Vo-Dinh T ldquoSERS of Organophosphorous Chemical Agentsrdquo Analytical Chemistry 59 2149-2153
(1987) 18 Norrod KL Sudnik LM Rousell D and Rowlen KL ldquoQuantitative Comparison of Five SERS Substrates
Sensitivity and Detection Limitrdquo Applied Spectroscopy 51 994-1001 (1997) 19 Lee Y and Farquharson S ldquoSERS Sample Vials Based on Sol-Gel Process for Trace Pesticide Analysisrdquo SPIE 4206
140-146 (2000) 20 Farquharson S and Lee Y ldquoTrace Drug Analysis by Surface-Enhanced Raman Spectroscopyrdquo SPIE 4200-16 (2000) 21 Lee Y Farquharson S and Rainey P M Surface-Enhanced Raman Sensor for Trace Chemical Detection in Water
SPIE 3857 76-84 (1999) 22 Lee Y Farquharson S Kwong H and Shahriari M ldquoSurface-Enhanced Raman Sensor for Surface-Enhanced Raman
Spectroscopyrdquo SPIE 3537 252-260 (1998) 23 see Material Safety Data Sheets for details 24 Farquharson S Smith W Carangelo R C and Brouillette C ldquoIndustrial Raman Providing Easy Immediate Cost
Effective Chemical Analysis Anywhererdquo SPIE 3859 14-23 (1999) 25 Bennett S Bane J Benford P and Pratt R ldquoEnvironmental Hazards of Chemical Agent Simulantsrdquo Aberdeen
Proving Ground Maryland Chemical Research and Development Center CRDC-TR-84055 (1984) 26 Mabey W and Mill T Critical Review of Hydrolysis of Organic Compounds in Water under Environmental Conditions
Journal of Physics and Chemistry Reference Data 7(2) 383-414 (1978) 27 Christesen S MacIver B Procell L Sorrick D Carabba M and Bello J ldquo Noninstrusive Analysis of Chemical Agent
Identification Sets Using a Portable Fiber-Optic Raman Spectrometerrdquo Applied Spectroscopy 53 850-855 (1999) 28 Meylan WM and Howard PH J Pharm Sci 84 83-92 (1995) 29 Jenkins A Uy O and Murray G ldquoPolymer-Based Lanthanide Luminescent Sensor for Detection of Hydrolysis Product
of the Nerve Agent Soman in Waterrdquo Analytical Chemistry 71 373-378 (1999) 30 Nassar A Lucas S and Hoffland L ldquoDetermination of Chemical Warfare Agent Degradation Products at Low-Part-
per-Billion Levels in Aqueous Samples and Sub-Part-per-Million Levels in Soils Using Capillary Electrophoresisrdquo Analytical Chemistry 71 1285-1292 (1999)
SPIE 2001-4575
62
Detection of bioagent signatures A comparison of electrolytic and metal-doped sol-gel surface-enhanced Raman media
Stuart Farquharson Wayne Smith and Yuan Lee
Real-Time Analyzers 87 Church Street East Hartford CT 06108
Susan Elliott and Jay F Sperry University of Rhode Island 45 Lower College Rd Kingston RI 02881
ABSTRACT Since September 11 2001 the threat of terrorist attacks and biological warfare within US borders has become a sobering reality In an effort to aid military personnel and the public at large we have been investigating the utility of surface-enhanced Raman spectroscopy (SERS) to provide rapid identification of chemical agents directly and biological agents through their chemical signatures This approach is based on the ability of Raman spectroscopy to identify molecular structure through the abundant vibration information provided in spectra and the ability of SERS to detect extremely low concentrations (eg part-per-billion) through the enhancement of Raman scattering by six orders of magnitude or more Towards the goal of developing a portable analyzer we have been studying the ability of two SER media to obtain continuous (ie reversible) and quantitative (ie reproducible) measurements Here we compare measurements of nucleic acid-bases adenosine monophosphate and ribonucleic acid extracted from Escherichia coli Bacillus subtilis and Staphylococcus aureus obtained by electrolytic SERS and metal-doped sol-gel SERS The capabilities of these SER media are summarized in terms of rapid detection of B anthracis and dipicolinic acid Keywords bioagent detection SERS RNA analysis bacterial analysis Raman spectroscopy
1 INTRODUCTION The recent distribution of anthrax through the US postal system and the subsequent infection and death of several postal and national media employees amplifies the need for methods to rapidly detect and identify this and other chemical and biological warfare agents (BWA) The primary methods currently used immunoassays for screening and nucleic acid (NA) sequencing for positive identification of BWAs (bacteria protozoa and viruses) have serious limitations123 Immunoassay methods employ competitive binding of the bioagent (as an antigen) and its labeled (eg fluorescence) conjugate for a limited number of antibodies Although this analysis method is fast and semi-quantitative other chemicals may compete for the antibodies interfere with the enzymatic reaction or interfere with the measurement (eg it fluoresces) resulting in a high number of false positive responses1 Furthermore the antibodies denature due to moisture and heat limiting shelf life and require sterile often refrigerated storage Positive identification of a BWA can be accomplished by sequencing deoxyribonucleic acid or ribonucleic acid (DNA and RNA)23 This requires enumeration of a nucleic acid sequence through polymerase chain reactions (PCR) or multiplication of the microorganism through culture growth to provide sufficient quantities of DNA or RNA for analysis Unfortunately PCR and culture growth require from several hours to several days23 Consequently a wide variety of technologies have been investigated for rapid identification of BWAs The Department of Defense is actively monitoring 200 such technologies4 This includes traditional methods such as gas chromatographic separation coupled with ion mobility spectrometry detection5 to exotic methods based on nature such as monitoring toxin induced color changes in fish scales6 Although all of these techniques have achieved varying degrees of success none are yet capable of detecting and identifying BWAs in 10 minutes or less Towards this goal we have been investigating the ability of SERS to detect sub-nanogram quantities of DNA or RNA (eliminating enumeration) determine relative NA base concentrations and identify BWA taxonomy
To whom correspondences should be addresses e-mailfarqureal-time-analyzerscom wwwreal-time-analyzerscom
SPIE 2001-4575
63
Raman spectroscopy has a rich history of investigating biochemical and biological processes7 Some of the earliest laser-Raman studies demonstrated that the five NA bases adenine (A) cytosine (C) guanine (G) thymine (T in DNA) and uracil (U in RNA) yielded distinct spectra with several bands suitable for identification and quantification8 Furthermore these studies included exceptional spectra of both DNA and RNA for which the NA bases as well as several phosphate bands were easily identified9 However since the Raman effect is very inefficient (very low conversion of incident radiation to inelastically scattered Raman radiation) these samples had to be highly concentrated Fortunately two phenomena exist that can increase the generation of Raman photons by six orders of magnitude or more known as the resonance Raman and surface-enhanced Raman effects1011 Resonance Raman scattering occurs when the laser excitation wavelength is in resonance with an electronic transition of a molecule (within the absorption envelope)10 Excitation at ultraviolet wavelengths has been used to obtain resonance Raman spectra of amino acids and nucleic acids in whole bacteria1213 For example excitation at 242 nm has been used to maximize the nucleic acid spectral band intensities and minimize the amino acids band intensities A peak at 1530 cm-1 was found to be proportional to the amount of the NA base cytosine while a peak at 1485 cm-1 was proportional to the combined amount of the NA bases adenine and guanine This quantitative behavior has been used to define an A+TG+C base-pair ratio and provide a level of bacterial identification as taxonomic markers13 In recent years SERS has also been used to analyze bacterial cell components14 including amino acids15 lipids16 nucleic acids151718 and the adenine derivatives192021 SERS has proven to be one of the most sensitive methods for trace chemical analysis through the detection of single molecules2223 including DNA (dye labeled 17-mer)24 Since its discovery in 197425 the mechanism responsible for the large increase in scattering efficiency has been the subject of considerable research2627 Briefly incident laser photons couple to free conducting electrons within a metal which confined by the particle surface collectively cause the electron cloud to resonate2628 These surface plasmons are known as the physical component of the SER effect These surface plasmons can transfer energy to the molecular vibrational modes of molecules through interactions with the molecular electron orbitals2629 This interaction is known as the chemical component of the SER effect This perturbation of the molecular polarizability generates surface-enhanced Raman photons26 A number of methods have been developed to produce surfaces or solutions containing one of these metals with optimum roughness or diameter to promote SERS30 These methods include preparation of activated electrodes in electrolytic cells 1131 activated silver and gold colloid reagents32 and metal coated substrates333435 Selecting a SER-active medium for chemical and biological agent detection requires consideration of the method of deployment and hence the method of sampling Chemical aerosols or airborne bacteria will require a collection device to concentrate and continuously present the sample to the SERS medium Poisoned water supplies will also require a flow through device for continuous monitoring or a grab-sample device for periodic analysis And contaminated surfaces will require a grab-sample extractive device A SERS-based device used for continuous monitoring (air or water) must be reversible and reproducible if quantitative measurements are desired while a SERS-based device used for periodic sampling (water or surfaces) must be reproducible Both reversible and reproducible measurements have been performed using electrolytic SERS (E-SERS)36 But this requires a three-electrode sample cell and an electrolyte of known concentration to perform the necessary oxidation-reduction cycles (ORCs) to re-activate the electrode surface with new uncontaminated sites from one measurement to the next Colloids are severely limited in that continuous measurements would require a continuous supply of colloids For periodic measurements vials of colloids one per measurement could be used However aggregate size and consequently SER intensity change with sample conditions (especially pH) and quantitative reproducible measurements are unlikely Substrates appear to have the greatest potential and designs range from silver evaporated on titania particles34 to periodic gold pyramids evaporated between polystyrene beads35 Most substrates require concentrating the sample on the surface through drying to obtain the largest signal enhancements in effect making the measurements irreproducible and irreversible However successful measurements using flow systems have been obtained with glass posts but manufacturing costs appear prohibitive In an effort to overcome these limitations we have developed metal-doped sol-gels to provide SERS measurements that are reproducible reversible and quantitative and yet not restricted to specific environments such as electrolytes solvents or evaporated surfaces3738 The porous silica network of the sol-gel offers a unique environment for stabilizing SER active metal particles and the high surface area increases the interaction between the analyte and metal particles The sol-gel can be coated on the end of fiber optics or on the internal walls of a glass flow tube for continuous measurements or standard glass sample vials for periodic measurements Previously we measured 100 mgL methylphosphonic acid (the primary hydrolysis product of nerve agents) in water with an estimated detection limit of 05 mgL (100 parts-per-billion) We have also
SPIE 2001-4575
64
demonstrated reversible and reproducible measurements of p-aminobenzoic acid (PABA) in a flow through system Here we investigate the ability of the sol-gel SERS (SG-SERS) to measure the NA bases adenosine monophosphate and RNA extracted from E coli B subtilis and S aureus The measurements are compared to those obtained by E-SERS
2 EXPERIMENTAL The inorganic chemicals and solvents used to prepare samples were spectroscopic grade and purchased from Aldrich (Milwaukee WI) Fisher (Pittsburgh PA) or Pfaltz amp Bauer (Waterbury CT) The nucleic acid bases and dipicolinic acid were purchased from Sigma (St Louis MO) Normal Raman samples were measured to establish enhancement factors In each case 1cm3 of sample was placed into a 1x1 cm glass cuvette weighed and measured Unpacked densities were typically 6-7 gcm3 For all SER measurements including RNA samples were prepared as ~01mgmL (see Figure captions for exact concentrations) in 01M KCl and buffered to a pH of 92 with Na2B4O7bullH2O Adenine pH dependence measurements used pH buffer solutions at 4 (potassium acid phthalate) 69 (potassium phosphate monobasicsodium phosphate dibasic) 92 (Na2B4O7bullH2O) and 104 (tris-hydroxymethyl amino methane) Escherichia coli Bacillus subtilis and Staphylococcus aureus cultures (250ml per 1000mL Erlenmeyer flask) were grown overnight in a Trypticase soy broth (TSB) medium containing 1 glucose in a shaking water bath at 37 oC The bacteria were harvested by centrifugation for 10 minutes at 8000 rpm in a GSA rotor at 5degC then washed once in 085 saline The gram-positive bacteria were concentrated to 20 ml and passed through a French pressure cell twice at 15000 psi to break open the cells RNA was extracted according to Protocol 44139 to ensure pristine samples for initial measurements Since this method takes approximately 4 hours a streamlined method was developed For vegetative bacteria the specimen was boiled for 30 sec in 1 ml of distilled water to lyse the cells and release the RNA For bacterial spores the specimen was first incubated in 1 ml of saline solution containing 02 mg lysozyme and phosphate-buffered to pH of 624 for 1 hr at 37 oC This solution was then boiled for 2-3 minutes in 4X loading buffer to release the RNA For both specimens RNA STAT-60TM was added to the supernatant which was centrifuged at 12000 g for 5 minutes to precipitate the ~15 water-soluble proteins This procedure allowed extracting RNA for SER analysis in ~ 10 minutes Electrophoresis shows high purity while the existence of chemicals that could interfere with the SER measurements is still under investigation The electrolytic sample cell has been described previously36 Briefly a three electrode design is incorporated into a Plexiglas structure containing a well for the reference electrode (a saturated calomel electrode Cole Parmer Vernon Hills IL) and a 5mL sample well containing the silver working electrode and platinum wire counter electrode (05 mm wire Alfa Ward Hill MA) A channel connecting the two wells contained a 2 mm diameter semi-porous membrane (10-20 micron pore Ace Glass) The silver electrode was made from a 3 mm length of 2 mm diameter silver wire (Alfa) soldered to a copper wire lead encased in a 4 mm diameter Pyrex tube A cap containing the silver electrode platinum wire and nitrogen purge and vent lines fixed the silver electrode surface 1 mm from a 1 mm thick glass plate attached to the bottom of the sample well The potentiostat used to control the three electrodes was built in-house and has been described in detail elsewhere36 A multifuntional analog digital and timing inputoutput interface card (DAQCard-1200 National Instruments) is used to both drive the electrolytic cell as well as read the current generated in the cell A LabVIEW software program is used set the oxidation potential reduction potential SER measurement potential hold times and sweep rates The amount of charge passed was plotted as a cyclic voltammogram For all spectra presented five oxidation-reduction cycles (ORCs) stepping from -03 VSCE to 03 VSCE and back to -3 VSCE at 50 mVsec were used The SG-SER measurements were accomplished by simply placing the identical samples prepared above into Simple SERS Sample VialsTM (RTA) These 2-mL glass vials are internally coated with ~ 01 micron thick silver-doped sol-gel A prototype Fourier transform Raman spectrometer (RTA) was used for these measurements40 The system consisted of a NdYAG laser (Brimrose or Spectra Physics) for excitation at 1064 nm an interferometer built by On-Line Technologies (OLT East Hartford CT) for frequency separation an uncooled InGaAs detector for signal detection (RTA) and an Intel 400 MHz Pentium II based laptop computer (Dell Round Rock TX) for interferometric control data acquisition (OLT) and analysis (LabVIEW by National Instruments Austin TX) Additional components included a Notch filter (Kaiser Ann Arbor MI) and interferometer entrance and exit optics (Edmund Scientific Barrington NJ) Fiber optics were used to deliver the excitation beam to the sample and the scattered radiation to the interferometer (2 meter lengths of 200 and 365 micron core diameter respectively Spectran Avon CT) A second Notch filter (Kaiser) was used as a beam splitter to direct the excitation beam along the same axis as the collected radiation A microscope object (20x magnification 04 numeric aperture Newport Irvine CA) was used to focus the beam into the sample and to collect the scattered radiation back along the same axis In this co-axial backscattering arrangement the excitation beam passed through the glass plate onto the silver
SPIE 2001-4575
65
electrode surface for E-SERS through the vial glass wall and into the silver-doped sol-gel film for SG-SERS or through the glass wall of the cuvette and into the solid sample for normal Raman spectroscopy All E-SERS and normal Raman spectra were obtained with 750 mW of laser power at the sample while all SG-SERS spectra were obtained with 75 mW of laser power at the system Incident powers above 200 mW in some cases degraded the sol-gel
3 RESULTS AND DISCUSSION The generation of surface-enhanced Raman scattering at electrode surfaces has been extensively researched and the optimum sample conditions are well developed2729 Several researches incorporated electrodes into flowing systems and demonstrated that continuous monitoring of chemicals is possible18 These successes suggested investigated the capability of E-SERS to measure the NA bases and RNA The E-SERS measurements also provided a benchmark to compare and evaluate SG-SERS measurements The molecular structure of adenine (as well as the other base pairs) which includes an aromatic nitrogen-containing heterocycle is ideally suited to interact with the surface plasmons and contribute substantially to the chemical component of the SER effect1119 Even with excitation at 1064 nm a 3-minute scan of 18x10-5M adenine yields high signal-to-noise (SN) E-SER spectra and all of the bands are revealed with clarity (Figure 1 Table 1) Spectra of similar quality were obtained by SG-SERS and the principal spectral bands are easily observed The identical 18x10-5M adenine sample was measured in the same 3-minute time frame but with 110th the laser power The lower power appears to reduce the SN The amount of adenine responsible for the SER spectra as well as enhancement factors for the two SER media can be determined The molecules producing the E-SERS spectrum are those on the electrode surface within the illumination area of the laser (The solution concentration only determines the number of molecules available to adsorb to the electrode surface) For the current experiments the laser illuminates an area of 28x10-7m2 or 56 x10-7m2 if we assume the ORCs increase the surface area by a factor of two Furthermore if we assume monolayer coverage on the electrode and each 3x5 angstrom molecule (lying flat) occupies 15x10-19m2 then there are ca 4x1012 molecules contributing to the Raman scattering This is ca twice the number of molecules measured at electrode surfaces using either differential capacitance-potential curve measurements or rapid linear sweep voltammetry (eg 3x1018 moleculesm2 for pyridine and pyrazine)29 Thus the adenine spectrum in Figure 1 is due to 87x10-10g (6x10-12 moles) A detection limit defined as a SN of 3 can also be calculated The SN for a 3-minute scan is 844 for the 735 cm-1 band suggesting a mass detection limit of 3x10-12g (2x10-14 moles) This is consistent with previous estimates for adenine by others of 25 x10-14 moles1530 However sub-monolayer concentrations must be measured to verify this The root-mean-squared (RMS) noise is measured between 4400-4600 cm-1 Since noise is distributed evenly throughout the spectrum when transformed this region does not have any
Wavenumbers (∆cm-1) Figure 1 A) Normal Raman spectrum of pure adenine powder B) E-SERS and C) SG-SERS of 18x10-5M adenine at pH 92 All spectra 8 cm-1 resolution 200 scans (3 min) and 1064 nm excitation A) and B) 750 mW C) 75 mW B) measurement potential of 11VSCE
A
B
C
A
B
C SG-SERS
pH 10
D
725
735
735
pH 4
Wavenumbers (∆cm-1) Figure 2 A) and C) E-SERS and B) and D) SG-SERS of adenine at A) and B) pH 104 and C) and D) pH 40 Note consistent appearance of bands at 1270 and 1375 cm-1 as the pH is changed to 10 for both SER media E-SERS used 750 mW SG-SERS used 75 mW of 1064 nm excitation
SPIE 2001-4575
66
contributions from signals or baseline offsets The measurement error is given as SplusmnRMS and for adenine this equals 234 The number of molecules contributing to the SG-SERS are those on the silver particles that are embedded in the sol-gel The total silver surface area can be determined from the average particle size (40 nm diameter) concentration (073 by weight based on molar conc and measured sol-gel density) and the scattering volume (a cylinder defined by the laser area 28x10-7m2 and sol-gel thickness10-4m) The 61x109 silver particles in this volume have a collective area of 31x10-5m2 However it may reasonably be assumed that at least half of the silver surface is in contact with the sol-gel matrix and unavailable for analyte interaction Then approximately 10x1014 molecules or 22x10-8g of adenine contribute to the SG-SER spectrum The slightly lower SN of 207 suggests a mass detection limit of 32x10-10g Determination of the enhancement factors for the two SER media requires estimating the number of adenine molecules contributing to the normal Raman spectrum Here a cylindrical scattering volume is assumed again based on the laser area (28x10-7m2) and the penetration depth (1x10-3 m)41 The density of the sample was measured at 064 gcm3 indicating that 18x10-4g (13x10-6 moles) of adenine produced the normal Raman signal The enhancement factor EF is defined by the following equation
EF = (ISERSINR)bull(MNRMSERS) bull(PNRPSERS) bull(TNRTSERS)12
where I is the spectral band intensity (here 735 cm-1) M is the sample mass P is the incident laser power and T is the measurement time (or number of scans) For the E-SERS measurement the enhancement factor is 22x105 (01780184) bull (18x10-487x10-10)) while the SG-SERS enhancement factor is 10x105 (0160184) bull(18x10-422x10-8) bull(75075) bull(315)12) The lower enhancement for the SG-SERS may be real or the available surface of the silver embedded in the sol-gel may have been overestimated In addition to enhancing the Raman scattering efficiency to an extent similar to E-SERS the SG-SER medium also yields an identical shift of the adenine ring-breathing mode from 725 cm-1 in the normal Raman to 735 cm-1 Furthermore in the course of optimizing the E-SERS sample conditions it was found that pH influenced the adenine interaction with the silver surface (Figure 2) In particular the relative band intensities of the pyrimidine ring skeletal vibrations at 1270 and 1375 cm-1 and the imidazol ring skeletal vibration at1335 cm-1 change At pH 4 adenine is protonated presumably the imidazol ring since the band at 1335 cm-1 increases in intensity while the pyrimidine bands are virtually absent Conversely at pH 10 the imidazol band decreases in intensity while the pyrimidine bands appear It is worth noting that the ring-breathing mode at 735 cm-1 changes little between pH 4 and 10 suggesting that the skeletal changes are more a function of molecule-plasmon interactions than reorientation of the molecule on the surface Measurements of the identical pH series of adenine samples by SG-SERS yielded virtually identical spectral changes This suggests that the sol-gel does not influence the measurement This is critical to reproducing measurements and performing quantitative analysis Next the remaining NA bases were measured by both E-SERS and SG-SERS and compared Previously we examined the optimum pH and electrode potentials for E-SERS measurements to determine if a common pH could be used that yielded good sensitivity for all the bases and if variations in potential could be used to provide an added degree of selectivity between the bases Primarily it was found that high quality spectra were obtained between pH 7 and 95 and that cytosine and uracil were best enhanced at potentials positive of the potential-of-zero charge (pzc ca 065VSCE for Ag) guanine and thymine near the pzc and adenine negative of the pzc In all cases the ring-breathing modes were the most intense and in general could be used to identify the NA bases (Figure 3 Table 1) Specifically adenine has an intense band at 735 cm-1 cytosine at 797 cm-1 guanine at 653 cm-1 thymine at 784 cm-1 and uracil at 800 cm-1 The adenine cytosine guanine and thymine bands are sufficiently separated that their contributions to DNA should be determinable Although adenine and guanine contributions to RNA should also be determinable cytosine and uracil are highly overlapped and unfortunately share the same potential dependence Alternate unique bands at 1183 cm-1 for cytosine and 1275 cm-1 for uracil might be suitable for calculating contributions The SG-SER spectra of the remaining NA bases faithfully reproduced the E-SER spectra In particular the primary identifying bands occur at virtually the same wavenumbers (see Table 1) However the spectra for both cytosine and thymine contain an intense band at ca 1040 cm-1 Initially this was attributed to the pH buffer but samples prepared without either the buffer or the 01M KCl electrolyte yielded identical spectra containing this band In fact the E-SER and SG-SER spectra of thymine are virtually identical except for this band Also the SG-SERS of guanine contains an intense band at 1551 cm-1 that is not observed in the E-SER spectrum This band may be due to a moderately intense band at 1553cm-1 in the normal Raman spectrum that is SG-SER active It was also found that the SG-SERS of cytosine was considerably better than the E-SERS while uracil showed the opposite relationship It is also worth noting that all of the SG-SERS were obtained with 110th the laser power Most importantly the primary ring-breathing modes in the SG-SER spectra are sufficiently intense and unique to be used in determining contributions to DNA and RNA as outlined above
SPIE 2001-4575
67
Table 1 Comparison of E-SER and SG-SER Spectral Band Positions for the NA Bases and Adenosine Monophosphate
Adenine Cytosine Guanine Thymine Uracil AMP E-SERS SG-SERS E-SERS SG-SERS E-SERS SG-SERS E-SERS SG-SERS E-SERS SG-SERS E-SERS SG-SERS1647 1638 1634 1656 1655 1630 1587 1585 1510 1580 1551 1539 1456 1456 1465 1460 1462 1480 1453 1459 1394 1398 1425 1431 1435 1399 1404 1392 1374 1375 1373 1383 1370 1335 1332 1311 1307 1333 1331 1353 1348 1331 1329 1265 1273 1280 1292 1278 1276 1275 1279 1271 1183 1195 1222 1232 1221 1219 1204 1205 1180 1144 1097 1033 1029 1038 1040 1035 1051 1037 1041 1035 963 963 957 1001 1000 961 944 884 819 817 859 866 735 737 797 799 784 782 800 800 72738 742 630 630 653 664 667 684 603 602 590 611 466 561
Bands unique to SG-SERS The next chemical to be analyzed by both E-SERS and SG-SERS was adenosine monophosphate (AMP) The E-SER spectrum yields bands due to the adenine chemical functionality at 727 961 1233 1279 1331 1381 and 1486 cm-1 In addition phosphate bands are observed at 860 1097 1453 1587 and 1705 cm-1 (Figure 5) Other researchers have noted that the ribose component does not appear to contribute to the spectrum19 The AMP spectrum also changes as a function of potential As the electrode is swept more positive (here from -09 to -03VSCE) the phosphate bands at 860 1097 1453 and 1587 cm-1 increase in intensity compared to the adenine bands while a band at 1705 cm-1 appears The adenine bands at 1233 1381 and 1486 cm-1 virtually disappear These potential dependent spectral changes are consistent with earlier studies that show that phosphate is attracted to silver at potentials positive of the pzc but repelled at potentials negative of the pzc19
Wavenumbers (∆cm-1) Figure 3 E-SERS of A) 21x10-3M cytosine at -03VSCE 1000 scans B) ~10x10-5M guanine at -06VSCE 500 scans C) 23x10-3M thymine at -06VSCE 500 scans and D) 12x10-3M uracil at -093VSCE 500 scans All spectra at pH 92 750 mW 1064 nm at 8 cm-1
Wavenumbers (∆cm-1) Figure 4 SG-SERS of A) 21x10-3M cytosine 200 scans B) ~10x10-5M guanine 200 scans C) 23x10-3M thymine 200 scans and D) 12x10-3M uracil 500 scans All spectra at pH 92 75 mW 1064 nm at 8 cm-1
A
B
C
A
B
C
D D
SPIE 2001-4575
68
The SG-SER spectrum of AMP is considerably different The adenine bands virtually disappear except for the two primary bands which shift to742 and 1329 cm-1 While the phosphate band at 1459 cm-1 has gained considerable intensity In addition two new intense bands appear at 684 and 1539 cm-1 as well as a moderately intense band at 1180 cm-1 The SG-SER spectrum has greater similarity to the E-SER spectrum at -03VSCE and suggests that the silver particles embedded in the sol-gel behave as if at a potential positive of the pzc RNA samples extracted from E coli B subtilis and S aureus were next examined by both E-SERS and SG-SERS E-SER spectra of these samples yielded quality spectra in 10 minutes in which all of the major features can be identified (Figures 6 and 7) This includes guanine at 650 cm-1 adenine at 791 cm-1 cytosine and uracil combining at 790 cm-1 and phosphate at 1100 1335 (in combination with adenine and guanine) 1465 and 1570 cm-1 Surprisingly adenine which demonstrated the greatest surface-enhanced Raman effect does not dominate the ring-breathing mode portion of the spectrum The intensities of the other base-pairs bands are of the same order of magnitude This suggests that when the base-pairs are linked together as in RNA they are enhanced in concert In fact the relative intensities are very similar to a normal Raman spectrum of E coli RNA which shows the combined cytosine and uracil band at ca twice the intensity of the adenine band and ca four times the intensity of the guanine band Unfortunately this means that the independent enhancement factors for the NA bases can not be used to estimate relative concentrations For example the relative 791 and 734 cm-1 bands for B subtilis would indicate that the cytosine andor uracil concentration was at least 20 times the adenine concentration whereas each of the four RNA bases are known to contribute 15-35 Nevertheless it is worth noting that the three RNA samples yield different relative band intensities that were reproduced in numerous measurements Although the relative concentrations of the NA bases for these samples have not been determined these differences can be quantified If it is assumed that the 650 cm-1 band represents 25 guanine the 791 cm-1 band represents 25 adenine and the 790 cm-1 50 cytosine plus uracil in the E coli RNA spectrum then the relative concentrations can be estimated for the other RNA samples To aid this calculation the three spectra were normalized to the phosphate band at 1100 cm-1 which has been shown to correlate to the total phosphate concentration and can be used as an internal standard In addition a simple baseline correction was applied (Figure 7) This yields 15 adenine 30 guanine and 55 cytosine plus uracil for B subtilis RNA and 18 adenine 25 guanine and 57 cytosine plus uracil for S aureus RNA The average SN of these measurements was 26 with an average error of 8 of the value (SplusmnN) It is also worth noting that the three RNA spectra show a marked shift in a band near 825 cm-1 This band is assigned to the symmetric stretch of the O-P-O ester linkage9 The band appears at 815 cm-1 for S aureus shifting to 820 cm-1 for B subtilis and 830 cm-1 for E coli Others have used the normal Raman intensity of the band at 815 cm-1 as a direct indication of the amount of A-class helix present while the intensity of the band at 830 cm-1 has been used as a direct indication of the amount of B-class helix present However the latter is more associated with DNA than RNA
Wavenumbers (∆cm-1) Figure 5 E-SER spectra of 020 mgmL adenosine monophosphate at A) -03 and B) -09VSCE and C) SG-SER spectra Conditions sample in 01M KCl buffered to pH 92 A) and B) 750 mW C) 75 mW of 1064 64 scans (1-min) at 8 cm-1
Wavenumbers (∆cm-1) Figure 6 E-SERS of 01 mgmL RNA from E coli 02 mgmL RNA from B subtilis and 02 mgmL RNA from S aureus Conditions 01M KCl pH 92 -03VSCE 750 mW of 1064 nm 640 scans (10 min) at 8 cm-1
AMP RNA
E coli A
B
C
B subtilis
S aureus
SPIE 2001-4575
69
SG-SER spectra of reasonable quality were also obtained for E coli and B subtilis especially the latter (Figure 8) However the spectra differ substantially from the E-SERS of the same samples Both SG-SER spectra are dominated by adenine at 735 cm-1 and a band at 1030 cm-1 Although unassigned the latter does appear in the RNA E-SER spectra Bands at 1105 and 1565 cm-1 are likely due to phosphate while bands at 1320 and 1470 cm-1 are less confidently assigned to phosphate They are significantly less intense and somewhat shifted from their SG-SERS counterparts (1335 and 1455 cm-1) A band at 670 cm-1 may be due to guanine which was observed at 664 cm-1 for SG-SERS of the pure sample However the SG-SER spectrum of AMP also had an intense 667 cm-1 band A number of other bands occur at 890 1070 1165 1245 1290 1420 1505 cm-1 and remain unassigned The SG-SER spectra are somewhat disappointing in that only adenine and guanine contributions can be positively identified This limits the ability to determine relative NA base concentrations and distinguish bacterial RNA However several of the unassigned bands may be due to the bases (eg 1030 and 1420 cm-1 due to cytosine) Further experiments will be required to clarify this point
Ecoli
B subtilis
S aureas
P G
A C+U
OPO
A B
Wavenumbers (∆cm-1) Wavenumbers (∆cm-1) Figure 7 SER spectra of RNA from A) B subtilis with contributions indicated and B) E coli B subtilis and S aureus with baseline correction and peak positions used to calculate contributions indicated G = guanine A = adenine C+U = cytosine plus uracil P = phosphate (backbone) OPO = phosphate ester linkage (A- vs B-class helix)
Wavenumbers (∆cm-1) Wavenumbers (∆cm-1) Figure 9 E-SER (-03VSCE) and SG-SER spectra of RNA from A) E coli and B) B subtilis Sample conditions as in Figure 6 E-SER spectra at 750 mW SG-SERS at 75 mW
A B E-SERS
SG-SERS
E-SERS
SG-SERS
SPIE 2001-4575
70
A final comparison was made between the two SER methods by measuring dipicolinic acid (DPA) This chemical may be invaluable as a test for spore forming bacteria specifically B anthracis 50 to 90 of B anthracis sporilates During spore formation dipicolinic acid is synthesized and once completed 10-15 of the dry spore weight is composed of the Ca2+ complex located in the spore core42 Heating in water can be used to initiate germination at which point the exosporium breaks and releases the Ca dipicolinate which becomes dipicolinic acid in water The structure of this chemical strongly
4 CONCLUSIONS Towards the goal of developing a method and technology to rapidly detect and identify bioagents we have been investigating surface-enhanced Raman spectroscopy as a tool to measure relative concentrations of nucleic acid bases in RNA and determine bioagent taxonomy Initially we investigated E-SERS since this method has been extensively researched and the optimum sample conditions are well developed However this method requires a three-electrode sample cell and electrolyte solution Incorporation of an electrolytic cell into sample systems used to analyze for BWA as aerosols in water or on surfaces can be designed using flow injection analysis technologies but cross-contamination and plugging of sample lines seems inevitable For this reason we also investigated metal-doped sol-gels as a SER-active medium Previous studies have shown this material to be active in all solvents particularly water capable of continuous measurements in flowing systems and reproducible (quantitative) between coated sample vials Here we compared SG-SER spectra to traditional E-SER spectra of the nucleic acid base pairs adenosine monophosphate and RNA High quality spectra of adenine cytosine guanine thymine and uracil were obtained by both E-SERS and SG-SERS Both methods yielded very similar spectra for the NA bases including a pH dependent study of adenine Enhancement factors and detection limits for adenine were determined as 2x105 and 16x10-11g and 1x105 and 12x10-10g for E-SERS and SG-SERS respectively (normalized to 75 mW and 10-min acquisition time) Fifty percent of the silver particle surface area in the sol-gel matrix was assumed covered by adenine which may have been overestimated yielding a lower EF and higher detection limit It should also be realized that each E-SER spectrum required several attempts to optimize the measurement conditions (pH electrode potential etc) While each SG-SER spectrum involved no sample preparation and often represents the first and only attempt to make the measurement Quality spectra of RNA extracted from Escherichia coli Bacillus subtilis and Staphylococcus aureus were obtained by E-SERS that were easily interpreted Bands due to adenine guanine cytosine plus uracil and phosphate were identified The SER band intensity of the NA bases in the RNA samples were of the same order of magnitude suggesting that their interaction with the silver surface is concerted as is their Raman enhancement Interestingly the relative SER band
Wavenumbers (∆cm-1) Figure 10 A) Raman spectrum of solid dipicolinic acid B) Ax20 C) electrolytic SERS of 6x10-3 M dipicolinic acid in 01 M KCl at a potential of +07VSCE and pH of 4 and D) sol-gel SERS of 6x10-3 M dipicolinic acid Conditions for A and C as in Figure 1 C) 100 mW of 1064 nm 50 scans 8 cm-1
A B (Ax20)
C
D
suggested that it would be SER active However the E-SER spectrum was unstable and varied considerably as a function of potential A consistent spectrum was obtained at +06VSCE (Figure 10) This potential is not recommended for measurement because the surface is actively dissolving in solution The SG-SER spectrum was considerably more stable of higher quality and easily reproduced Bands at 660 825 1010 1390 1430 1570 1590 and 3075 cm-1 were observed Enhancement factors were determined for the two media using the symmetric ring stretching mode at 995 and 1010 cm-1 for the normal Raman and SER spectra respectively E-SERS yielded an EF of 5x103 while SG-SERS yielded an EF of 2x105 for DPA The SN of the latter suggests a detection limit of 20x10-10g (based on adenine coverage 75 mW and 10-min) The differences in SER activity for these two media may be attributed to the combined electrolytic potential of the solution chemical and metal15 Again the E-SERS suggests that the SG-SERS is at a potential positive of the pzc While the instability in the E-SERS may also be associated with surface interactions of two carboxylic acid groups of dipicolinic acid during the ORCs
DPA
SPIE 2001-4575
71
intensities for RNA extracted from E coli are very similar to those measured by normal Raman spectroscopy Although the relative percent that each of the NA bases contributed to each RNA sample was not determined reproducible band intensities allowed noting the following trends The percent adenine decreases while the combined percent cytosine and guanine increase for both B subtilis and S aureus compared to E coli Quality spectra were also obtained for the RNA samples by SG-SERS but only a few bands were readily identified Calculations of NA base concentrations by SG-SERS will require further research In light of recent events we summarize the capabilities of these SERS media in terms of rapid detection of B anthracis and dipicolinic acid However these capabilities must be qualified First and foremost the level to which SERS can distinguish bacteria or viruses has not yet been determined Development of a database of both DNA and RNA base concentrations for BWAs and common bacteria to establish the level of taxonomic identification is ongoing Second rapid collection of aerosol water or surface samples is being addressed by others who report trapping particles on filters from 100 liters of air per minute Third although not presented here we have developed methods to extract RNA or DNA from cells and spores for SER analysis within 10 minutes Finally we assume a detection limit of 3600 spores per 100 liters of air is required although a 50 lethal dosage of anthrax has not been established With these qualifications a mass detection limit for RNA using SERS is estimated as follows A single measurement is performed in ca 20 minutes (140 liters collected in 14 min RNA extracted in 8 min spectral acquisition and analysis in 10 min) The average human breaths 7 liters per minute therefore the analyzer must at the very minimum detect 5000 spores in 140 liters of air One spore is approximately 2x10-18m3 (1x1x2 microm) and if a density of 075 gcm3 is assumed this corresponds to a mass of 15x10-12g Each spore contains 4-12 RNA or 12x10-13g RNA for 8 If we assume 23 of the RNA can be isolated for analysis during lysis then the proposed instrument must be able to detect 4x10-10g RNA from 5000 spores per 70 liters of air within 10 minutes As noted above the mass detection limits for adenine were estimated at 16x10-11g and 12x10-10g for E-SERS and SG-SERS respectively Although these detection limits suggests that RNA from 5000 spores is detectable with the current instrumentation it is highly likely that only a portion of an RNA segment (eg 120-nucleotide 5S rRNA) is in contact with the metal surface and will contribute to the SER effect The SN for the RNA spectra were 110th of the average SN for the four individual RNA bases suggesting a 10 contribution Furthermore effective taxonomy will likely require knowing the NA base concentrations to 1 of the value (eg 25plusmn025) Again the average measurement error for the bases is 12 These values suggest that the E-SERS is within a factor of 4 of the required detection limit whereas the SG-SERS detection limit must be improved by 25 times The same arguments can be applied to the detection of dipicolinic acid If we assume a spore releases 10 by weight DPA during germination then the proposed instrument must be able to detect 75x10-10g DPA from 5000 spores per 70 liters of air within 10 minutes The detection limit for SG-SERS was estimated at 20x10-10g and suggest that the vials are suitable to perform a rapid screen for anthrax A series of concentration dependent measurements are currently being performed to verify this assertion Finally we note that the measurements performed here employed an FT-Raman spectrometer This instrumentation was chosen over dispersive Raman instrumentation for the high wavelength accuracy afforded by the HeNe reference laser (Connes Advantage43) which would allow reliable spectral subtraction matching of observed spectra to stored library spectra and confident use of chemometric approaches Such data analysis is likely to be required to enhance BWA identification However this instrumentation sacrifices sensitivity due to a less efficient detector (InGaAs vs Si) less efficient Raman scattering and less efficient generation of plasmon modes Substantial improvements can be obtained using visible excitation and Si detection and these measurements are underway
5 ACKNOWLEDGEMENTS The authors are grateful to Drs D Cookmeyer and S Tove of the US Army Research Office (Contract Number DAAH04-96-C-0078) for their interest and support of this research The authors would also like top acknowledge Dr R Yin and J Jensen for supporting development of the metal-doped sol-gels (Contract Number DAAD13-01-C-0019) They also thank Dr Wilfred H Nelson for assistance in spectral interpretations
SPIE 2001-4575
72
6 REFERENCES 1 Roberts WL and Rainey PM Clin Chem 39 1872-1877 (1993) 2 Pasechnik VA CC Shone and P Hambleton Bioseperations 3 267-283 (1993) 3 Jackson PJ ME Hugh-Jones DM Adair G Green KK Hill CR Kuske LM Grinberg FA Abramova and P
Keim Proc Natl Acad Sci 95 1224-1229 (1998) 4 Jensen JL NC Wong W R Loerop SPIE 4775-03 (2001) 5 Snyder A Pet al SPIE 3853-15 (1999) 6 Danosky T R and McFadden P N in press (1997) 7 Woodruff WH Farquharson S Science 201 831-833 (1978) 8 Lord RC and Thomas GJJr Spectrochemica Acta 23A 2551-2591 (1967) 9 Thomas GJJr Biochim Biophys Acta 213 417-423 (1970) 10 Placzek G Handbuch der Radiologie 2 EMarx ed Akademische Verlagagescellschatt Liepzig 1934 UCRL
Trans No 526 (1959) 11 Jeanmaire DL and RP Van Duyne J Electroanalytical Chem 84 1-20 (1977) 12 Chada S Manoharan R Moenne-Loccoz P Nelson WH Peticolas WL and Sperry JF Applied Spectroscopy
47 38-43 (1993) 13 Manoharan R Ghiamati E Chada S Nelson WH and Sperry JF Applied Spectroscopy 47 2145-2150 (1993) 14 Todd EA Morris MD Applied Spectroscopy 48 545-548 (1994) 15 Wentrup-Byrne E Sarinas S and Fredericks PM Applied Spectroscopy 47 1192-1197 (1993) 16 Weldon MK VR Zhelyaskov and MD Morris Applied Spectroscopy 52 265-269 (1998) 17 Kneipp K and J Fleming J Mol Structure 145 173-179 (1986) 18 Pothier NJ and Force RK Applied Spectroscopy 46 147-151 (1992) 19 Ervin KM E Koglin JM Sequaris P Valenta and HW Nurnberg J Electroanal Chem 114 179-194 (1980) 20 Kim SK TH Joo SW Suh and MS Kim J Raman Spectrosc 17 381-386 (1986) 21 Pothier NJ and Force RK Analytical Chemistry 62 678-680 (1990) 22 Kneipp K Y Wang RR Dasari and MS Feld Applied Spectroscopy 49 780-784 (1995) 23 Nie S and Emory SR Science 275 1102 (1997) 24 Graham D WE Smith AMT Linacre CH Munro ND Watson and PC White Analytical Chemistry 69 4703-
4707 (1997) 25 Fleischmann M PJ Hendra and AJ McQuillan Chem Phys Lett 26 163-166 (1974) 26 Pettinger B J Chemical Phys 85 7442-7451 (1986) 27 Surface-Enhanced Raman Scattering Section Four Theory SPIE MS 10 M Kerker and B Thompson Eds (1990) 28 Wang D-S and Kerker M SPIE (M Kerker and B Thompson Eds) MS 10 417-429 (1990) 29 Weaver MJ Farquharson S Tadayyoni MA J Chem Phys 82 4867-4874 (1985) 30 Norrod KL Sudnik LM Rousell D and Rowlen KL Applied Spectroscopy 51 994-1001 (1997) 31 Farquharson S Weaver WJ Lay PA Magnuson RH and Taube H J Am Chem Soc 105 3350-3351 (1983) 32 Lee PC and Meisel D ldquoAdsorption and Surface-Enhanced Raman of Dyes on Silver and gold Solsrdquo J Phys Chem
86 3391-3395 (1982) 33 Li Y-S and Wang Y Applied Spectroscopy 46 142-146 (1992) 34 Bello JM DL Stokes and T Vo-Dinh Analytical Chemistry 61 1779-1783 (1989) 35 van Duyne RP J C Hulteen D A Treichel M T Smith M L Duval and T R Jensen J Phys Chem B1033854-
3863 (1999) 36 Farquharson S and W W Smith W H Nelson and J F Sperry SPIE 3533-27 207-214 (1998) 37 Lee Y H W Smith S Farquharson H C Kwon M R Shahriari and P M Rainey SPIE 3537 252-260 (1998) 38 Lee Y-H S Farquharson and P M Rainey SPIE 3857 76-84 (1999) 39 Current Protocols in Molecular Biology Wiley Interscience1003-1006 (1987) 40 Farquharson S Smith W Carangelo R C and Brouillette C SPIE 3859 14-23 (1999) 41 Chase D B and JF Rabolt Fourier Transform Raman Spectroscopy Acad Press Ch1 p 131 (1994) 42 Brock TD MT Madigan JM Martinko and J Parker Biology of Microorganisms 7th Ed Prentice Hall p 76-80
(1994) 43 Connes J Rev Opt Theor Instrum 40 45 (1961)
166
Chemical agent identification by surface-enhanced Raman spectroscopy
Stuart Farquharson and Paul Maksymiuk
Real-Time Analyzers 87 Church Street East Hartford CT 06108
Kate Ong and Steven D Christesen US Army SBCCOM Aberdeen Proving Ground MD 21010
ABSTRACT The recent distribution of anthrax through the US postal system and the subsequent infection and death of several postal and national media employees amplifies the need for methods to rapidly detect identify and quantify this and other chemical and biological warfare agents The US military has also identified water supplies as a likely method of warfare agent deployment and is funding the development of a Joint Service Agent Water Monitor (JSAWM) In an effort to aid military personnel and the public at large we are developing a portable analyzer capable of identifying and quantifying chemical agents rapidly either on-demand or continuously The approach is based on the ability of Raman spectroscopy to identify molecular structure through the abundant vibration information provided in spectra and the ability of surface-enhanced Raman spectroscopy (SERS) to detect extremely low concentrations (eg part-per-billion) through the enhancement of Raman scattering by six orders of magnitude or more A key element to the analyzer design is a new SER active medium that is capable of quantitative reversible measurements The medium consists of silver or gold nanoparticles incorporated into a sol-gel matrix The porous silica network offers a unique environment for stabilizing SER active metals and the high surface area increases the interaction between the analyte and metal particles Here we present the use of new sol-gels that also selectively enhance chemicals based on polarity and charge Base-line measurements of chemical agents and their hydrolysis products are presented and compared to the JSAWM goal of 30 micrograms per liter detection Keywords Chemical warfare agent hydrolysis product SERS Raman spectroscopy sol-gel nanoparticle
1 INTRODUCTION Since September 11 2001 the threat of terrorist attacks and biological warfare within US borders has become a sobering reality The simplicity in manufacturing ease of deployment and the relatively low cost of chemical warfare agents (CWAs) raises public concern that they may also be used by terrorists Indeed terrorists released sarin (GB) in the Tokyo subway in 19951 Countering terrorism and terrorist attacks requires recognizing likely deployment scenarios and having the required technology to rapidly detect the deployment event One method of deployment has been long identified by the US military distribution through water supplies To counter this threat the Department of Defense is funding or monitoring the capabilities of 200 technologies with the goal of developing a Joint Service Agent Water Monitor (JSAWM) that is field portable2 The JSAWM must be capable of identifying the chemical agents with no false-positives and quantifying the agents at microgram per liter concentrations within 10 minutes (Table 1)3 This includes the analysis of drinking water supplies distribution and storage systems as well as potable water supplies The earliest technologies developed for CWA detection were based on electrochemical ionization or colorimetric analysis (eg phosgene tape) Although these analyzers were easy to use they were not generally agent specific and suffered from false-positives4 More traditional laboratory methods have also been investigated and in particular combined gas chromatography and mass spectrometry (GCMS) has been very successful at eliminating false-positives56 However GCMS requires extraction repeated calibration and long analysis times (typically 20 to 60 minutes)6 making it labor intensive and less than desirable for field use More rapid analysis of agents in the solid liquid and gas phase has been To whom correspondence should be addressed emailfarqureal-time-analyzerscom
Vibrational Spectroscopy-based Sensor Systems Steven D Christesen Arthur J Sedlacek III EditorsProceedings of SPIE Vol 4557 (2002) copy 2002 SPIE middot0277-786X02$1500
167
Table 1 Chemical Agent Structures Hydrolysis Half-lives and JSAWM Thresholds Agent Short-Hand Chemical Structure Hydrolysis
Half-Life JSAWM
Thresholds
Sarin (GB) F-[O=P-CH3]-O-CH(CH3)2 213 hours 32 microgL Soman (GD) F-[O=P-CH3]-O-CH(CH3)-(C-(CH3)3) 23 hours 32 microgL Tabun (GA) (CH3)2-N-[O=P-CN]-O-C2H5 41 hours 32 microgL VX C2H5O-[O=P-CH3]-S-(CH2)2-N-(CH(CH2)2)2 821 hours 32 microgL EA2192 HO-[O=P-CH3]-S-(CH2)2-N-(CH(CH2)2)2 gt9 years 32 microgL Mustard (H) ClCH2CH2-S-CH2CH2Cl encapsulates 47 microgL Lewisite (L) ClCH=CH-As-Cl2 rapid 27 microgL HCN HCN rapid 20 mgL BZ C7NH12-O-[C=O]-COH(C6H5)2 23 microgL T-2 Toxin 87 microgL at pH 7 to 75 and 20 to 25 oC
demonstrated by vibrational spectroscopy7-10 Hoffland et al7 reported infrared absorbance spectra and absolute Raman cross sections for several chemical agents while Christesen measured Raman cross sections for sarin tabun mustard gas and VX (ethyl S-2-diisopropylamino ethyl methylphosphonothioate)11 Again however these techniques also have limitations Raman spectroscopy is simply not a very sensitive technique and detection limits are typically 01 (1000 ppm) While infrared spectroscopy would have limited value in analyzing poisoned water since the very strong infrared absorption of water would obscure most other chemicals present Nevertheless efforts to overcome these limitations have been demonstrated Braue and Pannella8 quantified the G-series nerve agents (tabun sarin and soman) in terms of infrared attenuated total reflectance using a circle-cell And Alak and Vo-Dinh demonstrated the possibility of surface-enhanced Raman spectroscopy (SERS) to identify CWAs by measuring several organophosphonates that simulate the nerve agents12 However quantitative measurements have not been demonstrated for the SER-active material used (silver coated on alumina particles) or other SER-active media13 Recently we developed silver-doped sol-gels to promote the SER effect14-17 The porous silica network of the sol-gel matrix offers a unique environment for stabilizing SER-active metal particles and the sol-gel provides a high surface area that effectively increases the number of molecules observed within the Raman scattering volume The silver-doped sol-gels have been coated on the internal walls of standard glass vials to allow rapid SER analysis of small sample quantities (lt 01 mL) without preparation We have used p-aminobenzoic acid (PABA) as a test chemical to demonstrate surface enhancements greater than 106 reversible measurements in a flowing system reproducible measurements from vial-to-vial and batch-to-batch and measurements in multiple solvents including water14-17 Recently we used these vials to measure Tabun (GB) and Sarin and several hydrolysis products pinacolyl methyl phosphonate (PMP from Soman) and methyl phosphonic acid (MPA from all G-agents Figure 1) Although a number of unique vibrational bands are observed (eg C-N stretch doublet and P-C stretch) the G-agents were only observed for 5 concentrations and all spectra required baseline corrections Figure 1 Surface enhanced Raman spectra of ~5 vv A) Tabun and B) Sarin C) 1 vv PMP and D) 10 ppm MPA using sol-gel sample vials 785 nm excitation 1-min scan and CCD detection Performed at Aberdeen Proving Ground
C A
Wavenumber (∆cm-1) Wavenumber (∆cm-1)
D B
790
545 C-N
2135 2190 P-C 770
760 1290
Proc SPIE Vol 4577
168
Nevertheless MPA was readily observed for a 10 ppm sample with an estimated detection limit of 04 ppm (based on a signal-to-noise ratio of 3 for the 760 cm-1 band intensity) This measurement provides encouragement in that SERS may satisfy the needs of a JSAWM Furthermore MPA is also a hydrolysis product of VX and V-gas and EA2192 (Figure 2) and may prove a valuable indicator of agent usage Figure 2 Hydrolysis of Sarin to form hydrofluoric acid (HF) methylphosphonic acid 1-methylethyl ester (MPAMME) methyl phosphonic acid (MPA) and 2-propanol With this initial albeit modest success we began analyzing chemicals with various sol-gel compositions that we have been developing Here we describe four sol-gel compositions that select for 1) polar-positive 2) polar-negative 3) weakly polar-positive and 4) weakly polar-negative chemical species The ability of these sol-gels to select and enhance Raman scattering is described for several test chemicals and MPA
2 EXPERIMENTAL The chemicals analyzed as well as all chemicals used to prepare the metal-doped sol-gels were obtained at their purest commercially available grade from Aldrich (Milwaukee WI) The sol-gel designed to select for polar-negative species was prepared from a silver amine complex tetramethyl orthosilicate (TMOS) and methanol After mixing 02 mL of the sol-gel solution was transferred into a glass vial (2 mL) dried and heated The incorporated silver ions were then reduced using dilute sodium borohydride The vials were washed and dried prior to the addition of a sample solution In a similar manner the sol-gel designed to select for polar-positive species was prepared from a gold salt TMOS and methanol The sol-gel designed to select for weakly polar-negative species was prepared from a silver amine complex tetraethyl orthosilicate (TEOS) and methanol And the last sol-gel designed to select for weakly polar-positive species was prepared from a gold salt TEOS and methanol All samples were prepared in a chemical hood and transferred into plain or SER-active vials for analysis Normal Raman spectral measurements employed 1-mL pure samples that were placed in a 1-cm3 cuvette and weighed This yielded a powder density that allowed accurate calculation of molecules in the optical collection field SERS measurements employed 1-mg sample per mL water concentrations unless otherwise stated Once prepared a 01 mL sample was placed into one of the four selective sample vials which in turn was placed into the sample compartment of a Raman spectrometer for analysis A prototype Fourier transform Raman spectrometer (RTA) was used for these measurements18 The system consisted of a NdYAG laser (Brimrose) for excitation at 1064 nm an interferometer built by On-Line Technologies (OLT East Hartford CT) for frequency separation an uncooled InGaAs detector for signal detection (RTA) and an Intel 400 MHz Pentium II based laptop computer (Dell Round Rock TX) for interferometric control data acquisition (OLT) and analysis (LabVIEW by National Instruments Austin TX) Additional components included a Notch filter (Kaiser Ann Arbor MI) and interferometer entrance and exit optics (Edmund Scientific Barrington NJ) Fiber optics were used to deliver the excitation beam to the sample and the scattered radiation to the interferometer (1 meter lengths of 200 and 365 micron core diameter respectively Spectran Avon CT) A second Notch filter (Kaiser) was used as a beam splitter to direct the excitation beam along the same axis as the collected radiation A microscope object (20x04 Newport Irvine CA) was used to focus the beam into the sample and to collect the scattered radiation back along the same axis In this co-axial backscattering arrangement the excitation beam was passed through the outside of a glass vial and focused onto the silver-doped sol-gel film (01-03 mm thickness) containing the sample
3 RESULTS AND DISCUSSION p-aminobenzoic acid (PABA) and phenyl acetylene (PA) and were used to refine the selectivity and SER-activity of the four different metal-doped sol-gels PABA is a popular chemical used to evaluate the performance of SER-active media Here the polar end groups can be used to test selectivity of the polar-negative and polar-positive sol-gels PA is essentially non-
2O+ H HF + +OH
OH3H C
OP
OF
CH
CH3
3
3H C
OCP
OOH
CH
CH3
3
3H C
O
CPHO
CH
CH
3
3
C
Sarin MPAMME MPA 2-propanol
Proc SPIE Vol 4577
169
polar but a high electron density in the cylindrical π cloud around the carbon-carbon triple bond allows testing the selectivity of the weakly polar-negative and weakly polar-positive sol-gels As Figure 3 illustrates PABA passes through the polar sol-gel and is enhanced by either the silver or gold particles At 1 mgml the concentration of neutral PABA is ca 20 times that of the ionized form (pKa = 48) For electropositive silver the PABA anion is expected to interact through the carboxylate group and the associated vibrational modes are expected to dominate the spectrum Conversely for electronegative gold either form of PABA is expected to interact through the amine group The clear differences in our spectra support this expectation Furthermore bands at 840 and 1405 cm-1 assigned to a COO- bend and stretch respectively are significantly more intense for silver than gold Additional bands at 1140 and 1195 cm-1 are assigned to CH bending modes while bands at 1450 1500 and 1605 cm-1 are assigned to ring vibrational modes A very similar SER spectrum for PABA on a silver-coated alumina substrate has previously been reported with similar assignments19 For the gold-doped sol-gel new bands appear at 690 1355 and 1585 cm-1 The first band is assigned to a ring-H bending mode the second band to a ring-N- stretching mode and the third band to a possible NH2 scissors mode or ring mode The second band is not observed in the normal Raman spectrum but infrared bands occur at this frequency for aromatic ring-secondary amine stretching modes The scissors mode occurs at this frequency in Raman spectra for several chemicals but is absent in the PABA Raman spectrum Alternatively this mode may be the1600 cm-1 ring mode that has been shifted by the gold interaction Again a very similar SER spectrum of PABA has been reported but surprisingly using silver (colloids)2021 not gold as the enhancement medium These researchers also assumed the primary interaction of PABA with silver was through the carboxylate anion and made assignments accordingly For example they assigned the 1359 cm-1 to a COO- stretch not to the amine group as we have They also favor the ring stretching mode assignment for the 1582 cm-1 band Finally it should be said that other researchers have argued that the most dominant band in the SER spectra at 1450 cm-1 a ring vibration mode suggests that PABA lies flat on the surface and the π-orbitals dominate the surface interaction22
Figure 3 SER spectra of A) PABA using polar-negative and B) polar-positive sol-gels and C) PA using weakly polar-negative and D) weakly polar-positive sol-gels PABA is 1 mgmL PA is 1 vv Spectral conditions 75 mw 1064 nm 100 scans (15 min) 8 cm-1 resolution Non-polar PA passes through the non-polar sol-gels and is also enhanced by both metals The spectra are easily understood For electropositive silver PA interacts through the cylindrical triple bond π electron cloud and a -CequivC- doublet occurs near 2000 cm-1 The interaction is reasonably strong since this band appears at 2112 cm-1 in the normal Raman spectrum For electronegative gold this interaction is unlikely and only very weak bands occur near 2000 cm-1 The remaining bands are at 1000 cm-1 1200 cm-1 doublet and 1595 cm-1 all appear in the normal Raman spectra at virtually the same frequencies and are assigned to the symmetric ring-breathing mode CH bending modes and the trigonal ring-breathing mode respectively The polarnon-polar selectivity of the polar-negative and weakly polar-negative sol-gels was tested by adding a 11 molar mixture of PABA and PA The selective enhancement is quite good (Figure 4) The spectrum obtained using the polar sol-gel represents 78 PABA and 22 PA while the spectrum obtained using the weakly polar sol-gel represents 9 PABA and 91 PA The band peak intensities at 2000 cm-1 for PA and 1450 cm-1 for PABA were used for these calculations and are expanded in Figure 4 for clarity
C A
Wavenumber (∆cm-1) Wavenumber (∆cm-1)
2NH COOH
D B
C CH
Proc SPIE Vol 4577
170
Figure 4 SERS of 11 MM of PABA and PA in A) polar-negative and C) weakly polar-negative sol-gels The lower traces compare the pure chemicals B) 1 mgml PABA in polar-negative sol-gel and D) 1 PA in weakly polar-negative sol-gel while the insets magnify the minority species for clarity (x5 in A and x10 in B) Spectral conditions as in Figure 3 Following this development of selective sol-gels that maintained SER activity we measured cyanide and MPA (Figure 5) Not surprisingly the best sensitivity for both hydrolysis products was obtained using the polar-negative sol-gel The interaction of the cyanide anion with the silver surface is sufficient to shift the CequivN stretch observed at 2080 cm-1 in the normal Raman spectrum to 2145 cm-1 in the surface-enhanced Raman spectrum Furthermore the band is substantially broadened This anion has been extensively studied by electrolytic SERS and this shift and broadening have been attributed to the formation of a tetrahedral Ag(CN)3
2- surface structure23 Figure 5 Surface-enhanced (upper traces) and normal Raman spectra (bottom traces) of A) CN- and B) MPA in silver-doped TMOS SERS conditions as in Figure 3 and 1 mgmL Note MPA yields two distinct spectra for neutral (top) and acidic pH (middle) The normal Raman spectra employed pure powders 500 scans and 900 mW of 1064 nm SER measurements of MPA with the polar-negative sol-gel yielded two unique spectral signatures that depended on solution pH (Figure 5) For more neutral solutions the P-C stretch of MPA at 762 cm-1 dominates and the CH2 stretch at 2922 cm-1 appears The SN is sufficiently high that the anti-Stokes Raman shift at -762 cm-1 is observed For deprotonated MPA an oxygen-surface mode appears at 325 cm-1 (as well as its anti-Stokes complement) suggesting a strong interaction This results in substantial enhancement of the P-O-C mode at 1051 cm-1 (upper trace) Others report that this mode dominates the infrared spectra of nerve agents measured in water8 Comparison of the two spectra suggests the following molecule-to-surface orientations The appearance of the oxygen-surface and P-O-C modes in the upper spectrum of Figure 5B indicates that the tetrahedral molecule interacts with the silver surface through the deprotonated oxygen and is oriented end-on The
A B
Wavenumber (∆cm-1)
C A
Wavenumber (∆cm-1) Wavenumber (∆cm-1)
D B
Wavenumber (∆cm-1)
Proc SPIE Vol 4577
171
dominance of the P-C and the CH2 stretches and the disappearance of the P-O-C mode in the upper spectrum suggest the molecule is oriented with the methyl-hydroxyl-hydroxyl face towards the surface However considerably more research must be performed to verify these points Both chemical forms were measured as a function of concentration to determine the onset of monolayer coverage and estimate expected detection limits (Figure 6) Below monolayer coverage the signal to concentration dependence should be linear and the SN of any spectral measurement in this range can be used to predict the detection limit In the spectra presented here the peak height was used as the signal while the noise as root-mean-squared (RMS) was measured between 4400-4600 cm-1 Since noise is distributed evenly throughout the spectrum when transformed this region was used since it does not have any contributions from signals or baseline offsets Figure 6 shows a series of spectra for MPA along with a plot of the 1050 cm-1 band intensity (the noise was identical for this series) as a function of sample concentration A clear discontinuity appears in the vicinity of 01 mgmL (19 ppm) indicating the onset of monolayer coverage A detection limit defined as a SN of 3 was calculated for the 01 and 005gmL samples at 24x10-4 and 25x10-4 gL respectively A more modest detection limit of 101x10-4 gL was obtained using the 760 cm-1 band in the second series of concentration measurements These detection limits correspond to 46 and 190 ppb for a 3-min scan and 75 mW of incident laser power
Figure 6 A) Concentration dependence of MPA SERS measured in silver-doped TMOS) B) Concentrations are 001 005 01 05 1 gL (188 94 188 94 188 ppm) I760 series (bull) and I1050 series (∆)
Enhancement factors for cyanide and methyl phosphonic acid can be determined by estimating the number of molecules contributing to the surface-enhanced and normal Raman spectra The enhancement factor EF can be defined by the following equation
EF = (ISERSINR)bull(MNRMSERS) bull(PNRPSERS) bull(TNRTSERS)12
where I is the spectral band intensity M is the sample mass P is the incident laser power and T is the measurement time (or number of scans) for the two measurements For the normal Raman spectra a cylindrical scattering volume is assumed based on the laser area (28x10-7m2 6x10-4m diameter spot) and the penetration depth (1x10-3 m)24 The density of KCN and MPA as powders were measured at 0572 and 0516 gcm3 indicating that 16x10-4 and 144x10-4 g produced the normal Raman signals in Figure 5 respectively The number of molecules contributing to the SER spectra are those on the silver particles embedded in the sol-gel The total silver surface area can be determined from the average particle size concentration and the scattering volume Previous scanning electron micrographs showed the average silver particle size to be 40 nm in diameter (335x10-23m3)17 The silver concentration is 012M based on the reactant molar concentrations and dilution factors And the scattering volume is 76x10-
11m3 again based on a cylindrical scattering volume defined by a laser area of 28x10-7m2 and a sol-gel thickness of 27x10-
4m This volume contains 123x10-6g of silver equivalent to 35x109 silver particles with a collective surface area of 18x10-
5m2 However it may reasonably be assumed that at least half of the silver surface is in contact with the sol-gel matrix and unavailable for analyte interaction If we assume monolayer coverage and that each CN molecule occupies 15x10-20m2 then approximately 62x1014 molecules or 27x10-8g of CN contribute to the SER spectrum (20x10-19m2 46x1013 molecules 74x10-9g for MPA) Accordingly the EF for cyanide equals 48x104 ((180599) bull(16x10-427x10-8) bull(90075) bull(500100)12) The EF for MPA is considerably higher at 87x106 ((60326) bull(144x10-474x10-9) bull(90075) bull(500200)12)
0
20
40
60
80
100
120
140
0 02 04 06 08 1 12[MPA] (mgmL)
I (76
0)
0
100
200
300
400
500
600
I (10
50)
Wavenumber (∆cm-1)
A B
Proc SPIE Vol 4577
172
4 CONCLUSIONS Here we present for the first time surface-enhanced Raman spectra of Tabun and Sarin obtained using silver-doped sol-gels However the inferior enhancement suggested employing SER-active sol-gels varying in composition to improve both detection limits and selectivity of the target analytes To this end we successfully demonstrated the capabilities of four sol-gels that select for 1) polar-positive 2) polar-negative 3) weakly polar-positive and 4) weakly polar-negative chemical species p-aminobenzoic acid was used to show that silver could be used to attract polar-negative chemicals or functional groups (carboxylate anion) and that gold could be used to attract polar-positive chemicals (amine) while a mixture of p-aminobenzoic acid and phenylacetylene was used to show that tetramethyl orthosilicate preferentially solvates polar chemicals and that tetraethyl orthosilicate preferentially solvates weakly polar chemicals This increased sample control was applied to cyanide and methyl phosphonic acid two hydrolysis products of chemical warfare agents Exceptional results were obtained for methyl phosphonic acid allowing measurement of 1x10-2 gL for a 3-min scan and 75 mW of incident laser power with an estimated detection limit of 245x10-4 gL and an enhancement factor of 87x106 However this detection limit is 76 times less sensitive than required for the JSAWM (32x10-6gL for the G-agents) Finally we note that the measurements performed here employed an FT-Raman spectrometer This instrumentation was chosen over dispersive Raman instrumentation for the high wavelength accuracy afforded by the HeNe reference laser (Connes Advantage25) which would allow reliable spectral subtraction matching of observed spectra to stored library spectra and confident use of chemometric approaches Such data analysis is likely to be required to identify the chemical agents as well as distinguish hydrolysis products However this instrumentation which employs 1064 nm excitation and InGaAs detection sacrifices sensitivity We believe that the measurement sensitivity can be improved by at least two orders of magnitude by using shorter laser excitation wavelengths (eg 532 nm) This would provide more efficient Raman scattering (fourth power dependence on laser excitation wavelength) more efficient generation of plasmon modes and allow using more efficient detector material (Si vs InGaAs) These modifications are underway
5 ACKNOWLEDGEMENTS The authors would like to thank Dr R Yin and J Jensen of the US Army for supporting this work (Contract Number DAAD13-01-C-0019) They would also like to thank Advanced Fuel Research for making their laboratory facilities available
6 REFERENCES 1 Tu Anthony ldquoOverview of Sarin Terrorist Incidents in Japan in 1994 and 1995rdquo 6th CBW Protection Symposium
Stockholm Sweden 10-15 May 1998 2 Jensen JL NC Wong W R Loerop SPIE 4775-03 (2001) 3 JSAWM Requirements at wwwsbccomapgeaarmymilRDAecbcrtPRODSERJSAWMjsawmhtml 4 ldquoThe Chemical Weapons Convention Redefines Analytical Challengerdquo Analytical Chemistry News amp Features June 1
397A (1998) 5 Johnston RL Hoefler CM Fargo JC and Moberley B ldquoThe Defense Nuclear Agencyrsquos ChemicalBiochemical
Weapons Agreements Verification Technology Research Development Test and Evaluation Program and its Requirements for On-Site Analysisrdquo AT-ONSITE 5-8 (1994)
6 Black RM Clarke RJ Read RW and Reid MT ldquoApplication of gas chromatography-mass spectrometry and gas chromatography-tandem mass spectrometry to the analysis of chemical warfare samples found to contain residues of the nerve agent sarin sulphur mustard and their degradation productsrdquo J Chromatography 662 301-321 (1994)
7 Hoffland LD Piffath RJ Bouck JBrdquoSpectral signatures of chemical agents and simulantsrdquo Optical Engineering 24 982-984 (1985)
8 Braue EHJ Pannella MGrdquoCIRCLE CELL FT-IR Analysis of Chemical Warfare Agents in Aqueous Solutionsrdquo Applied Spectroscopy 44 1513-1520 (1990)
9 Tseng C-H Mann CK and Vickers TJ ldquoDetermination of Organics on Metal Surfaces by Raman Spectroscopyrdquo Applied Spectroscopy 47 1767-1771 (1993)
10 Staff Reporter ldquoUS Army Tests Lidar to Detect Biological Toxinsrdquo Photonic Spectra pg 50 December 1998
Proc SPIE Vol 4577
Proc SPIE Vol 4577
173
11 Christesen SD Raman cross sections of chemical agents and simulants Applied Spectroscopy 42 318-321 (1988) 12 Alak AM and Vo-Dinh T ldquoSERS of Organophosphorous Chemical Agentsrdquo Analytical Chemistry 59 2149-2153
(1987) 13 Norrod KL Sudnik LM Rousell D and Rowlen KL ldquoQuantitative Comparison of Five SERS Substrates
Sensitivity and Detection Limitrdquo Applied Spectroscopy 51 994-1001 (1997) 14 Lee Y and Farquharson S ldquoSERS Sample Vials Based on Sol-Gel Process for Trace Pesticide Analysisrdquo SPIE 4206
140-146 (2000) 15 Farquharson S and Lee Y ldquoTrace Drug Analysis by Surface-Enhanced Raman Spectroscopyrdquo SPIE 4200-16 (2000) 16 Lee Y Farquharson S and Rainey P M Surface-Enhanced Raman Sensor for Trace Chemical Detection in Water
SPIE 3857 76-84 (1999) 17 Lee Y Farquharson S Kwong H and Shahriari M ldquoSurface-Enhanced Raman Sensor for Surface-Enhanced Raman
Spectroscopyrdquo SPIE 3537 252-260 (1998) 18 Farquharson S Smith W Carangelo R C and Brouillette C ldquoIndustrial Raman Providing Easy Immediate Cost
Effective Chemical Analysis Anywhererdquo SPIE 3859 14-23 (1999) 19 Narayanan VA JM Bello JD Stokes and T Vo-Dinh Analusis 19 307-310 (1991) 20 Laserna JJ E L Torres and JD Winefordner Analytica Chemica Acta 469-480 (1987) 21 Torres EL and JD Winefordner Analytical Chemistry 59 1626-1632 (1987) 22 Suh JS DP DiLella M Moskovits J Phys Chem 87 1540-1544 (1983) 23 Benner RE R Dornhaus R Chang and BL Laube Correlations in the Raman spectra of cyanide complexes adsorbed
at silver electrodes with voltammograms Surface Science 101 341 (1980) 24 Chase D B and JF Rabolt Fourier Transform Raman Spectroscopy Acad Press Ch1 p 131 (1994) 25 Connes J Rev Opt Theor Instrum 40 45 (1961)
APPLIED SPECTROSCOPY 351
focusing the 488 nm laser beam 2 mm deep into thebulk of the crystals to avoid contributions from potentialdepletion layers As shown in Fig 1b the dependence ofthe Raman shift on the degree of deuteration is almostperfectly linear and ts very well with D 5 22684cmR1 24526 where D is the degree of deuteration (in )and R is the spectral mean of the PO4 vibration in cm21A linear correlation coef cient of 0998 indicates an ex-cellent linear dependence of the Raman peak shift withdegree of deuteration This result shows that the shift ofthe PO 4 peak is simply caused by the linear increase inatomic mass due to isotope substitution which decreasesthe length of hydrogen-like bonds
This excellent linear dependence allows us to map thepro le of the DH exchange layer at the surface of DKDPcrystals by acquiring Raman spectra and determining theposition of the PO 4 peak for various depths This methodis preferable over other methods such as determining thestrength of the OD vibration directly (eg at 715 cm21)because the position of the most intense peak in the Ra-man spectrum can be measured more precisely than theintensity of some of the weakest peaks in the spectrumThis is demonstrated in Fig 2 where depth-dependentRaman spectra (Fig 2a) and the resulting exchange layerpro les for two DKDP crystals are shown (Fig 2b) Thespectra in Fig 2a were obtained from a depth scan of aDKDP crystal with 75 degree of deuteration in thebulk grown at 45 8C The spectra start out as DKDP with30 deuteration close to the surface and approach thebulk DKDP spectrum within a few micrometers of depthThe fact that the relative degree of deuteration does notextend to 0 D is due to the limited depth resolution ofthe Raman microprobe which averages over 4 mm indepth Figure 2b depicts the resulting DH exchange layerpro les for this and a second crystal grown at 63 8Crespectively Both crystals had the same exposure to am-bient conditions and their main difference is the temper-ature at which they were grown The different exchangelayer pro les indicate that crystals grown at differenttemperatures have differen t proton conductiv ities 13
which leads to a difference in their rate of deuteriumdepletion The parameters controlling this behavior arecurrently the objective of a detailed study the results ofwhich will be reported elsewhere
CONCLUSION
In conclusion we have shown that the shift of the to-tally symmetric PO 4 stretch mode in the Raman spectrumof DKDP crystals scales linearly with degree of deuter-ation This allows us to correlate Raman peak positionsto deuteration levels in these crystals We have presenteda new technique to determine DH diffusion pro les inDKDP frequency conversion crystals based on micro-Ra-man spectroscopy This technique is fast inexpensiveand works under various environmental conditionswhich will allow us to better understand and control deu-terium depletion in DKDP crystals
ACKNOWLEDGMENTS
We would like to thank M Runkel for rst discovering DKDP crack-ing R Floyd for providing DKDP crystals and L Chase and A Burn-ham for their support and helpful discussions This work was performedunder the auspices of the US Department of Energy by the University
of California Lawrence Livermore National Laboratory under contract W-7405-Eng-48
1 J J De Yoreo A K Burnham and P K Whitman Int Mat Rev47 113 (2002)
2 C E Barker R A Sacks B M Van Wonterghern J A Caird JR Murray J H Campbell K Kyle R B Ehrlich and N DNielsen Proc SPIE-Int Soc Opt Eng 2633 501 (1995)
3 T Suratwala paper to be published4 Cleveland Crystals Inc httpwwwclevelandcrystalscom
KDPshtmltable5 E A Popova I T Savatinova and I A Velichko Sov Phys Solid
State 12 1543 (1971)6 I P Kaminow R C C Leite and S P S Porto J Phys Chem
Solids 26 2085 (1965)7 J A Subramony B J Marquardt J W Macklin and B Kahr
Chem Mat 11 1312 (1999)8 H Tanaka M Tokunaga and I Tatsuzaki Solid State Commun
49 153 (1984)9 R J Nelmes G M Meyer and J E Tibballs J Phys C 15 59
(1982)10 M A Yakshin D W Kim Y S Kim Y Y Broslavets O E
Sidoryuk and S Goldstein Laser Physics 7 941 (1997)11 I Takenaga Y Tominaga S Endo and M Kobayashi Solid State
Commun 84 931 (1992)12 C Krenn personal communication13 M Sharon and A K Kalia J Solid State Chem 21 171 (1977)
Rapid Dipicolinic Acid Extractionfrom Bacillus Spores Detectedby Surface-EnhancedRaman Spectroscopy
STUART FARQUHARSON ALAND GIFT PAUL MAKSYMIUK andFRANK E INSCOREReal-Time Analyzers Inc East Hartford Connecticut06108
Index Headings Dipicolinic acid Bacillus spores Anthrax Surface-enhanced Raman spectroscopy
INTRODUCTION
The anxiety caused by the distribution of anthrax en-dospores through the US postal system in October 2001was exacerbated by the long time required for positiveidenti cation of the Bacillus anthracis spores and the un-known extent of their distribution Since that time manymethods capable of rapid eld analysis have been inves-tigated to augment or replace the laboratory method ofgrowing microorganisms in culture media which takesdays to perform12 Prominent among these approachesare polymerase chain reactions (PCR)3 immunoassaysand detection of calcium dipicolinate as a biochemicalsignature PCR employs primers to separate organism-
Received 10 October 2003 accepted 14 November 2003 Author to whom correspondence should be sent
352 Volume 58 Number 3 2004
speci c nucleic acid sequences (eg capsular protein en-coding gene for Bacillus anthracis)4 and polymerases toamplify the segment until it is detectable Recently am-pli cation times have been substantially reduced andcomplete analysis can now be performed in an hour orless Immunoassay methods are also being developed thatuse competitive binding of the bioagent (as an antigen)and its labeled conjugate for a limited number of anti-bodies Although analyses can be performed in under 30minutes a well-de ned anthrax antigen has not yet beenidenti ed5ndash7 and consequently the false-positive rate isunacceptably high8
A number of other methods are being developed witha focus on the detection of calcium dipicolinate (CaDPA)and its derivatives as a B anthracis signature This is sobecause only spore-forming bacteria contain CaDPA andthe most common potentially interfering spores such aspollen and mold spores do not Relatively fast methodshave been developed to chemically extract CaDPA andthen detect it directly by uorescence9 or indirectly byluminescence1011 In the latter case hot dodecylamine(DDA) has been used to extract dipicolinic acid (DPA)and terbium has been utilized to form a highly lumines-cent DPA complex11 Although measurements have beenperformed in as little as ve minutes it was found thatas many as three concentration-dependent complexes canform each with different lifetimes This coupled withthe fact that the Tb31 cation produces the same lumines-cence spectrum makes determinations of low spore con-centrations problematic
It has been long known that Raman spectra of Bacillispores are dominated by bands associated with CaDPA12
and that these spectra may provide a suitable anthrax sig-nature at the genus level13 Since that time considerableimprovements in Raman instrumentation have led to lab-oratory measurements of single Bacilli spores14 and to eld measurements of spores captured from a mail-sort-ing system15 However the single spore measurementsrequired complex instrumentation that is not ruggedwhile the eld measurements required milligrams of sam-ple Furthermore the Raman spectra of both measure-ments contained uorescence contributions that would in-crease uncertainty in quanti cation
In related research we demonstrated that nanogramquantities of DPA could be detected by uorescence-freesurface-enhanced Raman spectroscopy (SERS)16 We alsodemonstrated that microliter volumes of chemicals canbe detected by SERS using metal-doped sol-gel-packedglass capillaries17 Towards the goal of developing a rap-id eld SERS-based anthrax spore detector we havecombined our previous research and we now report thatDPA can be extracted from a 10 mg B cereus spore sam-ple using DDA in 1 minute and can be detected by SERSin an additional 1 minute
EXPERIMENTAL
Dipicolinic acid (26-pyridinedicarboxylic acid DPA)and dodecylamine (DDA) were used as received fromSigma-Aldrich (Milwaukee WI) Lyophilized B cereusspores prepared according to the literature13 were sup-plied by the University of Rhode Island and used as re-ceived Multiple particles approximately 01 mm3 each
were separated and weighed at 5 to 15 mg representing05 to 15 million spores The sample masses were con-sistent with a previous determination of spore density at0081 gmL that indicated a high degree of entrained air
All chemicals used to prepare the silver-doped sol-gelcoated capillaries were also obtained and used as receivedfrom Sigma-Aldrich According to previously publishedprocedures17 two precursor solutions were preparedmixed and then drawn into 1-mm-diameter glass capil-laries The silver amine precursor consisted of a 51 vvratio of 1 N AgNO 3 to 28 NH3OH while the alkoxideprecursor consisted of a 21 vv ratio of methanol to te-tramethyl orthosilicate The alkoxide precursors weremixed with silver amine precursor in an 81 vv ratioApproximately 015 mL was drawn into the capillarycoating a 15-mm length After sol-gel formation the in-corporated silver ions were reduced with dilute sodiumborohydride which was followed by a water wash to re-move residual reducing agent
A 100 mL drop of a 50 mM DDA solution in ethanolpre-heated to 78 8C was added to each of the B cereusparticles to digest the spore coat After 1 minute the re-sultant solution was drawn into a SER-active capillarythat was immediately xed horizontally to an XY posi-tioning stage (Conix Research Spring eld OR) just in-side the focal point of an f 07 aspheric lens The lensfocused the beam into the sample and collected the scat-tered radiation back along the same axis A dichroic lter(Omega Optical Brattleborough VT) was used to re ectthe excitation laser to the lens and pass the Raman scat-tered radiation collected by the lens An f 2 achromatwas used to collimate the laser beam exiting a 200-mm-core-diameter source ber optic while a second f 2 ach-romat was used to focus the scattered radiation into a 365mm ber optic (Spectran Avon CT) A short-pass lterwas placed in the excitation beam path to block the sil-icon Raman scattering generated in the source ber fromre ecting off sampling optics and reaching the detectorA long-pass lter was placed in the collection beam pathto block the sample Rayleigh scattering from reachingthe detector A 785 nm diode laser (Process InstrumentsInc model 785-600 Salt Lake City UT) was used todeliver 100 to 150 mW of power to the sample A Fouriertransform Raman spectrometer (Real-Time Analyzersmodel IRA-785 East Hartford CT) and a silicon photo-avalanche detector (Perkin Elmer model C30902S Stam-ford CT) were used to acquire the SER spectra
RESULTS AND DISCUSSION
As an initial experiment the SER spectrum of 1 gLof DPA in water was measured using the newly devel-oped silver-doped sol-gel-coated capillaries (Fig 1A) Atthis concentration a high signal-to-noise ratio (SN) isobtained in 1 min In fact a reasonable spectrum is ob-tained in the same time frame for 1 mgL (Fig 1B) TheSER spectra are reasonably similar to the normal Raman(NR) spectrum obtained for a saturated solution of DPAin 1 N KOH (Fig 1C) and the following band shifts areobserved (NR to SER) 647 to 657 cm21 817 to 815cm21 998 to 1008 cm21 1384 to 1382 cm21 1434 to1428 cm21 and 1569 to 1567 cm21 Many of these bandshave been previously assigned1213 such as 998 cm21 to
APPLIED SPECTROSCOPY 353
FIG 1 SERS of DPA in water using silver-doped sol-gel-coated glasscapillary for (A) 1 gL and (B) 1 mgL (C ) NR of saturated DPA in 1N KOH in a glass capillary Spectral conditions (A) and (B) 150 mWof 785 nm 1-min acquisition time (C) 450 mW of 785 nm 5-minacquisition time both 8 cm 21 resolution
FIG 2 SERS of DPA extracted from 10 mg B cereus particle using100 mL of 50 mM hot DDA acquired in (A) 1 minute and (B) 2 seconds(C ) Attempted SERS of 50 mM hot DDA in ethanol using silver-dopedsol-gel-coated glass capillary acquired in 1 min Spectral conditions150 mW of 785 nm 8 cm21 resolution
the symmetric ring stretch 1384 cm21 to the OndashCndashOsymmetric stretch 1428 cm21 to the symmetric ring CndashH bend and 1569 cm21 to the asymmetric OndashCndashOstretch
The rst B cereus samples consisted of 2 mg of sporesin 2 mL of 5 mM hot DDA The samples were main-tained at 78 8C for 40 min and while hot approximately10 mL was drawn into a SER-active capillary Since spec-tra of DPA were obtained for these initial samples small-er spore masses higher DDA concentrations and shorterheating periods were examined In due course it wasfound that 10 mg of spores could be digested by 100 mLof 50 mM hot DDA in one minute and detected (Fig2A) In fact the signal was suf ciently intense that it canbe observed in as little as two seconds (Fig 2B) Theamount of DPA that was extracted was estimated to bebetween 5 and 10 mgL by comparing the signal intensityof the 1008 cm21 band to that measured for DPA in waterThis is consistent with previous research that found thatthe majority of the DPA is extracted from spores usingDDA11 and that B cereus spores contain approximately10 DPA by weight18 The SN of 127 for the 1008 cm21
band in the 1-minute SER spectrum suggests a limit ofdetection of approximately 250 ng of B cereus sporesbased on a SN of 3 Finally it should be noted that DDAdid not produce a detectable SER spectrum as shown inFig 2C
CONCLUSION
We have demonstrated that by combining rapid extrac-tion of dipicolinic acid from Bacillus cereus spores withchemical identi cation by surface-enhanced Raman spec-troscopy as little as 10 mg of spores can be detected Infact the entire measurement from the time of adding hotdodecylamine to the spores to the time when the dipi-colinic acid SER spectrum is acquired and analyzedcould be performed in less than two minutes The abilityof this method to distinguish between spore-forming bac-teria such as Bacillus anthracis and non-DPA containingpowders could help prevent costly shutdowns associated
with the appearance of suspicious material or intentionalmailing of common substances as an anthrax hoax Thismethod could also prove useful in detecting the locationof anthrax endospores in mail distribution facilities if an-other veri ed attack should occur
Research continues to fully characterize the surface-enhanced Raman spectroscopy signal intensities as afunction of sample concentration and to explore otherextractants that do not require the use of elevated tem-perature
ACKNOWLEDGMENTS
The authors are grateful for the support of the National Science Foun-dation (DMI-0296116 and DMI-0215819) and the US Army(DAAD13-02-C-0015 Joint Service Agent Water Monitor program)The authors are indebted to Chetan Shende for preparing the sol-gelcapillaries The authors also thank James Gillespie Nicholas Fell andAugustus Fountain for providing important background informationMark Farquharson for laboratory support and Professor Jay Sperry ofthe University of Rhode Island for supplying B cereus spores
1 V A Pasechnik C C Shone and P Hambleton Bioseparations 3267 (1993)
2 P J Jackson M E Hugh-Jones D M Adair G Green K K HillC R Kuske L M Grinberg F A Abramova and P Keim ProcNatl Acad Sci USA 95 1224 (1998)
3 B R Glick and J J Pasternak Molecular Biology Principles andApplications of Recombinant DNA (ASM Press Washington DC1994)
4 C A Bell J R Uhl T L Had eld J C David R F Meyer TF Smith and F R Cockerill III J Clin Microbiol 40 2897(2002)
5 D L Gatto-Menking H Yu J G Bruno M T Goode M Millerand A W Zulich Biosens Bioelectron 10 501 (1995)
6 J J Quinlan and P M Foegeding J Rapid Methods AutomationMicrobiol 6 1 (1998)
7 A A Hindle and E A H Hall Analyst (Cambridge UK) 1241599 (1999)
8 M S Ascher US Department of Health amp Human Services(httpwwwhhsgovophppresentationsAscherdoc)
9 R Nudelman B V Bronk and S Efrima Appl Spectrosc 54 445(2000)
10 D L Rosen C Sharpless and L B McBrown Anal Chem 691082 (1997)
354 Volume 58 Number 3 2004
11 P M Pellegrino N F Fell Jr and J B Gillespie Anal ChimActa 455 167 (2002)
12 W H Woodruff T G Spiro and C Gilvarg Biochem BiophysRes Commun 58 197 (1974)
13 E Ghiamati R S Manoharan W H Nelson and J F SperryAppl Spectrosc 46 357 (1992)
14 A P Esposito C E Talley T Huser C W Hollars C M Schal-dach and S M Lane Appl Spectrosc 57 868 (2003)
15 S Farquharson L Grigely V Khitrov W W Smith J F Sperryand G Fenerty J Raman Spectrosc paper accep ted (2003)
16 S Farquharson W W Smith S Elliott and J F Sperry SPIE-IntSoc Opt Eng 3855 110 (1999)
17 S Farquharson and P Maksymiuk Appl Spectrosc 57 479(2003)
18 F W Janssen A J Lund and L E Anderson Science (Washing-ton DC) 127 26 (1958)
SPIE -2003-5269 117
pH dependence of methyl phosphonic acid dipicolinic acid and cyanide by surface-enhanced Raman spectroscopy
Stuart Farquharson Alan Gift Paul Maksymiuk Frank Inscore and Wayne Smith
Real-Time Analyzers 87 Church Street East Hartford CT 06108
ABSTRACT US and Coalition forces fighting terrorism in Afghanistan and Iraq must consider a wide range of attack scenarios in addition to car bombings Among these is the intentional poisoning of water supplies to obstruct military operations To counter such attacks the military is developing portable analyzers that can identify and quantify potential chemical agents in water supplies at microgram per liter concentrations within 10 minutes To aid this effort we have been investigating the value of a surface-enhanced Raman spectroscopy based portable analyzer In particular we have been developing silver-doped sol-gels to generate SER spectra of chemical agents and their hydrolysis products Here we present SER spectra of methyl phosphonic acid and cyanide as a function of pH an important factor affecting quantitation measurements which to our knowledge has not been examined In addition dipicolinic acid a chemical signature associated with anthrax-causing spores is also presented Keywords Chemical warfare agents agent detection agent hydrolysis SERS Raman spectroscopy homeland security
1 INTRODUCTION In the past decade the Unites States and its allies have been challenged by a different kind of warfare exemplified by the terrorist attacks of September 11 2001 Suicide bombings and the use of chemical agents are the norm and military personnel must consider a wide range of attack scenarios Among these is the intentional poisoning of water supplies to obstruct military operations in Afghanistan and Iraq Currently colorimetric paper is used to detect agents on-site while gas chromatography combined with mass spectrometry (GCMS) is used in mobile support laboratories However both methods have severe drawbacks The paper changes color in response to contact with many chemicals besides CWAs causing a high incidence of false positives1 while GCMS although very chemically specific requires hours to perform and constant re-calibration234 Military operations would be greatly aided by a portable analyzer that can identify and quantify potential chemical agents at concentrations that impact safety This includes the analysis of drinking water supplies distribution and storage systems To meet this goal the Department of Defense has been investigating numerous approaches under the auspices of the Joint Service Agent Water Monitor (JSAWM) program5 The JSAWM must be capable of identifying the chemical agents with no false-positives and quantifying the agents at microgram per liter concentrations within 10 minutes678 Recently we and others have been investigating the ability of surface-enhanced Raman spectroscopy (SERS) to measure chemical agents9-12 bioagents13-17 and their hydrolysis products in water SERS employs the interaction of surface plasmon modes of metal particles with target analytes to increase scattering efficiency by as much as 1 million times18 In 1987 the potential of this approach to measure CWAs was demonstrated by measuring a series of pesticides19 Several of these organophosphonates have chemical structures similar to CWAs in particular P=O functional groups In our studies we have been employing silver-doped sol-gels to promote the SER effect The porous silica network of the sol-gel matrix offers a unique environment for immobilizing and stabilizing SER-active metal particles20-23 The sol-gels have been coated on the internal walls of standard glass vials to allow rapid SER analysis of small sample quantities without preparation We have measured over 100 chemicals with enhancements of 104 to 106 demonstrated reversible measurements in a flowing system reproducible measurements from vial-to-vial and measurements in multiple solvents including water20-23 Previously we used these vials to perform preliminary measurements of cyanide (CN) methylphosphonic acid (MPA) and dipicolinic acid (DPA) MPA is a hydrolysis product of the nerve agents (eg sarin Reaction 1) and may be a valuable indicator of nerve agent usage particularly since the alkyl methylphosphonic acids are relatively more stable than their corresponding parent complexes24 DPA is
SPIE -2003-5269 118
a chemical signature of spore forming bacteria such as Bacillus anthracis And in light of the inability to rapidly detect the anthrax spores distributed through the US mail in October 2001 a number of methods are being developed to extract and analyze this signature Reaction 1 Stepwise hydrolysis of Sarin to form hydrofluoric acid (HF) isopropyl methylphosphonic acid (IMPA) then methyl phosphonic acid (MPA) and 2-propanol In our previous SERS investigations MPA and DPA were measured at 50 and 100 mgL respectively In both cases limits of detection (LOD) were estimated at 100 microgL providing encouragement in that SERS may satisfy the needs of the JSAWM Since it has been shown that pH can substantially influence the intensity of SER bands25 which would clearly influence quantitative analysis we undertook the present study to determine the severity of these effects for cyanide methyl phosphonic acid and dipicolinic acid Furthermore we previously observed a band at 1050 cm-1 for MPA6 possibly due to an anion formed at basic pH Here we investigate the source of this spectral anomaly
2 EXPERIMENTAL All chemicals including potassium cyanide methyl phosphonic acid dipicolinic acid and those used to prepare the silver-doped and gold-doped sol-gel coated vials were obtained and used as received from Sigma-Aldrich All samples were prepared in HPLC grade water (Fischer Scientific Fair Lawn NJ) for SERS measurements The pH of these samples was adjusted using dilute nitric acid or potassium hydroxide and verified using a pH electrode (Corning Inc Corning NY) that had been calibrated with pH 400 700 and 1000 buffered standards from Fischer Scientific Once prepared the samples were transferred into the silver-doped sol-gel vials (Real-Time Analyzers Inc East Hartford CT) The vials were coated in a manner similar to that previously reported by adding ammonium hydroxide to a solution of silver nitrate tetramethyl orthosilicate (TMOS) and methanol20 Gold-vials were coated by adding nitric acid to a solution of gold tetrachloride TMOS and methanol The two precursor solutions were prepared mixed and transferred to 2-mL glass vials dried and heated After sol-gel formation the incorporated metal ions were reduced with dilute sodium borohydride (1mgmL) which was followed by a water wash to remove residual reducing agent After the resultant analyte solution was introduced the SER-active vial was immediately fixed horizontally to an XY positioning stage (Conix Research Springfield OR) just inside the focal point of an f07 aspheric lens The lens focused the beam into the sample and collected the scattered radiation back along the same axis A dichroic filter (Omega Optical Brattleborough VT) was used to reflect the excitation laser to the lens and pass the Raman scattered radiation collected by the lens An f2 achromat was used to collimate the laser beam exiting a 200 microm core diameter source fiber optic while a second f2 achromat was used to focus the scattered radiation into a 365 microm fiber optic (Spectran Avon CT) A short pass filter was placed in the excitation beam path to block the silicon Raman scattering generated in the source fiber from reflecting off sampling optics and reaching the detector A long pass filter was placed in the collection beam path to block the sample Rayleigh scattering from reaching the detector A 785 nm diode laser (Process Instruments Inc model 785-600 Salt Lake City UT) was used to deliver 100 to 150 mW of power to the sample A Fourier transform Raman spectrometer (Real-Time Analyzers model IRA-785 East Hartford CT) and a silicon photo-avalanche detector (Perkin Elmer model C30902S Stamford CT) were used to acquire the SER spectra
3 RESULTS AND DISCUSSION In a previous study of MPA6 aimed at developing a concentration calibration curve and determining limits of detection (LOD) we observed an anomaly at 1050 cm-1 Since it was found that the band intensity changed as a function of concentration the band must be associated with a sample parameter Two possible parameters photon flux and pH are examined here The first parameter was investigated by irradiating a 1mgmL MPA sample in a SER-active vial with laser powers of 200 mW and above and monitoring spectral changes It was immediately found that the 1050 cm-1 band
2O+ H HF + +OH
OH3H C
OP
OF
CH
CH3
3
3H C
OCP
OOH
CH
CH3
3
3H C
O
CPHO
CH
CH
3
3
C
Sarin IMPA MPA 2-propanol
SPIE -2003-5269 119
appeared and grew as a function of time and that the higher the laser power the faster the growth Figure 1 shows the growth of the 1050 cm-1 band over the course of 30 minutes when using 150 mW of 785 nm excitation while Figure 2A shows that the growth can be fit with a first order exponential equation namely I1050 = 03+05e-013t Furthermore the 760 cm-1 band could be fit with a first order decay equation with an identical exponential rate constant ie I760 = 08-
08e-013t The rates represent classical first order kinetics and their correspondence allows one to conclude that MPA is being transformed one-for-one into a photo-generated product At this time the photoproduct has not been positively identified but phosphonic (phosphorous) acid and phosphonate are likely candidates since the symmetric P(OH)3 stretch occurs at ~1050cm-1 Our studies show that a reduction in laser power to 100 mW at the sample for MPA essentially eliminates this degradation process This laser power or lower was used for further measurements
Methyl phosphonic acid is a diprotic acid that stepwise dissociates into two anions MPA- and MPA= according to the following reactions26
MPA MPA- + H+ pKa1 = 212 Reaction 2
MPA- MPA= + H+ pKa2 = 729 Reaction 3 The relative concentrations of MPA MPA- and MPA= can be determined at any pH by expressing [MPA] and [MPA=] in terms of [MPA-] using Reactions 2 and 3 and summing all three to equal the total starting concentration here 2 mgmL (0021M MW = 9602) viz
[MPA] + [MPA-] + [MPA=] = 0021M Equation 1 substituting from Reactions 1 and 2
([H+][MPA-])K1a + [MPA-] + (K2a[MPA-])[H+] = 0021M Equation 2 rearranging [MPA-] = 0021M(1+[H+]K1a + K2a[H+]) Equation 3 The relative concentrations of MPA MPA- and MPA= as a function of pH are shown in Figure 3 It is worth noting that near neutral pH both MPA- and MPA= will be present To confirm that the SER signal followed this pH dependence a starting solution consisting of 20 mg of MPA in 10 mL HPLC grade water was prepared and brought to pH of 20 using dilute nitric acid From this solution 2 mL were added to a SER-active vial and the SER spectrum recorded At this pH a peak at 760 cm-1 was barely discernable The 2 mL solution was returned to the starting solution and the pH was re-measured to correct for any changes that the silver-doped sol-gel vials might cause In most cases the change was less than 02 pH units and the pH is reported as the before and after average Next the pH of the
Figure 2 A) Exponential growth of 1050 cm-1 band and B) exponential decay of 760 cm-1 band for spectral series in Figure 1
Figure 1 Growth of 1050 cm-1 band as a function of time due to exposure to 150 mW of 785 nm Spectra are 5 sec each collected every 100-sec from 0 to 30-min
B
570 770 970 1170 1370Raman Shift (cm-1)
Arbit
rary
Unit
s
MPA Photodegradation
600 800 1000 1200 1400 Wavenumber (cm-1)
A
0 10 20 30 0 10 20 30 time (min) time (min)
Ram
an In
tens
ity (r
elat
ive)
30
min
0
SPIE -2003-5269 120
starting solution was adjusted to 325 using dilute KOH Again 2 mL were added to a vial and the SER spectrum recorded At this pH a reasonably strong 760 cm-1 band was observed This process was repeated as spectra were recorded at pHs of 70 74 75 79 85 and 100 A total of 1 mL of KOH was added diluting the total concentration by 10 Next the pH of the starting solution was made acidic by adding dilute nitric acid dropwise This time spectra were recorded at pHs of 72 69 64 and 37 Figure 4 shows the SER spectra for representative pHs (spectra were left out to simplify the figure) while Figure 3 shows the 760 cm-1 peak intensities as a function of pH (The band intensities were adjusted to compensate for dilution effects caused by the addition of HNO3 and KOH then normalized to 0021 M for the most intense band observed at pH 37) It is clear from Figure 3 that the 760 cm-1 band follows the MPA- concentration as a function of pH and must be assigned to this anion No bands were observed that corresponded to MPA or MPA= The lack of an MPA SER spectrum may be due to the absence of an attraction between the neutral analyte and the electropositive silver surface The same reasoning suggests that a strong SER spectrum should be observed for MPA= but it is not and a satisfactory explanation has not been found
0000
0005
0010
0015
0020
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14pH
Con
cent
ratio
n [M
]
pK1 = 212 pK2 = 729
MPA- MPA=
MPA
Figure 4 SER spectra of 002M MPA as a function of pH Conditions 100 mW of 785 nm 36 scans (1 min) 8 cm-1 recorded 2 min after sample introduction pH 19 and 20 not apparent on this scale pH 69 and 74 near identical to 70 and 75 and not shown for clarity
00
02
04
06
08
10
12
14
16
18
0 5 10 15 20 25 30Measurement Number
Ram
an In
tens
ity (7
60 c
m-1
)
Figure 3 Concentration dependence of MPA MPA- and MPA= as a function of pH for a 002M sample Intensity of 760 cm-1
band from Figure 6 as a function of pH ( for increasing basic adjustment for increasing acidic adjustment error was measured at ~10 for pH 64)
Figure 5 SER spectra of 002M MPA at pH 64 measured around a vial at three heights (9 points per height) Conditions as in Figure 4 but 10-sec scans
Since these measurements involved the removal and replacement of the SER-active vial in the sample holder to remove and add sample variation in the intensity as a function of vial position was minimized by illuminating the exact same height along the vial wall But this does not account for variability of the SERS response of the sol-gel coating around the vial To analyze this effect a vial containing MPA at pH 64 was rotated at ~ 40o
intervals at the original height and 18rdquo above and below this value Figure 5 shows the intensity of the 760 cm-1
band for the 27 positions It was found that the average value was 137plusmn014 an RSD of 10 overall and 5 for each height An error bar is included in Figure 4 for the pH 64 measurement
SPIE -2003-5269 121
With the above analyses in mind a preliminary investigation of the SER spectral response for MPA (as MPA-) as a function of concentration was performed A single vial was used for these measurements beginning with 1 mgL followed by measurements of 10 100 and 1000 mgL In all cases the pH was ~7 and 3 positions around the vial were measured per concentration Since the 760 cm-1 band was not observed for concentrations of 1 or 10 mgL using 100 mW of 785 nm the laser power at the sample was raised to 200 mW beginning with the 10 mgL concentration Photo-degradation was largely avoided (and not observed) by exposing the sample for only 33 seconds per spectral acquisition Representative spectra for 10 100 and 1000 mgL are shown in Figure 6 while a plot of the 760 cm-1 band intensity as a function of concentration is shown in Figure 7 These values were also used to estimate limits of detection based on the signal-to-noise ratio (SN) of the 760 cm-1 band using 100 mW of 785 nm laser excitation and a 10-min acquisition time As summarized in the Figure 7 inset the lower the measured concentration the lower the predicted LOD Even if as estimated 210 microgL could be measured using the silver-doped sol-gel vials an improvement of a factor of 70 is still required to meet JSAWM goals of measuring 3 microgL in 10-minutes Similar to MPA DPA is a diprotic acid (pKa1 = 216 and pKa2 = 692) and variations in pH will effect the relative concentrations of DPA DPA- and DPA= and possibly the SER spectra and band intensities This could prove significant if an acid or base is used to denature anthrax spores with the goal of extracting and analyzing DPA The MPA pH study described above was mimicked for DPA except that the starting solution consisted of 20 mg of DPA in 20 mL HPLC grade water (60x10-3M MW = 1671) The initial solution had a pH of 245 which was made basic by dilute KOH to pHs of 355 433 487 559 1069 and 1166 SER spectra were recorded at each pH using 100 mW of 785 nm and a 44-sec acquisition time Next one drop of concentrated nitric acid was used to remake the solution acidic at a pH of 200 Again sequential pH measurements were performed at 383 510 735 and 822 The solution pH was made acidic a third time but to pH 219 171 then 135 Throughout this process no more than 20 drops of acid or base were added and therefore the concentration was diluted by no more than 10 Most of the spectral bands showed a minor decrease in intensity as a function of increasing pH values However the bands at 525 and 795 cm-1 showed the most dramatic changes which occurred at acid pH Figure 8 shows the SER spectra of DPA for the spectral region and pH range of interest The identity of the DPA species was determined by plotting the normalized peak intensities with the lowest value set to 0 and the highest to 0006 M as a function of pH and overlaying these values on a plot of the relative concentrations for DPA DPA- and DPA= as previously done for MPA (Figure 9) As can be seen the 525 cm-1 band clearly corresponds to DPA The correspondence of the 795 cm-1 band to this species is less clear as the band retains intensity until pH of 55 This can be attributed to contribution to the overlapping band at 810 cm-1 which does not change as a function of pH The fact that most bands are observed at all pHs suggest that the primary interaction with silver is through the ring nitrogen This is supported by the fact that the most intense band occurs at 1008 cm-1 attributed to a symmetric ring breathing mode and that this interaction has been characterized for pyridine in numerous papers27
0
1
2
3
4
5
6
7
0 200 400 600 800 1000 1200
MPA Concentration (mgL)
760
Ban
d In
tens
ity
Figure 6 SER spectra of MPA in water at A) 1000 B) 100 and C) 10 mgL Conditions pH of 7 silver-doped sol-gel coated vial 200 mW of 785 nm 33-sec 8 cm-1 resolution
Figure 7 Plot of SER intensity of 760 cm-1 band of MPA as a function of concentration using 200 mW of 785 nm Inset table includes average intensity LOD standard deviation and percent deviation for each concentration but for 100 mW and 10 min
conc (mgL) LOD ave stddev dev10 021 002 1021100 072 010 14301000 312 040 1280
A
B C
SPIE -2003-5269 122
The ability to detect dipicolinic acid (DPA) by SERS is demonstrated in Figure 10 Even at 1 mgL the primary bands are visible The intensity of the dominant ring breathing mode observed at 1008 cm-1 for DPA in water (pH = 25 to 55) is plotted as a function of concentration in Figure 11 These values were also used to estimate limits of detection based on the SN of the 1008 cm-1 band using 100 mW of 785 nm laser excitation and a 10-min acquisition time Again the lower the measured concentration the lower the predicted LOD (see Figure 11 inset) and detection of 160 microgL is possible Gastrointestinal anthrax requires significant more spores than inhalation anthrax28 and a limit of detection might be placed at 1 million spores in 1 liter of water or 10 microgL Since each spore contains ~10 CaDPA by weight29 a goal for DPA might be 1 microgL indicating that the present measurements must be improved by nearly two orders of magnitude Due to the increased hazards of handling HCN gas KCN salt was used for these experiments Nevertheless all sample preparations were performed in a chemical hood KCN completely dissolves in water but its conjugate acid HCN is formed and has a Ka of 615x10-1030 viz
HCN CN- + H+ pKa = 921 Reaction 4
Figure 8 SER spectra of 1 mgmL DPA as a function of pH Conditions 100 mW 785 nm 100 scans (44-sec)
Figure 9 Concentration dependence of DPA DPA- and DPA= as a function of pH for a 0006M sample Intensity of 525 () and 795 (diams) cm-1 bands from Fig 8 as a function of pH
pH
135 171 219 383
0
0001
0002
0003
0004
0005
0006
0007
0 2 4 6 8 10 12 14pH
Con
cent
ratio
n [M
]
DPADPA-DPA=795525
DPA DPA=DPA-
pK2 = 692pK1 = 216
795 525
0
05
1
15
2
25
0 200 400 600 800 1000 1200
DPA Concentration (mgL)
1008
Ban
d In
tens
ity
conc (mgL) lod-10min-100mw1 017
10 016100 103
1000 355
Figure 10 SER spectra of DPA in water at A) 1000 B) 100 C) 10 and D) 1 mgL Conditions pH of 25-55 silver-doped sol-gel coated vial 175 mW of 785 nm 1-min 8 cm-1 D) has been multiplied by x10 to make bands visible
A
B C
D
Figure 11 Plot of SER intensity of 1008 cm-1 band of DPA as a function of concentration using 175 mW of 785 nm Inset table includes LOD in mgL for each concentration but for 100 mW and 10 min
SPIE -2003-5269 123
Consequently the cyanide concentration must be determined for each initial KCN concentration Specifically the samples prepared with concentrations of 01 1 10 100 and 1000 mgL of KCN produced CN- concentrations of 63x10-3 033 69 89 and 964 mgL at pHs of 816 90 967 102 and 107 respectively The pH dependence for the HCN and CN- concentrations are shown in Figure 12 Thus as the amount of KCN added to the solution decreases so does the pH of the solution (becomes less basic) and according to Reaction 4 the relative amount of CN- to HCN also decreases For example in the preparation of a 01 mgL solution of KCN the pH is shifted from 7 for pure water to only 816 and only 63 of the starting material becomes CN- or 63x10-3 mgL In comparison for a solution of 1000 mgL the pH is shifted from 7 to 107 and 96 of the starting material becomes CN- This is significant since the cyanide ion is better able to adsorb onto the silver particles and become SERS active SER spectra of 10 100 and 1000 mgL of KCN samples are shown in Figure 13 and a corresponding plot of the 2100 cm-1 band intensity as a function of concentration is shown in Figure 14
000102030405060708091011
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14pH
Con
cent
ratio
n [m
gm
L]
pKa = 921
CN -HCN
The SER spectra of cyanide are dominated by the single CN stretch at 2100 cm-1 which occurs in normal Raman spectra of solutions at 2080 cm-1 However a low frequency mode occurs at 135 cm-1 due to a Ag-CN stretch (not shown) indicative of a strong surface interaction It is also observed that as the concentration decreases the CN stretch shifts to 2140 cm-1 This shift has been attribute to the formation of a tetrahedral Ag(CN)3
2- surface structure31 as well
Figure 14 Concentration dependence of KCN SERS measured under conditions in Fig 11 Concentrations are 1 01 and 001 mgml Intensities are measured for the CN stretch at 2100 cm-1 Inset table includes LOD in mgL for each concentration in Figs 13 and 15 but for 100 mW and 10 min
Figure 13 SER spectra of KCN in water at A) 1000 B) 100 and C) 10 mgL Conditions pHs of 107 102 and 97 silver-doped sol-gel coated vial 100 mW of 785 nm 1-min 8 cm-1 C) has been multiplied by x10 to make band visible
Figure 12 Concentration dependence of HCN and CN- as a function of pH for a 1 mgmL sample Calculated intensity of 2100 () cm-1 band for a 1 mgml sample at pHs of 816 90 967 102 and 107
A
B
C
0
20
40
60
80
100
120
140
0 200 400 600 800 1000 1200CN Concentration (mgL)
2100
Ban
d In
tens
ity
Figure 15 SER spectra of KCN in water at A) 10 B) 1 and C) 01 mgL Conditions pHs of 97 12 and 12 gold-doped sol-gel coated vial 100 mW of 785 nm 1-min 8 cm-1 C) has been multiplied by x10 to make band visible
A
B
C
conc (mgL) lod-10min-100mw Condition01 001 Au-pH 121 007 Au-pH 12
69 003 Au-pH 9769 016 Ag-pH 9789 022 Ag-pH 102946 113 Ag-pH 107
SPIE -2003-5269 124
as to CN adsorbed to two different surface sites32 Alternatively the 2140 cm-1 band could be attributed to HCN since this species dominates at lower concentrations However it is unlikely that this species would be attracted to the electropositive silver surface Further both peaks should be present at pHs between 85 and 105 but this is not observed It has also been suggested that at concentrations near and above monolayer coverage the CN- species is forced to adsorb end-on due to crowding and at lower concentrations the molecule can reorient to lie flat33 This suggests that the 2100 and 2140 cm-1 bands correspond to the end-on and flat orientations respectively As Figures 13 and 14 show the intensity of the CN stretch for the 89 mgL sample is nearly as intense as the 964 mgmL band This suggests that the Raman signal for the flat orientation is more enhanced However more extensive measurements are required to verify this point Since resent research has suggested that cyanide may be more effectively detected on gold measurements of KCN solutions were also performed using gold-doped sol-gel vials Preliminary measurements are shown in Figure 15 for samples prepared from 01 1 and 10 mgL KCN Since the pHs are 816 90 967 the resultant CN- concentrations are 63x10-3 033 and 69 mgL Initially only the highest concentration was observed and the signal intensity was significantly better than the equivalent concentration measured using silver In an effort to shift Reaction 4 to the left transforming HCN to CN- (Le Chatelierrsquos principle) KOH was added to the lower concentration samples producing solutions with pH 12 This effectively forces all of the cyanide in solution to be CN- or 01 and 10 mgL respectively More importantly the CN stretch is now observed in the SER spectra The band appears at 2125 cm-1 as has been previously reported for gold12 As calculated for MPA and DPA LODs can be estimated from this data For the three concentrations of cyanide on silver the LODs are 016 to 11 mgL for 100 mW of 785 nm laser excitation and a 10-min acquisition time For gold pH adjusted the LODs are10 to 70 microgL an improvement of more than 10 times silver Nevertheless either substrate is sufficient to meet the JSAWM goals of measuring 3 mgL in 10-minutes as the requirements form cyanide are much less stringent than the nerve agents
4 CONCLUSIONS Here we examined the surface-enhanced Raman spectral response for methyl phosphonic acid dipicolinic acid and cyanide as a function of pH It was determined that the most prominent peak at 760 cm-1 reaches a maximum intensity between pH 3 and 7 and corresponds to the MPA- species Neither the MPA nor MPA= species appear to generate a SER spectrum and consequently no spectra were observed below pH 2 or above pH 8 In this study we also found that higher laser powers could cause photodegradation of MPA signified by the exponential growth of a band at 1050 cm-1 which is tentatively assigned to phosphorous acid Unlike MPA DPA was observed at all pHs This is attributed to the dominant interaction of the pyridine functional group with silver Minor spectral changes were observed at acid pHs and were assigned to neutral DPA Like MPA SER spectra of cyanide were pH dependent No spectra were observed for the HCN species while CN- was best observed at pHs more basic than 8 Preliminary concentration studies for the three analytes allowed estimating limits of detection for MPA DPA and CN using 100 mW of 785 nm and a 10-min acquisition time of 210 165 and 70 microgL respectively Although the latter value suggests that cyanide can be measured at sufficiently low concentrations to meet JSAWM goal improvements by 100 to 200 times are required for MPA and DPA It is clear from this study that pH of the sample is important and must be taken into account when developing concentration calibration curves as well as fieldable analyzers Future work will include tailoring the sol-gel with specific functional groups to dictate sample pH and thereby optimize sensitivity
ACKNOWLEDGMENTS The authors are grateful for the support of the US Army (DAAD13-02-C-0015 Joint Service Agent Water Monitor program) The authors would also like to thank Dr Steve Christensen of the US Army SBCCOM for helpful discussions and Mr Chetan Shende of Real-Time Analyzers for assistance in development of the gold-doped sol-gels
REFERENCES 1 Erickson B Analytical Chemistry News amp Features June 1 397A (1998)
SPIE -2003-5269 125
2 Johnston RL Hoefler CM Fargo JC and Moberley B AT-ONSITE 5-8 (1994) 3 Black RM RJ Clarke RW Read and MT Reid J Chromatography 662 301-321 (1994) 4 USA Army Center for Health Promotion and Preventive Medicine (USACHPPM) Technical Guideline 230A 5 Jensen JL NC Wong W R Loerop SPIE 4775-03 (2001) 6 Hoenig SL Handbook of Chemical Warfare and Terrorism Greenwood Press Wesport CT (2002) 7 Munro NB SS Talmage GD Griffin LC Waters AP Watson JF King and V Hauschild Env Health
Persp 107 933-974 (1999) 8 Holstege CP Kirk M Sidell FR Crit Care Clin13 923-42 (1997) 9 Farquharson S P Maksymiuk K Ong and S Christesen SPIE 4577 166-173 (2001) 10 Lee Y and S Farquharson SPIE 4378 21-26 (2001) 11 Spencer KM J Sylvia S Clauson and J Janni SPIE 4577 158-165 (2001) 12 Tessier P S Christesen K Ong E Clemente A Lenhoff E Kaler and O Velev Applied Spectroscopy 56
1524-1530 (2002) 13 Farquharson S WW Smith S Elliott and JF Sperry SPIE 3533 207-214 (1998) 14 Farquharson S WW Smith S Elliott and JF Sperry SPIE 3855110-116 (1999) 15 Farquharson S WW Smith YH Lee S Elliott and JF Sperry SPIE 4575 62-72 (2002) 16 Guzelian AA J Sylvia J Janni S Clauson and KM Spencer SPIE 4577 182-192 (2001) 17 Shende C F Inscore A Gift P Maksymiuk and S Farquharson in press 18 Weaver MJ S Farquharson and MA Tadayyoni J Chem Phys 82 4867-4874 (1985) 19 Alak AM and T Vo-Dinh Analytical Chemistry 59 2149-2153 (1987) 20 Lee Y and S Farquharson SPIE 4206 140-146 (2000) 21 Farquharson S and Y Lee SPIE 4200-16 (2000) 22 Lee Y S Farquharson and P M Rainey SPIE 3857 76-84 (1999) 23 Lee Y S Farquharson H Kwong and M Shahriari SPIE 3537 252-260 (1998) 24 Wang J M Pumera G Collins and A Mulchandani Analytical Chemistry 74 6121-6125 (2002) 25 Dou X YM Jung Z-Q Cao and Y Ozaki Applied Spectroscopy 53 1440-1447 (1999) 26 Data supplied by S Christesen and K Ewing 27 Kerker M and B Thompson Eds SPIE MS 10 (1990) 28 Inglesby TV DA Henderson JG Bartlett JAMA 287 2236 (2002) 29 FW Janssen AJ Lund and LE Anderson Science 127 26 (1958) 30 Lide DR Ed Handbook of Chemistry and Physics CRC Press 77th Ed (1996-7) 31 Billmann J G Kovacs and A Otto Surf Sci 92 153 (1980) 32 Murray CA and S Bodoff Phys Rev B 32 671 (1985) 33 Kellogg D and J Pemberton J Phys Chem 91 1120 (1987)
SPIE-2003-5269 16
Chemical agent detection by surface-enhanced Raman spectroscopy
Stuart Farquharson Alan Gift Paul Maksymiuk Frank Inscore and Wayne Smith
Real-Time Analyzers 87 Church Street East Hartford CT 06108
Kevin Morrisey and Steven D Christesen US Army SBCCOM Aberdeen Proving Ground MD 21010
ABSTRACT
In the past decade the Unites States and its allies have been challenged by a different kind of warfare exemplified by the terrorist attacks of September 11 2001 Although suicide bombings are the most often used form of terror military personnel must consider a wide range of attack scenarios Among these is the intentional poisoning of water supplies to obstruct military operations in Afghanistan and Iraq To counter such attacks the military is developing portable analyzers that can identify and quantify potential chemical agents in water supplies at microgram per liter concentrations within 10 minutes To aid this effort we have been investigating the value of a surface-enhanced Raman spectroscopy based portable analyzer In particular we have been developing silver-doped sol-gels to generate SER spectra of chemical agents and their hydrolysis products Here we present SER spectra of several chemical agents measured in a generic tap water Repeat measurements were performed to establish statistical error associated with SERS obtained using the sol-gel coated vials Keywords Chemical agents chemical agent detection SERS Raman spectroscopy
1 INTRODUCTION
In the autumn of 2001 terrorism within US borders became a sobering reality While extensive efforts are being implemented to secure the homeland US and Coalition forces in Afghanistan and Iraq are constantly faced with terrorist attacks In addition to car-bombings the military has identified several non-traditional attack scenarios including poisoning of water supplies by chemical warfare agents (CWAs) To counter this threat the Department of Defense has been investigating numerous approaches to detect such attacks under the auspices of the Joint Service Agent Water Monitor (JSAWM) program1 The JSAWM must be capable of identifying the chemical agents with no false-positives and quantifying the agents at microgram per liter concentrations within 10 minutes2 This includes the analysis of drinking water supplies distribution and storage systems Currently colorimetric paper is used to detect agents on-site while gas chromatography combined with mass spectrometry (GCMS) is used in mobile support laboratories However both methods have severe drawbacks The paper changes color in response to contact with many chemicals besides CWAs causing a high incidence of false positives3 while GCMS although very chemically specific requires up to an hour to perform and regular re-calibration456 Vibrational spectroscopy has also been investigated as a potential method of rapidly detecting CWAs7-
11 as early as 1985 in the case of infrared spectroscopy8 and 1988 in the case of Raman spectroscopy11 Again however these techniques also have limitations when it comes to measuring trace poisons in water Infrared spectra would be dominated by the very strong absorption of water which would obscure absorptions by most other chemicals present Whereas Raman spectroscopy is simply not a very sensitive technique and detection limits are typically grams per liter Surface-enhanced Raman spectroscopy (SERS) offers several advantages over conventional vibrational methods and that may provide the necessary sensitivity required for detecting trace quantities of chemical agents in water SERS employs the interaction of surface plasmon modes of metal particles with target analytes to increase scattering efficiency by as much as 1 million times12 In 1987 the potential of this approach to measure CWAs was demonstrated by measuring a series of pesticides13 Several of these organophosphonates have chemical structures similar to CWAs in particular P=O functional groups In the past few years we and others have further explored the ability of SERS to detect CWAs14-17 and even bioagents 18-21 We have been employing silver-doped sol-gels to promote the SER effect
SPIE-2003-5269 17
in these studies The porous silica network of the sol-gel matrix offers a unique environment for immobilizing and stabilizing SER-active metal particles22-25 The silver-doped sol-gels have been coated on the internal walls of standard glass vials to allow rapid SER analysis of small sample quantities without preparation We have measured over 100 chemicals with enhancements of 104 to 106 demonstrated reversible measurements in a flowing system reproducible measurements from vial-to-vial and measurements in multiple solvents including water21-25 Previously we used these vials to perform preliminary measurements of cyanide (CN) and methylphosphonic acid (MPA) Most of the nerve agents form MPA during hydrolysis while Tabun forms CN a chemical agent in its own right In another paper including in these proceedings we examined the limits of detection (LOD) for MPA by measuring a series of concentrations down to 50 mgL and estimated a limit of detection of 100 microgL26 These measurements provide encouragement in that SERS may satisfy the needs of the JSAWM To further establish the viability of SERS in particular silver-doped sol-gels here we present analysis of cyanide mustard and VX in tap water The measurements performed at the US Armyrsquos Edgewood Chemical Biological Center Aberdeen MD also included numerous repeat measurements to establish reproducibility
2 EXPERIMENTAL 2a General All chemicals including potassium cyanide 2-chloroethylethyl sulfide and those used to prepare the silver-doped and gold-doped sol-gel coated vials were obtained and used as received from Sigma-Aldrich All samples were prepared in a chemical hood using HPLC grade water unless otherwise noted (Fischer Scientific Fair Lawn NJ) for SERS measurements Once prepared the samples were transferred into the silver-doped sol-gel vials (Real-Time Analyzers Inc East Hartford CT) The vials were coated in a manner similar to that previously reported by adding ammonium hydroxide to a solution of silver nitrate tetramethyl orthosilicate (TMOS) and methanol22 The two precursor solutions were prepared mixed and transferred to 2-mL glass vials dried and heated After sol-gel formation the incorporated metal ions were reduced with dilute sodium borohydride (1mgmL) which is followed by a water wash to remove residual reducing agent After the resultant analyte solution was introduced the SER-active vial was fixed horizontally to an XY positioning stage (Conix Research Springfield OR) just inside the focal point of an f07 aspheric lens The lens focused the beam into the sample and collected the scattered radiation back along the same axis A dichroic filter (Omega Optical Brattleborough VT) was used to reflect the excitation laser to the lens and pass the Raman scattered radiation collected by the lens An f2 achromat was used to collimate the laser beam exiting a 200 microm core diameter source fiber optic while a second f2 achromat was used to focus the scattered radiation into a 365 microm fiber optic (Spectran Avon CT) A short pass filter was placed in the excitation beam path to block the silicon Raman scattering generated in the source fiber from reflecting off sampling optics and reaching the detector A long pass filter was placed in the collection beam path to block the sample Rayleigh scattering from reaching the detector A 785 nm diode laser (Process Instruments Inc model 785-600 Salt Lake City UT) was used to deliver 100 to 150 mW of power to the sample A Fourier transform Raman spectrometer (Real-Time Analyzers model IRA-785 East Hartford CT) and a silicon photo-avalanche detector (Perkin Elmer model C30902S Stamford CT) were used to acquire the SER spectra
2b Edgewood Chemical Biological Center The surface-enhanced Raman spectral measurements at the US Armyrsquos Chemical Biological center presented here were all performed on September 12 2003 To expedite measurements a plate was machined to hold up to 12 SER-active sample vials (Figure 1) The plate fit a standard XY plate reader that could be programmed Pure KCN bis-(2-chloroethyl)sulfide (distilled mustard gas HD) and ethyl S-2-diisopropylamino ethyl methylphosphonothioate (VX) were obtained on-site and used to prepare 1 gL tap water solutions in a chemical hood with appropriate safety equipment Simulated tap water was prepared by adding 10 chemicals most often found in tap water at appropriate concentrations to distilled water (Table 1) SER measurements were also performed in a chemical hood For added safety the FT-Raman instrument was placed outside the laboratory and 30 foot fiber optic and electrical cables were used to allow remote SERS measurements and plate manipulation For each experiment 1gL samples were prepared and added to 9 individual vials which were then loaded on the plate In some cases a tenth vial was included as a blank
Table 1 Generic Tap Water Composition
Compound mgL NaHCO3 100 CaSO4 27 MgSO4bull7H2O 67 NaNO3 1 Fulvic Acid 1 K2HPO4 07 KH2PO4 03 (NH4)2HSO4 001 NaCl 001 FeSO4 0001 pH 76-78
SPIE-2003-5269 18
A software program was written that allowed selecting the sequence that the vials were measured the number of positions along the length of the vials to measure (1 to 5) and the number of scans to co-add During sample analysis the program displayed the vial being analyzed the point being analyzed and the spectrum as it was being acquired Once all the data was collected a second software program was written to rapidly analyze the data The spectra collected for all the vials on a plate could be loaded at one time and then the spectra for each point could be displayed simultaneously or separately The user could then select the Raman peak to analyze in terms of peak height or area This was accomplished by selecting points on either side of the peak to define a baseline of zero The peak height or area could then be computed for all of the spectra loaded and then exported to a spreadsheet for statistical analysis
Figure 1 A) Vial Holder 6 slots to hold 2 vials each end-to-end B) Measurement Configuration Program user selects vials to measure sequence number of points per vial (1 to 5) and number of scans per point C) Spectral Acquisition Program shows spectrum being collected which vial and position D) Spectra Analysis Program user selects spectra to analyze by plate vial and point (s) as well as two wavenumbers defining the peak and the baseline to subtract The image is of 5 repeat measurements of 10 mgL KCN in generic tap water 16 sec each 100 mW of 785 nm
3 RESULTS AND DISCUSSION Raman and surface-enhanced Raman spectra were obtained for potassium cyanide bis-(2-chloroethyl)sulfide and ethyl S-2-diisopropylamino ethyl methylphosphonothioate representing three classes of chemical agents cyanides mustards and nerve agents respectively Spectra were also obtained for 2-chloroethyl ethyl sulfide (CEES) a structural analogue to HD which was included in the study to aid in assigning spectral bands KCN salt was used for cyanide experiments to avoid the increased hazards of handling HCN gas KCN completely dissolves in water forming its conjugate acid HCN according to its Ka of 615x10-1027 and at a concentration of 1 mgmL results in a pH 107 solution This is significant in that only CN- appears to interact sufficiently with silver to produce a SER spectrum and no spectral signal is observed below pH 726 Figure 2 shows the SER and normal Raman spectra for KCN The SERS spectrum of 1mgml KCN in water shows a single intense somewhat broad feature at
A B
C D
SPIE-2003-5269 19
2100 cm-1 assigned to the single CequivN stretch The band is much sharper in the normal Raman spectra of the solid KCN salt at 2074 cm-1 This band does broaden and shift to 2080 cm-1 in solution (not shown) However the observed SERS frequency is attributed to interaction with silver and low frequency mode at 135 cm-1 attributed to a Ag-CN stretch (not shown) supports this conclusion
cm-1 band to a C-S stretch but the authors concede that it is in fact more likely a C-Cl stretch28 It appears that the most intense Raman bands at 648 692 and 747 cm-1 shift to 620 660 and 720 cm-1 in the SER spectra and are tentatively assigned as above The width of these bands suggests that they overlap underlying spectral features Additional bands in the Raman spectra occur at 972 1034 1049 1263 1286 1423 1442 2865 2935 and 2960 cm-1 Corresponding bands occur in the SER spectra at 964 1015 1054 1286 1410 1447 2865 and 2935 cm-1 Most of these bands are associated with alkane modes specifically the bands at approximately 1040 cm-1 to a C-C stretch 1290 cm-1 to a CH2 in-phase twist 1440 cm-1 to a CH2 wag 2865 cm-1 to a symmetric CH2 stretch and 2965 cm-1 to an asymmetric CH2 stretch The Raman and SER spectra of sulfur mustard were measured at the Edgewood center (Figure 4) Both spectra are largely similar to CEES The C-Cl and C-S bands in the Raman spectrum of HD now occur at 640 655 700 739 and 760 cm-1 and are more resolved possibly due to the increased molecular symmetry Theoretical calculations indicate that the first three bands are due to C-Cl stretching modes and the latter two to C-S stretching modes28 Only the C-Cl bands maintain significant intensity in the SER spectra occurring at 624 and 643 cm-1 which is attributed to the
Prior to measurements of HD CEES was examined by Raman and SER spectroscopy (Figure 3) CEES also known as half-mustard is essentially identical to HD except one of the chlorine end atoms is replaced by a hydrogen atom Again although not as toxic as HD CEES is a blister agent and dilute aqueous samples were prepared using appropriate safety equipment Both the Raman and SER spectra of CEES are similar and dominated by bands between 600 and 800 cm-1 These are associated with C-Cl and C-S stretching modes which are tentatively assigned to 648 and 747 cm-1 in the Raman spectra respectively The shoulder at 630 cm-1 the overlapped band at 660 cm-1 and the strong band at 692 cm-1 could also be due to these modes or their asymmetric counterparts It is worth noting that theoretical calculations assign the 692
Figure 2 A) SER and B) NR spectra of KCN Conditions A) 1 mgml in tap water 100 mW of 785 nm at sample 1-min acquisition time B) solid 300 mW of 785 nm 5-min All spectra are 8 cm-1 resolution
A
B
Figure 3 A) SER and B) NR spectra of CEES Conditions A) 1 vv (10 mgml) in MeOH 100 mW of 785 nm 1-min acquisition time B) neat 300 mW of 785 nm 5-min
A
B
Cl-CH2-CH2-S-CH2-CH3
A
B
Figure 4 A) SER and B) NR of HD Conditions A) 1mgml in tap water B) pure both 100 mW of 785 nm 1-min
Cl-CH2-CH2-S-CH2-CH2-Cl
SPIE-2003-5269 20
expected strong interaction between chlorine and silver and adds support to the assignment of this band to a C-Cl stretch Weaker overlapping bands occur at 670 692 and 724 cm-1 the latter possibly due to C-S stretching modes Again the alkane modes are apparent in the normal Raman spectra of HD but only a broad feature at 1300 to 1450 cm-1 suggests CH2 contributions in the SER spectrum Although the observed bands in the VX spectrum have not been assigned (Figure 4) a computer generated Raman spectrum29 predicts many of the same features with surprising accuracy and are used here Two intense bands at 460 and 530 cm-1 closely match predicted bands at 463 and 546 cm-1 assigned to a CH3-P=O bend and a PO2CS wag Three highly overlapped bands occur at 694 745 and 771 cm-1 matching predicted bands at 713 730 and 760 cm-1 The first
Table 2 Measured SER peak heights for the CN stretch at 2100 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
Figure 5 A) SERS and B) NR spectra of VX Conditions A) 1 vv (10 mgml) in MeOH B) pure sample both 100 mW of 785 nm 1-min acquisition time
A
B
two have been assigned to a C-S stretch and CH2bend respectively while the latter has been attributed to either a P-C stretch or an O-C-C stretch Although the 745 cm-1 band may alternatively be assigned to a C-S stretch based on the previous measurements of CEES and HD The relatively intense bands at 890 1106 1218 1445 and 1465 cm-1 also match predicted bands at 880 1108 1216 1440 and 1464 cm-1 that are assigned to a C-C stretch CH3 rock N-C3 stretch various C-H3 bends and C-H bends respectively Both the computer generated and the measured spectra contain numerous other less intense bands One is worth mentioning A unique band appears at 370 cm-1 that is predicted at 368 cm-1 and corresponds to an O-P=O bend The surface-enhanced Raman spectrum of VX is also rich with spectral features It has the unique low frequency band at 370 cm-1 as well as a second band at 380 cm-1 that is assigned tothe S-P-O bend predicted in the normal Raman spectrum at 388 cm-1 Based on the measured and predicted normal Raman spectra the following SERS assignments are given 460 cm-1 to the CH3-P=O bend 544 cm-1 to the PO2CS wag 738 cm-1 to a C-S stretch (based on arguments above) 890 cm-1
to a C-C stretch 1101 cm-1 to a CH3 rock and 1456
cm-1 to a C-H bend The ability of SERS to measure chemical agents in water containing real-world chemical interferents was tested by using the generic tap water described in Table 1 The ability to reproduce measurements was accomplished by preparing three separate water stock solutions which were used to prepare three sample solutions each of 1 mgmL KCN in the generic tap water The samples defined as vials 1-9 were then measured at 4 points per vial in a semi-random fashion such that errors associated with stock solution preparation and errors associated with instrument drift could be identified No trends were apparent that signified such systematic errors Each spectrum collected consisted of 20 averaged scans taking 16 seconds at 8 cm-1 resolution The laser power at the sample was measured periodically during the day and it ranged from 102 to 105 mW spacccc
spot Vial 1 Vial 2 Vial 41 13754 13004 128192 13519 12692 12909
1 3 13541 12721 126394 13462 12648 12651
ave 13569 12766 12755Vial 6 Vial 3 Vial 5
1 11586 14068 10757 2 11236 14402 115122 3 11376 14546 11559
4 10894 11763 11214ave 11273 13695 11261
Vial 7 Vial8 Vial 91 11198 15804 11023
2 11243 15716 88753 3 11562 15216 11218
4 11655 15084 11077ave 11415 15455 10548
CN AVG STDEV ERR1 mgmL pk ht 12526 1560 1245
stock solution
SPIE-2003-5269 21
The CN measurements consisting of 36 data points produced an average height of 125 for the 2100 cm-1 peak with a standard deviation of 156 or 125 (Table 2) The HD and VX measurements were performed precisely the same way (Tables 3 and 4) For HD the 624 cm-1 peak was used for analysis and it had an average height of 53 with a standard deviation of 068 or 129 while for VX the 544 cm-1 peak was used for analysis and it had an average height of 1051 with a standard deviation of 308 or 293 The greater error in the VX measurements can be somewhat attributed to Vial 4 which produced lower SER signal intensities But removing this vial from the data set changes the standard deviation to 233 only a modest improvement
spot Vial 1 Vial 2 Vial 41 525 609 4682 45 675 484
1 3 527 644 5094 559 807 669
ave 51525 68375 5325Vial 6 Vial 3 Vial 5
1 521 574 575 2 536 53 4492 3 509 508 451
4 565 594 379ave 53275 5515 4635
Vial 7 Vial 8 Vial 91 457 589 413
2 583 497 4273 3 544 505 4
4 528 5 509ave 528 52275 43725
HD AVG STDEV ERR1 mgmL pk ht 530 068 1291
stock solution
spot Vial 1 Vial 2 Vial 41 1464 1034 4672 1485 99 61
1 3 1491 989 5684 1041 777 553
ave 137025 9475 5495Vial 6 Vial 3 Vial 5
1 1058 942 1293 2 697 121 965
2 3 727 1206 11124 689 1254 746
ave 79275 1153 1029Vial 7 Vial 8 Vial 9
1 1127 783 163 2 1358 812 1614
3 3 1371 908 15124 1097 875 1397
ave 123825 8445 153825
VX AVG STDEV ERR1 mgmL pk ht 1051 308 2925
stock solution
4 CONCLUSIONS In this paper we examined the ability of surface-enhanced Raman spectroscopy to reproducibly measure CN HD and VX in tap water without chemical interference Both normal and surface-enhanced Raman spectra were examined to select unique bands suitable to identify and quantify these chemical agents For SER measurements the 2100 cm-1 C-N stretch was used for CN the 624 cm-1 C-Cl stretch was used for HD and the 544 cm-1 PO2CS wag was used for VX It was determined that 1 mgmL samples of each of these chemicals measured 36 times in glass vials coated with a silver-doped sol-gel reproduced measurements with standard deviations of 125 129 and 293 It was further found that the 10 chemicals added to simulate generic tap water did not interfere with or alter the SER spectra It should be noted that the concentrations used in this study were considerably greater than those required by the JSAWM program Current work involves improving SER sensitivity and designing sampling systems with better reproducibility This includes the development of fractal silver and gold structures within the sol-gel matrix and the development of chemically selective sol-gels
ACKNOWLEDGMENTS The authors are grateful for the support of the US Army (DAAD13-02-C-0015 Joint Service Agent Water Monitor program) The authors would also like to thank Janet Jensen Ronald Crosier and Kristina Gonser for helpful discussions
Table 4 Measured SER peak heights for the VX band at 544 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
Table 3 Measured SER peak heights for the HD band at 624 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
SPIE-2003-5269 22
REFERENCES 1 Jensen JL NC Wong W R Loerop SPIE 4775-03 (2001) 2 JSAWM Requirements at wwwsbccomapgeaarmymilRDAecbcrtPRODSERJSAWMjsawmhtml 3 Erickson B ldquoThe Chemical Weapons Convention Redefines Analytical Challengerdquo Anal Chem 70 397A (1998) 4 Johnston RL Hoefler CM Fargo JC and Moberley B ldquoThe Defense Nuclear Agencyrsquos
ChemicalBiochemical Weapons Agreements Verification Technology Research Development Test and Evaluation Program and its Requirements for On-Site Analysisrdquo AT-ONSITE 5-8 (1994)
5 Black RM Clarke RJ Read RW and Reid MT ldquoApplication of GC-MS and GC-tandem MS to the analysis of chemical warfare samples found to contain residues of the nerve agent sarin sulphur mustard and their degradation productsrdquo J Chrom 662 301-321 (1994)
6 USA Army Center for Health Promotion and Preventive Medicine (USACHPPM) Technical Guideline 230A 7 Hoffland LD Piffath RJ Bouck JB rdquoSpectral signatures of chemical agents and simulantsrdquo Optical
Engineering 24 982-984 (1985) 8 Braue EHJ Pannella MGrdquoCIRCLE CELL FT-IR Analysis of Chemical Warfare Agents in Aqueous Solutionsrdquo
App Spectrosc 44 1513-1520 (1990) 9 Tseng C-H Mann CK and Vickers TJ ldquoDetermination of Organics on Metal Surfaces by Raman
Spectroscopyrdquo App Spectrosc 47 1767-1771 (1993) 10 Staff Reporter ldquoUS Army Tests Lidar to Detect Biological Toxinsrdquo Photonic Spectra pg 50 December 1998 11 Christesen SD Raman cross sections of chemical agents and simulants App Spectrosc 42 318-321 (1988) 12 Weaver MJ Farquharson S Tadayyoni MA ldquoSurface-enhancement factors for Raman scattering at silver electrodesrdquo J Chem Phys 82 4867-4874 (1985) 13 Alak AM and Vo-Dinh T ldquoSERS of Organophosphorous Chemical Agentsrdquo Anal Chem 59 2149-2153 (1987) 14 Lee Y Farquharson S ldquoRapid chemical agent identification by SERSrdquo SPIE 4378 21-26 (2001) 15 Farquharson S Maksymiuk P Ong K Christesen S ldquoChemical agent identification by surface-enhanced Raman spectroscopyrdquo SPIE 4577 166-173 (2001) 16 Spencer KM Sylvia J Clauson S and Janni J ldquoSurface Enhanced Raman as a Water Monitor for Warfare
Agents in Waterrdquo SPIE 4577 158-165 (2001) 17 Tessier P Christesen S Ong K Clemente E Lenhoff A Kaler E Velev O ldquoOn-line spectroscopic characterization of sodium cyanide with nanostructured gold surface-enhanced Raman spectroscopy substratesrdquo
App Spectrosc 56 1524-1530 (2002) 18 Farquharson S W W Smith S Elliott and J F Sperry Rapid biological agent identification by surface- enhanced Raman spectroscopy SPIE 3855110-116 (1999) 19 Farquharson S WW Smith YH Lee S Elliott and J F Sperry Detection of biological signatures A
comparison of electrolytic and metal-doped sol-gel surface-enhanced Raman media SPIE 4575 62-72 (2002) 20 Guzelian AA Sylvia J Janni J Clauson S and Spencer KM ldquoSERS of whole cell bacteria and trace levels of biological moleculesrdquo SPIE 4577 182-192 (2001) 21 Shende C Inscore F Gift A Maksymiuk P Farquharson S ldquoRapid dipicolinic acid extraction from Bacillus spores detected by surface-enhanced Raman spectroscopyrdquo App Spectrosc 58 accepted 22 Lee Y and Farquharson S ldquoSERS Sample Vials Based on Sol-Gel Process for Trace Pesticide Analysisrdquo SPIE
4206 140-146 (2000) 23 Farquharson S and Lee Y ldquoTrace Drug Analysis by SERSrdquo SPIE 4200-16 (2000) 24 Lee Y Farquharson S and Rainey P M Surface-Enhanced Raman Sensor for Trace Chemical Detection in
Water SPIE 3857 76-84 (1999) 25 Lee Y Farquharson S Kwong H and Shahriari M ldquoSol-Gel Chemical Sensor for Surface-Enhanced Raman
Spectroscopyrdquo SPIE 3537 252-260 (1998) 26 Farquharson S A Gift P Maksymiuk F Inscore and W Smith ldquopH dependence of methyl phosphonic acid
dipicolinic acid and cyanide by surface-enhanced Raman spectroscopyrdquo SPIE 5269-19 (2003) 27 Lide DR Ed Handbook of Chemistry and Physics CRC Press 77th Ed (1996-7) 28 Sosa C RJ Bartlett K KuBulat and WB Person ldquoA theoretical study of harmonic vibrational frequencies and
infrared intensities of XCH2CH2SCH2CH2X and XCH2CH2SH (= H Cl)rdquo J Phys Chem 93 577-588 (1993) 29 Hameka HF and JO Jenson ldquoComputer-generated predictions of the structure and the IR and Raman spectra of
VXrdquo ERDEC-TR-065 May 1993
- Chemagents Appendicespdf
-
- SPIE2001-4575-Bioagent-SERSapdf
-
- Detection of bioagent signatures A comparison of electrolytic and metal-doped sol-gel surface-enhanced Raman media
- Stuart Farquharson Wayne Smith and Yuan Lee
-
- Real-Time Analyzers 87 Church Street East Hartford CT 06108
-
- ABSTRACT
- 4 CONCLUSIONS
- 5 ACKNOWLEDGEMENTS
- 6 REFERENCES
-
SPIE-4378-2001
21
Rapid chemical agent identification by surface-enhanced Raman spectroscopy
Yuan-Hsiang Lee and Stuart Farquharson
Real-Time Analyzers 87 Church Street East Hartford CT 06108
ABSTRACT Although the Chemical Weapons Convention prohibits the development production stockpiling and use of chemical warfare agents (CWAs) the use of these agents persists due to their low cost simplicity in manufacturing and ease of deployment These attributes make these weapons especially attractive to low technology countries and terrorists The military and the public at large require portable fast sensitive and accurate analyzers to provide early warning of the use of chemical weapons Traditional laboratory analyzers such as the combination of gas chromatography and mass spectrometry although sensitive and accurate are large and require up to an hour per analysis New chemical specific analyzers such as immunoassays and molecular recognition sensors are portable fast and sensitive but are plagued by false-positives (response to interferents) To overcome these limitations we have been investigating the potential of surface-enhanced Raman spectroscopy (SERS) to identify and quantify chemical warfare agents in either the gas or liquid phase The approach is based on the extreme sensitivity of SERS demonstrated by single molecule detection a new SERS material that we have developed to allow reproducible and reversible measurements and the molecular specific information provided by Raman spectroscopy Here we present SER spectra of chemical agent simulants in both the liquid and gas phase as well as CWA hydrolysis products Keywords Chemical warfare agent simulant hydrolysis product SERS Raman spectroscopy sol-gels vapor
1 INTRODUCTION Chemical warfare has been banned since the 1925 Geneva Protocol yet the use of chemical agents has persisted1 This can be attributed to the simplicity in manufacturing ease of deployment and the relatively low cost of chemical warfare agents (CWAs) These attributes make these weapons especially attractive to low technology countries and terrorists Well known examples include the large-scale use of tabun (GA) during the Iran-Iraq war (1984-1948)2 and the release of sarin (GB) in the Tokyo subway in 1995 The latter is the first documented terrorist use of a chemical weapon34 This ever-present threat was again substantiated by the United Nations Special Commissions report that described Iraqrsquos facilities for nerve agents anthrax and small pox production5-7 These uses of chemical weapons have motivated the development of fast and accurate analytical techniques to warn soldiers and the public at large The development of these analytical techniques is challenging in that these techniques must not only measure extremely low concentrations quickly (microgramliter in lt 1minute) but must also be capable of measuring both gas phase and liquid phase to be effective The latter is required since chemical agents can also be used to poison water supplies89 The earliest technologies developed for CWA detection were based on electrochemical ionization or colorimetric analysis (eg phosgene tape) Although these analyzers were easy to use they were not generally agent specific and suffered from false-positives1 More traditional laboratory methods have also been investigated and in particular combined gas chromatography and mass spectrometry (GCMS) has been very successful at eliminating false-positives1011 However GCMS requires extraction repeated calibration and long analysis times (typically 20 to 60 minutes)11 making it labor intensive and less than desirable for field use More rapid analysis of agents in the solid liquid and gas phase has been demonstrated by vibrational spectroscopy12-15 Hoffland et al12 reported infrared absorbance spectra and absolute Raman cross sections for several chemical agents while Christesen measured Raman cross sections for sarin tabun mustard gas and VX (ethyl S-2-diisopropylamino ethyl methylphosphonothioate)16 Again however these techniques also have limitations Raman spectroscopy is simply not a very sensitive technique and detection limits are typically 01 (1000 ppm) While infrared spectroscopy would have limited value in analyzing poisoned water since the very strong infrared To whom correspondence should be addressed emailfarqureal-time-analyzerscom
SPIE-4378-2001
22
absorption of water would obscure most other chemicals present Nevertheless efforts to overcome these limitations have been demonstrated Braue and Pannella13 quantified the G-series nerve agents (tabun sarin and soman) in terms of infrared attenuated total reflectance using a circle-cell And Alak and Vo-Dinh demonstrated the possibility of surface-enhanced Raman spectroscopy (SERS) to identify CWAs by measuring several organophosphonates that simulate the nerve agents17 However quantitative measurements have not been demonstrated for the SER-active material used (silver coated on alumina particles) or other SER-active media18 Recently we developed silver-doped sol-gels to promote the SER effect19-22 The porous silica network of the sol-gel matrix offers a unique environment for stabilizing SER-active metal particles and the sol-gel provides a high surface area that effectively increases the number of molecules observed within the Raman scattering volume The silver-doped sol-gels have been coated on the internal walls of standard glass vials to allow rapid SER analysis of small sample quantities (lt 01 mL) without preparation We have used p-aminobenzoic acid (PABA) as a test chemical to demonstrate surface enhancements greater than 107 reversible measurements in a flowing system reproducible measurements from vial-to-vial and batch-to-batch and measurements in multiple solvents including water19-22 Here we present preliminary measurements of chemical agent simulants in both the liquid and gas phases as well as chemical agent hydrolysis products using our SER-active vials
2 EXPERIMENTAL The chemical agent simulants employed were obtained at their purest commercially available grade from Aldrich (Milwaukee WI) and were dissolved in water or methanol for analysis All chemicals used to prepare the silver-doped sol-gels were spectroscopic grade and also purchased from Aldrich The sol-gel vials were coated in a manner similar to that previously reported by adding ammonium hydroxide to a solution of silver nitrate tetramethyl orthosilicate and methanol22 After mixing 02 mL of the sol-gel solution was transferred into a glass vial (2 mL) dried and heated The incorporated silver ions were then reduced using dilute sodium borohydride The vials were washed and dried prior to the addition of a sample solution The patent pending SER-active vials are commercially available from Real-Time Analyzers (Simple SERS Sample Vials RTA East Hartford CT) Dimethyl metylphosphonate (DMMP) pinacolyl methylphosphonate (PMP) and methylphosphonic acid (MPA) were prepared in aqueous solution while 2-chloroethyl ethyl sulfide (CEES) was prepared in methanol at 1 mM for SERS measurements Neat samples were employed for normal Raman measurements All samples were prepared in a chemical hood and transferred into plain or SER-active vials for analysis Special precaution was followed for CEES since it is a severe blistering agent23 Once prepared the vial was placed into the sample compartment of a Raman spectrometer for analysis A prototype Fourier transform Raman spectrometer (RTA) was used for these measurements24 The system consisted of a NdYAG laser (Brimrose) for excitation at 1064 nm an interferometer built by On-Line Technologies (OLT East Hartford CT) for frequency separation an uncooled InGaAs detector for signal detection (RTA) and an Intel 400 MHz Pentium II based laptop computer (Dell Round Rock TX) for interferometric control data acquisition (OLT) and analysis (LabVIEW by National Instruments Austin TX) Additional components included a Notch filter (Kaiser Ann Arbor MI) and interferometer entrance and exit optics (Edmund Scientific Barrington NJ) Fiber optics were used to deliver the excitation beam to the sample and the scattered radiation to the interferometer (1 meter lengths of 200 and 365 micron core diameter respectively Spectran Avon CT) A second Notch filter (Kaiser) was used as a beam splitter to direct the excitation beam along the same axis as the collected radiation A microscope object (20x04 Newport Irvine CA) was used to focus the beam into the sample and to collect the scattered radiation back along the same axis In this co-axial backscattering arrangement the excitation beam was passed through the outside of a glass vial and focused onto the silver-doped sol-gel film (01 mm thickness) containing the sample
3 RESULTS AND DISCUSSION As a prelude to chemical agent measurements in water we evaluated the quantitative performance of the SER-active vials by measuring PABA over the concentration range from 10-7 M to 10-2 M Figure 1 shows the spectra for 7 35 and 70 micromolar concentrations while Figure 2 shows a plot of the 1450 cm-1 band intensity as a function of concentration The SER response is linear over nearly three orders of magnitude to just over 10-4M at which point the band intensity suggests that the silver surface is becoming saturated
SPIE-4378-2001
23
In an effort to demonstrate the broad capabilities of the SER-active vials to measure chemical agents spectra of a nerve agent simulant dimethyl methylphosphonate a mustard gas simulant 2-chloroethyl ethyl sulfide and hydrolysis products pinacolyl methylphosphonate and methylphosphonic acid were collected DMMP is widely used by the US Army as a chemical warfare simulant because its chemical structure volatility and water solubility are similar to those of nerve agents25 DMMP is completely miscible and stable in water at room temperature26 Figure 3 compares the SER spectrum to the normal Raman spectrum of DMMP A number of the normal Raman bands are SER-active such as the P-C stretching mode which shifts from 715 to 735 cm-1 and the C-H stretching modes at 2855 2930 2960 and 3000 cm-1 which shift slightly Surprisingly the P=O stretching mode at 1250 cm-1 virtually disappears However the most dramatic change is the appearance of an intense triplet in the SER spectrum near 1000 cm-1 The bands at 1000 cm-1 1030 cm-1 and 1075 cm-1 likely involved the P-O-C bond This is supported by the nearly identical triplets observed for the SER spectra of fonofos and fonofoxon1719 It is also worth noting that a band appears at 425 cm-1 in the SER spectrum that may be unique to DMMP and useful for identification The enhancement factor is estimated at 120000 based on the normal Raman and SER P-C band intensity taking into account the difference in sample concentrations and spectral acquisition conditions A detection limit based on a signal-to-noise ratio of 3 can be estimated at 16 ppm
10-7 10-6 10-5 10-4 10-3 10-2 10-110-2
10-1
100
101
102
Figure 1 SER spectra of A) 70 B) 35 and C) 7 micromolar p-amino benzoic acid in water Conditions 80 mW of 1064 nm laser excitation 100 averaged scans (15 min) at 8 cm-1 resolution
Figure 2 SER spectral intensity for p-aminobenzoic acid as a function of concentration using RTA SER-active vials
Concentration (M)
Figure 3 A) SER and B) normal Raman spectra of dimethyl methylphosphonate Conditions SERS as in Figure 1 normal Raman 500 mW and 200 scans
Figure 4 A) SER and B) normal Raman spectra of 2-chloroethyl ethyl sulfide Conditions as in Figure 3
Wavenumbers (∆cm-1)
CH3O-P-OCH3
=
O
_
CH3
CH3O-P-OCH3
=
O
_
CH3
Cl-CH2-CH2-S-CH2-CH3 A
B
A
B
A
B
C
Wavenumbers (∆cm-1)
Wavenumbers (∆cm-1)
500 1000 1500 2000
SPIE-4378-2001
24
2-Chloroethyl ethyl sulfide a blister agent simulant has a chemical structure similar to the mustard gas (Cl-CH2-CH2-S-CH2-CH2-Cl) with only one terminal chlorine Due to its low solubility in water CEES was dissolved in methanol for the SER measurement Again the prominent Raman modes are SER-active and even maintain relative intensity (Figure 4) The primary difference is that the SER bands appear to broaden such that the triplet near 700 cm-1 becomes a doublet and the shoulders at 2875 and 2970 cm-1 become less defined Again the latter bands are assigned to C-H stretching modes A single band at 700 cm-1 which is attributed to the C-S-C asymmetric stretch dominates the reported infrared spectrum of mustard gas12 A corresponding symmetric stretch is reported at 705 cm-1 in the Raman spectrum of mustard gas27 Here a corresponding symmetric stretch appears but as a doublet at 700 and 755 cm-1 presumably due to the loss in symmetry for CEES The band at 655cm-1 can also be confidently assigned to a C-Cl stretch The SER spectral bands at 620 and 730 cm-1 are probably due to the same modes ie C-Cl and C-S-C stretches respectively The enhancement factor for CEES was somewhat less than DMMP at approximately 62000 as is the estimated detection limit of 22 ppm The ability to rapidly detect trace quantities of chemical agents in the gas phase would be invaluable as an early warning system Although the Raman scattering cross-sections for the nerve agents suggest that remote detection by Raman-based LIDAR is unlikely16 a SER-based system for perimeter monitoring could prove successful As a preliminary measurement we prepared a 10 by volume solution of CEES in methanol exposed a SER-active vial to the equilibrium vapor phase in a sealed jar and monitored the SER spectrum as a function of time Initially the vial was removed through a transfer chamber every hour to record the SER spectrum After ten hours spectra were recorded only every ten hours As illustrated by Figure 5 the sol-gel performed as a dosimeter in that the spectra increased as a function of exposure time The most intense SER bands at 620 and 2930 cm-1 are discernable in the first few hours The spectrum after 40 hours is nearly identical to the solution phase spectrum except for a diminished intensity of the 730 cm-1 band This may be due to methanol solvation effects or surface-orientation effects Based on the relative concentrations of methanol and CEES and their partial pressures we estimate the equilibrium concentration of CEES to between 1 and 2 micromolar Although not shown this concentration could be detected in one hour As previously stated the analysis of chemical agents in water is important in identifying poisoned water It is also important to decommissioning activities in which agents are destroyed by hydrolysis (acid or base) Furthermore any analytical technique used must be capable of distinguishing between parent CWA and hydrolysis products to assess safety or effectiveness of decommissioning For example soman has a hydrolysis half-life of ~23 hours at ambient temperatures and neutral pH28 and forms hydrofluoric acid (somewhat toxic) and pinacolyl methylphosphonate (relatively non-toxic)2930 PMP further hydrolyzes to form methyl phosphonic acid and 33-dimethyl-2-butanol (both non-toxic) The structural similarities between soman PMP and MPA are expected to produce similar Raman as well as SER spectra Figure 6 compares PMP and MPA but not the highly toxic parent CWA soman As with DMMP the P-C stretch the P-O-C mode and C-H stretches are readily apparent Yet it is worth noting that the band positions are reasonably different The former two bands appear at 764 and 1042 cm-1 for MPA while they are at 788 and 1032 cm-1 for PMP More importantly a unique band at 546 cm-1 as yet unassigned appears in the PMP spectrum
Figure 6 SER spectra of A) methyl phosphonic acid and B) pinacolyl methylphosphanate (note unique band at 546 cm-1) Spectral conditions as in Figure 1
Figure 5 SER spectra of 2-chloroethyl ethyl sulfide vapor as a function of time (10 hour increments to top which is 40 hours) Bottom trace is a blank Spectral conditions as in Figure 1
A
B
Wavenumbers (∆cm-1) Wavenumbers (∆cm-1)
HO-P-OH
=
O
CH3
_
HO-P-OH
=
O
=
O
CH3
_
CH3
_HO-P-O-CH-C-CH3
=O
CH3
_
CH3
_CH3_CH3
_
HO-P-O-CH-C-CH3
=O
=O
CH3
_
CH3
_
CH3
_
CH3
_CH3_CH3
_CH3
_
SPIE-4378-2001
25
4 CONCLUSIONS We have successfully measured the SER spectra of chemical agent simulants dimethyl metylphosphonate and 2-chloroethyl ethyl sulfide and chemical agent hydrolysis products pinacolyl methylphosphonate and methylphosphonic acid using silver-doped sol-gel coated sample vials Measurements were obtained in both aqueous and gas phase The P-C stretching mode was SER-active for all four chemicals allowing identification by class Within this group each chemical contained at least one unique spectral band that could be used for identification (Table 1) Furthermore these bands do not appear to coincide with SER spectra reported for organophosphorus pesticides the most likely source of false-positives Although surface enhancement factors appear to be an order of magnitude better than those previously presented in the literature for similar chemicals17 measurement sensitivity needs to be improved by 1 to 2 orders of magnitude to provide adequate warning of chemical agent use Current research efforts to increase surface-enhancement optical collection efficiency and instrument design are being pursued to achieve the required sensitivity
Table 1 Enhancement factors detection limits and unique SER bands fro chemicals studied Agent Simulant Enhancement Detection limit Unique bands (cm-1)
Dimethyl methylphosphonate 120000 90 microM (16 ppm) 425 2-Chloroethyl ethyl sulfide 62000 60 microM (22 ppm) 620
Methylphosphonic acid 110000 3 microM (60 ppb) 764 1042 Pinacolyl methylphosphonate 150000 70 microM (14 ppm) 546 788 1032
5 ACKNOWLEDGEMENTS The authors would like to thank Drs Janet Jensen and Steven Christesen of Aberdeen Proving Ground for encouraging this work They would also like to thank Advanced Fuel Research for making their laboratory facilities available
6 REFERENCES 1 ldquoThe Chemical Weapons Convention ndash A Guided Tour the Organization for the Prohibition of Chemical Weaponsrdquo at
httpwwwopcwnlguidehtm 2 Robinson JP and J Goldblat Chemical Warfare In The Iraq-Iran War Stockholm International Peace Research
Institute Fact Sheet at httpprojectssiprisecbwresearchfactsheet-1984html (1984) 3 ldquoChemistry of GB (Sarin)rdquo at httpwwwmitretekorgmissionenvenechemicalagentssarinhtml 4 Tu Anthony ldquoOverview of Sarin Terrorist Incidents in Japan in 1994 and 1995rdquo 6th CBW Protection Symposium
Stockholm Sweden 10-15 May 1998 5 Staff Reporter ldquoGoing out with a bangrdquo Newsweek June 28 1999 6 See UNSCOM reports in httpwwwunorgdeptsunscom (1999) 7 Treven T Saddamrsquos Secrets Harper Collins (1999) 8 ldquoDecaying Sarin-filled Rockets Spark Fearsrdquo Janersquos Defense Weekly 25(20)3 (1996) 9 ldquoThe Chemical Weapons Convention Redefines Analytical Challengerdquo Analytical Chemistry News amp Features June 1
397A (1998) 10 Johnston RL Hoefler CM Fargo JC and Moberley B ldquoThe Defense Nuclear Agencyrsquos ChemicalBiochemical
Weapons Agreements Verification Technology Research Development Test and Evaluation Program and its Requirements for On-Site Analysisrdquo AT-ONSITE 5-8 (1994)
11 Black RM Clarke RJ Read RW and Reid MT ldquoApplication of gas chromatography-mass spectrometry and gas chromatography-tandem mass spectrometry to the analysis of chemical warfare samples found to contain residues of the nerve agent sarin sulphur mustard and their degradation productsrdquo J Chromatography 662 301-321 (1994)
12 Hoffland LD Piffath RJ Bouck JBrdquoSpectral signatures of chemical agents and simulantsrdquo Optical Engineering 24 982-984 (1985)
13 Braue EHJ Pannella MGrdquoCIRCLE CELL FT-IR Analysis of Chemical Warfare Agents in Aqueous Solutionsrdquo Applied Spectroscopy 44 1513-1520 (1990)
SPIE-4378-2001
26
14 Tseng C-H Mann CK and Vickers TJ ldquoDetermination of Organics on Metal Surfaces by Raman Spectroscopyrdquo
Applied Spectroscopy 47 1767-1771 (1993) 15 Staff Reporter ldquoUS Army Tests Lidar to Detect Biological Toxinsrdquo Photonic Spectra pg 50 December 1998 16 Christesen SD Raman cross sections of chemical agents and simulants Applied Spectroscopy 42 318-321 (1988) 17 Alak AM and Vo-Dinh T ldquoSERS of Organophosphorous Chemical Agentsrdquo Analytical Chemistry 59 2149-2153
(1987) 18 Norrod KL Sudnik LM Rousell D and Rowlen KL ldquoQuantitative Comparison of Five SERS Substrates
Sensitivity and Detection Limitrdquo Applied Spectroscopy 51 994-1001 (1997) 19 Lee Y and Farquharson S ldquoSERS Sample Vials Based on Sol-Gel Process for Trace Pesticide Analysisrdquo SPIE 4206
140-146 (2000) 20 Farquharson S and Lee Y ldquoTrace Drug Analysis by Surface-Enhanced Raman Spectroscopyrdquo SPIE 4200-16 (2000) 21 Lee Y Farquharson S and Rainey P M Surface-Enhanced Raman Sensor for Trace Chemical Detection in Water
SPIE 3857 76-84 (1999) 22 Lee Y Farquharson S Kwong H and Shahriari M ldquoSurface-Enhanced Raman Sensor for Surface-Enhanced Raman
Spectroscopyrdquo SPIE 3537 252-260 (1998) 23 see Material Safety Data Sheets for details 24 Farquharson S Smith W Carangelo R C and Brouillette C ldquoIndustrial Raman Providing Easy Immediate Cost
Effective Chemical Analysis Anywhererdquo SPIE 3859 14-23 (1999) 25 Bennett S Bane J Benford P and Pratt R ldquoEnvironmental Hazards of Chemical Agent Simulantsrdquo Aberdeen
Proving Ground Maryland Chemical Research and Development Center CRDC-TR-84055 (1984) 26 Mabey W and Mill T Critical Review of Hydrolysis of Organic Compounds in Water under Environmental Conditions
Journal of Physics and Chemistry Reference Data 7(2) 383-414 (1978) 27 Christesen S MacIver B Procell L Sorrick D Carabba M and Bello J ldquo Noninstrusive Analysis of Chemical Agent
Identification Sets Using a Portable Fiber-Optic Raman Spectrometerrdquo Applied Spectroscopy 53 850-855 (1999) 28 Meylan WM and Howard PH J Pharm Sci 84 83-92 (1995) 29 Jenkins A Uy O and Murray G ldquoPolymer-Based Lanthanide Luminescent Sensor for Detection of Hydrolysis Product
of the Nerve Agent Soman in Waterrdquo Analytical Chemistry 71 373-378 (1999) 30 Nassar A Lucas S and Hoffland L ldquoDetermination of Chemical Warfare Agent Degradation Products at Low-Part-
per-Billion Levels in Aqueous Samples and Sub-Part-per-Million Levels in Soils Using Capillary Electrophoresisrdquo Analytical Chemistry 71 1285-1292 (1999)
SPIE 2001-4575
62
Detection of bioagent signatures A comparison of electrolytic and metal-doped sol-gel surface-enhanced Raman media
Stuart Farquharson Wayne Smith and Yuan Lee
Real-Time Analyzers 87 Church Street East Hartford CT 06108
Susan Elliott and Jay F Sperry University of Rhode Island 45 Lower College Rd Kingston RI 02881
ABSTRACT Since September 11 2001 the threat of terrorist attacks and biological warfare within US borders has become a sobering reality In an effort to aid military personnel and the public at large we have been investigating the utility of surface-enhanced Raman spectroscopy (SERS) to provide rapid identification of chemical agents directly and biological agents through their chemical signatures This approach is based on the ability of Raman spectroscopy to identify molecular structure through the abundant vibration information provided in spectra and the ability of SERS to detect extremely low concentrations (eg part-per-billion) through the enhancement of Raman scattering by six orders of magnitude or more Towards the goal of developing a portable analyzer we have been studying the ability of two SER media to obtain continuous (ie reversible) and quantitative (ie reproducible) measurements Here we compare measurements of nucleic acid-bases adenosine monophosphate and ribonucleic acid extracted from Escherichia coli Bacillus subtilis and Staphylococcus aureus obtained by electrolytic SERS and metal-doped sol-gel SERS The capabilities of these SER media are summarized in terms of rapid detection of B anthracis and dipicolinic acid Keywords bioagent detection SERS RNA analysis bacterial analysis Raman spectroscopy
1 INTRODUCTION The recent distribution of anthrax through the US postal system and the subsequent infection and death of several postal and national media employees amplifies the need for methods to rapidly detect and identify this and other chemical and biological warfare agents (BWA) The primary methods currently used immunoassays for screening and nucleic acid (NA) sequencing for positive identification of BWAs (bacteria protozoa and viruses) have serious limitations123 Immunoassay methods employ competitive binding of the bioagent (as an antigen) and its labeled (eg fluorescence) conjugate for a limited number of antibodies Although this analysis method is fast and semi-quantitative other chemicals may compete for the antibodies interfere with the enzymatic reaction or interfere with the measurement (eg it fluoresces) resulting in a high number of false positive responses1 Furthermore the antibodies denature due to moisture and heat limiting shelf life and require sterile often refrigerated storage Positive identification of a BWA can be accomplished by sequencing deoxyribonucleic acid or ribonucleic acid (DNA and RNA)23 This requires enumeration of a nucleic acid sequence through polymerase chain reactions (PCR) or multiplication of the microorganism through culture growth to provide sufficient quantities of DNA or RNA for analysis Unfortunately PCR and culture growth require from several hours to several days23 Consequently a wide variety of technologies have been investigated for rapid identification of BWAs The Department of Defense is actively monitoring 200 such technologies4 This includes traditional methods such as gas chromatographic separation coupled with ion mobility spectrometry detection5 to exotic methods based on nature such as monitoring toxin induced color changes in fish scales6 Although all of these techniques have achieved varying degrees of success none are yet capable of detecting and identifying BWAs in 10 minutes or less Towards this goal we have been investigating the ability of SERS to detect sub-nanogram quantities of DNA or RNA (eliminating enumeration) determine relative NA base concentrations and identify BWA taxonomy
To whom correspondences should be addresses e-mailfarqureal-time-analyzerscom wwwreal-time-analyzerscom
SPIE 2001-4575
63
Raman spectroscopy has a rich history of investigating biochemical and biological processes7 Some of the earliest laser-Raman studies demonstrated that the five NA bases adenine (A) cytosine (C) guanine (G) thymine (T in DNA) and uracil (U in RNA) yielded distinct spectra with several bands suitable for identification and quantification8 Furthermore these studies included exceptional spectra of both DNA and RNA for which the NA bases as well as several phosphate bands were easily identified9 However since the Raman effect is very inefficient (very low conversion of incident radiation to inelastically scattered Raman radiation) these samples had to be highly concentrated Fortunately two phenomena exist that can increase the generation of Raman photons by six orders of magnitude or more known as the resonance Raman and surface-enhanced Raman effects1011 Resonance Raman scattering occurs when the laser excitation wavelength is in resonance with an electronic transition of a molecule (within the absorption envelope)10 Excitation at ultraviolet wavelengths has been used to obtain resonance Raman spectra of amino acids and nucleic acids in whole bacteria1213 For example excitation at 242 nm has been used to maximize the nucleic acid spectral band intensities and minimize the amino acids band intensities A peak at 1530 cm-1 was found to be proportional to the amount of the NA base cytosine while a peak at 1485 cm-1 was proportional to the combined amount of the NA bases adenine and guanine This quantitative behavior has been used to define an A+TG+C base-pair ratio and provide a level of bacterial identification as taxonomic markers13 In recent years SERS has also been used to analyze bacterial cell components14 including amino acids15 lipids16 nucleic acids151718 and the adenine derivatives192021 SERS has proven to be one of the most sensitive methods for trace chemical analysis through the detection of single molecules2223 including DNA (dye labeled 17-mer)24 Since its discovery in 197425 the mechanism responsible for the large increase in scattering efficiency has been the subject of considerable research2627 Briefly incident laser photons couple to free conducting electrons within a metal which confined by the particle surface collectively cause the electron cloud to resonate2628 These surface plasmons are known as the physical component of the SER effect These surface plasmons can transfer energy to the molecular vibrational modes of molecules through interactions with the molecular electron orbitals2629 This interaction is known as the chemical component of the SER effect This perturbation of the molecular polarizability generates surface-enhanced Raman photons26 A number of methods have been developed to produce surfaces or solutions containing one of these metals with optimum roughness or diameter to promote SERS30 These methods include preparation of activated electrodes in electrolytic cells 1131 activated silver and gold colloid reagents32 and metal coated substrates333435 Selecting a SER-active medium for chemical and biological agent detection requires consideration of the method of deployment and hence the method of sampling Chemical aerosols or airborne bacteria will require a collection device to concentrate and continuously present the sample to the SERS medium Poisoned water supplies will also require a flow through device for continuous monitoring or a grab-sample device for periodic analysis And contaminated surfaces will require a grab-sample extractive device A SERS-based device used for continuous monitoring (air or water) must be reversible and reproducible if quantitative measurements are desired while a SERS-based device used for periodic sampling (water or surfaces) must be reproducible Both reversible and reproducible measurements have been performed using electrolytic SERS (E-SERS)36 But this requires a three-electrode sample cell and an electrolyte of known concentration to perform the necessary oxidation-reduction cycles (ORCs) to re-activate the electrode surface with new uncontaminated sites from one measurement to the next Colloids are severely limited in that continuous measurements would require a continuous supply of colloids For periodic measurements vials of colloids one per measurement could be used However aggregate size and consequently SER intensity change with sample conditions (especially pH) and quantitative reproducible measurements are unlikely Substrates appear to have the greatest potential and designs range from silver evaporated on titania particles34 to periodic gold pyramids evaporated between polystyrene beads35 Most substrates require concentrating the sample on the surface through drying to obtain the largest signal enhancements in effect making the measurements irreproducible and irreversible However successful measurements using flow systems have been obtained with glass posts but manufacturing costs appear prohibitive In an effort to overcome these limitations we have developed metal-doped sol-gels to provide SERS measurements that are reproducible reversible and quantitative and yet not restricted to specific environments such as electrolytes solvents or evaporated surfaces3738 The porous silica network of the sol-gel offers a unique environment for stabilizing SER active metal particles and the high surface area increases the interaction between the analyte and metal particles The sol-gel can be coated on the end of fiber optics or on the internal walls of a glass flow tube for continuous measurements or standard glass sample vials for periodic measurements Previously we measured 100 mgL methylphosphonic acid (the primary hydrolysis product of nerve agents) in water with an estimated detection limit of 05 mgL (100 parts-per-billion) We have also
SPIE 2001-4575
64
demonstrated reversible and reproducible measurements of p-aminobenzoic acid (PABA) in a flow through system Here we investigate the ability of the sol-gel SERS (SG-SERS) to measure the NA bases adenosine monophosphate and RNA extracted from E coli B subtilis and S aureus The measurements are compared to those obtained by E-SERS
2 EXPERIMENTAL The inorganic chemicals and solvents used to prepare samples were spectroscopic grade and purchased from Aldrich (Milwaukee WI) Fisher (Pittsburgh PA) or Pfaltz amp Bauer (Waterbury CT) The nucleic acid bases and dipicolinic acid were purchased from Sigma (St Louis MO) Normal Raman samples were measured to establish enhancement factors In each case 1cm3 of sample was placed into a 1x1 cm glass cuvette weighed and measured Unpacked densities were typically 6-7 gcm3 For all SER measurements including RNA samples were prepared as ~01mgmL (see Figure captions for exact concentrations) in 01M KCl and buffered to a pH of 92 with Na2B4O7bullH2O Adenine pH dependence measurements used pH buffer solutions at 4 (potassium acid phthalate) 69 (potassium phosphate monobasicsodium phosphate dibasic) 92 (Na2B4O7bullH2O) and 104 (tris-hydroxymethyl amino methane) Escherichia coli Bacillus subtilis and Staphylococcus aureus cultures (250ml per 1000mL Erlenmeyer flask) were grown overnight in a Trypticase soy broth (TSB) medium containing 1 glucose in a shaking water bath at 37 oC The bacteria were harvested by centrifugation for 10 minutes at 8000 rpm in a GSA rotor at 5degC then washed once in 085 saline The gram-positive bacteria were concentrated to 20 ml and passed through a French pressure cell twice at 15000 psi to break open the cells RNA was extracted according to Protocol 44139 to ensure pristine samples for initial measurements Since this method takes approximately 4 hours a streamlined method was developed For vegetative bacteria the specimen was boiled for 30 sec in 1 ml of distilled water to lyse the cells and release the RNA For bacterial spores the specimen was first incubated in 1 ml of saline solution containing 02 mg lysozyme and phosphate-buffered to pH of 624 for 1 hr at 37 oC This solution was then boiled for 2-3 minutes in 4X loading buffer to release the RNA For both specimens RNA STAT-60TM was added to the supernatant which was centrifuged at 12000 g for 5 minutes to precipitate the ~15 water-soluble proteins This procedure allowed extracting RNA for SER analysis in ~ 10 minutes Electrophoresis shows high purity while the existence of chemicals that could interfere with the SER measurements is still under investigation The electrolytic sample cell has been described previously36 Briefly a three electrode design is incorporated into a Plexiglas structure containing a well for the reference electrode (a saturated calomel electrode Cole Parmer Vernon Hills IL) and a 5mL sample well containing the silver working electrode and platinum wire counter electrode (05 mm wire Alfa Ward Hill MA) A channel connecting the two wells contained a 2 mm diameter semi-porous membrane (10-20 micron pore Ace Glass) The silver electrode was made from a 3 mm length of 2 mm diameter silver wire (Alfa) soldered to a copper wire lead encased in a 4 mm diameter Pyrex tube A cap containing the silver electrode platinum wire and nitrogen purge and vent lines fixed the silver electrode surface 1 mm from a 1 mm thick glass plate attached to the bottom of the sample well The potentiostat used to control the three electrodes was built in-house and has been described in detail elsewhere36 A multifuntional analog digital and timing inputoutput interface card (DAQCard-1200 National Instruments) is used to both drive the electrolytic cell as well as read the current generated in the cell A LabVIEW software program is used set the oxidation potential reduction potential SER measurement potential hold times and sweep rates The amount of charge passed was plotted as a cyclic voltammogram For all spectra presented five oxidation-reduction cycles (ORCs) stepping from -03 VSCE to 03 VSCE and back to -3 VSCE at 50 mVsec were used The SG-SER measurements were accomplished by simply placing the identical samples prepared above into Simple SERS Sample VialsTM (RTA) These 2-mL glass vials are internally coated with ~ 01 micron thick silver-doped sol-gel A prototype Fourier transform Raman spectrometer (RTA) was used for these measurements40 The system consisted of a NdYAG laser (Brimrose or Spectra Physics) for excitation at 1064 nm an interferometer built by On-Line Technologies (OLT East Hartford CT) for frequency separation an uncooled InGaAs detector for signal detection (RTA) and an Intel 400 MHz Pentium II based laptop computer (Dell Round Rock TX) for interferometric control data acquisition (OLT) and analysis (LabVIEW by National Instruments Austin TX) Additional components included a Notch filter (Kaiser Ann Arbor MI) and interferometer entrance and exit optics (Edmund Scientific Barrington NJ) Fiber optics were used to deliver the excitation beam to the sample and the scattered radiation to the interferometer (2 meter lengths of 200 and 365 micron core diameter respectively Spectran Avon CT) A second Notch filter (Kaiser) was used as a beam splitter to direct the excitation beam along the same axis as the collected radiation A microscope object (20x magnification 04 numeric aperture Newport Irvine CA) was used to focus the beam into the sample and to collect the scattered radiation back along the same axis In this co-axial backscattering arrangement the excitation beam passed through the glass plate onto the silver
SPIE 2001-4575
65
electrode surface for E-SERS through the vial glass wall and into the silver-doped sol-gel film for SG-SERS or through the glass wall of the cuvette and into the solid sample for normal Raman spectroscopy All E-SERS and normal Raman spectra were obtained with 750 mW of laser power at the sample while all SG-SERS spectra were obtained with 75 mW of laser power at the system Incident powers above 200 mW in some cases degraded the sol-gel
3 RESULTS AND DISCUSSION The generation of surface-enhanced Raman scattering at electrode surfaces has been extensively researched and the optimum sample conditions are well developed2729 Several researches incorporated electrodes into flowing systems and demonstrated that continuous monitoring of chemicals is possible18 These successes suggested investigated the capability of E-SERS to measure the NA bases and RNA The E-SERS measurements also provided a benchmark to compare and evaluate SG-SERS measurements The molecular structure of adenine (as well as the other base pairs) which includes an aromatic nitrogen-containing heterocycle is ideally suited to interact with the surface plasmons and contribute substantially to the chemical component of the SER effect1119 Even with excitation at 1064 nm a 3-minute scan of 18x10-5M adenine yields high signal-to-noise (SN) E-SER spectra and all of the bands are revealed with clarity (Figure 1 Table 1) Spectra of similar quality were obtained by SG-SERS and the principal spectral bands are easily observed The identical 18x10-5M adenine sample was measured in the same 3-minute time frame but with 110th the laser power The lower power appears to reduce the SN The amount of adenine responsible for the SER spectra as well as enhancement factors for the two SER media can be determined The molecules producing the E-SERS spectrum are those on the electrode surface within the illumination area of the laser (The solution concentration only determines the number of molecules available to adsorb to the electrode surface) For the current experiments the laser illuminates an area of 28x10-7m2 or 56 x10-7m2 if we assume the ORCs increase the surface area by a factor of two Furthermore if we assume monolayer coverage on the electrode and each 3x5 angstrom molecule (lying flat) occupies 15x10-19m2 then there are ca 4x1012 molecules contributing to the Raman scattering This is ca twice the number of molecules measured at electrode surfaces using either differential capacitance-potential curve measurements or rapid linear sweep voltammetry (eg 3x1018 moleculesm2 for pyridine and pyrazine)29 Thus the adenine spectrum in Figure 1 is due to 87x10-10g (6x10-12 moles) A detection limit defined as a SN of 3 can also be calculated The SN for a 3-minute scan is 844 for the 735 cm-1 band suggesting a mass detection limit of 3x10-12g (2x10-14 moles) This is consistent with previous estimates for adenine by others of 25 x10-14 moles1530 However sub-monolayer concentrations must be measured to verify this The root-mean-squared (RMS) noise is measured between 4400-4600 cm-1 Since noise is distributed evenly throughout the spectrum when transformed this region does not have any
Wavenumbers (∆cm-1) Figure 1 A) Normal Raman spectrum of pure adenine powder B) E-SERS and C) SG-SERS of 18x10-5M adenine at pH 92 All spectra 8 cm-1 resolution 200 scans (3 min) and 1064 nm excitation A) and B) 750 mW C) 75 mW B) measurement potential of 11VSCE
A
B
C
A
B
C SG-SERS
pH 10
D
725
735
735
pH 4
Wavenumbers (∆cm-1) Figure 2 A) and C) E-SERS and B) and D) SG-SERS of adenine at A) and B) pH 104 and C) and D) pH 40 Note consistent appearance of bands at 1270 and 1375 cm-1 as the pH is changed to 10 for both SER media E-SERS used 750 mW SG-SERS used 75 mW of 1064 nm excitation
SPIE 2001-4575
66
contributions from signals or baseline offsets The measurement error is given as SplusmnRMS and for adenine this equals 234 The number of molecules contributing to the SG-SERS are those on the silver particles that are embedded in the sol-gel The total silver surface area can be determined from the average particle size (40 nm diameter) concentration (073 by weight based on molar conc and measured sol-gel density) and the scattering volume (a cylinder defined by the laser area 28x10-7m2 and sol-gel thickness10-4m) The 61x109 silver particles in this volume have a collective area of 31x10-5m2 However it may reasonably be assumed that at least half of the silver surface is in contact with the sol-gel matrix and unavailable for analyte interaction Then approximately 10x1014 molecules or 22x10-8g of adenine contribute to the SG-SER spectrum The slightly lower SN of 207 suggests a mass detection limit of 32x10-10g Determination of the enhancement factors for the two SER media requires estimating the number of adenine molecules contributing to the normal Raman spectrum Here a cylindrical scattering volume is assumed again based on the laser area (28x10-7m2) and the penetration depth (1x10-3 m)41 The density of the sample was measured at 064 gcm3 indicating that 18x10-4g (13x10-6 moles) of adenine produced the normal Raman signal The enhancement factor EF is defined by the following equation
EF = (ISERSINR)bull(MNRMSERS) bull(PNRPSERS) bull(TNRTSERS)12
where I is the spectral band intensity (here 735 cm-1) M is the sample mass P is the incident laser power and T is the measurement time (or number of scans) For the E-SERS measurement the enhancement factor is 22x105 (01780184) bull (18x10-487x10-10)) while the SG-SERS enhancement factor is 10x105 (0160184) bull(18x10-422x10-8) bull(75075) bull(315)12) The lower enhancement for the SG-SERS may be real or the available surface of the silver embedded in the sol-gel may have been overestimated In addition to enhancing the Raman scattering efficiency to an extent similar to E-SERS the SG-SER medium also yields an identical shift of the adenine ring-breathing mode from 725 cm-1 in the normal Raman to 735 cm-1 Furthermore in the course of optimizing the E-SERS sample conditions it was found that pH influenced the adenine interaction with the silver surface (Figure 2) In particular the relative band intensities of the pyrimidine ring skeletal vibrations at 1270 and 1375 cm-1 and the imidazol ring skeletal vibration at1335 cm-1 change At pH 4 adenine is protonated presumably the imidazol ring since the band at 1335 cm-1 increases in intensity while the pyrimidine bands are virtually absent Conversely at pH 10 the imidazol band decreases in intensity while the pyrimidine bands appear It is worth noting that the ring-breathing mode at 735 cm-1 changes little between pH 4 and 10 suggesting that the skeletal changes are more a function of molecule-plasmon interactions than reorientation of the molecule on the surface Measurements of the identical pH series of adenine samples by SG-SERS yielded virtually identical spectral changes This suggests that the sol-gel does not influence the measurement This is critical to reproducing measurements and performing quantitative analysis Next the remaining NA bases were measured by both E-SERS and SG-SERS and compared Previously we examined the optimum pH and electrode potentials for E-SERS measurements to determine if a common pH could be used that yielded good sensitivity for all the bases and if variations in potential could be used to provide an added degree of selectivity between the bases Primarily it was found that high quality spectra were obtained between pH 7 and 95 and that cytosine and uracil were best enhanced at potentials positive of the potential-of-zero charge (pzc ca 065VSCE for Ag) guanine and thymine near the pzc and adenine negative of the pzc In all cases the ring-breathing modes were the most intense and in general could be used to identify the NA bases (Figure 3 Table 1) Specifically adenine has an intense band at 735 cm-1 cytosine at 797 cm-1 guanine at 653 cm-1 thymine at 784 cm-1 and uracil at 800 cm-1 The adenine cytosine guanine and thymine bands are sufficiently separated that their contributions to DNA should be determinable Although adenine and guanine contributions to RNA should also be determinable cytosine and uracil are highly overlapped and unfortunately share the same potential dependence Alternate unique bands at 1183 cm-1 for cytosine and 1275 cm-1 for uracil might be suitable for calculating contributions The SG-SER spectra of the remaining NA bases faithfully reproduced the E-SER spectra In particular the primary identifying bands occur at virtually the same wavenumbers (see Table 1) However the spectra for both cytosine and thymine contain an intense band at ca 1040 cm-1 Initially this was attributed to the pH buffer but samples prepared without either the buffer or the 01M KCl electrolyte yielded identical spectra containing this band In fact the E-SER and SG-SER spectra of thymine are virtually identical except for this band Also the SG-SERS of guanine contains an intense band at 1551 cm-1 that is not observed in the E-SER spectrum This band may be due to a moderately intense band at 1553cm-1 in the normal Raman spectrum that is SG-SER active It was also found that the SG-SERS of cytosine was considerably better than the E-SERS while uracil showed the opposite relationship It is also worth noting that all of the SG-SERS were obtained with 110th the laser power Most importantly the primary ring-breathing modes in the SG-SER spectra are sufficiently intense and unique to be used in determining contributions to DNA and RNA as outlined above
SPIE 2001-4575
67
Table 1 Comparison of E-SER and SG-SER Spectral Band Positions for the NA Bases and Adenosine Monophosphate
Adenine Cytosine Guanine Thymine Uracil AMP E-SERS SG-SERS E-SERS SG-SERS E-SERS SG-SERS E-SERS SG-SERS E-SERS SG-SERS E-SERS SG-SERS1647 1638 1634 1656 1655 1630 1587 1585 1510 1580 1551 1539 1456 1456 1465 1460 1462 1480 1453 1459 1394 1398 1425 1431 1435 1399 1404 1392 1374 1375 1373 1383 1370 1335 1332 1311 1307 1333 1331 1353 1348 1331 1329 1265 1273 1280 1292 1278 1276 1275 1279 1271 1183 1195 1222 1232 1221 1219 1204 1205 1180 1144 1097 1033 1029 1038 1040 1035 1051 1037 1041 1035 963 963 957 1001 1000 961 944 884 819 817 859 866 735 737 797 799 784 782 800 800 72738 742 630 630 653 664 667 684 603 602 590 611 466 561
Bands unique to SG-SERS The next chemical to be analyzed by both E-SERS and SG-SERS was adenosine monophosphate (AMP) The E-SER spectrum yields bands due to the adenine chemical functionality at 727 961 1233 1279 1331 1381 and 1486 cm-1 In addition phosphate bands are observed at 860 1097 1453 1587 and 1705 cm-1 (Figure 5) Other researchers have noted that the ribose component does not appear to contribute to the spectrum19 The AMP spectrum also changes as a function of potential As the electrode is swept more positive (here from -09 to -03VSCE) the phosphate bands at 860 1097 1453 and 1587 cm-1 increase in intensity compared to the adenine bands while a band at 1705 cm-1 appears The adenine bands at 1233 1381 and 1486 cm-1 virtually disappear These potential dependent spectral changes are consistent with earlier studies that show that phosphate is attracted to silver at potentials positive of the pzc but repelled at potentials negative of the pzc19
Wavenumbers (∆cm-1) Figure 3 E-SERS of A) 21x10-3M cytosine at -03VSCE 1000 scans B) ~10x10-5M guanine at -06VSCE 500 scans C) 23x10-3M thymine at -06VSCE 500 scans and D) 12x10-3M uracil at -093VSCE 500 scans All spectra at pH 92 750 mW 1064 nm at 8 cm-1
Wavenumbers (∆cm-1) Figure 4 SG-SERS of A) 21x10-3M cytosine 200 scans B) ~10x10-5M guanine 200 scans C) 23x10-3M thymine 200 scans and D) 12x10-3M uracil 500 scans All spectra at pH 92 75 mW 1064 nm at 8 cm-1
A
B
C
A
B
C
D D
SPIE 2001-4575
68
The SG-SER spectrum of AMP is considerably different The adenine bands virtually disappear except for the two primary bands which shift to742 and 1329 cm-1 While the phosphate band at 1459 cm-1 has gained considerable intensity In addition two new intense bands appear at 684 and 1539 cm-1 as well as a moderately intense band at 1180 cm-1 The SG-SER spectrum has greater similarity to the E-SER spectrum at -03VSCE and suggests that the silver particles embedded in the sol-gel behave as if at a potential positive of the pzc RNA samples extracted from E coli B subtilis and S aureus were next examined by both E-SERS and SG-SERS E-SER spectra of these samples yielded quality spectra in 10 minutes in which all of the major features can be identified (Figures 6 and 7) This includes guanine at 650 cm-1 adenine at 791 cm-1 cytosine and uracil combining at 790 cm-1 and phosphate at 1100 1335 (in combination with adenine and guanine) 1465 and 1570 cm-1 Surprisingly adenine which demonstrated the greatest surface-enhanced Raman effect does not dominate the ring-breathing mode portion of the spectrum The intensities of the other base-pairs bands are of the same order of magnitude This suggests that when the base-pairs are linked together as in RNA they are enhanced in concert In fact the relative intensities are very similar to a normal Raman spectrum of E coli RNA which shows the combined cytosine and uracil band at ca twice the intensity of the adenine band and ca four times the intensity of the guanine band Unfortunately this means that the independent enhancement factors for the NA bases can not be used to estimate relative concentrations For example the relative 791 and 734 cm-1 bands for B subtilis would indicate that the cytosine andor uracil concentration was at least 20 times the adenine concentration whereas each of the four RNA bases are known to contribute 15-35 Nevertheless it is worth noting that the three RNA samples yield different relative band intensities that were reproduced in numerous measurements Although the relative concentrations of the NA bases for these samples have not been determined these differences can be quantified If it is assumed that the 650 cm-1 band represents 25 guanine the 791 cm-1 band represents 25 adenine and the 790 cm-1 50 cytosine plus uracil in the E coli RNA spectrum then the relative concentrations can be estimated for the other RNA samples To aid this calculation the three spectra were normalized to the phosphate band at 1100 cm-1 which has been shown to correlate to the total phosphate concentration and can be used as an internal standard In addition a simple baseline correction was applied (Figure 7) This yields 15 adenine 30 guanine and 55 cytosine plus uracil for B subtilis RNA and 18 adenine 25 guanine and 57 cytosine plus uracil for S aureus RNA The average SN of these measurements was 26 with an average error of 8 of the value (SplusmnN) It is also worth noting that the three RNA spectra show a marked shift in a band near 825 cm-1 This band is assigned to the symmetric stretch of the O-P-O ester linkage9 The band appears at 815 cm-1 for S aureus shifting to 820 cm-1 for B subtilis and 830 cm-1 for E coli Others have used the normal Raman intensity of the band at 815 cm-1 as a direct indication of the amount of A-class helix present while the intensity of the band at 830 cm-1 has been used as a direct indication of the amount of B-class helix present However the latter is more associated with DNA than RNA
Wavenumbers (∆cm-1) Figure 5 E-SER spectra of 020 mgmL adenosine monophosphate at A) -03 and B) -09VSCE and C) SG-SER spectra Conditions sample in 01M KCl buffered to pH 92 A) and B) 750 mW C) 75 mW of 1064 64 scans (1-min) at 8 cm-1
Wavenumbers (∆cm-1) Figure 6 E-SERS of 01 mgmL RNA from E coli 02 mgmL RNA from B subtilis and 02 mgmL RNA from S aureus Conditions 01M KCl pH 92 -03VSCE 750 mW of 1064 nm 640 scans (10 min) at 8 cm-1
AMP RNA
E coli A
B
C
B subtilis
S aureus
SPIE 2001-4575
69
SG-SER spectra of reasonable quality were also obtained for E coli and B subtilis especially the latter (Figure 8) However the spectra differ substantially from the E-SERS of the same samples Both SG-SER spectra are dominated by adenine at 735 cm-1 and a band at 1030 cm-1 Although unassigned the latter does appear in the RNA E-SER spectra Bands at 1105 and 1565 cm-1 are likely due to phosphate while bands at 1320 and 1470 cm-1 are less confidently assigned to phosphate They are significantly less intense and somewhat shifted from their SG-SERS counterparts (1335 and 1455 cm-1) A band at 670 cm-1 may be due to guanine which was observed at 664 cm-1 for SG-SERS of the pure sample However the SG-SER spectrum of AMP also had an intense 667 cm-1 band A number of other bands occur at 890 1070 1165 1245 1290 1420 1505 cm-1 and remain unassigned The SG-SER spectra are somewhat disappointing in that only adenine and guanine contributions can be positively identified This limits the ability to determine relative NA base concentrations and distinguish bacterial RNA However several of the unassigned bands may be due to the bases (eg 1030 and 1420 cm-1 due to cytosine) Further experiments will be required to clarify this point
Ecoli
B subtilis
S aureas
P G
A C+U
OPO
A B
Wavenumbers (∆cm-1) Wavenumbers (∆cm-1) Figure 7 SER spectra of RNA from A) B subtilis with contributions indicated and B) E coli B subtilis and S aureus with baseline correction and peak positions used to calculate contributions indicated G = guanine A = adenine C+U = cytosine plus uracil P = phosphate (backbone) OPO = phosphate ester linkage (A- vs B-class helix)
Wavenumbers (∆cm-1) Wavenumbers (∆cm-1) Figure 9 E-SER (-03VSCE) and SG-SER spectra of RNA from A) E coli and B) B subtilis Sample conditions as in Figure 6 E-SER spectra at 750 mW SG-SERS at 75 mW
A B E-SERS
SG-SERS
E-SERS
SG-SERS
SPIE 2001-4575
70
A final comparison was made between the two SER methods by measuring dipicolinic acid (DPA) This chemical may be invaluable as a test for spore forming bacteria specifically B anthracis 50 to 90 of B anthracis sporilates During spore formation dipicolinic acid is synthesized and once completed 10-15 of the dry spore weight is composed of the Ca2+ complex located in the spore core42 Heating in water can be used to initiate germination at which point the exosporium breaks and releases the Ca dipicolinate which becomes dipicolinic acid in water The structure of this chemical strongly
4 CONCLUSIONS Towards the goal of developing a method and technology to rapidly detect and identify bioagents we have been investigating surface-enhanced Raman spectroscopy as a tool to measure relative concentrations of nucleic acid bases in RNA and determine bioagent taxonomy Initially we investigated E-SERS since this method has been extensively researched and the optimum sample conditions are well developed However this method requires a three-electrode sample cell and electrolyte solution Incorporation of an electrolytic cell into sample systems used to analyze for BWA as aerosols in water or on surfaces can be designed using flow injection analysis technologies but cross-contamination and plugging of sample lines seems inevitable For this reason we also investigated metal-doped sol-gels as a SER-active medium Previous studies have shown this material to be active in all solvents particularly water capable of continuous measurements in flowing systems and reproducible (quantitative) between coated sample vials Here we compared SG-SER spectra to traditional E-SER spectra of the nucleic acid base pairs adenosine monophosphate and RNA High quality spectra of adenine cytosine guanine thymine and uracil were obtained by both E-SERS and SG-SERS Both methods yielded very similar spectra for the NA bases including a pH dependent study of adenine Enhancement factors and detection limits for adenine were determined as 2x105 and 16x10-11g and 1x105 and 12x10-10g for E-SERS and SG-SERS respectively (normalized to 75 mW and 10-min acquisition time) Fifty percent of the silver particle surface area in the sol-gel matrix was assumed covered by adenine which may have been overestimated yielding a lower EF and higher detection limit It should also be realized that each E-SER spectrum required several attempts to optimize the measurement conditions (pH electrode potential etc) While each SG-SER spectrum involved no sample preparation and often represents the first and only attempt to make the measurement Quality spectra of RNA extracted from Escherichia coli Bacillus subtilis and Staphylococcus aureus were obtained by E-SERS that were easily interpreted Bands due to adenine guanine cytosine plus uracil and phosphate were identified The SER band intensity of the NA bases in the RNA samples were of the same order of magnitude suggesting that their interaction with the silver surface is concerted as is their Raman enhancement Interestingly the relative SER band
Wavenumbers (∆cm-1) Figure 10 A) Raman spectrum of solid dipicolinic acid B) Ax20 C) electrolytic SERS of 6x10-3 M dipicolinic acid in 01 M KCl at a potential of +07VSCE and pH of 4 and D) sol-gel SERS of 6x10-3 M dipicolinic acid Conditions for A and C as in Figure 1 C) 100 mW of 1064 nm 50 scans 8 cm-1
A B (Ax20)
C
D
suggested that it would be SER active However the E-SER spectrum was unstable and varied considerably as a function of potential A consistent spectrum was obtained at +06VSCE (Figure 10) This potential is not recommended for measurement because the surface is actively dissolving in solution The SG-SER spectrum was considerably more stable of higher quality and easily reproduced Bands at 660 825 1010 1390 1430 1570 1590 and 3075 cm-1 were observed Enhancement factors were determined for the two media using the symmetric ring stretching mode at 995 and 1010 cm-1 for the normal Raman and SER spectra respectively E-SERS yielded an EF of 5x103 while SG-SERS yielded an EF of 2x105 for DPA The SN of the latter suggests a detection limit of 20x10-10g (based on adenine coverage 75 mW and 10-min) The differences in SER activity for these two media may be attributed to the combined electrolytic potential of the solution chemical and metal15 Again the E-SERS suggests that the SG-SERS is at a potential positive of the pzc While the instability in the E-SERS may also be associated with surface interactions of two carboxylic acid groups of dipicolinic acid during the ORCs
DPA
SPIE 2001-4575
71
intensities for RNA extracted from E coli are very similar to those measured by normal Raman spectroscopy Although the relative percent that each of the NA bases contributed to each RNA sample was not determined reproducible band intensities allowed noting the following trends The percent adenine decreases while the combined percent cytosine and guanine increase for both B subtilis and S aureus compared to E coli Quality spectra were also obtained for the RNA samples by SG-SERS but only a few bands were readily identified Calculations of NA base concentrations by SG-SERS will require further research In light of recent events we summarize the capabilities of these SERS media in terms of rapid detection of B anthracis and dipicolinic acid However these capabilities must be qualified First and foremost the level to which SERS can distinguish bacteria or viruses has not yet been determined Development of a database of both DNA and RNA base concentrations for BWAs and common bacteria to establish the level of taxonomic identification is ongoing Second rapid collection of aerosol water or surface samples is being addressed by others who report trapping particles on filters from 100 liters of air per minute Third although not presented here we have developed methods to extract RNA or DNA from cells and spores for SER analysis within 10 minutes Finally we assume a detection limit of 3600 spores per 100 liters of air is required although a 50 lethal dosage of anthrax has not been established With these qualifications a mass detection limit for RNA using SERS is estimated as follows A single measurement is performed in ca 20 minutes (140 liters collected in 14 min RNA extracted in 8 min spectral acquisition and analysis in 10 min) The average human breaths 7 liters per minute therefore the analyzer must at the very minimum detect 5000 spores in 140 liters of air One spore is approximately 2x10-18m3 (1x1x2 microm) and if a density of 075 gcm3 is assumed this corresponds to a mass of 15x10-12g Each spore contains 4-12 RNA or 12x10-13g RNA for 8 If we assume 23 of the RNA can be isolated for analysis during lysis then the proposed instrument must be able to detect 4x10-10g RNA from 5000 spores per 70 liters of air within 10 minutes As noted above the mass detection limits for adenine were estimated at 16x10-11g and 12x10-10g for E-SERS and SG-SERS respectively Although these detection limits suggests that RNA from 5000 spores is detectable with the current instrumentation it is highly likely that only a portion of an RNA segment (eg 120-nucleotide 5S rRNA) is in contact with the metal surface and will contribute to the SER effect The SN for the RNA spectra were 110th of the average SN for the four individual RNA bases suggesting a 10 contribution Furthermore effective taxonomy will likely require knowing the NA base concentrations to 1 of the value (eg 25plusmn025) Again the average measurement error for the bases is 12 These values suggest that the E-SERS is within a factor of 4 of the required detection limit whereas the SG-SERS detection limit must be improved by 25 times The same arguments can be applied to the detection of dipicolinic acid If we assume a spore releases 10 by weight DPA during germination then the proposed instrument must be able to detect 75x10-10g DPA from 5000 spores per 70 liters of air within 10 minutes The detection limit for SG-SERS was estimated at 20x10-10g and suggest that the vials are suitable to perform a rapid screen for anthrax A series of concentration dependent measurements are currently being performed to verify this assertion Finally we note that the measurements performed here employed an FT-Raman spectrometer This instrumentation was chosen over dispersive Raman instrumentation for the high wavelength accuracy afforded by the HeNe reference laser (Connes Advantage43) which would allow reliable spectral subtraction matching of observed spectra to stored library spectra and confident use of chemometric approaches Such data analysis is likely to be required to enhance BWA identification However this instrumentation sacrifices sensitivity due to a less efficient detector (InGaAs vs Si) less efficient Raman scattering and less efficient generation of plasmon modes Substantial improvements can be obtained using visible excitation and Si detection and these measurements are underway
5 ACKNOWLEDGEMENTS The authors are grateful to Drs D Cookmeyer and S Tove of the US Army Research Office (Contract Number DAAH04-96-C-0078) for their interest and support of this research The authors would also like top acknowledge Dr R Yin and J Jensen for supporting development of the metal-doped sol-gels (Contract Number DAAD13-01-C-0019) They also thank Dr Wilfred H Nelson for assistance in spectral interpretations
SPIE 2001-4575
72
6 REFERENCES 1 Roberts WL and Rainey PM Clin Chem 39 1872-1877 (1993) 2 Pasechnik VA CC Shone and P Hambleton Bioseperations 3 267-283 (1993) 3 Jackson PJ ME Hugh-Jones DM Adair G Green KK Hill CR Kuske LM Grinberg FA Abramova and P
Keim Proc Natl Acad Sci 95 1224-1229 (1998) 4 Jensen JL NC Wong W R Loerop SPIE 4775-03 (2001) 5 Snyder A Pet al SPIE 3853-15 (1999) 6 Danosky T R and McFadden P N in press (1997) 7 Woodruff WH Farquharson S Science 201 831-833 (1978) 8 Lord RC and Thomas GJJr Spectrochemica Acta 23A 2551-2591 (1967) 9 Thomas GJJr Biochim Biophys Acta 213 417-423 (1970) 10 Placzek G Handbuch der Radiologie 2 EMarx ed Akademische Verlagagescellschatt Liepzig 1934 UCRL
Trans No 526 (1959) 11 Jeanmaire DL and RP Van Duyne J Electroanalytical Chem 84 1-20 (1977) 12 Chada S Manoharan R Moenne-Loccoz P Nelson WH Peticolas WL and Sperry JF Applied Spectroscopy
47 38-43 (1993) 13 Manoharan R Ghiamati E Chada S Nelson WH and Sperry JF Applied Spectroscopy 47 2145-2150 (1993) 14 Todd EA Morris MD Applied Spectroscopy 48 545-548 (1994) 15 Wentrup-Byrne E Sarinas S and Fredericks PM Applied Spectroscopy 47 1192-1197 (1993) 16 Weldon MK VR Zhelyaskov and MD Morris Applied Spectroscopy 52 265-269 (1998) 17 Kneipp K and J Fleming J Mol Structure 145 173-179 (1986) 18 Pothier NJ and Force RK Applied Spectroscopy 46 147-151 (1992) 19 Ervin KM E Koglin JM Sequaris P Valenta and HW Nurnberg J Electroanal Chem 114 179-194 (1980) 20 Kim SK TH Joo SW Suh and MS Kim J Raman Spectrosc 17 381-386 (1986) 21 Pothier NJ and Force RK Analytical Chemistry 62 678-680 (1990) 22 Kneipp K Y Wang RR Dasari and MS Feld Applied Spectroscopy 49 780-784 (1995) 23 Nie S and Emory SR Science 275 1102 (1997) 24 Graham D WE Smith AMT Linacre CH Munro ND Watson and PC White Analytical Chemistry 69 4703-
4707 (1997) 25 Fleischmann M PJ Hendra and AJ McQuillan Chem Phys Lett 26 163-166 (1974) 26 Pettinger B J Chemical Phys 85 7442-7451 (1986) 27 Surface-Enhanced Raman Scattering Section Four Theory SPIE MS 10 M Kerker and B Thompson Eds (1990) 28 Wang D-S and Kerker M SPIE (M Kerker and B Thompson Eds) MS 10 417-429 (1990) 29 Weaver MJ Farquharson S Tadayyoni MA J Chem Phys 82 4867-4874 (1985) 30 Norrod KL Sudnik LM Rousell D and Rowlen KL Applied Spectroscopy 51 994-1001 (1997) 31 Farquharson S Weaver WJ Lay PA Magnuson RH and Taube H J Am Chem Soc 105 3350-3351 (1983) 32 Lee PC and Meisel D ldquoAdsorption and Surface-Enhanced Raman of Dyes on Silver and gold Solsrdquo J Phys Chem
86 3391-3395 (1982) 33 Li Y-S and Wang Y Applied Spectroscopy 46 142-146 (1992) 34 Bello JM DL Stokes and T Vo-Dinh Analytical Chemistry 61 1779-1783 (1989) 35 van Duyne RP J C Hulteen D A Treichel M T Smith M L Duval and T R Jensen J Phys Chem B1033854-
3863 (1999) 36 Farquharson S and W W Smith W H Nelson and J F Sperry SPIE 3533-27 207-214 (1998) 37 Lee Y H W Smith S Farquharson H C Kwon M R Shahriari and P M Rainey SPIE 3537 252-260 (1998) 38 Lee Y-H S Farquharson and P M Rainey SPIE 3857 76-84 (1999) 39 Current Protocols in Molecular Biology Wiley Interscience1003-1006 (1987) 40 Farquharson S Smith W Carangelo R C and Brouillette C SPIE 3859 14-23 (1999) 41 Chase D B and JF Rabolt Fourier Transform Raman Spectroscopy Acad Press Ch1 p 131 (1994) 42 Brock TD MT Madigan JM Martinko and J Parker Biology of Microorganisms 7th Ed Prentice Hall p 76-80
(1994) 43 Connes J Rev Opt Theor Instrum 40 45 (1961)
166
Chemical agent identification by surface-enhanced Raman spectroscopy
Stuart Farquharson and Paul Maksymiuk
Real-Time Analyzers 87 Church Street East Hartford CT 06108
Kate Ong and Steven D Christesen US Army SBCCOM Aberdeen Proving Ground MD 21010
ABSTRACT The recent distribution of anthrax through the US postal system and the subsequent infection and death of several postal and national media employees amplifies the need for methods to rapidly detect identify and quantify this and other chemical and biological warfare agents The US military has also identified water supplies as a likely method of warfare agent deployment and is funding the development of a Joint Service Agent Water Monitor (JSAWM) In an effort to aid military personnel and the public at large we are developing a portable analyzer capable of identifying and quantifying chemical agents rapidly either on-demand or continuously The approach is based on the ability of Raman spectroscopy to identify molecular structure through the abundant vibration information provided in spectra and the ability of surface-enhanced Raman spectroscopy (SERS) to detect extremely low concentrations (eg part-per-billion) through the enhancement of Raman scattering by six orders of magnitude or more A key element to the analyzer design is a new SER active medium that is capable of quantitative reversible measurements The medium consists of silver or gold nanoparticles incorporated into a sol-gel matrix The porous silica network offers a unique environment for stabilizing SER active metals and the high surface area increases the interaction between the analyte and metal particles Here we present the use of new sol-gels that also selectively enhance chemicals based on polarity and charge Base-line measurements of chemical agents and their hydrolysis products are presented and compared to the JSAWM goal of 30 micrograms per liter detection Keywords Chemical warfare agent hydrolysis product SERS Raman spectroscopy sol-gel nanoparticle
1 INTRODUCTION Since September 11 2001 the threat of terrorist attacks and biological warfare within US borders has become a sobering reality The simplicity in manufacturing ease of deployment and the relatively low cost of chemical warfare agents (CWAs) raises public concern that they may also be used by terrorists Indeed terrorists released sarin (GB) in the Tokyo subway in 19951 Countering terrorism and terrorist attacks requires recognizing likely deployment scenarios and having the required technology to rapidly detect the deployment event One method of deployment has been long identified by the US military distribution through water supplies To counter this threat the Department of Defense is funding or monitoring the capabilities of 200 technologies with the goal of developing a Joint Service Agent Water Monitor (JSAWM) that is field portable2 The JSAWM must be capable of identifying the chemical agents with no false-positives and quantifying the agents at microgram per liter concentrations within 10 minutes (Table 1)3 This includes the analysis of drinking water supplies distribution and storage systems as well as potable water supplies The earliest technologies developed for CWA detection were based on electrochemical ionization or colorimetric analysis (eg phosgene tape) Although these analyzers were easy to use they were not generally agent specific and suffered from false-positives4 More traditional laboratory methods have also been investigated and in particular combined gas chromatography and mass spectrometry (GCMS) has been very successful at eliminating false-positives56 However GCMS requires extraction repeated calibration and long analysis times (typically 20 to 60 minutes)6 making it labor intensive and less than desirable for field use More rapid analysis of agents in the solid liquid and gas phase has been To whom correspondence should be addressed emailfarqureal-time-analyzerscom
Vibrational Spectroscopy-based Sensor Systems Steven D Christesen Arthur J Sedlacek III EditorsProceedings of SPIE Vol 4557 (2002) copy 2002 SPIE middot0277-786X02$1500
167
Table 1 Chemical Agent Structures Hydrolysis Half-lives and JSAWM Thresholds Agent Short-Hand Chemical Structure Hydrolysis
Half-Life JSAWM
Thresholds
Sarin (GB) F-[O=P-CH3]-O-CH(CH3)2 213 hours 32 microgL Soman (GD) F-[O=P-CH3]-O-CH(CH3)-(C-(CH3)3) 23 hours 32 microgL Tabun (GA) (CH3)2-N-[O=P-CN]-O-C2H5 41 hours 32 microgL VX C2H5O-[O=P-CH3]-S-(CH2)2-N-(CH(CH2)2)2 821 hours 32 microgL EA2192 HO-[O=P-CH3]-S-(CH2)2-N-(CH(CH2)2)2 gt9 years 32 microgL Mustard (H) ClCH2CH2-S-CH2CH2Cl encapsulates 47 microgL Lewisite (L) ClCH=CH-As-Cl2 rapid 27 microgL HCN HCN rapid 20 mgL BZ C7NH12-O-[C=O]-COH(C6H5)2 23 microgL T-2 Toxin 87 microgL at pH 7 to 75 and 20 to 25 oC
demonstrated by vibrational spectroscopy7-10 Hoffland et al7 reported infrared absorbance spectra and absolute Raman cross sections for several chemical agents while Christesen measured Raman cross sections for sarin tabun mustard gas and VX (ethyl S-2-diisopropylamino ethyl methylphosphonothioate)11 Again however these techniques also have limitations Raman spectroscopy is simply not a very sensitive technique and detection limits are typically 01 (1000 ppm) While infrared spectroscopy would have limited value in analyzing poisoned water since the very strong infrared absorption of water would obscure most other chemicals present Nevertheless efforts to overcome these limitations have been demonstrated Braue and Pannella8 quantified the G-series nerve agents (tabun sarin and soman) in terms of infrared attenuated total reflectance using a circle-cell And Alak and Vo-Dinh demonstrated the possibility of surface-enhanced Raman spectroscopy (SERS) to identify CWAs by measuring several organophosphonates that simulate the nerve agents12 However quantitative measurements have not been demonstrated for the SER-active material used (silver coated on alumina particles) or other SER-active media13 Recently we developed silver-doped sol-gels to promote the SER effect14-17 The porous silica network of the sol-gel matrix offers a unique environment for stabilizing SER-active metal particles and the sol-gel provides a high surface area that effectively increases the number of molecules observed within the Raman scattering volume The silver-doped sol-gels have been coated on the internal walls of standard glass vials to allow rapid SER analysis of small sample quantities (lt 01 mL) without preparation We have used p-aminobenzoic acid (PABA) as a test chemical to demonstrate surface enhancements greater than 106 reversible measurements in a flowing system reproducible measurements from vial-to-vial and batch-to-batch and measurements in multiple solvents including water14-17 Recently we used these vials to measure Tabun (GB) and Sarin and several hydrolysis products pinacolyl methyl phosphonate (PMP from Soman) and methyl phosphonic acid (MPA from all G-agents Figure 1) Although a number of unique vibrational bands are observed (eg C-N stretch doublet and P-C stretch) the G-agents were only observed for 5 concentrations and all spectra required baseline corrections Figure 1 Surface enhanced Raman spectra of ~5 vv A) Tabun and B) Sarin C) 1 vv PMP and D) 10 ppm MPA using sol-gel sample vials 785 nm excitation 1-min scan and CCD detection Performed at Aberdeen Proving Ground
C A
Wavenumber (∆cm-1) Wavenumber (∆cm-1)
D B
790
545 C-N
2135 2190 P-C 770
760 1290
Proc SPIE Vol 4577
168
Nevertheless MPA was readily observed for a 10 ppm sample with an estimated detection limit of 04 ppm (based on a signal-to-noise ratio of 3 for the 760 cm-1 band intensity) This measurement provides encouragement in that SERS may satisfy the needs of a JSAWM Furthermore MPA is also a hydrolysis product of VX and V-gas and EA2192 (Figure 2) and may prove a valuable indicator of agent usage Figure 2 Hydrolysis of Sarin to form hydrofluoric acid (HF) methylphosphonic acid 1-methylethyl ester (MPAMME) methyl phosphonic acid (MPA) and 2-propanol With this initial albeit modest success we began analyzing chemicals with various sol-gel compositions that we have been developing Here we describe four sol-gel compositions that select for 1) polar-positive 2) polar-negative 3) weakly polar-positive and 4) weakly polar-negative chemical species The ability of these sol-gels to select and enhance Raman scattering is described for several test chemicals and MPA
2 EXPERIMENTAL The chemicals analyzed as well as all chemicals used to prepare the metal-doped sol-gels were obtained at their purest commercially available grade from Aldrich (Milwaukee WI) The sol-gel designed to select for polar-negative species was prepared from a silver amine complex tetramethyl orthosilicate (TMOS) and methanol After mixing 02 mL of the sol-gel solution was transferred into a glass vial (2 mL) dried and heated The incorporated silver ions were then reduced using dilute sodium borohydride The vials were washed and dried prior to the addition of a sample solution In a similar manner the sol-gel designed to select for polar-positive species was prepared from a gold salt TMOS and methanol The sol-gel designed to select for weakly polar-negative species was prepared from a silver amine complex tetraethyl orthosilicate (TEOS) and methanol And the last sol-gel designed to select for weakly polar-positive species was prepared from a gold salt TEOS and methanol All samples were prepared in a chemical hood and transferred into plain or SER-active vials for analysis Normal Raman spectral measurements employed 1-mL pure samples that were placed in a 1-cm3 cuvette and weighed This yielded a powder density that allowed accurate calculation of molecules in the optical collection field SERS measurements employed 1-mg sample per mL water concentrations unless otherwise stated Once prepared a 01 mL sample was placed into one of the four selective sample vials which in turn was placed into the sample compartment of a Raman spectrometer for analysis A prototype Fourier transform Raman spectrometer (RTA) was used for these measurements18 The system consisted of a NdYAG laser (Brimrose) for excitation at 1064 nm an interferometer built by On-Line Technologies (OLT East Hartford CT) for frequency separation an uncooled InGaAs detector for signal detection (RTA) and an Intel 400 MHz Pentium II based laptop computer (Dell Round Rock TX) for interferometric control data acquisition (OLT) and analysis (LabVIEW by National Instruments Austin TX) Additional components included a Notch filter (Kaiser Ann Arbor MI) and interferometer entrance and exit optics (Edmund Scientific Barrington NJ) Fiber optics were used to deliver the excitation beam to the sample and the scattered radiation to the interferometer (1 meter lengths of 200 and 365 micron core diameter respectively Spectran Avon CT) A second Notch filter (Kaiser) was used as a beam splitter to direct the excitation beam along the same axis as the collected radiation A microscope object (20x04 Newport Irvine CA) was used to focus the beam into the sample and to collect the scattered radiation back along the same axis In this co-axial backscattering arrangement the excitation beam was passed through the outside of a glass vial and focused onto the silver-doped sol-gel film (01-03 mm thickness) containing the sample
3 RESULTS AND DISCUSSION p-aminobenzoic acid (PABA) and phenyl acetylene (PA) and were used to refine the selectivity and SER-activity of the four different metal-doped sol-gels PABA is a popular chemical used to evaluate the performance of SER-active media Here the polar end groups can be used to test selectivity of the polar-negative and polar-positive sol-gels PA is essentially non-
2O+ H HF + +OH
OH3H C
OP
OF
CH
CH3
3
3H C
OCP
OOH
CH
CH3
3
3H C
O
CPHO
CH
CH
3
3
C
Sarin MPAMME MPA 2-propanol
Proc SPIE Vol 4577
169
polar but a high electron density in the cylindrical π cloud around the carbon-carbon triple bond allows testing the selectivity of the weakly polar-negative and weakly polar-positive sol-gels As Figure 3 illustrates PABA passes through the polar sol-gel and is enhanced by either the silver or gold particles At 1 mgml the concentration of neutral PABA is ca 20 times that of the ionized form (pKa = 48) For electropositive silver the PABA anion is expected to interact through the carboxylate group and the associated vibrational modes are expected to dominate the spectrum Conversely for electronegative gold either form of PABA is expected to interact through the amine group The clear differences in our spectra support this expectation Furthermore bands at 840 and 1405 cm-1 assigned to a COO- bend and stretch respectively are significantly more intense for silver than gold Additional bands at 1140 and 1195 cm-1 are assigned to CH bending modes while bands at 1450 1500 and 1605 cm-1 are assigned to ring vibrational modes A very similar SER spectrum for PABA on a silver-coated alumina substrate has previously been reported with similar assignments19 For the gold-doped sol-gel new bands appear at 690 1355 and 1585 cm-1 The first band is assigned to a ring-H bending mode the second band to a ring-N- stretching mode and the third band to a possible NH2 scissors mode or ring mode The second band is not observed in the normal Raman spectrum but infrared bands occur at this frequency for aromatic ring-secondary amine stretching modes The scissors mode occurs at this frequency in Raman spectra for several chemicals but is absent in the PABA Raman spectrum Alternatively this mode may be the1600 cm-1 ring mode that has been shifted by the gold interaction Again a very similar SER spectrum of PABA has been reported but surprisingly using silver (colloids)2021 not gold as the enhancement medium These researchers also assumed the primary interaction of PABA with silver was through the carboxylate anion and made assignments accordingly For example they assigned the 1359 cm-1 to a COO- stretch not to the amine group as we have They also favor the ring stretching mode assignment for the 1582 cm-1 band Finally it should be said that other researchers have argued that the most dominant band in the SER spectra at 1450 cm-1 a ring vibration mode suggests that PABA lies flat on the surface and the π-orbitals dominate the surface interaction22
Figure 3 SER spectra of A) PABA using polar-negative and B) polar-positive sol-gels and C) PA using weakly polar-negative and D) weakly polar-positive sol-gels PABA is 1 mgmL PA is 1 vv Spectral conditions 75 mw 1064 nm 100 scans (15 min) 8 cm-1 resolution Non-polar PA passes through the non-polar sol-gels and is also enhanced by both metals The spectra are easily understood For electropositive silver PA interacts through the cylindrical triple bond π electron cloud and a -CequivC- doublet occurs near 2000 cm-1 The interaction is reasonably strong since this band appears at 2112 cm-1 in the normal Raman spectrum For electronegative gold this interaction is unlikely and only very weak bands occur near 2000 cm-1 The remaining bands are at 1000 cm-1 1200 cm-1 doublet and 1595 cm-1 all appear in the normal Raman spectra at virtually the same frequencies and are assigned to the symmetric ring-breathing mode CH bending modes and the trigonal ring-breathing mode respectively The polarnon-polar selectivity of the polar-negative and weakly polar-negative sol-gels was tested by adding a 11 molar mixture of PABA and PA The selective enhancement is quite good (Figure 4) The spectrum obtained using the polar sol-gel represents 78 PABA and 22 PA while the spectrum obtained using the weakly polar sol-gel represents 9 PABA and 91 PA The band peak intensities at 2000 cm-1 for PA and 1450 cm-1 for PABA were used for these calculations and are expanded in Figure 4 for clarity
C A
Wavenumber (∆cm-1) Wavenumber (∆cm-1)
2NH COOH
D B
C CH
Proc SPIE Vol 4577
170
Figure 4 SERS of 11 MM of PABA and PA in A) polar-negative and C) weakly polar-negative sol-gels The lower traces compare the pure chemicals B) 1 mgml PABA in polar-negative sol-gel and D) 1 PA in weakly polar-negative sol-gel while the insets magnify the minority species for clarity (x5 in A and x10 in B) Spectral conditions as in Figure 3 Following this development of selective sol-gels that maintained SER activity we measured cyanide and MPA (Figure 5) Not surprisingly the best sensitivity for both hydrolysis products was obtained using the polar-negative sol-gel The interaction of the cyanide anion with the silver surface is sufficient to shift the CequivN stretch observed at 2080 cm-1 in the normal Raman spectrum to 2145 cm-1 in the surface-enhanced Raman spectrum Furthermore the band is substantially broadened This anion has been extensively studied by electrolytic SERS and this shift and broadening have been attributed to the formation of a tetrahedral Ag(CN)3
2- surface structure23 Figure 5 Surface-enhanced (upper traces) and normal Raman spectra (bottom traces) of A) CN- and B) MPA in silver-doped TMOS SERS conditions as in Figure 3 and 1 mgmL Note MPA yields two distinct spectra for neutral (top) and acidic pH (middle) The normal Raman spectra employed pure powders 500 scans and 900 mW of 1064 nm SER measurements of MPA with the polar-negative sol-gel yielded two unique spectral signatures that depended on solution pH (Figure 5) For more neutral solutions the P-C stretch of MPA at 762 cm-1 dominates and the CH2 stretch at 2922 cm-1 appears The SN is sufficiently high that the anti-Stokes Raman shift at -762 cm-1 is observed For deprotonated MPA an oxygen-surface mode appears at 325 cm-1 (as well as its anti-Stokes complement) suggesting a strong interaction This results in substantial enhancement of the P-O-C mode at 1051 cm-1 (upper trace) Others report that this mode dominates the infrared spectra of nerve agents measured in water8 Comparison of the two spectra suggests the following molecule-to-surface orientations The appearance of the oxygen-surface and P-O-C modes in the upper spectrum of Figure 5B indicates that the tetrahedral molecule interacts with the silver surface through the deprotonated oxygen and is oriented end-on The
A B
Wavenumber (∆cm-1)
C A
Wavenumber (∆cm-1) Wavenumber (∆cm-1)
D B
Wavenumber (∆cm-1)
Proc SPIE Vol 4577
171
dominance of the P-C and the CH2 stretches and the disappearance of the P-O-C mode in the upper spectrum suggest the molecule is oriented with the methyl-hydroxyl-hydroxyl face towards the surface However considerably more research must be performed to verify these points Both chemical forms were measured as a function of concentration to determine the onset of monolayer coverage and estimate expected detection limits (Figure 6) Below monolayer coverage the signal to concentration dependence should be linear and the SN of any spectral measurement in this range can be used to predict the detection limit In the spectra presented here the peak height was used as the signal while the noise as root-mean-squared (RMS) was measured between 4400-4600 cm-1 Since noise is distributed evenly throughout the spectrum when transformed this region was used since it does not have any contributions from signals or baseline offsets Figure 6 shows a series of spectra for MPA along with a plot of the 1050 cm-1 band intensity (the noise was identical for this series) as a function of sample concentration A clear discontinuity appears in the vicinity of 01 mgmL (19 ppm) indicating the onset of monolayer coverage A detection limit defined as a SN of 3 was calculated for the 01 and 005gmL samples at 24x10-4 and 25x10-4 gL respectively A more modest detection limit of 101x10-4 gL was obtained using the 760 cm-1 band in the second series of concentration measurements These detection limits correspond to 46 and 190 ppb for a 3-min scan and 75 mW of incident laser power
Figure 6 A) Concentration dependence of MPA SERS measured in silver-doped TMOS) B) Concentrations are 001 005 01 05 1 gL (188 94 188 94 188 ppm) I760 series (bull) and I1050 series (∆)
Enhancement factors for cyanide and methyl phosphonic acid can be determined by estimating the number of molecules contributing to the surface-enhanced and normal Raman spectra The enhancement factor EF can be defined by the following equation
EF = (ISERSINR)bull(MNRMSERS) bull(PNRPSERS) bull(TNRTSERS)12
where I is the spectral band intensity M is the sample mass P is the incident laser power and T is the measurement time (or number of scans) for the two measurements For the normal Raman spectra a cylindrical scattering volume is assumed based on the laser area (28x10-7m2 6x10-4m diameter spot) and the penetration depth (1x10-3 m)24 The density of KCN and MPA as powders were measured at 0572 and 0516 gcm3 indicating that 16x10-4 and 144x10-4 g produced the normal Raman signals in Figure 5 respectively The number of molecules contributing to the SER spectra are those on the silver particles embedded in the sol-gel The total silver surface area can be determined from the average particle size concentration and the scattering volume Previous scanning electron micrographs showed the average silver particle size to be 40 nm in diameter (335x10-23m3)17 The silver concentration is 012M based on the reactant molar concentrations and dilution factors And the scattering volume is 76x10-
11m3 again based on a cylindrical scattering volume defined by a laser area of 28x10-7m2 and a sol-gel thickness of 27x10-
4m This volume contains 123x10-6g of silver equivalent to 35x109 silver particles with a collective surface area of 18x10-
5m2 However it may reasonably be assumed that at least half of the silver surface is in contact with the sol-gel matrix and unavailable for analyte interaction If we assume monolayer coverage and that each CN molecule occupies 15x10-20m2 then approximately 62x1014 molecules or 27x10-8g of CN contribute to the SER spectrum (20x10-19m2 46x1013 molecules 74x10-9g for MPA) Accordingly the EF for cyanide equals 48x104 ((180599) bull(16x10-427x10-8) bull(90075) bull(500100)12) The EF for MPA is considerably higher at 87x106 ((60326) bull(144x10-474x10-9) bull(90075) bull(500200)12)
0
20
40
60
80
100
120
140
0 02 04 06 08 1 12[MPA] (mgmL)
I (76
0)
0
100
200
300
400
500
600
I (10
50)
Wavenumber (∆cm-1)
A B
Proc SPIE Vol 4577
172
4 CONCLUSIONS Here we present for the first time surface-enhanced Raman spectra of Tabun and Sarin obtained using silver-doped sol-gels However the inferior enhancement suggested employing SER-active sol-gels varying in composition to improve both detection limits and selectivity of the target analytes To this end we successfully demonstrated the capabilities of four sol-gels that select for 1) polar-positive 2) polar-negative 3) weakly polar-positive and 4) weakly polar-negative chemical species p-aminobenzoic acid was used to show that silver could be used to attract polar-negative chemicals or functional groups (carboxylate anion) and that gold could be used to attract polar-positive chemicals (amine) while a mixture of p-aminobenzoic acid and phenylacetylene was used to show that tetramethyl orthosilicate preferentially solvates polar chemicals and that tetraethyl orthosilicate preferentially solvates weakly polar chemicals This increased sample control was applied to cyanide and methyl phosphonic acid two hydrolysis products of chemical warfare agents Exceptional results were obtained for methyl phosphonic acid allowing measurement of 1x10-2 gL for a 3-min scan and 75 mW of incident laser power with an estimated detection limit of 245x10-4 gL and an enhancement factor of 87x106 However this detection limit is 76 times less sensitive than required for the JSAWM (32x10-6gL for the G-agents) Finally we note that the measurements performed here employed an FT-Raman spectrometer This instrumentation was chosen over dispersive Raman instrumentation for the high wavelength accuracy afforded by the HeNe reference laser (Connes Advantage25) which would allow reliable spectral subtraction matching of observed spectra to stored library spectra and confident use of chemometric approaches Such data analysis is likely to be required to identify the chemical agents as well as distinguish hydrolysis products However this instrumentation which employs 1064 nm excitation and InGaAs detection sacrifices sensitivity We believe that the measurement sensitivity can be improved by at least two orders of magnitude by using shorter laser excitation wavelengths (eg 532 nm) This would provide more efficient Raman scattering (fourth power dependence on laser excitation wavelength) more efficient generation of plasmon modes and allow using more efficient detector material (Si vs InGaAs) These modifications are underway
5 ACKNOWLEDGEMENTS The authors would like to thank Dr R Yin and J Jensen of the US Army for supporting this work (Contract Number DAAD13-01-C-0019) They would also like to thank Advanced Fuel Research for making their laboratory facilities available
6 REFERENCES 1 Tu Anthony ldquoOverview of Sarin Terrorist Incidents in Japan in 1994 and 1995rdquo 6th CBW Protection Symposium
Stockholm Sweden 10-15 May 1998 2 Jensen JL NC Wong W R Loerop SPIE 4775-03 (2001) 3 JSAWM Requirements at wwwsbccomapgeaarmymilRDAecbcrtPRODSERJSAWMjsawmhtml 4 ldquoThe Chemical Weapons Convention Redefines Analytical Challengerdquo Analytical Chemistry News amp Features June 1
397A (1998) 5 Johnston RL Hoefler CM Fargo JC and Moberley B ldquoThe Defense Nuclear Agencyrsquos ChemicalBiochemical
Weapons Agreements Verification Technology Research Development Test and Evaluation Program and its Requirements for On-Site Analysisrdquo AT-ONSITE 5-8 (1994)
6 Black RM Clarke RJ Read RW and Reid MT ldquoApplication of gas chromatography-mass spectrometry and gas chromatography-tandem mass spectrometry to the analysis of chemical warfare samples found to contain residues of the nerve agent sarin sulphur mustard and their degradation productsrdquo J Chromatography 662 301-321 (1994)
7 Hoffland LD Piffath RJ Bouck JBrdquoSpectral signatures of chemical agents and simulantsrdquo Optical Engineering 24 982-984 (1985)
8 Braue EHJ Pannella MGrdquoCIRCLE CELL FT-IR Analysis of Chemical Warfare Agents in Aqueous Solutionsrdquo Applied Spectroscopy 44 1513-1520 (1990)
9 Tseng C-H Mann CK and Vickers TJ ldquoDetermination of Organics on Metal Surfaces by Raman Spectroscopyrdquo Applied Spectroscopy 47 1767-1771 (1993)
10 Staff Reporter ldquoUS Army Tests Lidar to Detect Biological Toxinsrdquo Photonic Spectra pg 50 December 1998
Proc SPIE Vol 4577
Proc SPIE Vol 4577
173
11 Christesen SD Raman cross sections of chemical agents and simulants Applied Spectroscopy 42 318-321 (1988) 12 Alak AM and Vo-Dinh T ldquoSERS of Organophosphorous Chemical Agentsrdquo Analytical Chemistry 59 2149-2153
(1987) 13 Norrod KL Sudnik LM Rousell D and Rowlen KL ldquoQuantitative Comparison of Five SERS Substrates
Sensitivity and Detection Limitrdquo Applied Spectroscopy 51 994-1001 (1997) 14 Lee Y and Farquharson S ldquoSERS Sample Vials Based on Sol-Gel Process for Trace Pesticide Analysisrdquo SPIE 4206
140-146 (2000) 15 Farquharson S and Lee Y ldquoTrace Drug Analysis by Surface-Enhanced Raman Spectroscopyrdquo SPIE 4200-16 (2000) 16 Lee Y Farquharson S and Rainey P M Surface-Enhanced Raman Sensor for Trace Chemical Detection in Water
SPIE 3857 76-84 (1999) 17 Lee Y Farquharson S Kwong H and Shahriari M ldquoSurface-Enhanced Raman Sensor for Surface-Enhanced Raman
Spectroscopyrdquo SPIE 3537 252-260 (1998) 18 Farquharson S Smith W Carangelo R C and Brouillette C ldquoIndustrial Raman Providing Easy Immediate Cost
Effective Chemical Analysis Anywhererdquo SPIE 3859 14-23 (1999) 19 Narayanan VA JM Bello JD Stokes and T Vo-Dinh Analusis 19 307-310 (1991) 20 Laserna JJ E L Torres and JD Winefordner Analytica Chemica Acta 469-480 (1987) 21 Torres EL and JD Winefordner Analytical Chemistry 59 1626-1632 (1987) 22 Suh JS DP DiLella M Moskovits J Phys Chem 87 1540-1544 (1983) 23 Benner RE R Dornhaus R Chang and BL Laube Correlations in the Raman spectra of cyanide complexes adsorbed
at silver electrodes with voltammograms Surface Science 101 341 (1980) 24 Chase D B and JF Rabolt Fourier Transform Raman Spectroscopy Acad Press Ch1 p 131 (1994) 25 Connes J Rev Opt Theor Instrum 40 45 (1961)
APPLIED SPECTROSCOPY 351
focusing the 488 nm laser beam 2 mm deep into thebulk of the crystals to avoid contributions from potentialdepletion layers As shown in Fig 1b the dependence ofthe Raman shift on the degree of deuteration is almostperfectly linear and ts very well with D 5 22684cmR1 24526 where D is the degree of deuteration (in )and R is the spectral mean of the PO4 vibration in cm21A linear correlation coef cient of 0998 indicates an ex-cellent linear dependence of the Raman peak shift withdegree of deuteration This result shows that the shift ofthe PO 4 peak is simply caused by the linear increase inatomic mass due to isotope substitution which decreasesthe length of hydrogen-like bonds
This excellent linear dependence allows us to map thepro le of the DH exchange layer at the surface of DKDPcrystals by acquiring Raman spectra and determining theposition of the PO 4 peak for various depths This methodis preferable over other methods such as determining thestrength of the OD vibration directly (eg at 715 cm21)because the position of the most intense peak in the Ra-man spectrum can be measured more precisely than theintensity of some of the weakest peaks in the spectrumThis is demonstrated in Fig 2 where depth-dependentRaman spectra (Fig 2a) and the resulting exchange layerpro les for two DKDP crystals are shown (Fig 2b) Thespectra in Fig 2a were obtained from a depth scan of aDKDP crystal with 75 degree of deuteration in thebulk grown at 45 8C The spectra start out as DKDP with30 deuteration close to the surface and approach thebulk DKDP spectrum within a few micrometers of depthThe fact that the relative degree of deuteration does notextend to 0 D is due to the limited depth resolution ofthe Raman microprobe which averages over 4 mm indepth Figure 2b depicts the resulting DH exchange layerpro les for this and a second crystal grown at 63 8Crespectively Both crystals had the same exposure to am-bient conditions and their main difference is the temper-ature at which they were grown The different exchangelayer pro les indicate that crystals grown at differenttemperatures have differen t proton conductiv ities 13
which leads to a difference in their rate of deuteriumdepletion The parameters controlling this behavior arecurrently the objective of a detailed study the results ofwhich will be reported elsewhere
CONCLUSION
In conclusion we have shown that the shift of the to-tally symmetric PO 4 stretch mode in the Raman spectrumof DKDP crystals scales linearly with degree of deuter-ation This allows us to correlate Raman peak positionsto deuteration levels in these crystals We have presenteda new technique to determine DH diffusion pro les inDKDP frequency conversion crystals based on micro-Ra-man spectroscopy This technique is fast inexpensiveand works under various environmental conditionswhich will allow us to better understand and control deu-terium depletion in DKDP crystals
ACKNOWLEDGMENTS
We would like to thank M Runkel for rst discovering DKDP crack-ing R Floyd for providing DKDP crystals and L Chase and A Burn-ham for their support and helpful discussions This work was performedunder the auspices of the US Department of Energy by the University
of California Lawrence Livermore National Laboratory under contract W-7405-Eng-48
1 J J De Yoreo A K Burnham and P K Whitman Int Mat Rev47 113 (2002)
2 C E Barker R A Sacks B M Van Wonterghern J A Caird JR Murray J H Campbell K Kyle R B Ehrlich and N DNielsen Proc SPIE-Int Soc Opt Eng 2633 501 (1995)
3 T Suratwala paper to be published4 Cleveland Crystals Inc httpwwwclevelandcrystalscom
KDPshtmltable5 E A Popova I T Savatinova and I A Velichko Sov Phys Solid
State 12 1543 (1971)6 I P Kaminow R C C Leite and S P S Porto J Phys Chem
Solids 26 2085 (1965)7 J A Subramony B J Marquardt J W Macklin and B Kahr
Chem Mat 11 1312 (1999)8 H Tanaka M Tokunaga and I Tatsuzaki Solid State Commun
49 153 (1984)9 R J Nelmes G M Meyer and J E Tibballs J Phys C 15 59
(1982)10 M A Yakshin D W Kim Y S Kim Y Y Broslavets O E
Sidoryuk and S Goldstein Laser Physics 7 941 (1997)11 I Takenaga Y Tominaga S Endo and M Kobayashi Solid State
Commun 84 931 (1992)12 C Krenn personal communication13 M Sharon and A K Kalia J Solid State Chem 21 171 (1977)
Rapid Dipicolinic Acid Extractionfrom Bacillus Spores Detectedby Surface-EnhancedRaman Spectroscopy
STUART FARQUHARSON ALAND GIFT PAUL MAKSYMIUK andFRANK E INSCOREReal-Time Analyzers Inc East Hartford Connecticut06108
Index Headings Dipicolinic acid Bacillus spores Anthrax Surface-enhanced Raman spectroscopy
INTRODUCTION
The anxiety caused by the distribution of anthrax en-dospores through the US postal system in October 2001was exacerbated by the long time required for positiveidenti cation of the Bacillus anthracis spores and the un-known extent of their distribution Since that time manymethods capable of rapid eld analysis have been inves-tigated to augment or replace the laboratory method ofgrowing microorganisms in culture media which takesdays to perform12 Prominent among these approachesare polymerase chain reactions (PCR)3 immunoassaysand detection of calcium dipicolinate as a biochemicalsignature PCR employs primers to separate organism-
Received 10 October 2003 accepted 14 November 2003 Author to whom correspondence should be sent
352 Volume 58 Number 3 2004
speci c nucleic acid sequences (eg capsular protein en-coding gene for Bacillus anthracis)4 and polymerases toamplify the segment until it is detectable Recently am-pli cation times have been substantially reduced andcomplete analysis can now be performed in an hour orless Immunoassay methods are also being developed thatuse competitive binding of the bioagent (as an antigen)and its labeled conjugate for a limited number of anti-bodies Although analyses can be performed in under 30minutes a well-de ned anthrax antigen has not yet beenidenti ed5ndash7 and consequently the false-positive rate isunacceptably high8
A number of other methods are being developed witha focus on the detection of calcium dipicolinate (CaDPA)and its derivatives as a B anthracis signature This is sobecause only spore-forming bacteria contain CaDPA andthe most common potentially interfering spores such aspollen and mold spores do not Relatively fast methodshave been developed to chemically extract CaDPA andthen detect it directly by uorescence9 or indirectly byluminescence1011 In the latter case hot dodecylamine(DDA) has been used to extract dipicolinic acid (DPA)and terbium has been utilized to form a highly lumines-cent DPA complex11 Although measurements have beenperformed in as little as ve minutes it was found thatas many as three concentration-dependent complexes canform each with different lifetimes This coupled withthe fact that the Tb31 cation produces the same lumines-cence spectrum makes determinations of low spore con-centrations problematic
It has been long known that Raman spectra of Bacillispores are dominated by bands associated with CaDPA12
and that these spectra may provide a suitable anthrax sig-nature at the genus level13 Since that time considerableimprovements in Raman instrumentation have led to lab-oratory measurements of single Bacilli spores14 and to eld measurements of spores captured from a mail-sort-ing system15 However the single spore measurementsrequired complex instrumentation that is not ruggedwhile the eld measurements required milligrams of sam-ple Furthermore the Raman spectra of both measure-ments contained uorescence contributions that would in-crease uncertainty in quanti cation
In related research we demonstrated that nanogramquantities of DPA could be detected by uorescence-freesurface-enhanced Raman spectroscopy (SERS)16 We alsodemonstrated that microliter volumes of chemicals canbe detected by SERS using metal-doped sol-gel-packedglass capillaries17 Towards the goal of developing a rap-id eld SERS-based anthrax spore detector we havecombined our previous research and we now report thatDPA can be extracted from a 10 mg B cereus spore sam-ple using DDA in 1 minute and can be detected by SERSin an additional 1 minute
EXPERIMENTAL
Dipicolinic acid (26-pyridinedicarboxylic acid DPA)and dodecylamine (DDA) were used as received fromSigma-Aldrich (Milwaukee WI) Lyophilized B cereusspores prepared according to the literature13 were sup-plied by the University of Rhode Island and used as re-ceived Multiple particles approximately 01 mm3 each
were separated and weighed at 5 to 15 mg representing05 to 15 million spores The sample masses were con-sistent with a previous determination of spore density at0081 gmL that indicated a high degree of entrained air
All chemicals used to prepare the silver-doped sol-gelcoated capillaries were also obtained and used as receivedfrom Sigma-Aldrich According to previously publishedprocedures17 two precursor solutions were preparedmixed and then drawn into 1-mm-diameter glass capil-laries The silver amine precursor consisted of a 51 vvratio of 1 N AgNO 3 to 28 NH3OH while the alkoxideprecursor consisted of a 21 vv ratio of methanol to te-tramethyl orthosilicate The alkoxide precursors weremixed with silver amine precursor in an 81 vv ratioApproximately 015 mL was drawn into the capillarycoating a 15-mm length After sol-gel formation the in-corporated silver ions were reduced with dilute sodiumborohydride which was followed by a water wash to re-move residual reducing agent
A 100 mL drop of a 50 mM DDA solution in ethanolpre-heated to 78 8C was added to each of the B cereusparticles to digest the spore coat After 1 minute the re-sultant solution was drawn into a SER-active capillarythat was immediately xed horizontally to an XY posi-tioning stage (Conix Research Spring eld OR) just in-side the focal point of an f 07 aspheric lens The lensfocused the beam into the sample and collected the scat-tered radiation back along the same axis A dichroic lter(Omega Optical Brattleborough VT) was used to re ectthe excitation laser to the lens and pass the Raman scat-tered radiation collected by the lens An f 2 achromatwas used to collimate the laser beam exiting a 200-mm-core-diameter source ber optic while a second f 2 ach-romat was used to focus the scattered radiation into a 365mm ber optic (Spectran Avon CT) A short-pass lterwas placed in the excitation beam path to block the sil-icon Raman scattering generated in the source ber fromre ecting off sampling optics and reaching the detectorA long-pass lter was placed in the collection beam pathto block the sample Rayleigh scattering from reachingthe detector A 785 nm diode laser (Process InstrumentsInc model 785-600 Salt Lake City UT) was used todeliver 100 to 150 mW of power to the sample A Fouriertransform Raman spectrometer (Real-Time Analyzersmodel IRA-785 East Hartford CT) and a silicon photo-avalanche detector (Perkin Elmer model C30902S Stam-ford CT) were used to acquire the SER spectra
RESULTS AND DISCUSSION
As an initial experiment the SER spectrum of 1 gLof DPA in water was measured using the newly devel-oped silver-doped sol-gel-coated capillaries (Fig 1A) Atthis concentration a high signal-to-noise ratio (SN) isobtained in 1 min In fact a reasonable spectrum is ob-tained in the same time frame for 1 mgL (Fig 1B) TheSER spectra are reasonably similar to the normal Raman(NR) spectrum obtained for a saturated solution of DPAin 1 N KOH (Fig 1C) and the following band shifts areobserved (NR to SER) 647 to 657 cm21 817 to 815cm21 998 to 1008 cm21 1384 to 1382 cm21 1434 to1428 cm21 and 1569 to 1567 cm21 Many of these bandshave been previously assigned1213 such as 998 cm21 to
APPLIED SPECTROSCOPY 353
FIG 1 SERS of DPA in water using silver-doped sol-gel-coated glasscapillary for (A) 1 gL and (B) 1 mgL (C ) NR of saturated DPA in 1N KOH in a glass capillary Spectral conditions (A) and (B) 150 mWof 785 nm 1-min acquisition time (C) 450 mW of 785 nm 5-minacquisition time both 8 cm 21 resolution
FIG 2 SERS of DPA extracted from 10 mg B cereus particle using100 mL of 50 mM hot DDA acquired in (A) 1 minute and (B) 2 seconds(C ) Attempted SERS of 50 mM hot DDA in ethanol using silver-dopedsol-gel-coated glass capillary acquired in 1 min Spectral conditions150 mW of 785 nm 8 cm21 resolution
the symmetric ring stretch 1384 cm21 to the OndashCndashOsymmetric stretch 1428 cm21 to the symmetric ring CndashH bend and 1569 cm21 to the asymmetric OndashCndashOstretch
The rst B cereus samples consisted of 2 mg of sporesin 2 mL of 5 mM hot DDA The samples were main-tained at 78 8C for 40 min and while hot approximately10 mL was drawn into a SER-active capillary Since spec-tra of DPA were obtained for these initial samples small-er spore masses higher DDA concentrations and shorterheating periods were examined In due course it wasfound that 10 mg of spores could be digested by 100 mLof 50 mM hot DDA in one minute and detected (Fig2A) In fact the signal was suf ciently intense that it canbe observed in as little as two seconds (Fig 2B) Theamount of DPA that was extracted was estimated to bebetween 5 and 10 mgL by comparing the signal intensityof the 1008 cm21 band to that measured for DPA in waterThis is consistent with previous research that found thatthe majority of the DPA is extracted from spores usingDDA11 and that B cereus spores contain approximately10 DPA by weight18 The SN of 127 for the 1008 cm21
band in the 1-minute SER spectrum suggests a limit ofdetection of approximately 250 ng of B cereus sporesbased on a SN of 3 Finally it should be noted that DDAdid not produce a detectable SER spectrum as shown inFig 2C
CONCLUSION
We have demonstrated that by combining rapid extrac-tion of dipicolinic acid from Bacillus cereus spores withchemical identi cation by surface-enhanced Raman spec-troscopy as little as 10 mg of spores can be detected Infact the entire measurement from the time of adding hotdodecylamine to the spores to the time when the dipi-colinic acid SER spectrum is acquired and analyzedcould be performed in less than two minutes The abilityof this method to distinguish between spore-forming bac-teria such as Bacillus anthracis and non-DPA containingpowders could help prevent costly shutdowns associated
with the appearance of suspicious material or intentionalmailing of common substances as an anthrax hoax Thismethod could also prove useful in detecting the locationof anthrax endospores in mail distribution facilities if an-other veri ed attack should occur
Research continues to fully characterize the surface-enhanced Raman spectroscopy signal intensities as afunction of sample concentration and to explore otherextractants that do not require the use of elevated tem-perature
ACKNOWLEDGMENTS
The authors are grateful for the support of the National Science Foun-dation (DMI-0296116 and DMI-0215819) and the US Army(DAAD13-02-C-0015 Joint Service Agent Water Monitor program)The authors are indebted to Chetan Shende for preparing the sol-gelcapillaries The authors also thank James Gillespie Nicholas Fell andAugustus Fountain for providing important background informationMark Farquharson for laboratory support and Professor Jay Sperry ofthe University of Rhode Island for supplying B cereus spores
1 V A Pasechnik C C Shone and P Hambleton Bioseparations 3267 (1993)
2 P J Jackson M E Hugh-Jones D M Adair G Green K K HillC R Kuske L M Grinberg F A Abramova and P Keim ProcNatl Acad Sci USA 95 1224 (1998)
3 B R Glick and J J Pasternak Molecular Biology Principles andApplications of Recombinant DNA (ASM Press Washington DC1994)
4 C A Bell J R Uhl T L Had eld J C David R F Meyer TF Smith and F R Cockerill III J Clin Microbiol 40 2897(2002)
5 D L Gatto-Menking H Yu J G Bruno M T Goode M Millerand A W Zulich Biosens Bioelectron 10 501 (1995)
6 J J Quinlan and P M Foegeding J Rapid Methods AutomationMicrobiol 6 1 (1998)
7 A A Hindle and E A H Hall Analyst (Cambridge UK) 1241599 (1999)
8 M S Ascher US Department of Health amp Human Services(httpwwwhhsgovophppresentationsAscherdoc)
9 R Nudelman B V Bronk and S Efrima Appl Spectrosc 54 445(2000)
10 D L Rosen C Sharpless and L B McBrown Anal Chem 691082 (1997)
354 Volume 58 Number 3 2004
11 P M Pellegrino N F Fell Jr and J B Gillespie Anal ChimActa 455 167 (2002)
12 W H Woodruff T G Spiro and C Gilvarg Biochem BiophysRes Commun 58 197 (1974)
13 E Ghiamati R S Manoharan W H Nelson and J F SperryAppl Spectrosc 46 357 (1992)
14 A P Esposito C E Talley T Huser C W Hollars C M Schal-dach and S M Lane Appl Spectrosc 57 868 (2003)
15 S Farquharson L Grigely V Khitrov W W Smith J F Sperryand G Fenerty J Raman Spectrosc paper accep ted (2003)
16 S Farquharson W W Smith S Elliott and J F Sperry SPIE-IntSoc Opt Eng 3855 110 (1999)
17 S Farquharson and P Maksymiuk Appl Spectrosc 57 479(2003)
18 F W Janssen A J Lund and L E Anderson Science (Washing-ton DC) 127 26 (1958)
SPIE -2003-5269 117
pH dependence of methyl phosphonic acid dipicolinic acid and cyanide by surface-enhanced Raman spectroscopy
Stuart Farquharson Alan Gift Paul Maksymiuk Frank Inscore and Wayne Smith
Real-Time Analyzers 87 Church Street East Hartford CT 06108
ABSTRACT US and Coalition forces fighting terrorism in Afghanistan and Iraq must consider a wide range of attack scenarios in addition to car bombings Among these is the intentional poisoning of water supplies to obstruct military operations To counter such attacks the military is developing portable analyzers that can identify and quantify potential chemical agents in water supplies at microgram per liter concentrations within 10 minutes To aid this effort we have been investigating the value of a surface-enhanced Raman spectroscopy based portable analyzer In particular we have been developing silver-doped sol-gels to generate SER spectra of chemical agents and their hydrolysis products Here we present SER spectra of methyl phosphonic acid and cyanide as a function of pH an important factor affecting quantitation measurements which to our knowledge has not been examined In addition dipicolinic acid a chemical signature associated with anthrax-causing spores is also presented Keywords Chemical warfare agents agent detection agent hydrolysis SERS Raman spectroscopy homeland security
1 INTRODUCTION In the past decade the Unites States and its allies have been challenged by a different kind of warfare exemplified by the terrorist attacks of September 11 2001 Suicide bombings and the use of chemical agents are the norm and military personnel must consider a wide range of attack scenarios Among these is the intentional poisoning of water supplies to obstruct military operations in Afghanistan and Iraq Currently colorimetric paper is used to detect agents on-site while gas chromatography combined with mass spectrometry (GCMS) is used in mobile support laboratories However both methods have severe drawbacks The paper changes color in response to contact with many chemicals besides CWAs causing a high incidence of false positives1 while GCMS although very chemically specific requires hours to perform and constant re-calibration234 Military operations would be greatly aided by a portable analyzer that can identify and quantify potential chemical agents at concentrations that impact safety This includes the analysis of drinking water supplies distribution and storage systems To meet this goal the Department of Defense has been investigating numerous approaches under the auspices of the Joint Service Agent Water Monitor (JSAWM) program5 The JSAWM must be capable of identifying the chemical agents with no false-positives and quantifying the agents at microgram per liter concentrations within 10 minutes678 Recently we and others have been investigating the ability of surface-enhanced Raman spectroscopy (SERS) to measure chemical agents9-12 bioagents13-17 and their hydrolysis products in water SERS employs the interaction of surface plasmon modes of metal particles with target analytes to increase scattering efficiency by as much as 1 million times18 In 1987 the potential of this approach to measure CWAs was demonstrated by measuring a series of pesticides19 Several of these organophosphonates have chemical structures similar to CWAs in particular P=O functional groups In our studies we have been employing silver-doped sol-gels to promote the SER effect The porous silica network of the sol-gel matrix offers a unique environment for immobilizing and stabilizing SER-active metal particles20-23 The sol-gels have been coated on the internal walls of standard glass vials to allow rapid SER analysis of small sample quantities without preparation We have measured over 100 chemicals with enhancements of 104 to 106 demonstrated reversible measurements in a flowing system reproducible measurements from vial-to-vial and measurements in multiple solvents including water20-23 Previously we used these vials to perform preliminary measurements of cyanide (CN) methylphosphonic acid (MPA) and dipicolinic acid (DPA) MPA is a hydrolysis product of the nerve agents (eg sarin Reaction 1) and may be a valuable indicator of nerve agent usage particularly since the alkyl methylphosphonic acids are relatively more stable than their corresponding parent complexes24 DPA is
SPIE -2003-5269 118
a chemical signature of spore forming bacteria such as Bacillus anthracis And in light of the inability to rapidly detect the anthrax spores distributed through the US mail in October 2001 a number of methods are being developed to extract and analyze this signature Reaction 1 Stepwise hydrolysis of Sarin to form hydrofluoric acid (HF) isopropyl methylphosphonic acid (IMPA) then methyl phosphonic acid (MPA) and 2-propanol In our previous SERS investigations MPA and DPA were measured at 50 and 100 mgL respectively In both cases limits of detection (LOD) were estimated at 100 microgL providing encouragement in that SERS may satisfy the needs of the JSAWM Since it has been shown that pH can substantially influence the intensity of SER bands25 which would clearly influence quantitative analysis we undertook the present study to determine the severity of these effects for cyanide methyl phosphonic acid and dipicolinic acid Furthermore we previously observed a band at 1050 cm-1 for MPA6 possibly due to an anion formed at basic pH Here we investigate the source of this spectral anomaly
2 EXPERIMENTAL All chemicals including potassium cyanide methyl phosphonic acid dipicolinic acid and those used to prepare the silver-doped and gold-doped sol-gel coated vials were obtained and used as received from Sigma-Aldrich All samples were prepared in HPLC grade water (Fischer Scientific Fair Lawn NJ) for SERS measurements The pH of these samples was adjusted using dilute nitric acid or potassium hydroxide and verified using a pH electrode (Corning Inc Corning NY) that had been calibrated with pH 400 700 and 1000 buffered standards from Fischer Scientific Once prepared the samples were transferred into the silver-doped sol-gel vials (Real-Time Analyzers Inc East Hartford CT) The vials were coated in a manner similar to that previously reported by adding ammonium hydroxide to a solution of silver nitrate tetramethyl orthosilicate (TMOS) and methanol20 Gold-vials were coated by adding nitric acid to a solution of gold tetrachloride TMOS and methanol The two precursor solutions were prepared mixed and transferred to 2-mL glass vials dried and heated After sol-gel formation the incorporated metal ions were reduced with dilute sodium borohydride (1mgmL) which was followed by a water wash to remove residual reducing agent After the resultant analyte solution was introduced the SER-active vial was immediately fixed horizontally to an XY positioning stage (Conix Research Springfield OR) just inside the focal point of an f07 aspheric lens The lens focused the beam into the sample and collected the scattered radiation back along the same axis A dichroic filter (Omega Optical Brattleborough VT) was used to reflect the excitation laser to the lens and pass the Raman scattered radiation collected by the lens An f2 achromat was used to collimate the laser beam exiting a 200 microm core diameter source fiber optic while a second f2 achromat was used to focus the scattered radiation into a 365 microm fiber optic (Spectran Avon CT) A short pass filter was placed in the excitation beam path to block the silicon Raman scattering generated in the source fiber from reflecting off sampling optics and reaching the detector A long pass filter was placed in the collection beam path to block the sample Rayleigh scattering from reaching the detector A 785 nm diode laser (Process Instruments Inc model 785-600 Salt Lake City UT) was used to deliver 100 to 150 mW of power to the sample A Fourier transform Raman spectrometer (Real-Time Analyzers model IRA-785 East Hartford CT) and a silicon photo-avalanche detector (Perkin Elmer model C30902S Stamford CT) were used to acquire the SER spectra
3 RESULTS AND DISCUSSION In a previous study of MPA6 aimed at developing a concentration calibration curve and determining limits of detection (LOD) we observed an anomaly at 1050 cm-1 Since it was found that the band intensity changed as a function of concentration the band must be associated with a sample parameter Two possible parameters photon flux and pH are examined here The first parameter was investigated by irradiating a 1mgmL MPA sample in a SER-active vial with laser powers of 200 mW and above and monitoring spectral changes It was immediately found that the 1050 cm-1 band
2O+ H HF + +OH
OH3H C
OP
OF
CH
CH3
3
3H C
OCP
OOH
CH
CH3
3
3H C
O
CPHO
CH
CH
3
3
C
Sarin IMPA MPA 2-propanol
SPIE -2003-5269 119
appeared and grew as a function of time and that the higher the laser power the faster the growth Figure 1 shows the growth of the 1050 cm-1 band over the course of 30 minutes when using 150 mW of 785 nm excitation while Figure 2A shows that the growth can be fit with a first order exponential equation namely I1050 = 03+05e-013t Furthermore the 760 cm-1 band could be fit with a first order decay equation with an identical exponential rate constant ie I760 = 08-
08e-013t The rates represent classical first order kinetics and their correspondence allows one to conclude that MPA is being transformed one-for-one into a photo-generated product At this time the photoproduct has not been positively identified but phosphonic (phosphorous) acid and phosphonate are likely candidates since the symmetric P(OH)3 stretch occurs at ~1050cm-1 Our studies show that a reduction in laser power to 100 mW at the sample for MPA essentially eliminates this degradation process This laser power or lower was used for further measurements
Methyl phosphonic acid is a diprotic acid that stepwise dissociates into two anions MPA- and MPA= according to the following reactions26
MPA MPA- + H+ pKa1 = 212 Reaction 2
MPA- MPA= + H+ pKa2 = 729 Reaction 3 The relative concentrations of MPA MPA- and MPA= can be determined at any pH by expressing [MPA] and [MPA=] in terms of [MPA-] using Reactions 2 and 3 and summing all three to equal the total starting concentration here 2 mgmL (0021M MW = 9602) viz
[MPA] + [MPA-] + [MPA=] = 0021M Equation 1 substituting from Reactions 1 and 2
([H+][MPA-])K1a + [MPA-] + (K2a[MPA-])[H+] = 0021M Equation 2 rearranging [MPA-] = 0021M(1+[H+]K1a + K2a[H+]) Equation 3 The relative concentrations of MPA MPA- and MPA= as a function of pH are shown in Figure 3 It is worth noting that near neutral pH both MPA- and MPA= will be present To confirm that the SER signal followed this pH dependence a starting solution consisting of 20 mg of MPA in 10 mL HPLC grade water was prepared and brought to pH of 20 using dilute nitric acid From this solution 2 mL were added to a SER-active vial and the SER spectrum recorded At this pH a peak at 760 cm-1 was barely discernable The 2 mL solution was returned to the starting solution and the pH was re-measured to correct for any changes that the silver-doped sol-gel vials might cause In most cases the change was less than 02 pH units and the pH is reported as the before and after average Next the pH of the
Figure 2 A) Exponential growth of 1050 cm-1 band and B) exponential decay of 760 cm-1 band for spectral series in Figure 1
Figure 1 Growth of 1050 cm-1 band as a function of time due to exposure to 150 mW of 785 nm Spectra are 5 sec each collected every 100-sec from 0 to 30-min
B
570 770 970 1170 1370Raman Shift (cm-1)
Arbit
rary
Unit
s
MPA Photodegradation
600 800 1000 1200 1400 Wavenumber (cm-1)
A
0 10 20 30 0 10 20 30 time (min) time (min)
Ram
an In
tens
ity (r
elat
ive)
30
min
0
SPIE -2003-5269 120
starting solution was adjusted to 325 using dilute KOH Again 2 mL were added to a vial and the SER spectrum recorded At this pH a reasonably strong 760 cm-1 band was observed This process was repeated as spectra were recorded at pHs of 70 74 75 79 85 and 100 A total of 1 mL of KOH was added diluting the total concentration by 10 Next the pH of the starting solution was made acidic by adding dilute nitric acid dropwise This time spectra were recorded at pHs of 72 69 64 and 37 Figure 4 shows the SER spectra for representative pHs (spectra were left out to simplify the figure) while Figure 3 shows the 760 cm-1 peak intensities as a function of pH (The band intensities were adjusted to compensate for dilution effects caused by the addition of HNO3 and KOH then normalized to 0021 M for the most intense band observed at pH 37) It is clear from Figure 3 that the 760 cm-1 band follows the MPA- concentration as a function of pH and must be assigned to this anion No bands were observed that corresponded to MPA or MPA= The lack of an MPA SER spectrum may be due to the absence of an attraction between the neutral analyte and the electropositive silver surface The same reasoning suggests that a strong SER spectrum should be observed for MPA= but it is not and a satisfactory explanation has not been found
0000
0005
0010
0015
0020
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14pH
Con
cent
ratio
n [M
]
pK1 = 212 pK2 = 729
MPA- MPA=
MPA
Figure 4 SER spectra of 002M MPA as a function of pH Conditions 100 mW of 785 nm 36 scans (1 min) 8 cm-1 recorded 2 min after sample introduction pH 19 and 20 not apparent on this scale pH 69 and 74 near identical to 70 and 75 and not shown for clarity
00
02
04
06
08
10
12
14
16
18
0 5 10 15 20 25 30Measurement Number
Ram
an In
tens
ity (7
60 c
m-1
)
Figure 3 Concentration dependence of MPA MPA- and MPA= as a function of pH for a 002M sample Intensity of 760 cm-1
band from Figure 6 as a function of pH ( for increasing basic adjustment for increasing acidic adjustment error was measured at ~10 for pH 64)
Figure 5 SER spectra of 002M MPA at pH 64 measured around a vial at three heights (9 points per height) Conditions as in Figure 4 but 10-sec scans
Since these measurements involved the removal and replacement of the SER-active vial in the sample holder to remove and add sample variation in the intensity as a function of vial position was minimized by illuminating the exact same height along the vial wall But this does not account for variability of the SERS response of the sol-gel coating around the vial To analyze this effect a vial containing MPA at pH 64 was rotated at ~ 40o
intervals at the original height and 18rdquo above and below this value Figure 5 shows the intensity of the 760 cm-1
band for the 27 positions It was found that the average value was 137plusmn014 an RSD of 10 overall and 5 for each height An error bar is included in Figure 4 for the pH 64 measurement
SPIE -2003-5269 121
With the above analyses in mind a preliminary investigation of the SER spectral response for MPA (as MPA-) as a function of concentration was performed A single vial was used for these measurements beginning with 1 mgL followed by measurements of 10 100 and 1000 mgL In all cases the pH was ~7 and 3 positions around the vial were measured per concentration Since the 760 cm-1 band was not observed for concentrations of 1 or 10 mgL using 100 mW of 785 nm the laser power at the sample was raised to 200 mW beginning with the 10 mgL concentration Photo-degradation was largely avoided (and not observed) by exposing the sample for only 33 seconds per spectral acquisition Representative spectra for 10 100 and 1000 mgL are shown in Figure 6 while a plot of the 760 cm-1 band intensity as a function of concentration is shown in Figure 7 These values were also used to estimate limits of detection based on the signal-to-noise ratio (SN) of the 760 cm-1 band using 100 mW of 785 nm laser excitation and a 10-min acquisition time As summarized in the Figure 7 inset the lower the measured concentration the lower the predicted LOD Even if as estimated 210 microgL could be measured using the silver-doped sol-gel vials an improvement of a factor of 70 is still required to meet JSAWM goals of measuring 3 microgL in 10-minutes Similar to MPA DPA is a diprotic acid (pKa1 = 216 and pKa2 = 692) and variations in pH will effect the relative concentrations of DPA DPA- and DPA= and possibly the SER spectra and band intensities This could prove significant if an acid or base is used to denature anthrax spores with the goal of extracting and analyzing DPA The MPA pH study described above was mimicked for DPA except that the starting solution consisted of 20 mg of DPA in 20 mL HPLC grade water (60x10-3M MW = 1671) The initial solution had a pH of 245 which was made basic by dilute KOH to pHs of 355 433 487 559 1069 and 1166 SER spectra were recorded at each pH using 100 mW of 785 nm and a 44-sec acquisition time Next one drop of concentrated nitric acid was used to remake the solution acidic at a pH of 200 Again sequential pH measurements were performed at 383 510 735 and 822 The solution pH was made acidic a third time but to pH 219 171 then 135 Throughout this process no more than 20 drops of acid or base were added and therefore the concentration was diluted by no more than 10 Most of the spectral bands showed a minor decrease in intensity as a function of increasing pH values However the bands at 525 and 795 cm-1 showed the most dramatic changes which occurred at acid pH Figure 8 shows the SER spectra of DPA for the spectral region and pH range of interest The identity of the DPA species was determined by plotting the normalized peak intensities with the lowest value set to 0 and the highest to 0006 M as a function of pH and overlaying these values on a plot of the relative concentrations for DPA DPA- and DPA= as previously done for MPA (Figure 9) As can be seen the 525 cm-1 band clearly corresponds to DPA The correspondence of the 795 cm-1 band to this species is less clear as the band retains intensity until pH of 55 This can be attributed to contribution to the overlapping band at 810 cm-1 which does not change as a function of pH The fact that most bands are observed at all pHs suggest that the primary interaction with silver is through the ring nitrogen This is supported by the fact that the most intense band occurs at 1008 cm-1 attributed to a symmetric ring breathing mode and that this interaction has been characterized for pyridine in numerous papers27
0
1
2
3
4
5
6
7
0 200 400 600 800 1000 1200
MPA Concentration (mgL)
760
Ban
d In
tens
ity
Figure 6 SER spectra of MPA in water at A) 1000 B) 100 and C) 10 mgL Conditions pH of 7 silver-doped sol-gel coated vial 200 mW of 785 nm 33-sec 8 cm-1 resolution
Figure 7 Plot of SER intensity of 760 cm-1 band of MPA as a function of concentration using 200 mW of 785 nm Inset table includes average intensity LOD standard deviation and percent deviation for each concentration but for 100 mW and 10 min
conc (mgL) LOD ave stddev dev10 021 002 1021100 072 010 14301000 312 040 1280
A
B C
SPIE -2003-5269 122
The ability to detect dipicolinic acid (DPA) by SERS is demonstrated in Figure 10 Even at 1 mgL the primary bands are visible The intensity of the dominant ring breathing mode observed at 1008 cm-1 for DPA in water (pH = 25 to 55) is plotted as a function of concentration in Figure 11 These values were also used to estimate limits of detection based on the SN of the 1008 cm-1 band using 100 mW of 785 nm laser excitation and a 10-min acquisition time Again the lower the measured concentration the lower the predicted LOD (see Figure 11 inset) and detection of 160 microgL is possible Gastrointestinal anthrax requires significant more spores than inhalation anthrax28 and a limit of detection might be placed at 1 million spores in 1 liter of water or 10 microgL Since each spore contains ~10 CaDPA by weight29 a goal for DPA might be 1 microgL indicating that the present measurements must be improved by nearly two orders of magnitude Due to the increased hazards of handling HCN gas KCN salt was used for these experiments Nevertheless all sample preparations were performed in a chemical hood KCN completely dissolves in water but its conjugate acid HCN is formed and has a Ka of 615x10-1030 viz
HCN CN- + H+ pKa = 921 Reaction 4
Figure 8 SER spectra of 1 mgmL DPA as a function of pH Conditions 100 mW 785 nm 100 scans (44-sec)
Figure 9 Concentration dependence of DPA DPA- and DPA= as a function of pH for a 0006M sample Intensity of 525 () and 795 (diams) cm-1 bands from Fig 8 as a function of pH
pH
135 171 219 383
0
0001
0002
0003
0004
0005
0006
0007
0 2 4 6 8 10 12 14pH
Con
cent
ratio
n [M
]
DPADPA-DPA=795525
DPA DPA=DPA-
pK2 = 692pK1 = 216
795 525
0
05
1
15
2
25
0 200 400 600 800 1000 1200
DPA Concentration (mgL)
1008
Ban
d In
tens
ity
conc (mgL) lod-10min-100mw1 017
10 016100 103
1000 355
Figure 10 SER spectra of DPA in water at A) 1000 B) 100 C) 10 and D) 1 mgL Conditions pH of 25-55 silver-doped sol-gel coated vial 175 mW of 785 nm 1-min 8 cm-1 D) has been multiplied by x10 to make bands visible
A
B C
D
Figure 11 Plot of SER intensity of 1008 cm-1 band of DPA as a function of concentration using 175 mW of 785 nm Inset table includes LOD in mgL for each concentration but for 100 mW and 10 min
SPIE -2003-5269 123
Consequently the cyanide concentration must be determined for each initial KCN concentration Specifically the samples prepared with concentrations of 01 1 10 100 and 1000 mgL of KCN produced CN- concentrations of 63x10-3 033 69 89 and 964 mgL at pHs of 816 90 967 102 and 107 respectively The pH dependence for the HCN and CN- concentrations are shown in Figure 12 Thus as the amount of KCN added to the solution decreases so does the pH of the solution (becomes less basic) and according to Reaction 4 the relative amount of CN- to HCN also decreases For example in the preparation of a 01 mgL solution of KCN the pH is shifted from 7 for pure water to only 816 and only 63 of the starting material becomes CN- or 63x10-3 mgL In comparison for a solution of 1000 mgL the pH is shifted from 7 to 107 and 96 of the starting material becomes CN- This is significant since the cyanide ion is better able to adsorb onto the silver particles and become SERS active SER spectra of 10 100 and 1000 mgL of KCN samples are shown in Figure 13 and a corresponding plot of the 2100 cm-1 band intensity as a function of concentration is shown in Figure 14
000102030405060708091011
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14pH
Con
cent
ratio
n [m
gm
L]
pKa = 921
CN -HCN
The SER spectra of cyanide are dominated by the single CN stretch at 2100 cm-1 which occurs in normal Raman spectra of solutions at 2080 cm-1 However a low frequency mode occurs at 135 cm-1 due to a Ag-CN stretch (not shown) indicative of a strong surface interaction It is also observed that as the concentration decreases the CN stretch shifts to 2140 cm-1 This shift has been attribute to the formation of a tetrahedral Ag(CN)3
2- surface structure31 as well
Figure 14 Concentration dependence of KCN SERS measured under conditions in Fig 11 Concentrations are 1 01 and 001 mgml Intensities are measured for the CN stretch at 2100 cm-1 Inset table includes LOD in mgL for each concentration in Figs 13 and 15 but for 100 mW and 10 min
Figure 13 SER spectra of KCN in water at A) 1000 B) 100 and C) 10 mgL Conditions pHs of 107 102 and 97 silver-doped sol-gel coated vial 100 mW of 785 nm 1-min 8 cm-1 C) has been multiplied by x10 to make band visible
Figure 12 Concentration dependence of HCN and CN- as a function of pH for a 1 mgmL sample Calculated intensity of 2100 () cm-1 band for a 1 mgml sample at pHs of 816 90 967 102 and 107
A
B
C
0
20
40
60
80
100
120
140
0 200 400 600 800 1000 1200CN Concentration (mgL)
2100
Ban
d In
tens
ity
Figure 15 SER spectra of KCN in water at A) 10 B) 1 and C) 01 mgL Conditions pHs of 97 12 and 12 gold-doped sol-gel coated vial 100 mW of 785 nm 1-min 8 cm-1 C) has been multiplied by x10 to make band visible
A
B
C
conc (mgL) lod-10min-100mw Condition01 001 Au-pH 121 007 Au-pH 12
69 003 Au-pH 9769 016 Ag-pH 9789 022 Ag-pH 102946 113 Ag-pH 107
SPIE -2003-5269 124
as to CN adsorbed to two different surface sites32 Alternatively the 2140 cm-1 band could be attributed to HCN since this species dominates at lower concentrations However it is unlikely that this species would be attracted to the electropositive silver surface Further both peaks should be present at pHs between 85 and 105 but this is not observed It has also been suggested that at concentrations near and above monolayer coverage the CN- species is forced to adsorb end-on due to crowding and at lower concentrations the molecule can reorient to lie flat33 This suggests that the 2100 and 2140 cm-1 bands correspond to the end-on and flat orientations respectively As Figures 13 and 14 show the intensity of the CN stretch for the 89 mgL sample is nearly as intense as the 964 mgmL band This suggests that the Raman signal for the flat orientation is more enhanced However more extensive measurements are required to verify this point Since resent research has suggested that cyanide may be more effectively detected on gold measurements of KCN solutions were also performed using gold-doped sol-gel vials Preliminary measurements are shown in Figure 15 for samples prepared from 01 1 and 10 mgL KCN Since the pHs are 816 90 967 the resultant CN- concentrations are 63x10-3 033 and 69 mgL Initially only the highest concentration was observed and the signal intensity was significantly better than the equivalent concentration measured using silver In an effort to shift Reaction 4 to the left transforming HCN to CN- (Le Chatelierrsquos principle) KOH was added to the lower concentration samples producing solutions with pH 12 This effectively forces all of the cyanide in solution to be CN- or 01 and 10 mgL respectively More importantly the CN stretch is now observed in the SER spectra The band appears at 2125 cm-1 as has been previously reported for gold12 As calculated for MPA and DPA LODs can be estimated from this data For the three concentrations of cyanide on silver the LODs are 016 to 11 mgL for 100 mW of 785 nm laser excitation and a 10-min acquisition time For gold pH adjusted the LODs are10 to 70 microgL an improvement of more than 10 times silver Nevertheless either substrate is sufficient to meet the JSAWM goals of measuring 3 mgL in 10-minutes as the requirements form cyanide are much less stringent than the nerve agents
4 CONCLUSIONS Here we examined the surface-enhanced Raman spectral response for methyl phosphonic acid dipicolinic acid and cyanide as a function of pH It was determined that the most prominent peak at 760 cm-1 reaches a maximum intensity between pH 3 and 7 and corresponds to the MPA- species Neither the MPA nor MPA= species appear to generate a SER spectrum and consequently no spectra were observed below pH 2 or above pH 8 In this study we also found that higher laser powers could cause photodegradation of MPA signified by the exponential growth of a band at 1050 cm-1 which is tentatively assigned to phosphorous acid Unlike MPA DPA was observed at all pHs This is attributed to the dominant interaction of the pyridine functional group with silver Minor spectral changes were observed at acid pHs and were assigned to neutral DPA Like MPA SER spectra of cyanide were pH dependent No spectra were observed for the HCN species while CN- was best observed at pHs more basic than 8 Preliminary concentration studies for the three analytes allowed estimating limits of detection for MPA DPA and CN using 100 mW of 785 nm and a 10-min acquisition time of 210 165 and 70 microgL respectively Although the latter value suggests that cyanide can be measured at sufficiently low concentrations to meet JSAWM goal improvements by 100 to 200 times are required for MPA and DPA It is clear from this study that pH of the sample is important and must be taken into account when developing concentration calibration curves as well as fieldable analyzers Future work will include tailoring the sol-gel with specific functional groups to dictate sample pH and thereby optimize sensitivity
ACKNOWLEDGMENTS The authors are grateful for the support of the US Army (DAAD13-02-C-0015 Joint Service Agent Water Monitor program) The authors would also like to thank Dr Steve Christensen of the US Army SBCCOM for helpful discussions and Mr Chetan Shende of Real-Time Analyzers for assistance in development of the gold-doped sol-gels
REFERENCES 1 Erickson B Analytical Chemistry News amp Features June 1 397A (1998)
SPIE -2003-5269 125
2 Johnston RL Hoefler CM Fargo JC and Moberley B AT-ONSITE 5-8 (1994) 3 Black RM RJ Clarke RW Read and MT Reid J Chromatography 662 301-321 (1994) 4 USA Army Center for Health Promotion and Preventive Medicine (USACHPPM) Technical Guideline 230A 5 Jensen JL NC Wong W R Loerop SPIE 4775-03 (2001) 6 Hoenig SL Handbook of Chemical Warfare and Terrorism Greenwood Press Wesport CT (2002) 7 Munro NB SS Talmage GD Griffin LC Waters AP Watson JF King and V Hauschild Env Health
Persp 107 933-974 (1999) 8 Holstege CP Kirk M Sidell FR Crit Care Clin13 923-42 (1997) 9 Farquharson S P Maksymiuk K Ong and S Christesen SPIE 4577 166-173 (2001) 10 Lee Y and S Farquharson SPIE 4378 21-26 (2001) 11 Spencer KM J Sylvia S Clauson and J Janni SPIE 4577 158-165 (2001) 12 Tessier P S Christesen K Ong E Clemente A Lenhoff E Kaler and O Velev Applied Spectroscopy 56
1524-1530 (2002) 13 Farquharson S WW Smith S Elliott and JF Sperry SPIE 3533 207-214 (1998) 14 Farquharson S WW Smith S Elliott and JF Sperry SPIE 3855110-116 (1999) 15 Farquharson S WW Smith YH Lee S Elliott and JF Sperry SPIE 4575 62-72 (2002) 16 Guzelian AA J Sylvia J Janni S Clauson and KM Spencer SPIE 4577 182-192 (2001) 17 Shende C F Inscore A Gift P Maksymiuk and S Farquharson in press 18 Weaver MJ S Farquharson and MA Tadayyoni J Chem Phys 82 4867-4874 (1985) 19 Alak AM and T Vo-Dinh Analytical Chemistry 59 2149-2153 (1987) 20 Lee Y and S Farquharson SPIE 4206 140-146 (2000) 21 Farquharson S and Y Lee SPIE 4200-16 (2000) 22 Lee Y S Farquharson and P M Rainey SPIE 3857 76-84 (1999) 23 Lee Y S Farquharson H Kwong and M Shahriari SPIE 3537 252-260 (1998) 24 Wang J M Pumera G Collins and A Mulchandani Analytical Chemistry 74 6121-6125 (2002) 25 Dou X YM Jung Z-Q Cao and Y Ozaki Applied Spectroscopy 53 1440-1447 (1999) 26 Data supplied by S Christesen and K Ewing 27 Kerker M and B Thompson Eds SPIE MS 10 (1990) 28 Inglesby TV DA Henderson JG Bartlett JAMA 287 2236 (2002) 29 FW Janssen AJ Lund and LE Anderson Science 127 26 (1958) 30 Lide DR Ed Handbook of Chemistry and Physics CRC Press 77th Ed (1996-7) 31 Billmann J G Kovacs and A Otto Surf Sci 92 153 (1980) 32 Murray CA and S Bodoff Phys Rev B 32 671 (1985) 33 Kellogg D and J Pemberton J Phys Chem 91 1120 (1987)
SPIE-2003-5269 16
Chemical agent detection by surface-enhanced Raman spectroscopy
Stuart Farquharson Alan Gift Paul Maksymiuk Frank Inscore and Wayne Smith
Real-Time Analyzers 87 Church Street East Hartford CT 06108
Kevin Morrisey and Steven D Christesen US Army SBCCOM Aberdeen Proving Ground MD 21010
ABSTRACT
In the past decade the Unites States and its allies have been challenged by a different kind of warfare exemplified by the terrorist attacks of September 11 2001 Although suicide bombings are the most often used form of terror military personnel must consider a wide range of attack scenarios Among these is the intentional poisoning of water supplies to obstruct military operations in Afghanistan and Iraq To counter such attacks the military is developing portable analyzers that can identify and quantify potential chemical agents in water supplies at microgram per liter concentrations within 10 minutes To aid this effort we have been investigating the value of a surface-enhanced Raman spectroscopy based portable analyzer In particular we have been developing silver-doped sol-gels to generate SER spectra of chemical agents and their hydrolysis products Here we present SER spectra of several chemical agents measured in a generic tap water Repeat measurements were performed to establish statistical error associated with SERS obtained using the sol-gel coated vials Keywords Chemical agents chemical agent detection SERS Raman spectroscopy
1 INTRODUCTION
In the autumn of 2001 terrorism within US borders became a sobering reality While extensive efforts are being implemented to secure the homeland US and Coalition forces in Afghanistan and Iraq are constantly faced with terrorist attacks In addition to car-bombings the military has identified several non-traditional attack scenarios including poisoning of water supplies by chemical warfare agents (CWAs) To counter this threat the Department of Defense has been investigating numerous approaches to detect such attacks under the auspices of the Joint Service Agent Water Monitor (JSAWM) program1 The JSAWM must be capable of identifying the chemical agents with no false-positives and quantifying the agents at microgram per liter concentrations within 10 minutes2 This includes the analysis of drinking water supplies distribution and storage systems Currently colorimetric paper is used to detect agents on-site while gas chromatography combined with mass spectrometry (GCMS) is used in mobile support laboratories However both methods have severe drawbacks The paper changes color in response to contact with many chemicals besides CWAs causing a high incidence of false positives3 while GCMS although very chemically specific requires up to an hour to perform and regular re-calibration456 Vibrational spectroscopy has also been investigated as a potential method of rapidly detecting CWAs7-
11 as early as 1985 in the case of infrared spectroscopy8 and 1988 in the case of Raman spectroscopy11 Again however these techniques also have limitations when it comes to measuring trace poisons in water Infrared spectra would be dominated by the very strong absorption of water which would obscure absorptions by most other chemicals present Whereas Raman spectroscopy is simply not a very sensitive technique and detection limits are typically grams per liter Surface-enhanced Raman spectroscopy (SERS) offers several advantages over conventional vibrational methods and that may provide the necessary sensitivity required for detecting trace quantities of chemical agents in water SERS employs the interaction of surface plasmon modes of metal particles with target analytes to increase scattering efficiency by as much as 1 million times12 In 1987 the potential of this approach to measure CWAs was demonstrated by measuring a series of pesticides13 Several of these organophosphonates have chemical structures similar to CWAs in particular P=O functional groups In the past few years we and others have further explored the ability of SERS to detect CWAs14-17 and even bioagents 18-21 We have been employing silver-doped sol-gels to promote the SER effect
SPIE-2003-5269 17
in these studies The porous silica network of the sol-gel matrix offers a unique environment for immobilizing and stabilizing SER-active metal particles22-25 The silver-doped sol-gels have been coated on the internal walls of standard glass vials to allow rapid SER analysis of small sample quantities without preparation We have measured over 100 chemicals with enhancements of 104 to 106 demonstrated reversible measurements in a flowing system reproducible measurements from vial-to-vial and measurements in multiple solvents including water21-25 Previously we used these vials to perform preliminary measurements of cyanide (CN) and methylphosphonic acid (MPA) Most of the nerve agents form MPA during hydrolysis while Tabun forms CN a chemical agent in its own right In another paper including in these proceedings we examined the limits of detection (LOD) for MPA by measuring a series of concentrations down to 50 mgL and estimated a limit of detection of 100 microgL26 These measurements provide encouragement in that SERS may satisfy the needs of the JSAWM To further establish the viability of SERS in particular silver-doped sol-gels here we present analysis of cyanide mustard and VX in tap water The measurements performed at the US Armyrsquos Edgewood Chemical Biological Center Aberdeen MD also included numerous repeat measurements to establish reproducibility
2 EXPERIMENTAL 2a General All chemicals including potassium cyanide 2-chloroethylethyl sulfide and those used to prepare the silver-doped and gold-doped sol-gel coated vials were obtained and used as received from Sigma-Aldrich All samples were prepared in a chemical hood using HPLC grade water unless otherwise noted (Fischer Scientific Fair Lawn NJ) for SERS measurements Once prepared the samples were transferred into the silver-doped sol-gel vials (Real-Time Analyzers Inc East Hartford CT) The vials were coated in a manner similar to that previously reported by adding ammonium hydroxide to a solution of silver nitrate tetramethyl orthosilicate (TMOS) and methanol22 The two precursor solutions were prepared mixed and transferred to 2-mL glass vials dried and heated After sol-gel formation the incorporated metal ions were reduced with dilute sodium borohydride (1mgmL) which is followed by a water wash to remove residual reducing agent After the resultant analyte solution was introduced the SER-active vial was fixed horizontally to an XY positioning stage (Conix Research Springfield OR) just inside the focal point of an f07 aspheric lens The lens focused the beam into the sample and collected the scattered radiation back along the same axis A dichroic filter (Omega Optical Brattleborough VT) was used to reflect the excitation laser to the lens and pass the Raman scattered radiation collected by the lens An f2 achromat was used to collimate the laser beam exiting a 200 microm core diameter source fiber optic while a second f2 achromat was used to focus the scattered radiation into a 365 microm fiber optic (Spectran Avon CT) A short pass filter was placed in the excitation beam path to block the silicon Raman scattering generated in the source fiber from reflecting off sampling optics and reaching the detector A long pass filter was placed in the collection beam path to block the sample Rayleigh scattering from reaching the detector A 785 nm diode laser (Process Instruments Inc model 785-600 Salt Lake City UT) was used to deliver 100 to 150 mW of power to the sample A Fourier transform Raman spectrometer (Real-Time Analyzers model IRA-785 East Hartford CT) and a silicon photo-avalanche detector (Perkin Elmer model C30902S Stamford CT) were used to acquire the SER spectra
2b Edgewood Chemical Biological Center The surface-enhanced Raman spectral measurements at the US Armyrsquos Chemical Biological center presented here were all performed on September 12 2003 To expedite measurements a plate was machined to hold up to 12 SER-active sample vials (Figure 1) The plate fit a standard XY plate reader that could be programmed Pure KCN bis-(2-chloroethyl)sulfide (distilled mustard gas HD) and ethyl S-2-diisopropylamino ethyl methylphosphonothioate (VX) were obtained on-site and used to prepare 1 gL tap water solutions in a chemical hood with appropriate safety equipment Simulated tap water was prepared by adding 10 chemicals most often found in tap water at appropriate concentrations to distilled water (Table 1) SER measurements were also performed in a chemical hood For added safety the FT-Raman instrument was placed outside the laboratory and 30 foot fiber optic and electrical cables were used to allow remote SERS measurements and plate manipulation For each experiment 1gL samples were prepared and added to 9 individual vials which were then loaded on the plate In some cases a tenth vial was included as a blank
Table 1 Generic Tap Water Composition
Compound mgL NaHCO3 100 CaSO4 27 MgSO4bull7H2O 67 NaNO3 1 Fulvic Acid 1 K2HPO4 07 KH2PO4 03 (NH4)2HSO4 001 NaCl 001 FeSO4 0001 pH 76-78
SPIE-2003-5269 18
A software program was written that allowed selecting the sequence that the vials were measured the number of positions along the length of the vials to measure (1 to 5) and the number of scans to co-add During sample analysis the program displayed the vial being analyzed the point being analyzed and the spectrum as it was being acquired Once all the data was collected a second software program was written to rapidly analyze the data The spectra collected for all the vials on a plate could be loaded at one time and then the spectra for each point could be displayed simultaneously or separately The user could then select the Raman peak to analyze in terms of peak height or area This was accomplished by selecting points on either side of the peak to define a baseline of zero The peak height or area could then be computed for all of the spectra loaded and then exported to a spreadsheet for statistical analysis
Figure 1 A) Vial Holder 6 slots to hold 2 vials each end-to-end B) Measurement Configuration Program user selects vials to measure sequence number of points per vial (1 to 5) and number of scans per point C) Spectral Acquisition Program shows spectrum being collected which vial and position D) Spectra Analysis Program user selects spectra to analyze by plate vial and point (s) as well as two wavenumbers defining the peak and the baseline to subtract The image is of 5 repeat measurements of 10 mgL KCN in generic tap water 16 sec each 100 mW of 785 nm
3 RESULTS AND DISCUSSION Raman and surface-enhanced Raman spectra were obtained for potassium cyanide bis-(2-chloroethyl)sulfide and ethyl S-2-diisopropylamino ethyl methylphosphonothioate representing three classes of chemical agents cyanides mustards and nerve agents respectively Spectra were also obtained for 2-chloroethyl ethyl sulfide (CEES) a structural analogue to HD which was included in the study to aid in assigning spectral bands KCN salt was used for cyanide experiments to avoid the increased hazards of handling HCN gas KCN completely dissolves in water forming its conjugate acid HCN according to its Ka of 615x10-1027 and at a concentration of 1 mgmL results in a pH 107 solution This is significant in that only CN- appears to interact sufficiently with silver to produce a SER spectrum and no spectral signal is observed below pH 726 Figure 2 shows the SER and normal Raman spectra for KCN The SERS spectrum of 1mgml KCN in water shows a single intense somewhat broad feature at
A B
C D
SPIE-2003-5269 19
2100 cm-1 assigned to the single CequivN stretch The band is much sharper in the normal Raman spectra of the solid KCN salt at 2074 cm-1 This band does broaden and shift to 2080 cm-1 in solution (not shown) However the observed SERS frequency is attributed to interaction with silver and low frequency mode at 135 cm-1 attributed to a Ag-CN stretch (not shown) supports this conclusion
cm-1 band to a C-S stretch but the authors concede that it is in fact more likely a C-Cl stretch28 It appears that the most intense Raman bands at 648 692 and 747 cm-1 shift to 620 660 and 720 cm-1 in the SER spectra and are tentatively assigned as above The width of these bands suggests that they overlap underlying spectral features Additional bands in the Raman spectra occur at 972 1034 1049 1263 1286 1423 1442 2865 2935 and 2960 cm-1 Corresponding bands occur in the SER spectra at 964 1015 1054 1286 1410 1447 2865 and 2935 cm-1 Most of these bands are associated with alkane modes specifically the bands at approximately 1040 cm-1 to a C-C stretch 1290 cm-1 to a CH2 in-phase twist 1440 cm-1 to a CH2 wag 2865 cm-1 to a symmetric CH2 stretch and 2965 cm-1 to an asymmetric CH2 stretch The Raman and SER spectra of sulfur mustard were measured at the Edgewood center (Figure 4) Both spectra are largely similar to CEES The C-Cl and C-S bands in the Raman spectrum of HD now occur at 640 655 700 739 and 760 cm-1 and are more resolved possibly due to the increased molecular symmetry Theoretical calculations indicate that the first three bands are due to C-Cl stretching modes and the latter two to C-S stretching modes28 Only the C-Cl bands maintain significant intensity in the SER spectra occurring at 624 and 643 cm-1 which is attributed to the
Prior to measurements of HD CEES was examined by Raman and SER spectroscopy (Figure 3) CEES also known as half-mustard is essentially identical to HD except one of the chlorine end atoms is replaced by a hydrogen atom Again although not as toxic as HD CEES is a blister agent and dilute aqueous samples were prepared using appropriate safety equipment Both the Raman and SER spectra of CEES are similar and dominated by bands between 600 and 800 cm-1 These are associated with C-Cl and C-S stretching modes which are tentatively assigned to 648 and 747 cm-1 in the Raman spectra respectively The shoulder at 630 cm-1 the overlapped band at 660 cm-1 and the strong band at 692 cm-1 could also be due to these modes or their asymmetric counterparts It is worth noting that theoretical calculations assign the 692
Figure 2 A) SER and B) NR spectra of KCN Conditions A) 1 mgml in tap water 100 mW of 785 nm at sample 1-min acquisition time B) solid 300 mW of 785 nm 5-min All spectra are 8 cm-1 resolution
A
B
Figure 3 A) SER and B) NR spectra of CEES Conditions A) 1 vv (10 mgml) in MeOH 100 mW of 785 nm 1-min acquisition time B) neat 300 mW of 785 nm 5-min
A
B
Cl-CH2-CH2-S-CH2-CH3
A
B
Figure 4 A) SER and B) NR of HD Conditions A) 1mgml in tap water B) pure both 100 mW of 785 nm 1-min
Cl-CH2-CH2-S-CH2-CH2-Cl
SPIE-2003-5269 20
expected strong interaction between chlorine and silver and adds support to the assignment of this band to a C-Cl stretch Weaker overlapping bands occur at 670 692 and 724 cm-1 the latter possibly due to C-S stretching modes Again the alkane modes are apparent in the normal Raman spectra of HD but only a broad feature at 1300 to 1450 cm-1 suggests CH2 contributions in the SER spectrum Although the observed bands in the VX spectrum have not been assigned (Figure 4) a computer generated Raman spectrum29 predicts many of the same features with surprising accuracy and are used here Two intense bands at 460 and 530 cm-1 closely match predicted bands at 463 and 546 cm-1 assigned to a CH3-P=O bend and a PO2CS wag Three highly overlapped bands occur at 694 745 and 771 cm-1 matching predicted bands at 713 730 and 760 cm-1 The first
Table 2 Measured SER peak heights for the CN stretch at 2100 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
Figure 5 A) SERS and B) NR spectra of VX Conditions A) 1 vv (10 mgml) in MeOH B) pure sample both 100 mW of 785 nm 1-min acquisition time
A
B
two have been assigned to a C-S stretch and CH2bend respectively while the latter has been attributed to either a P-C stretch or an O-C-C stretch Although the 745 cm-1 band may alternatively be assigned to a C-S stretch based on the previous measurements of CEES and HD The relatively intense bands at 890 1106 1218 1445 and 1465 cm-1 also match predicted bands at 880 1108 1216 1440 and 1464 cm-1 that are assigned to a C-C stretch CH3 rock N-C3 stretch various C-H3 bends and C-H bends respectively Both the computer generated and the measured spectra contain numerous other less intense bands One is worth mentioning A unique band appears at 370 cm-1 that is predicted at 368 cm-1 and corresponds to an O-P=O bend The surface-enhanced Raman spectrum of VX is also rich with spectral features It has the unique low frequency band at 370 cm-1 as well as a second band at 380 cm-1 that is assigned tothe S-P-O bend predicted in the normal Raman spectrum at 388 cm-1 Based on the measured and predicted normal Raman spectra the following SERS assignments are given 460 cm-1 to the CH3-P=O bend 544 cm-1 to the PO2CS wag 738 cm-1 to a C-S stretch (based on arguments above) 890 cm-1
to a C-C stretch 1101 cm-1 to a CH3 rock and 1456
cm-1 to a C-H bend The ability of SERS to measure chemical agents in water containing real-world chemical interferents was tested by using the generic tap water described in Table 1 The ability to reproduce measurements was accomplished by preparing three separate water stock solutions which were used to prepare three sample solutions each of 1 mgmL KCN in the generic tap water The samples defined as vials 1-9 were then measured at 4 points per vial in a semi-random fashion such that errors associated with stock solution preparation and errors associated with instrument drift could be identified No trends were apparent that signified such systematic errors Each spectrum collected consisted of 20 averaged scans taking 16 seconds at 8 cm-1 resolution The laser power at the sample was measured periodically during the day and it ranged from 102 to 105 mW spacccc
spot Vial 1 Vial 2 Vial 41 13754 13004 128192 13519 12692 12909
1 3 13541 12721 126394 13462 12648 12651
ave 13569 12766 12755Vial 6 Vial 3 Vial 5
1 11586 14068 10757 2 11236 14402 115122 3 11376 14546 11559
4 10894 11763 11214ave 11273 13695 11261
Vial 7 Vial8 Vial 91 11198 15804 11023
2 11243 15716 88753 3 11562 15216 11218
4 11655 15084 11077ave 11415 15455 10548
CN AVG STDEV ERR1 mgmL pk ht 12526 1560 1245
stock solution
SPIE-2003-5269 21
The CN measurements consisting of 36 data points produced an average height of 125 for the 2100 cm-1 peak with a standard deviation of 156 or 125 (Table 2) The HD and VX measurements were performed precisely the same way (Tables 3 and 4) For HD the 624 cm-1 peak was used for analysis and it had an average height of 53 with a standard deviation of 068 or 129 while for VX the 544 cm-1 peak was used for analysis and it had an average height of 1051 with a standard deviation of 308 or 293 The greater error in the VX measurements can be somewhat attributed to Vial 4 which produced lower SER signal intensities But removing this vial from the data set changes the standard deviation to 233 only a modest improvement
spot Vial 1 Vial 2 Vial 41 525 609 4682 45 675 484
1 3 527 644 5094 559 807 669
ave 51525 68375 5325Vial 6 Vial 3 Vial 5
1 521 574 575 2 536 53 4492 3 509 508 451
4 565 594 379ave 53275 5515 4635
Vial 7 Vial 8 Vial 91 457 589 413
2 583 497 4273 3 544 505 4
4 528 5 509ave 528 52275 43725
HD AVG STDEV ERR1 mgmL pk ht 530 068 1291
stock solution
spot Vial 1 Vial 2 Vial 41 1464 1034 4672 1485 99 61
1 3 1491 989 5684 1041 777 553
ave 137025 9475 5495Vial 6 Vial 3 Vial 5
1 1058 942 1293 2 697 121 965
2 3 727 1206 11124 689 1254 746
ave 79275 1153 1029Vial 7 Vial 8 Vial 9
1 1127 783 163 2 1358 812 1614
3 3 1371 908 15124 1097 875 1397
ave 123825 8445 153825
VX AVG STDEV ERR1 mgmL pk ht 1051 308 2925
stock solution
4 CONCLUSIONS In this paper we examined the ability of surface-enhanced Raman spectroscopy to reproducibly measure CN HD and VX in tap water without chemical interference Both normal and surface-enhanced Raman spectra were examined to select unique bands suitable to identify and quantify these chemical agents For SER measurements the 2100 cm-1 C-N stretch was used for CN the 624 cm-1 C-Cl stretch was used for HD and the 544 cm-1 PO2CS wag was used for VX It was determined that 1 mgmL samples of each of these chemicals measured 36 times in glass vials coated with a silver-doped sol-gel reproduced measurements with standard deviations of 125 129 and 293 It was further found that the 10 chemicals added to simulate generic tap water did not interfere with or alter the SER spectra It should be noted that the concentrations used in this study were considerably greater than those required by the JSAWM program Current work involves improving SER sensitivity and designing sampling systems with better reproducibility This includes the development of fractal silver and gold structures within the sol-gel matrix and the development of chemically selective sol-gels
ACKNOWLEDGMENTS The authors are grateful for the support of the US Army (DAAD13-02-C-0015 Joint Service Agent Water Monitor program) The authors would also like to thank Janet Jensen Ronald Crosier and Kristina Gonser for helpful discussions
Table 4 Measured SER peak heights for the VX band at 544 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
Table 3 Measured SER peak heights for the HD band at 624 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
SPIE-2003-5269 22
REFERENCES 1 Jensen JL NC Wong W R Loerop SPIE 4775-03 (2001) 2 JSAWM Requirements at wwwsbccomapgeaarmymilRDAecbcrtPRODSERJSAWMjsawmhtml 3 Erickson B ldquoThe Chemical Weapons Convention Redefines Analytical Challengerdquo Anal Chem 70 397A (1998) 4 Johnston RL Hoefler CM Fargo JC and Moberley B ldquoThe Defense Nuclear Agencyrsquos
ChemicalBiochemical Weapons Agreements Verification Technology Research Development Test and Evaluation Program and its Requirements for On-Site Analysisrdquo AT-ONSITE 5-8 (1994)
5 Black RM Clarke RJ Read RW and Reid MT ldquoApplication of GC-MS and GC-tandem MS to the analysis of chemical warfare samples found to contain residues of the nerve agent sarin sulphur mustard and their degradation productsrdquo J Chrom 662 301-321 (1994)
6 USA Army Center for Health Promotion and Preventive Medicine (USACHPPM) Technical Guideline 230A 7 Hoffland LD Piffath RJ Bouck JB rdquoSpectral signatures of chemical agents and simulantsrdquo Optical
Engineering 24 982-984 (1985) 8 Braue EHJ Pannella MGrdquoCIRCLE CELL FT-IR Analysis of Chemical Warfare Agents in Aqueous Solutionsrdquo
App Spectrosc 44 1513-1520 (1990) 9 Tseng C-H Mann CK and Vickers TJ ldquoDetermination of Organics on Metal Surfaces by Raman
Spectroscopyrdquo App Spectrosc 47 1767-1771 (1993) 10 Staff Reporter ldquoUS Army Tests Lidar to Detect Biological Toxinsrdquo Photonic Spectra pg 50 December 1998 11 Christesen SD Raman cross sections of chemical agents and simulants App Spectrosc 42 318-321 (1988) 12 Weaver MJ Farquharson S Tadayyoni MA ldquoSurface-enhancement factors for Raman scattering at silver electrodesrdquo J Chem Phys 82 4867-4874 (1985) 13 Alak AM and Vo-Dinh T ldquoSERS of Organophosphorous Chemical Agentsrdquo Anal Chem 59 2149-2153 (1987) 14 Lee Y Farquharson S ldquoRapid chemical agent identification by SERSrdquo SPIE 4378 21-26 (2001) 15 Farquharson S Maksymiuk P Ong K Christesen S ldquoChemical agent identification by surface-enhanced Raman spectroscopyrdquo SPIE 4577 166-173 (2001) 16 Spencer KM Sylvia J Clauson S and Janni J ldquoSurface Enhanced Raman as a Water Monitor for Warfare
Agents in Waterrdquo SPIE 4577 158-165 (2001) 17 Tessier P Christesen S Ong K Clemente E Lenhoff A Kaler E Velev O ldquoOn-line spectroscopic characterization of sodium cyanide with nanostructured gold surface-enhanced Raman spectroscopy substratesrdquo
App Spectrosc 56 1524-1530 (2002) 18 Farquharson S W W Smith S Elliott and J F Sperry Rapid biological agent identification by surface- enhanced Raman spectroscopy SPIE 3855110-116 (1999) 19 Farquharson S WW Smith YH Lee S Elliott and J F Sperry Detection of biological signatures A
comparison of electrolytic and metal-doped sol-gel surface-enhanced Raman media SPIE 4575 62-72 (2002) 20 Guzelian AA Sylvia J Janni J Clauson S and Spencer KM ldquoSERS of whole cell bacteria and trace levels of biological moleculesrdquo SPIE 4577 182-192 (2001) 21 Shende C Inscore F Gift A Maksymiuk P Farquharson S ldquoRapid dipicolinic acid extraction from Bacillus spores detected by surface-enhanced Raman spectroscopyrdquo App Spectrosc 58 accepted 22 Lee Y and Farquharson S ldquoSERS Sample Vials Based on Sol-Gel Process for Trace Pesticide Analysisrdquo SPIE
4206 140-146 (2000) 23 Farquharson S and Lee Y ldquoTrace Drug Analysis by SERSrdquo SPIE 4200-16 (2000) 24 Lee Y Farquharson S and Rainey P M Surface-Enhanced Raman Sensor for Trace Chemical Detection in
Water SPIE 3857 76-84 (1999) 25 Lee Y Farquharson S Kwong H and Shahriari M ldquoSol-Gel Chemical Sensor for Surface-Enhanced Raman
Spectroscopyrdquo SPIE 3537 252-260 (1998) 26 Farquharson S A Gift P Maksymiuk F Inscore and W Smith ldquopH dependence of methyl phosphonic acid
dipicolinic acid and cyanide by surface-enhanced Raman spectroscopyrdquo SPIE 5269-19 (2003) 27 Lide DR Ed Handbook of Chemistry and Physics CRC Press 77th Ed (1996-7) 28 Sosa C RJ Bartlett K KuBulat and WB Person ldquoA theoretical study of harmonic vibrational frequencies and
infrared intensities of XCH2CH2SCH2CH2X and XCH2CH2SH (= H Cl)rdquo J Phys Chem 93 577-588 (1993) 29 Hameka HF and JO Jenson ldquoComputer-generated predictions of the structure and the IR and Raman spectra of
VXrdquo ERDEC-TR-065 May 1993
- Chemagents Appendicespdf
-
- SPIE2001-4575-Bioagent-SERSapdf
-
- Detection of bioagent signatures A comparison of electrolytic and metal-doped sol-gel surface-enhanced Raman media
- Stuart Farquharson Wayne Smith and Yuan Lee
-
- Real-Time Analyzers 87 Church Street East Hartford CT 06108
-
- ABSTRACT
- 4 CONCLUSIONS
- 5 ACKNOWLEDGEMENTS
- 6 REFERENCES
-
SPIE-4378-2001
22
absorption of water would obscure most other chemicals present Nevertheless efforts to overcome these limitations have been demonstrated Braue and Pannella13 quantified the G-series nerve agents (tabun sarin and soman) in terms of infrared attenuated total reflectance using a circle-cell And Alak and Vo-Dinh demonstrated the possibility of surface-enhanced Raman spectroscopy (SERS) to identify CWAs by measuring several organophosphonates that simulate the nerve agents17 However quantitative measurements have not been demonstrated for the SER-active material used (silver coated on alumina particles) or other SER-active media18 Recently we developed silver-doped sol-gels to promote the SER effect19-22 The porous silica network of the sol-gel matrix offers a unique environment for stabilizing SER-active metal particles and the sol-gel provides a high surface area that effectively increases the number of molecules observed within the Raman scattering volume The silver-doped sol-gels have been coated on the internal walls of standard glass vials to allow rapid SER analysis of small sample quantities (lt 01 mL) without preparation We have used p-aminobenzoic acid (PABA) as a test chemical to demonstrate surface enhancements greater than 107 reversible measurements in a flowing system reproducible measurements from vial-to-vial and batch-to-batch and measurements in multiple solvents including water19-22 Here we present preliminary measurements of chemical agent simulants in both the liquid and gas phases as well as chemical agent hydrolysis products using our SER-active vials
2 EXPERIMENTAL The chemical agent simulants employed were obtained at their purest commercially available grade from Aldrich (Milwaukee WI) and were dissolved in water or methanol for analysis All chemicals used to prepare the silver-doped sol-gels were spectroscopic grade and also purchased from Aldrich The sol-gel vials were coated in a manner similar to that previously reported by adding ammonium hydroxide to a solution of silver nitrate tetramethyl orthosilicate and methanol22 After mixing 02 mL of the sol-gel solution was transferred into a glass vial (2 mL) dried and heated The incorporated silver ions were then reduced using dilute sodium borohydride The vials were washed and dried prior to the addition of a sample solution The patent pending SER-active vials are commercially available from Real-Time Analyzers (Simple SERS Sample Vials RTA East Hartford CT) Dimethyl metylphosphonate (DMMP) pinacolyl methylphosphonate (PMP) and methylphosphonic acid (MPA) were prepared in aqueous solution while 2-chloroethyl ethyl sulfide (CEES) was prepared in methanol at 1 mM for SERS measurements Neat samples were employed for normal Raman measurements All samples were prepared in a chemical hood and transferred into plain or SER-active vials for analysis Special precaution was followed for CEES since it is a severe blistering agent23 Once prepared the vial was placed into the sample compartment of a Raman spectrometer for analysis A prototype Fourier transform Raman spectrometer (RTA) was used for these measurements24 The system consisted of a NdYAG laser (Brimrose) for excitation at 1064 nm an interferometer built by On-Line Technologies (OLT East Hartford CT) for frequency separation an uncooled InGaAs detector for signal detection (RTA) and an Intel 400 MHz Pentium II based laptop computer (Dell Round Rock TX) for interferometric control data acquisition (OLT) and analysis (LabVIEW by National Instruments Austin TX) Additional components included a Notch filter (Kaiser Ann Arbor MI) and interferometer entrance and exit optics (Edmund Scientific Barrington NJ) Fiber optics were used to deliver the excitation beam to the sample and the scattered radiation to the interferometer (1 meter lengths of 200 and 365 micron core diameter respectively Spectran Avon CT) A second Notch filter (Kaiser) was used as a beam splitter to direct the excitation beam along the same axis as the collected radiation A microscope object (20x04 Newport Irvine CA) was used to focus the beam into the sample and to collect the scattered radiation back along the same axis In this co-axial backscattering arrangement the excitation beam was passed through the outside of a glass vial and focused onto the silver-doped sol-gel film (01 mm thickness) containing the sample
3 RESULTS AND DISCUSSION As a prelude to chemical agent measurements in water we evaluated the quantitative performance of the SER-active vials by measuring PABA over the concentration range from 10-7 M to 10-2 M Figure 1 shows the spectra for 7 35 and 70 micromolar concentrations while Figure 2 shows a plot of the 1450 cm-1 band intensity as a function of concentration The SER response is linear over nearly three orders of magnitude to just over 10-4M at which point the band intensity suggests that the silver surface is becoming saturated
SPIE-4378-2001
23
In an effort to demonstrate the broad capabilities of the SER-active vials to measure chemical agents spectra of a nerve agent simulant dimethyl methylphosphonate a mustard gas simulant 2-chloroethyl ethyl sulfide and hydrolysis products pinacolyl methylphosphonate and methylphosphonic acid were collected DMMP is widely used by the US Army as a chemical warfare simulant because its chemical structure volatility and water solubility are similar to those of nerve agents25 DMMP is completely miscible and stable in water at room temperature26 Figure 3 compares the SER spectrum to the normal Raman spectrum of DMMP A number of the normal Raman bands are SER-active such as the P-C stretching mode which shifts from 715 to 735 cm-1 and the C-H stretching modes at 2855 2930 2960 and 3000 cm-1 which shift slightly Surprisingly the P=O stretching mode at 1250 cm-1 virtually disappears However the most dramatic change is the appearance of an intense triplet in the SER spectrum near 1000 cm-1 The bands at 1000 cm-1 1030 cm-1 and 1075 cm-1 likely involved the P-O-C bond This is supported by the nearly identical triplets observed for the SER spectra of fonofos and fonofoxon1719 It is also worth noting that a band appears at 425 cm-1 in the SER spectrum that may be unique to DMMP and useful for identification The enhancement factor is estimated at 120000 based on the normal Raman and SER P-C band intensity taking into account the difference in sample concentrations and spectral acquisition conditions A detection limit based on a signal-to-noise ratio of 3 can be estimated at 16 ppm
10-7 10-6 10-5 10-4 10-3 10-2 10-110-2
10-1
100
101
102
Figure 1 SER spectra of A) 70 B) 35 and C) 7 micromolar p-amino benzoic acid in water Conditions 80 mW of 1064 nm laser excitation 100 averaged scans (15 min) at 8 cm-1 resolution
Figure 2 SER spectral intensity for p-aminobenzoic acid as a function of concentration using RTA SER-active vials
Concentration (M)
Figure 3 A) SER and B) normal Raman spectra of dimethyl methylphosphonate Conditions SERS as in Figure 1 normal Raman 500 mW and 200 scans
Figure 4 A) SER and B) normal Raman spectra of 2-chloroethyl ethyl sulfide Conditions as in Figure 3
Wavenumbers (∆cm-1)
CH3O-P-OCH3
=
O
_
CH3
CH3O-P-OCH3
=
O
_
CH3
Cl-CH2-CH2-S-CH2-CH3 A
B
A
B
A
B
C
Wavenumbers (∆cm-1)
Wavenumbers (∆cm-1)
500 1000 1500 2000
SPIE-4378-2001
24
2-Chloroethyl ethyl sulfide a blister agent simulant has a chemical structure similar to the mustard gas (Cl-CH2-CH2-S-CH2-CH2-Cl) with only one terminal chlorine Due to its low solubility in water CEES was dissolved in methanol for the SER measurement Again the prominent Raman modes are SER-active and even maintain relative intensity (Figure 4) The primary difference is that the SER bands appear to broaden such that the triplet near 700 cm-1 becomes a doublet and the shoulders at 2875 and 2970 cm-1 become less defined Again the latter bands are assigned to C-H stretching modes A single band at 700 cm-1 which is attributed to the C-S-C asymmetric stretch dominates the reported infrared spectrum of mustard gas12 A corresponding symmetric stretch is reported at 705 cm-1 in the Raman spectrum of mustard gas27 Here a corresponding symmetric stretch appears but as a doublet at 700 and 755 cm-1 presumably due to the loss in symmetry for CEES The band at 655cm-1 can also be confidently assigned to a C-Cl stretch The SER spectral bands at 620 and 730 cm-1 are probably due to the same modes ie C-Cl and C-S-C stretches respectively The enhancement factor for CEES was somewhat less than DMMP at approximately 62000 as is the estimated detection limit of 22 ppm The ability to rapidly detect trace quantities of chemical agents in the gas phase would be invaluable as an early warning system Although the Raman scattering cross-sections for the nerve agents suggest that remote detection by Raman-based LIDAR is unlikely16 a SER-based system for perimeter monitoring could prove successful As a preliminary measurement we prepared a 10 by volume solution of CEES in methanol exposed a SER-active vial to the equilibrium vapor phase in a sealed jar and monitored the SER spectrum as a function of time Initially the vial was removed through a transfer chamber every hour to record the SER spectrum After ten hours spectra were recorded only every ten hours As illustrated by Figure 5 the sol-gel performed as a dosimeter in that the spectra increased as a function of exposure time The most intense SER bands at 620 and 2930 cm-1 are discernable in the first few hours The spectrum after 40 hours is nearly identical to the solution phase spectrum except for a diminished intensity of the 730 cm-1 band This may be due to methanol solvation effects or surface-orientation effects Based on the relative concentrations of methanol and CEES and their partial pressures we estimate the equilibrium concentration of CEES to between 1 and 2 micromolar Although not shown this concentration could be detected in one hour As previously stated the analysis of chemical agents in water is important in identifying poisoned water It is also important to decommissioning activities in which agents are destroyed by hydrolysis (acid or base) Furthermore any analytical technique used must be capable of distinguishing between parent CWA and hydrolysis products to assess safety or effectiveness of decommissioning For example soman has a hydrolysis half-life of ~23 hours at ambient temperatures and neutral pH28 and forms hydrofluoric acid (somewhat toxic) and pinacolyl methylphosphonate (relatively non-toxic)2930 PMP further hydrolyzes to form methyl phosphonic acid and 33-dimethyl-2-butanol (both non-toxic) The structural similarities between soman PMP and MPA are expected to produce similar Raman as well as SER spectra Figure 6 compares PMP and MPA but not the highly toxic parent CWA soman As with DMMP the P-C stretch the P-O-C mode and C-H stretches are readily apparent Yet it is worth noting that the band positions are reasonably different The former two bands appear at 764 and 1042 cm-1 for MPA while they are at 788 and 1032 cm-1 for PMP More importantly a unique band at 546 cm-1 as yet unassigned appears in the PMP spectrum
Figure 6 SER spectra of A) methyl phosphonic acid and B) pinacolyl methylphosphanate (note unique band at 546 cm-1) Spectral conditions as in Figure 1
Figure 5 SER spectra of 2-chloroethyl ethyl sulfide vapor as a function of time (10 hour increments to top which is 40 hours) Bottom trace is a blank Spectral conditions as in Figure 1
A
B
Wavenumbers (∆cm-1) Wavenumbers (∆cm-1)
HO-P-OH
=
O
CH3
_
HO-P-OH
=
O
=
O
CH3
_
CH3
_HO-P-O-CH-C-CH3
=O
CH3
_
CH3
_CH3_CH3
_
HO-P-O-CH-C-CH3
=O
=O
CH3
_
CH3
_
CH3
_
CH3
_CH3_CH3
_CH3
_
SPIE-4378-2001
25
4 CONCLUSIONS We have successfully measured the SER spectra of chemical agent simulants dimethyl metylphosphonate and 2-chloroethyl ethyl sulfide and chemical agent hydrolysis products pinacolyl methylphosphonate and methylphosphonic acid using silver-doped sol-gel coated sample vials Measurements were obtained in both aqueous and gas phase The P-C stretching mode was SER-active for all four chemicals allowing identification by class Within this group each chemical contained at least one unique spectral band that could be used for identification (Table 1) Furthermore these bands do not appear to coincide with SER spectra reported for organophosphorus pesticides the most likely source of false-positives Although surface enhancement factors appear to be an order of magnitude better than those previously presented in the literature for similar chemicals17 measurement sensitivity needs to be improved by 1 to 2 orders of magnitude to provide adequate warning of chemical agent use Current research efforts to increase surface-enhancement optical collection efficiency and instrument design are being pursued to achieve the required sensitivity
Table 1 Enhancement factors detection limits and unique SER bands fro chemicals studied Agent Simulant Enhancement Detection limit Unique bands (cm-1)
Dimethyl methylphosphonate 120000 90 microM (16 ppm) 425 2-Chloroethyl ethyl sulfide 62000 60 microM (22 ppm) 620
Methylphosphonic acid 110000 3 microM (60 ppb) 764 1042 Pinacolyl methylphosphonate 150000 70 microM (14 ppm) 546 788 1032
5 ACKNOWLEDGEMENTS The authors would like to thank Drs Janet Jensen and Steven Christesen of Aberdeen Proving Ground for encouraging this work They would also like to thank Advanced Fuel Research for making their laboratory facilities available
6 REFERENCES 1 ldquoThe Chemical Weapons Convention ndash A Guided Tour the Organization for the Prohibition of Chemical Weaponsrdquo at
httpwwwopcwnlguidehtm 2 Robinson JP and J Goldblat Chemical Warfare In The Iraq-Iran War Stockholm International Peace Research
Institute Fact Sheet at httpprojectssiprisecbwresearchfactsheet-1984html (1984) 3 ldquoChemistry of GB (Sarin)rdquo at httpwwwmitretekorgmissionenvenechemicalagentssarinhtml 4 Tu Anthony ldquoOverview of Sarin Terrorist Incidents in Japan in 1994 and 1995rdquo 6th CBW Protection Symposium
Stockholm Sweden 10-15 May 1998 5 Staff Reporter ldquoGoing out with a bangrdquo Newsweek June 28 1999 6 See UNSCOM reports in httpwwwunorgdeptsunscom (1999) 7 Treven T Saddamrsquos Secrets Harper Collins (1999) 8 ldquoDecaying Sarin-filled Rockets Spark Fearsrdquo Janersquos Defense Weekly 25(20)3 (1996) 9 ldquoThe Chemical Weapons Convention Redefines Analytical Challengerdquo Analytical Chemistry News amp Features June 1
397A (1998) 10 Johnston RL Hoefler CM Fargo JC and Moberley B ldquoThe Defense Nuclear Agencyrsquos ChemicalBiochemical
Weapons Agreements Verification Technology Research Development Test and Evaluation Program and its Requirements for On-Site Analysisrdquo AT-ONSITE 5-8 (1994)
11 Black RM Clarke RJ Read RW and Reid MT ldquoApplication of gas chromatography-mass spectrometry and gas chromatography-tandem mass spectrometry to the analysis of chemical warfare samples found to contain residues of the nerve agent sarin sulphur mustard and their degradation productsrdquo J Chromatography 662 301-321 (1994)
12 Hoffland LD Piffath RJ Bouck JBrdquoSpectral signatures of chemical agents and simulantsrdquo Optical Engineering 24 982-984 (1985)
13 Braue EHJ Pannella MGrdquoCIRCLE CELL FT-IR Analysis of Chemical Warfare Agents in Aqueous Solutionsrdquo Applied Spectroscopy 44 1513-1520 (1990)
SPIE-4378-2001
26
14 Tseng C-H Mann CK and Vickers TJ ldquoDetermination of Organics on Metal Surfaces by Raman Spectroscopyrdquo
Applied Spectroscopy 47 1767-1771 (1993) 15 Staff Reporter ldquoUS Army Tests Lidar to Detect Biological Toxinsrdquo Photonic Spectra pg 50 December 1998 16 Christesen SD Raman cross sections of chemical agents and simulants Applied Spectroscopy 42 318-321 (1988) 17 Alak AM and Vo-Dinh T ldquoSERS of Organophosphorous Chemical Agentsrdquo Analytical Chemistry 59 2149-2153
(1987) 18 Norrod KL Sudnik LM Rousell D and Rowlen KL ldquoQuantitative Comparison of Five SERS Substrates
Sensitivity and Detection Limitrdquo Applied Spectroscopy 51 994-1001 (1997) 19 Lee Y and Farquharson S ldquoSERS Sample Vials Based on Sol-Gel Process for Trace Pesticide Analysisrdquo SPIE 4206
140-146 (2000) 20 Farquharson S and Lee Y ldquoTrace Drug Analysis by Surface-Enhanced Raman Spectroscopyrdquo SPIE 4200-16 (2000) 21 Lee Y Farquharson S and Rainey P M Surface-Enhanced Raman Sensor for Trace Chemical Detection in Water
SPIE 3857 76-84 (1999) 22 Lee Y Farquharson S Kwong H and Shahriari M ldquoSurface-Enhanced Raman Sensor for Surface-Enhanced Raman
Spectroscopyrdquo SPIE 3537 252-260 (1998) 23 see Material Safety Data Sheets for details 24 Farquharson S Smith W Carangelo R C and Brouillette C ldquoIndustrial Raman Providing Easy Immediate Cost
Effective Chemical Analysis Anywhererdquo SPIE 3859 14-23 (1999) 25 Bennett S Bane J Benford P and Pratt R ldquoEnvironmental Hazards of Chemical Agent Simulantsrdquo Aberdeen
Proving Ground Maryland Chemical Research and Development Center CRDC-TR-84055 (1984) 26 Mabey W and Mill T Critical Review of Hydrolysis of Organic Compounds in Water under Environmental Conditions
Journal of Physics and Chemistry Reference Data 7(2) 383-414 (1978) 27 Christesen S MacIver B Procell L Sorrick D Carabba M and Bello J ldquo Noninstrusive Analysis of Chemical Agent
Identification Sets Using a Portable Fiber-Optic Raman Spectrometerrdquo Applied Spectroscopy 53 850-855 (1999) 28 Meylan WM and Howard PH J Pharm Sci 84 83-92 (1995) 29 Jenkins A Uy O and Murray G ldquoPolymer-Based Lanthanide Luminescent Sensor for Detection of Hydrolysis Product
of the Nerve Agent Soman in Waterrdquo Analytical Chemistry 71 373-378 (1999) 30 Nassar A Lucas S and Hoffland L ldquoDetermination of Chemical Warfare Agent Degradation Products at Low-Part-
per-Billion Levels in Aqueous Samples and Sub-Part-per-Million Levels in Soils Using Capillary Electrophoresisrdquo Analytical Chemistry 71 1285-1292 (1999)
SPIE 2001-4575
62
Detection of bioagent signatures A comparison of electrolytic and metal-doped sol-gel surface-enhanced Raman media
Stuart Farquharson Wayne Smith and Yuan Lee
Real-Time Analyzers 87 Church Street East Hartford CT 06108
Susan Elliott and Jay F Sperry University of Rhode Island 45 Lower College Rd Kingston RI 02881
ABSTRACT Since September 11 2001 the threat of terrorist attacks and biological warfare within US borders has become a sobering reality In an effort to aid military personnel and the public at large we have been investigating the utility of surface-enhanced Raman spectroscopy (SERS) to provide rapid identification of chemical agents directly and biological agents through their chemical signatures This approach is based on the ability of Raman spectroscopy to identify molecular structure through the abundant vibration information provided in spectra and the ability of SERS to detect extremely low concentrations (eg part-per-billion) through the enhancement of Raman scattering by six orders of magnitude or more Towards the goal of developing a portable analyzer we have been studying the ability of two SER media to obtain continuous (ie reversible) and quantitative (ie reproducible) measurements Here we compare measurements of nucleic acid-bases adenosine monophosphate and ribonucleic acid extracted from Escherichia coli Bacillus subtilis and Staphylococcus aureus obtained by electrolytic SERS and metal-doped sol-gel SERS The capabilities of these SER media are summarized in terms of rapid detection of B anthracis and dipicolinic acid Keywords bioagent detection SERS RNA analysis bacterial analysis Raman spectroscopy
1 INTRODUCTION The recent distribution of anthrax through the US postal system and the subsequent infection and death of several postal and national media employees amplifies the need for methods to rapidly detect and identify this and other chemical and biological warfare agents (BWA) The primary methods currently used immunoassays for screening and nucleic acid (NA) sequencing for positive identification of BWAs (bacteria protozoa and viruses) have serious limitations123 Immunoassay methods employ competitive binding of the bioagent (as an antigen) and its labeled (eg fluorescence) conjugate for a limited number of antibodies Although this analysis method is fast and semi-quantitative other chemicals may compete for the antibodies interfere with the enzymatic reaction or interfere with the measurement (eg it fluoresces) resulting in a high number of false positive responses1 Furthermore the antibodies denature due to moisture and heat limiting shelf life and require sterile often refrigerated storage Positive identification of a BWA can be accomplished by sequencing deoxyribonucleic acid or ribonucleic acid (DNA and RNA)23 This requires enumeration of a nucleic acid sequence through polymerase chain reactions (PCR) or multiplication of the microorganism through culture growth to provide sufficient quantities of DNA or RNA for analysis Unfortunately PCR and culture growth require from several hours to several days23 Consequently a wide variety of technologies have been investigated for rapid identification of BWAs The Department of Defense is actively monitoring 200 such technologies4 This includes traditional methods such as gas chromatographic separation coupled with ion mobility spectrometry detection5 to exotic methods based on nature such as monitoring toxin induced color changes in fish scales6 Although all of these techniques have achieved varying degrees of success none are yet capable of detecting and identifying BWAs in 10 minutes or less Towards this goal we have been investigating the ability of SERS to detect sub-nanogram quantities of DNA or RNA (eliminating enumeration) determine relative NA base concentrations and identify BWA taxonomy
To whom correspondences should be addresses e-mailfarqureal-time-analyzerscom wwwreal-time-analyzerscom
SPIE 2001-4575
63
Raman spectroscopy has a rich history of investigating biochemical and biological processes7 Some of the earliest laser-Raman studies demonstrated that the five NA bases adenine (A) cytosine (C) guanine (G) thymine (T in DNA) and uracil (U in RNA) yielded distinct spectra with several bands suitable for identification and quantification8 Furthermore these studies included exceptional spectra of both DNA and RNA for which the NA bases as well as several phosphate bands were easily identified9 However since the Raman effect is very inefficient (very low conversion of incident radiation to inelastically scattered Raman radiation) these samples had to be highly concentrated Fortunately two phenomena exist that can increase the generation of Raman photons by six orders of magnitude or more known as the resonance Raman and surface-enhanced Raman effects1011 Resonance Raman scattering occurs when the laser excitation wavelength is in resonance with an electronic transition of a molecule (within the absorption envelope)10 Excitation at ultraviolet wavelengths has been used to obtain resonance Raman spectra of amino acids and nucleic acids in whole bacteria1213 For example excitation at 242 nm has been used to maximize the nucleic acid spectral band intensities and minimize the amino acids band intensities A peak at 1530 cm-1 was found to be proportional to the amount of the NA base cytosine while a peak at 1485 cm-1 was proportional to the combined amount of the NA bases adenine and guanine This quantitative behavior has been used to define an A+TG+C base-pair ratio and provide a level of bacterial identification as taxonomic markers13 In recent years SERS has also been used to analyze bacterial cell components14 including amino acids15 lipids16 nucleic acids151718 and the adenine derivatives192021 SERS has proven to be one of the most sensitive methods for trace chemical analysis through the detection of single molecules2223 including DNA (dye labeled 17-mer)24 Since its discovery in 197425 the mechanism responsible for the large increase in scattering efficiency has been the subject of considerable research2627 Briefly incident laser photons couple to free conducting electrons within a metal which confined by the particle surface collectively cause the electron cloud to resonate2628 These surface plasmons are known as the physical component of the SER effect These surface plasmons can transfer energy to the molecular vibrational modes of molecules through interactions with the molecular electron orbitals2629 This interaction is known as the chemical component of the SER effect This perturbation of the molecular polarizability generates surface-enhanced Raman photons26 A number of methods have been developed to produce surfaces or solutions containing one of these metals with optimum roughness or diameter to promote SERS30 These methods include preparation of activated electrodes in electrolytic cells 1131 activated silver and gold colloid reagents32 and metal coated substrates333435 Selecting a SER-active medium for chemical and biological agent detection requires consideration of the method of deployment and hence the method of sampling Chemical aerosols or airborne bacteria will require a collection device to concentrate and continuously present the sample to the SERS medium Poisoned water supplies will also require a flow through device for continuous monitoring or a grab-sample device for periodic analysis And contaminated surfaces will require a grab-sample extractive device A SERS-based device used for continuous monitoring (air or water) must be reversible and reproducible if quantitative measurements are desired while a SERS-based device used for periodic sampling (water or surfaces) must be reproducible Both reversible and reproducible measurements have been performed using electrolytic SERS (E-SERS)36 But this requires a three-electrode sample cell and an electrolyte of known concentration to perform the necessary oxidation-reduction cycles (ORCs) to re-activate the electrode surface with new uncontaminated sites from one measurement to the next Colloids are severely limited in that continuous measurements would require a continuous supply of colloids For periodic measurements vials of colloids one per measurement could be used However aggregate size and consequently SER intensity change with sample conditions (especially pH) and quantitative reproducible measurements are unlikely Substrates appear to have the greatest potential and designs range from silver evaporated on titania particles34 to periodic gold pyramids evaporated between polystyrene beads35 Most substrates require concentrating the sample on the surface through drying to obtain the largest signal enhancements in effect making the measurements irreproducible and irreversible However successful measurements using flow systems have been obtained with glass posts but manufacturing costs appear prohibitive In an effort to overcome these limitations we have developed metal-doped sol-gels to provide SERS measurements that are reproducible reversible and quantitative and yet not restricted to specific environments such as electrolytes solvents or evaporated surfaces3738 The porous silica network of the sol-gel offers a unique environment for stabilizing SER active metal particles and the high surface area increases the interaction between the analyte and metal particles The sol-gel can be coated on the end of fiber optics or on the internal walls of a glass flow tube for continuous measurements or standard glass sample vials for periodic measurements Previously we measured 100 mgL methylphosphonic acid (the primary hydrolysis product of nerve agents) in water with an estimated detection limit of 05 mgL (100 parts-per-billion) We have also
SPIE 2001-4575
64
demonstrated reversible and reproducible measurements of p-aminobenzoic acid (PABA) in a flow through system Here we investigate the ability of the sol-gel SERS (SG-SERS) to measure the NA bases adenosine monophosphate and RNA extracted from E coli B subtilis and S aureus The measurements are compared to those obtained by E-SERS
2 EXPERIMENTAL The inorganic chemicals and solvents used to prepare samples were spectroscopic grade and purchased from Aldrich (Milwaukee WI) Fisher (Pittsburgh PA) or Pfaltz amp Bauer (Waterbury CT) The nucleic acid bases and dipicolinic acid were purchased from Sigma (St Louis MO) Normal Raman samples were measured to establish enhancement factors In each case 1cm3 of sample was placed into a 1x1 cm glass cuvette weighed and measured Unpacked densities were typically 6-7 gcm3 For all SER measurements including RNA samples were prepared as ~01mgmL (see Figure captions for exact concentrations) in 01M KCl and buffered to a pH of 92 with Na2B4O7bullH2O Adenine pH dependence measurements used pH buffer solutions at 4 (potassium acid phthalate) 69 (potassium phosphate monobasicsodium phosphate dibasic) 92 (Na2B4O7bullH2O) and 104 (tris-hydroxymethyl amino methane) Escherichia coli Bacillus subtilis and Staphylococcus aureus cultures (250ml per 1000mL Erlenmeyer flask) were grown overnight in a Trypticase soy broth (TSB) medium containing 1 glucose in a shaking water bath at 37 oC The bacteria were harvested by centrifugation for 10 minutes at 8000 rpm in a GSA rotor at 5degC then washed once in 085 saline The gram-positive bacteria were concentrated to 20 ml and passed through a French pressure cell twice at 15000 psi to break open the cells RNA was extracted according to Protocol 44139 to ensure pristine samples for initial measurements Since this method takes approximately 4 hours a streamlined method was developed For vegetative bacteria the specimen was boiled for 30 sec in 1 ml of distilled water to lyse the cells and release the RNA For bacterial spores the specimen was first incubated in 1 ml of saline solution containing 02 mg lysozyme and phosphate-buffered to pH of 624 for 1 hr at 37 oC This solution was then boiled for 2-3 minutes in 4X loading buffer to release the RNA For both specimens RNA STAT-60TM was added to the supernatant which was centrifuged at 12000 g for 5 minutes to precipitate the ~15 water-soluble proteins This procedure allowed extracting RNA for SER analysis in ~ 10 minutes Electrophoresis shows high purity while the existence of chemicals that could interfere with the SER measurements is still under investigation The electrolytic sample cell has been described previously36 Briefly a three electrode design is incorporated into a Plexiglas structure containing a well for the reference electrode (a saturated calomel electrode Cole Parmer Vernon Hills IL) and a 5mL sample well containing the silver working electrode and platinum wire counter electrode (05 mm wire Alfa Ward Hill MA) A channel connecting the two wells contained a 2 mm diameter semi-porous membrane (10-20 micron pore Ace Glass) The silver electrode was made from a 3 mm length of 2 mm diameter silver wire (Alfa) soldered to a copper wire lead encased in a 4 mm diameter Pyrex tube A cap containing the silver electrode platinum wire and nitrogen purge and vent lines fixed the silver electrode surface 1 mm from a 1 mm thick glass plate attached to the bottom of the sample well The potentiostat used to control the three electrodes was built in-house and has been described in detail elsewhere36 A multifuntional analog digital and timing inputoutput interface card (DAQCard-1200 National Instruments) is used to both drive the electrolytic cell as well as read the current generated in the cell A LabVIEW software program is used set the oxidation potential reduction potential SER measurement potential hold times and sweep rates The amount of charge passed was plotted as a cyclic voltammogram For all spectra presented five oxidation-reduction cycles (ORCs) stepping from -03 VSCE to 03 VSCE and back to -3 VSCE at 50 mVsec were used The SG-SER measurements were accomplished by simply placing the identical samples prepared above into Simple SERS Sample VialsTM (RTA) These 2-mL glass vials are internally coated with ~ 01 micron thick silver-doped sol-gel A prototype Fourier transform Raman spectrometer (RTA) was used for these measurements40 The system consisted of a NdYAG laser (Brimrose or Spectra Physics) for excitation at 1064 nm an interferometer built by On-Line Technologies (OLT East Hartford CT) for frequency separation an uncooled InGaAs detector for signal detection (RTA) and an Intel 400 MHz Pentium II based laptop computer (Dell Round Rock TX) for interferometric control data acquisition (OLT) and analysis (LabVIEW by National Instruments Austin TX) Additional components included a Notch filter (Kaiser Ann Arbor MI) and interferometer entrance and exit optics (Edmund Scientific Barrington NJ) Fiber optics were used to deliver the excitation beam to the sample and the scattered radiation to the interferometer (2 meter lengths of 200 and 365 micron core diameter respectively Spectran Avon CT) A second Notch filter (Kaiser) was used as a beam splitter to direct the excitation beam along the same axis as the collected radiation A microscope object (20x magnification 04 numeric aperture Newport Irvine CA) was used to focus the beam into the sample and to collect the scattered radiation back along the same axis In this co-axial backscattering arrangement the excitation beam passed through the glass plate onto the silver
SPIE 2001-4575
65
electrode surface for E-SERS through the vial glass wall and into the silver-doped sol-gel film for SG-SERS or through the glass wall of the cuvette and into the solid sample for normal Raman spectroscopy All E-SERS and normal Raman spectra were obtained with 750 mW of laser power at the sample while all SG-SERS spectra were obtained with 75 mW of laser power at the system Incident powers above 200 mW in some cases degraded the sol-gel
3 RESULTS AND DISCUSSION The generation of surface-enhanced Raman scattering at electrode surfaces has been extensively researched and the optimum sample conditions are well developed2729 Several researches incorporated electrodes into flowing systems and demonstrated that continuous monitoring of chemicals is possible18 These successes suggested investigated the capability of E-SERS to measure the NA bases and RNA The E-SERS measurements also provided a benchmark to compare and evaluate SG-SERS measurements The molecular structure of adenine (as well as the other base pairs) which includes an aromatic nitrogen-containing heterocycle is ideally suited to interact with the surface plasmons and contribute substantially to the chemical component of the SER effect1119 Even with excitation at 1064 nm a 3-minute scan of 18x10-5M adenine yields high signal-to-noise (SN) E-SER spectra and all of the bands are revealed with clarity (Figure 1 Table 1) Spectra of similar quality were obtained by SG-SERS and the principal spectral bands are easily observed The identical 18x10-5M adenine sample was measured in the same 3-minute time frame but with 110th the laser power The lower power appears to reduce the SN The amount of adenine responsible for the SER spectra as well as enhancement factors for the two SER media can be determined The molecules producing the E-SERS spectrum are those on the electrode surface within the illumination area of the laser (The solution concentration only determines the number of molecules available to adsorb to the electrode surface) For the current experiments the laser illuminates an area of 28x10-7m2 or 56 x10-7m2 if we assume the ORCs increase the surface area by a factor of two Furthermore if we assume monolayer coverage on the electrode and each 3x5 angstrom molecule (lying flat) occupies 15x10-19m2 then there are ca 4x1012 molecules contributing to the Raman scattering This is ca twice the number of molecules measured at electrode surfaces using either differential capacitance-potential curve measurements or rapid linear sweep voltammetry (eg 3x1018 moleculesm2 for pyridine and pyrazine)29 Thus the adenine spectrum in Figure 1 is due to 87x10-10g (6x10-12 moles) A detection limit defined as a SN of 3 can also be calculated The SN for a 3-minute scan is 844 for the 735 cm-1 band suggesting a mass detection limit of 3x10-12g (2x10-14 moles) This is consistent with previous estimates for adenine by others of 25 x10-14 moles1530 However sub-monolayer concentrations must be measured to verify this The root-mean-squared (RMS) noise is measured between 4400-4600 cm-1 Since noise is distributed evenly throughout the spectrum when transformed this region does not have any
Wavenumbers (∆cm-1) Figure 1 A) Normal Raman spectrum of pure adenine powder B) E-SERS and C) SG-SERS of 18x10-5M adenine at pH 92 All spectra 8 cm-1 resolution 200 scans (3 min) and 1064 nm excitation A) and B) 750 mW C) 75 mW B) measurement potential of 11VSCE
A
B
C
A
B
C SG-SERS
pH 10
D
725
735
735
pH 4
Wavenumbers (∆cm-1) Figure 2 A) and C) E-SERS and B) and D) SG-SERS of adenine at A) and B) pH 104 and C) and D) pH 40 Note consistent appearance of bands at 1270 and 1375 cm-1 as the pH is changed to 10 for both SER media E-SERS used 750 mW SG-SERS used 75 mW of 1064 nm excitation
SPIE 2001-4575
66
contributions from signals or baseline offsets The measurement error is given as SplusmnRMS and for adenine this equals 234 The number of molecules contributing to the SG-SERS are those on the silver particles that are embedded in the sol-gel The total silver surface area can be determined from the average particle size (40 nm diameter) concentration (073 by weight based on molar conc and measured sol-gel density) and the scattering volume (a cylinder defined by the laser area 28x10-7m2 and sol-gel thickness10-4m) The 61x109 silver particles in this volume have a collective area of 31x10-5m2 However it may reasonably be assumed that at least half of the silver surface is in contact with the sol-gel matrix and unavailable for analyte interaction Then approximately 10x1014 molecules or 22x10-8g of adenine contribute to the SG-SER spectrum The slightly lower SN of 207 suggests a mass detection limit of 32x10-10g Determination of the enhancement factors for the two SER media requires estimating the number of adenine molecules contributing to the normal Raman spectrum Here a cylindrical scattering volume is assumed again based on the laser area (28x10-7m2) and the penetration depth (1x10-3 m)41 The density of the sample was measured at 064 gcm3 indicating that 18x10-4g (13x10-6 moles) of adenine produced the normal Raman signal The enhancement factor EF is defined by the following equation
EF = (ISERSINR)bull(MNRMSERS) bull(PNRPSERS) bull(TNRTSERS)12
where I is the spectral band intensity (here 735 cm-1) M is the sample mass P is the incident laser power and T is the measurement time (or number of scans) For the E-SERS measurement the enhancement factor is 22x105 (01780184) bull (18x10-487x10-10)) while the SG-SERS enhancement factor is 10x105 (0160184) bull(18x10-422x10-8) bull(75075) bull(315)12) The lower enhancement for the SG-SERS may be real or the available surface of the silver embedded in the sol-gel may have been overestimated In addition to enhancing the Raman scattering efficiency to an extent similar to E-SERS the SG-SER medium also yields an identical shift of the adenine ring-breathing mode from 725 cm-1 in the normal Raman to 735 cm-1 Furthermore in the course of optimizing the E-SERS sample conditions it was found that pH influenced the adenine interaction with the silver surface (Figure 2) In particular the relative band intensities of the pyrimidine ring skeletal vibrations at 1270 and 1375 cm-1 and the imidazol ring skeletal vibration at1335 cm-1 change At pH 4 adenine is protonated presumably the imidazol ring since the band at 1335 cm-1 increases in intensity while the pyrimidine bands are virtually absent Conversely at pH 10 the imidazol band decreases in intensity while the pyrimidine bands appear It is worth noting that the ring-breathing mode at 735 cm-1 changes little between pH 4 and 10 suggesting that the skeletal changes are more a function of molecule-plasmon interactions than reorientation of the molecule on the surface Measurements of the identical pH series of adenine samples by SG-SERS yielded virtually identical spectral changes This suggests that the sol-gel does not influence the measurement This is critical to reproducing measurements and performing quantitative analysis Next the remaining NA bases were measured by both E-SERS and SG-SERS and compared Previously we examined the optimum pH and electrode potentials for E-SERS measurements to determine if a common pH could be used that yielded good sensitivity for all the bases and if variations in potential could be used to provide an added degree of selectivity between the bases Primarily it was found that high quality spectra were obtained between pH 7 and 95 and that cytosine and uracil were best enhanced at potentials positive of the potential-of-zero charge (pzc ca 065VSCE for Ag) guanine and thymine near the pzc and adenine negative of the pzc In all cases the ring-breathing modes were the most intense and in general could be used to identify the NA bases (Figure 3 Table 1) Specifically adenine has an intense band at 735 cm-1 cytosine at 797 cm-1 guanine at 653 cm-1 thymine at 784 cm-1 and uracil at 800 cm-1 The adenine cytosine guanine and thymine bands are sufficiently separated that their contributions to DNA should be determinable Although adenine and guanine contributions to RNA should also be determinable cytosine and uracil are highly overlapped and unfortunately share the same potential dependence Alternate unique bands at 1183 cm-1 for cytosine and 1275 cm-1 for uracil might be suitable for calculating contributions The SG-SER spectra of the remaining NA bases faithfully reproduced the E-SER spectra In particular the primary identifying bands occur at virtually the same wavenumbers (see Table 1) However the spectra for both cytosine and thymine contain an intense band at ca 1040 cm-1 Initially this was attributed to the pH buffer but samples prepared without either the buffer or the 01M KCl electrolyte yielded identical spectra containing this band In fact the E-SER and SG-SER spectra of thymine are virtually identical except for this band Also the SG-SERS of guanine contains an intense band at 1551 cm-1 that is not observed in the E-SER spectrum This band may be due to a moderately intense band at 1553cm-1 in the normal Raman spectrum that is SG-SER active It was also found that the SG-SERS of cytosine was considerably better than the E-SERS while uracil showed the opposite relationship It is also worth noting that all of the SG-SERS were obtained with 110th the laser power Most importantly the primary ring-breathing modes in the SG-SER spectra are sufficiently intense and unique to be used in determining contributions to DNA and RNA as outlined above
SPIE 2001-4575
67
Table 1 Comparison of E-SER and SG-SER Spectral Band Positions for the NA Bases and Adenosine Monophosphate
Adenine Cytosine Guanine Thymine Uracil AMP E-SERS SG-SERS E-SERS SG-SERS E-SERS SG-SERS E-SERS SG-SERS E-SERS SG-SERS E-SERS SG-SERS1647 1638 1634 1656 1655 1630 1587 1585 1510 1580 1551 1539 1456 1456 1465 1460 1462 1480 1453 1459 1394 1398 1425 1431 1435 1399 1404 1392 1374 1375 1373 1383 1370 1335 1332 1311 1307 1333 1331 1353 1348 1331 1329 1265 1273 1280 1292 1278 1276 1275 1279 1271 1183 1195 1222 1232 1221 1219 1204 1205 1180 1144 1097 1033 1029 1038 1040 1035 1051 1037 1041 1035 963 963 957 1001 1000 961 944 884 819 817 859 866 735 737 797 799 784 782 800 800 72738 742 630 630 653 664 667 684 603 602 590 611 466 561
Bands unique to SG-SERS The next chemical to be analyzed by both E-SERS and SG-SERS was adenosine monophosphate (AMP) The E-SER spectrum yields bands due to the adenine chemical functionality at 727 961 1233 1279 1331 1381 and 1486 cm-1 In addition phosphate bands are observed at 860 1097 1453 1587 and 1705 cm-1 (Figure 5) Other researchers have noted that the ribose component does not appear to contribute to the spectrum19 The AMP spectrum also changes as a function of potential As the electrode is swept more positive (here from -09 to -03VSCE) the phosphate bands at 860 1097 1453 and 1587 cm-1 increase in intensity compared to the adenine bands while a band at 1705 cm-1 appears The adenine bands at 1233 1381 and 1486 cm-1 virtually disappear These potential dependent spectral changes are consistent with earlier studies that show that phosphate is attracted to silver at potentials positive of the pzc but repelled at potentials negative of the pzc19
Wavenumbers (∆cm-1) Figure 3 E-SERS of A) 21x10-3M cytosine at -03VSCE 1000 scans B) ~10x10-5M guanine at -06VSCE 500 scans C) 23x10-3M thymine at -06VSCE 500 scans and D) 12x10-3M uracil at -093VSCE 500 scans All spectra at pH 92 750 mW 1064 nm at 8 cm-1
Wavenumbers (∆cm-1) Figure 4 SG-SERS of A) 21x10-3M cytosine 200 scans B) ~10x10-5M guanine 200 scans C) 23x10-3M thymine 200 scans and D) 12x10-3M uracil 500 scans All spectra at pH 92 75 mW 1064 nm at 8 cm-1
A
B
C
A
B
C
D D
SPIE 2001-4575
68
The SG-SER spectrum of AMP is considerably different The adenine bands virtually disappear except for the two primary bands which shift to742 and 1329 cm-1 While the phosphate band at 1459 cm-1 has gained considerable intensity In addition two new intense bands appear at 684 and 1539 cm-1 as well as a moderately intense band at 1180 cm-1 The SG-SER spectrum has greater similarity to the E-SER spectrum at -03VSCE and suggests that the silver particles embedded in the sol-gel behave as if at a potential positive of the pzc RNA samples extracted from E coli B subtilis and S aureus were next examined by both E-SERS and SG-SERS E-SER spectra of these samples yielded quality spectra in 10 minutes in which all of the major features can be identified (Figures 6 and 7) This includes guanine at 650 cm-1 adenine at 791 cm-1 cytosine and uracil combining at 790 cm-1 and phosphate at 1100 1335 (in combination with adenine and guanine) 1465 and 1570 cm-1 Surprisingly adenine which demonstrated the greatest surface-enhanced Raman effect does not dominate the ring-breathing mode portion of the spectrum The intensities of the other base-pairs bands are of the same order of magnitude This suggests that when the base-pairs are linked together as in RNA they are enhanced in concert In fact the relative intensities are very similar to a normal Raman spectrum of E coli RNA which shows the combined cytosine and uracil band at ca twice the intensity of the adenine band and ca four times the intensity of the guanine band Unfortunately this means that the independent enhancement factors for the NA bases can not be used to estimate relative concentrations For example the relative 791 and 734 cm-1 bands for B subtilis would indicate that the cytosine andor uracil concentration was at least 20 times the adenine concentration whereas each of the four RNA bases are known to contribute 15-35 Nevertheless it is worth noting that the three RNA samples yield different relative band intensities that were reproduced in numerous measurements Although the relative concentrations of the NA bases for these samples have not been determined these differences can be quantified If it is assumed that the 650 cm-1 band represents 25 guanine the 791 cm-1 band represents 25 adenine and the 790 cm-1 50 cytosine plus uracil in the E coli RNA spectrum then the relative concentrations can be estimated for the other RNA samples To aid this calculation the three spectra were normalized to the phosphate band at 1100 cm-1 which has been shown to correlate to the total phosphate concentration and can be used as an internal standard In addition a simple baseline correction was applied (Figure 7) This yields 15 adenine 30 guanine and 55 cytosine plus uracil for B subtilis RNA and 18 adenine 25 guanine and 57 cytosine plus uracil for S aureus RNA The average SN of these measurements was 26 with an average error of 8 of the value (SplusmnN) It is also worth noting that the three RNA spectra show a marked shift in a band near 825 cm-1 This band is assigned to the symmetric stretch of the O-P-O ester linkage9 The band appears at 815 cm-1 for S aureus shifting to 820 cm-1 for B subtilis and 830 cm-1 for E coli Others have used the normal Raman intensity of the band at 815 cm-1 as a direct indication of the amount of A-class helix present while the intensity of the band at 830 cm-1 has been used as a direct indication of the amount of B-class helix present However the latter is more associated with DNA than RNA
Wavenumbers (∆cm-1) Figure 5 E-SER spectra of 020 mgmL adenosine monophosphate at A) -03 and B) -09VSCE and C) SG-SER spectra Conditions sample in 01M KCl buffered to pH 92 A) and B) 750 mW C) 75 mW of 1064 64 scans (1-min) at 8 cm-1
Wavenumbers (∆cm-1) Figure 6 E-SERS of 01 mgmL RNA from E coli 02 mgmL RNA from B subtilis and 02 mgmL RNA from S aureus Conditions 01M KCl pH 92 -03VSCE 750 mW of 1064 nm 640 scans (10 min) at 8 cm-1
AMP RNA
E coli A
B
C
B subtilis
S aureus
SPIE 2001-4575
69
SG-SER spectra of reasonable quality were also obtained for E coli and B subtilis especially the latter (Figure 8) However the spectra differ substantially from the E-SERS of the same samples Both SG-SER spectra are dominated by adenine at 735 cm-1 and a band at 1030 cm-1 Although unassigned the latter does appear in the RNA E-SER spectra Bands at 1105 and 1565 cm-1 are likely due to phosphate while bands at 1320 and 1470 cm-1 are less confidently assigned to phosphate They are significantly less intense and somewhat shifted from their SG-SERS counterparts (1335 and 1455 cm-1) A band at 670 cm-1 may be due to guanine which was observed at 664 cm-1 for SG-SERS of the pure sample However the SG-SER spectrum of AMP also had an intense 667 cm-1 band A number of other bands occur at 890 1070 1165 1245 1290 1420 1505 cm-1 and remain unassigned The SG-SER spectra are somewhat disappointing in that only adenine and guanine contributions can be positively identified This limits the ability to determine relative NA base concentrations and distinguish bacterial RNA However several of the unassigned bands may be due to the bases (eg 1030 and 1420 cm-1 due to cytosine) Further experiments will be required to clarify this point
Ecoli
B subtilis
S aureas
P G
A C+U
OPO
A B
Wavenumbers (∆cm-1) Wavenumbers (∆cm-1) Figure 7 SER spectra of RNA from A) B subtilis with contributions indicated and B) E coli B subtilis and S aureus with baseline correction and peak positions used to calculate contributions indicated G = guanine A = adenine C+U = cytosine plus uracil P = phosphate (backbone) OPO = phosphate ester linkage (A- vs B-class helix)
Wavenumbers (∆cm-1) Wavenumbers (∆cm-1) Figure 9 E-SER (-03VSCE) and SG-SER spectra of RNA from A) E coli and B) B subtilis Sample conditions as in Figure 6 E-SER spectra at 750 mW SG-SERS at 75 mW
A B E-SERS
SG-SERS
E-SERS
SG-SERS
SPIE 2001-4575
70
A final comparison was made between the two SER methods by measuring dipicolinic acid (DPA) This chemical may be invaluable as a test for spore forming bacteria specifically B anthracis 50 to 90 of B anthracis sporilates During spore formation dipicolinic acid is synthesized and once completed 10-15 of the dry spore weight is composed of the Ca2+ complex located in the spore core42 Heating in water can be used to initiate germination at which point the exosporium breaks and releases the Ca dipicolinate which becomes dipicolinic acid in water The structure of this chemical strongly
4 CONCLUSIONS Towards the goal of developing a method and technology to rapidly detect and identify bioagents we have been investigating surface-enhanced Raman spectroscopy as a tool to measure relative concentrations of nucleic acid bases in RNA and determine bioagent taxonomy Initially we investigated E-SERS since this method has been extensively researched and the optimum sample conditions are well developed However this method requires a three-electrode sample cell and electrolyte solution Incorporation of an electrolytic cell into sample systems used to analyze for BWA as aerosols in water or on surfaces can be designed using flow injection analysis technologies but cross-contamination and plugging of sample lines seems inevitable For this reason we also investigated metal-doped sol-gels as a SER-active medium Previous studies have shown this material to be active in all solvents particularly water capable of continuous measurements in flowing systems and reproducible (quantitative) between coated sample vials Here we compared SG-SER spectra to traditional E-SER spectra of the nucleic acid base pairs adenosine monophosphate and RNA High quality spectra of adenine cytosine guanine thymine and uracil were obtained by both E-SERS and SG-SERS Both methods yielded very similar spectra for the NA bases including a pH dependent study of adenine Enhancement factors and detection limits for adenine were determined as 2x105 and 16x10-11g and 1x105 and 12x10-10g for E-SERS and SG-SERS respectively (normalized to 75 mW and 10-min acquisition time) Fifty percent of the silver particle surface area in the sol-gel matrix was assumed covered by adenine which may have been overestimated yielding a lower EF and higher detection limit It should also be realized that each E-SER spectrum required several attempts to optimize the measurement conditions (pH electrode potential etc) While each SG-SER spectrum involved no sample preparation and often represents the first and only attempt to make the measurement Quality spectra of RNA extracted from Escherichia coli Bacillus subtilis and Staphylococcus aureus were obtained by E-SERS that were easily interpreted Bands due to adenine guanine cytosine plus uracil and phosphate were identified The SER band intensity of the NA bases in the RNA samples were of the same order of magnitude suggesting that their interaction with the silver surface is concerted as is their Raman enhancement Interestingly the relative SER band
Wavenumbers (∆cm-1) Figure 10 A) Raman spectrum of solid dipicolinic acid B) Ax20 C) electrolytic SERS of 6x10-3 M dipicolinic acid in 01 M KCl at a potential of +07VSCE and pH of 4 and D) sol-gel SERS of 6x10-3 M dipicolinic acid Conditions for A and C as in Figure 1 C) 100 mW of 1064 nm 50 scans 8 cm-1
A B (Ax20)
C
D
suggested that it would be SER active However the E-SER spectrum was unstable and varied considerably as a function of potential A consistent spectrum was obtained at +06VSCE (Figure 10) This potential is not recommended for measurement because the surface is actively dissolving in solution The SG-SER spectrum was considerably more stable of higher quality and easily reproduced Bands at 660 825 1010 1390 1430 1570 1590 and 3075 cm-1 were observed Enhancement factors were determined for the two media using the symmetric ring stretching mode at 995 and 1010 cm-1 for the normal Raman and SER spectra respectively E-SERS yielded an EF of 5x103 while SG-SERS yielded an EF of 2x105 for DPA The SN of the latter suggests a detection limit of 20x10-10g (based on adenine coverage 75 mW and 10-min) The differences in SER activity for these two media may be attributed to the combined electrolytic potential of the solution chemical and metal15 Again the E-SERS suggests that the SG-SERS is at a potential positive of the pzc While the instability in the E-SERS may also be associated with surface interactions of two carboxylic acid groups of dipicolinic acid during the ORCs
DPA
SPIE 2001-4575
71
intensities for RNA extracted from E coli are very similar to those measured by normal Raman spectroscopy Although the relative percent that each of the NA bases contributed to each RNA sample was not determined reproducible band intensities allowed noting the following trends The percent adenine decreases while the combined percent cytosine and guanine increase for both B subtilis and S aureus compared to E coli Quality spectra were also obtained for the RNA samples by SG-SERS but only a few bands were readily identified Calculations of NA base concentrations by SG-SERS will require further research In light of recent events we summarize the capabilities of these SERS media in terms of rapid detection of B anthracis and dipicolinic acid However these capabilities must be qualified First and foremost the level to which SERS can distinguish bacteria or viruses has not yet been determined Development of a database of both DNA and RNA base concentrations for BWAs and common bacteria to establish the level of taxonomic identification is ongoing Second rapid collection of aerosol water or surface samples is being addressed by others who report trapping particles on filters from 100 liters of air per minute Third although not presented here we have developed methods to extract RNA or DNA from cells and spores for SER analysis within 10 minutes Finally we assume a detection limit of 3600 spores per 100 liters of air is required although a 50 lethal dosage of anthrax has not been established With these qualifications a mass detection limit for RNA using SERS is estimated as follows A single measurement is performed in ca 20 minutes (140 liters collected in 14 min RNA extracted in 8 min spectral acquisition and analysis in 10 min) The average human breaths 7 liters per minute therefore the analyzer must at the very minimum detect 5000 spores in 140 liters of air One spore is approximately 2x10-18m3 (1x1x2 microm) and if a density of 075 gcm3 is assumed this corresponds to a mass of 15x10-12g Each spore contains 4-12 RNA or 12x10-13g RNA for 8 If we assume 23 of the RNA can be isolated for analysis during lysis then the proposed instrument must be able to detect 4x10-10g RNA from 5000 spores per 70 liters of air within 10 minutes As noted above the mass detection limits for adenine were estimated at 16x10-11g and 12x10-10g for E-SERS and SG-SERS respectively Although these detection limits suggests that RNA from 5000 spores is detectable with the current instrumentation it is highly likely that only a portion of an RNA segment (eg 120-nucleotide 5S rRNA) is in contact with the metal surface and will contribute to the SER effect The SN for the RNA spectra were 110th of the average SN for the four individual RNA bases suggesting a 10 contribution Furthermore effective taxonomy will likely require knowing the NA base concentrations to 1 of the value (eg 25plusmn025) Again the average measurement error for the bases is 12 These values suggest that the E-SERS is within a factor of 4 of the required detection limit whereas the SG-SERS detection limit must be improved by 25 times The same arguments can be applied to the detection of dipicolinic acid If we assume a spore releases 10 by weight DPA during germination then the proposed instrument must be able to detect 75x10-10g DPA from 5000 spores per 70 liters of air within 10 minutes The detection limit for SG-SERS was estimated at 20x10-10g and suggest that the vials are suitable to perform a rapid screen for anthrax A series of concentration dependent measurements are currently being performed to verify this assertion Finally we note that the measurements performed here employed an FT-Raman spectrometer This instrumentation was chosen over dispersive Raman instrumentation for the high wavelength accuracy afforded by the HeNe reference laser (Connes Advantage43) which would allow reliable spectral subtraction matching of observed spectra to stored library spectra and confident use of chemometric approaches Such data analysis is likely to be required to enhance BWA identification However this instrumentation sacrifices sensitivity due to a less efficient detector (InGaAs vs Si) less efficient Raman scattering and less efficient generation of plasmon modes Substantial improvements can be obtained using visible excitation and Si detection and these measurements are underway
5 ACKNOWLEDGEMENTS The authors are grateful to Drs D Cookmeyer and S Tove of the US Army Research Office (Contract Number DAAH04-96-C-0078) for their interest and support of this research The authors would also like top acknowledge Dr R Yin and J Jensen for supporting development of the metal-doped sol-gels (Contract Number DAAD13-01-C-0019) They also thank Dr Wilfred H Nelson for assistance in spectral interpretations
SPIE 2001-4575
72
6 REFERENCES 1 Roberts WL and Rainey PM Clin Chem 39 1872-1877 (1993) 2 Pasechnik VA CC Shone and P Hambleton Bioseperations 3 267-283 (1993) 3 Jackson PJ ME Hugh-Jones DM Adair G Green KK Hill CR Kuske LM Grinberg FA Abramova and P
Keim Proc Natl Acad Sci 95 1224-1229 (1998) 4 Jensen JL NC Wong W R Loerop SPIE 4775-03 (2001) 5 Snyder A Pet al SPIE 3853-15 (1999) 6 Danosky T R and McFadden P N in press (1997) 7 Woodruff WH Farquharson S Science 201 831-833 (1978) 8 Lord RC and Thomas GJJr Spectrochemica Acta 23A 2551-2591 (1967) 9 Thomas GJJr Biochim Biophys Acta 213 417-423 (1970) 10 Placzek G Handbuch der Radiologie 2 EMarx ed Akademische Verlagagescellschatt Liepzig 1934 UCRL
Trans No 526 (1959) 11 Jeanmaire DL and RP Van Duyne J Electroanalytical Chem 84 1-20 (1977) 12 Chada S Manoharan R Moenne-Loccoz P Nelson WH Peticolas WL and Sperry JF Applied Spectroscopy
47 38-43 (1993) 13 Manoharan R Ghiamati E Chada S Nelson WH and Sperry JF Applied Spectroscopy 47 2145-2150 (1993) 14 Todd EA Morris MD Applied Spectroscopy 48 545-548 (1994) 15 Wentrup-Byrne E Sarinas S and Fredericks PM Applied Spectroscopy 47 1192-1197 (1993) 16 Weldon MK VR Zhelyaskov and MD Morris Applied Spectroscopy 52 265-269 (1998) 17 Kneipp K and J Fleming J Mol Structure 145 173-179 (1986) 18 Pothier NJ and Force RK Applied Spectroscopy 46 147-151 (1992) 19 Ervin KM E Koglin JM Sequaris P Valenta and HW Nurnberg J Electroanal Chem 114 179-194 (1980) 20 Kim SK TH Joo SW Suh and MS Kim J Raman Spectrosc 17 381-386 (1986) 21 Pothier NJ and Force RK Analytical Chemistry 62 678-680 (1990) 22 Kneipp K Y Wang RR Dasari and MS Feld Applied Spectroscopy 49 780-784 (1995) 23 Nie S and Emory SR Science 275 1102 (1997) 24 Graham D WE Smith AMT Linacre CH Munro ND Watson and PC White Analytical Chemistry 69 4703-
4707 (1997) 25 Fleischmann M PJ Hendra and AJ McQuillan Chem Phys Lett 26 163-166 (1974) 26 Pettinger B J Chemical Phys 85 7442-7451 (1986) 27 Surface-Enhanced Raman Scattering Section Four Theory SPIE MS 10 M Kerker and B Thompson Eds (1990) 28 Wang D-S and Kerker M SPIE (M Kerker and B Thompson Eds) MS 10 417-429 (1990) 29 Weaver MJ Farquharson S Tadayyoni MA J Chem Phys 82 4867-4874 (1985) 30 Norrod KL Sudnik LM Rousell D and Rowlen KL Applied Spectroscopy 51 994-1001 (1997) 31 Farquharson S Weaver WJ Lay PA Magnuson RH and Taube H J Am Chem Soc 105 3350-3351 (1983) 32 Lee PC and Meisel D ldquoAdsorption and Surface-Enhanced Raman of Dyes on Silver and gold Solsrdquo J Phys Chem
86 3391-3395 (1982) 33 Li Y-S and Wang Y Applied Spectroscopy 46 142-146 (1992) 34 Bello JM DL Stokes and T Vo-Dinh Analytical Chemistry 61 1779-1783 (1989) 35 van Duyne RP J C Hulteen D A Treichel M T Smith M L Duval and T R Jensen J Phys Chem B1033854-
3863 (1999) 36 Farquharson S and W W Smith W H Nelson and J F Sperry SPIE 3533-27 207-214 (1998) 37 Lee Y H W Smith S Farquharson H C Kwon M R Shahriari and P M Rainey SPIE 3537 252-260 (1998) 38 Lee Y-H S Farquharson and P M Rainey SPIE 3857 76-84 (1999) 39 Current Protocols in Molecular Biology Wiley Interscience1003-1006 (1987) 40 Farquharson S Smith W Carangelo R C and Brouillette C SPIE 3859 14-23 (1999) 41 Chase D B and JF Rabolt Fourier Transform Raman Spectroscopy Acad Press Ch1 p 131 (1994) 42 Brock TD MT Madigan JM Martinko and J Parker Biology of Microorganisms 7th Ed Prentice Hall p 76-80
(1994) 43 Connes J Rev Opt Theor Instrum 40 45 (1961)
166
Chemical agent identification by surface-enhanced Raman spectroscopy
Stuart Farquharson and Paul Maksymiuk
Real-Time Analyzers 87 Church Street East Hartford CT 06108
Kate Ong and Steven D Christesen US Army SBCCOM Aberdeen Proving Ground MD 21010
ABSTRACT The recent distribution of anthrax through the US postal system and the subsequent infection and death of several postal and national media employees amplifies the need for methods to rapidly detect identify and quantify this and other chemical and biological warfare agents The US military has also identified water supplies as a likely method of warfare agent deployment and is funding the development of a Joint Service Agent Water Monitor (JSAWM) In an effort to aid military personnel and the public at large we are developing a portable analyzer capable of identifying and quantifying chemical agents rapidly either on-demand or continuously The approach is based on the ability of Raman spectroscopy to identify molecular structure through the abundant vibration information provided in spectra and the ability of surface-enhanced Raman spectroscopy (SERS) to detect extremely low concentrations (eg part-per-billion) through the enhancement of Raman scattering by six orders of magnitude or more A key element to the analyzer design is a new SER active medium that is capable of quantitative reversible measurements The medium consists of silver or gold nanoparticles incorporated into a sol-gel matrix The porous silica network offers a unique environment for stabilizing SER active metals and the high surface area increases the interaction between the analyte and metal particles Here we present the use of new sol-gels that also selectively enhance chemicals based on polarity and charge Base-line measurements of chemical agents and their hydrolysis products are presented and compared to the JSAWM goal of 30 micrograms per liter detection Keywords Chemical warfare agent hydrolysis product SERS Raman spectroscopy sol-gel nanoparticle
1 INTRODUCTION Since September 11 2001 the threat of terrorist attacks and biological warfare within US borders has become a sobering reality The simplicity in manufacturing ease of deployment and the relatively low cost of chemical warfare agents (CWAs) raises public concern that they may also be used by terrorists Indeed terrorists released sarin (GB) in the Tokyo subway in 19951 Countering terrorism and terrorist attacks requires recognizing likely deployment scenarios and having the required technology to rapidly detect the deployment event One method of deployment has been long identified by the US military distribution through water supplies To counter this threat the Department of Defense is funding or monitoring the capabilities of 200 technologies with the goal of developing a Joint Service Agent Water Monitor (JSAWM) that is field portable2 The JSAWM must be capable of identifying the chemical agents with no false-positives and quantifying the agents at microgram per liter concentrations within 10 minutes (Table 1)3 This includes the analysis of drinking water supplies distribution and storage systems as well as potable water supplies The earliest technologies developed for CWA detection were based on electrochemical ionization or colorimetric analysis (eg phosgene tape) Although these analyzers were easy to use they were not generally agent specific and suffered from false-positives4 More traditional laboratory methods have also been investigated and in particular combined gas chromatography and mass spectrometry (GCMS) has been very successful at eliminating false-positives56 However GCMS requires extraction repeated calibration and long analysis times (typically 20 to 60 minutes)6 making it labor intensive and less than desirable for field use More rapid analysis of agents in the solid liquid and gas phase has been To whom correspondence should be addressed emailfarqureal-time-analyzerscom
Vibrational Spectroscopy-based Sensor Systems Steven D Christesen Arthur J Sedlacek III EditorsProceedings of SPIE Vol 4557 (2002) copy 2002 SPIE middot0277-786X02$1500
167
Table 1 Chemical Agent Structures Hydrolysis Half-lives and JSAWM Thresholds Agent Short-Hand Chemical Structure Hydrolysis
Half-Life JSAWM
Thresholds
Sarin (GB) F-[O=P-CH3]-O-CH(CH3)2 213 hours 32 microgL Soman (GD) F-[O=P-CH3]-O-CH(CH3)-(C-(CH3)3) 23 hours 32 microgL Tabun (GA) (CH3)2-N-[O=P-CN]-O-C2H5 41 hours 32 microgL VX C2H5O-[O=P-CH3]-S-(CH2)2-N-(CH(CH2)2)2 821 hours 32 microgL EA2192 HO-[O=P-CH3]-S-(CH2)2-N-(CH(CH2)2)2 gt9 years 32 microgL Mustard (H) ClCH2CH2-S-CH2CH2Cl encapsulates 47 microgL Lewisite (L) ClCH=CH-As-Cl2 rapid 27 microgL HCN HCN rapid 20 mgL BZ C7NH12-O-[C=O]-COH(C6H5)2 23 microgL T-2 Toxin 87 microgL at pH 7 to 75 and 20 to 25 oC
demonstrated by vibrational spectroscopy7-10 Hoffland et al7 reported infrared absorbance spectra and absolute Raman cross sections for several chemical agents while Christesen measured Raman cross sections for sarin tabun mustard gas and VX (ethyl S-2-diisopropylamino ethyl methylphosphonothioate)11 Again however these techniques also have limitations Raman spectroscopy is simply not a very sensitive technique and detection limits are typically 01 (1000 ppm) While infrared spectroscopy would have limited value in analyzing poisoned water since the very strong infrared absorption of water would obscure most other chemicals present Nevertheless efforts to overcome these limitations have been demonstrated Braue and Pannella8 quantified the G-series nerve agents (tabun sarin and soman) in terms of infrared attenuated total reflectance using a circle-cell And Alak and Vo-Dinh demonstrated the possibility of surface-enhanced Raman spectroscopy (SERS) to identify CWAs by measuring several organophosphonates that simulate the nerve agents12 However quantitative measurements have not been demonstrated for the SER-active material used (silver coated on alumina particles) or other SER-active media13 Recently we developed silver-doped sol-gels to promote the SER effect14-17 The porous silica network of the sol-gel matrix offers a unique environment for stabilizing SER-active metal particles and the sol-gel provides a high surface area that effectively increases the number of molecules observed within the Raman scattering volume The silver-doped sol-gels have been coated on the internal walls of standard glass vials to allow rapid SER analysis of small sample quantities (lt 01 mL) without preparation We have used p-aminobenzoic acid (PABA) as a test chemical to demonstrate surface enhancements greater than 106 reversible measurements in a flowing system reproducible measurements from vial-to-vial and batch-to-batch and measurements in multiple solvents including water14-17 Recently we used these vials to measure Tabun (GB) and Sarin and several hydrolysis products pinacolyl methyl phosphonate (PMP from Soman) and methyl phosphonic acid (MPA from all G-agents Figure 1) Although a number of unique vibrational bands are observed (eg C-N stretch doublet and P-C stretch) the G-agents were only observed for 5 concentrations and all spectra required baseline corrections Figure 1 Surface enhanced Raman spectra of ~5 vv A) Tabun and B) Sarin C) 1 vv PMP and D) 10 ppm MPA using sol-gel sample vials 785 nm excitation 1-min scan and CCD detection Performed at Aberdeen Proving Ground
C A
Wavenumber (∆cm-1) Wavenumber (∆cm-1)
D B
790
545 C-N
2135 2190 P-C 770
760 1290
Proc SPIE Vol 4577
168
Nevertheless MPA was readily observed for a 10 ppm sample with an estimated detection limit of 04 ppm (based on a signal-to-noise ratio of 3 for the 760 cm-1 band intensity) This measurement provides encouragement in that SERS may satisfy the needs of a JSAWM Furthermore MPA is also a hydrolysis product of VX and V-gas and EA2192 (Figure 2) and may prove a valuable indicator of agent usage Figure 2 Hydrolysis of Sarin to form hydrofluoric acid (HF) methylphosphonic acid 1-methylethyl ester (MPAMME) methyl phosphonic acid (MPA) and 2-propanol With this initial albeit modest success we began analyzing chemicals with various sol-gel compositions that we have been developing Here we describe four sol-gel compositions that select for 1) polar-positive 2) polar-negative 3) weakly polar-positive and 4) weakly polar-negative chemical species The ability of these sol-gels to select and enhance Raman scattering is described for several test chemicals and MPA
2 EXPERIMENTAL The chemicals analyzed as well as all chemicals used to prepare the metal-doped sol-gels were obtained at their purest commercially available grade from Aldrich (Milwaukee WI) The sol-gel designed to select for polar-negative species was prepared from a silver amine complex tetramethyl orthosilicate (TMOS) and methanol After mixing 02 mL of the sol-gel solution was transferred into a glass vial (2 mL) dried and heated The incorporated silver ions were then reduced using dilute sodium borohydride The vials were washed and dried prior to the addition of a sample solution In a similar manner the sol-gel designed to select for polar-positive species was prepared from a gold salt TMOS and methanol The sol-gel designed to select for weakly polar-negative species was prepared from a silver amine complex tetraethyl orthosilicate (TEOS) and methanol And the last sol-gel designed to select for weakly polar-positive species was prepared from a gold salt TEOS and methanol All samples were prepared in a chemical hood and transferred into plain or SER-active vials for analysis Normal Raman spectral measurements employed 1-mL pure samples that were placed in a 1-cm3 cuvette and weighed This yielded a powder density that allowed accurate calculation of molecules in the optical collection field SERS measurements employed 1-mg sample per mL water concentrations unless otherwise stated Once prepared a 01 mL sample was placed into one of the four selective sample vials which in turn was placed into the sample compartment of a Raman spectrometer for analysis A prototype Fourier transform Raman spectrometer (RTA) was used for these measurements18 The system consisted of a NdYAG laser (Brimrose) for excitation at 1064 nm an interferometer built by On-Line Technologies (OLT East Hartford CT) for frequency separation an uncooled InGaAs detector for signal detection (RTA) and an Intel 400 MHz Pentium II based laptop computer (Dell Round Rock TX) for interferometric control data acquisition (OLT) and analysis (LabVIEW by National Instruments Austin TX) Additional components included a Notch filter (Kaiser Ann Arbor MI) and interferometer entrance and exit optics (Edmund Scientific Barrington NJ) Fiber optics were used to deliver the excitation beam to the sample and the scattered radiation to the interferometer (1 meter lengths of 200 and 365 micron core diameter respectively Spectran Avon CT) A second Notch filter (Kaiser) was used as a beam splitter to direct the excitation beam along the same axis as the collected radiation A microscope object (20x04 Newport Irvine CA) was used to focus the beam into the sample and to collect the scattered radiation back along the same axis In this co-axial backscattering arrangement the excitation beam was passed through the outside of a glass vial and focused onto the silver-doped sol-gel film (01-03 mm thickness) containing the sample
3 RESULTS AND DISCUSSION p-aminobenzoic acid (PABA) and phenyl acetylene (PA) and were used to refine the selectivity and SER-activity of the four different metal-doped sol-gels PABA is a popular chemical used to evaluate the performance of SER-active media Here the polar end groups can be used to test selectivity of the polar-negative and polar-positive sol-gels PA is essentially non-
2O+ H HF + +OH
OH3H C
OP
OF
CH
CH3
3
3H C
OCP
OOH
CH
CH3
3
3H C
O
CPHO
CH
CH
3
3
C
Sarin MPAMME MPA 2-propanol
Proc SPIE Vol 4577
169
polar but a high electron density in the cylindrical π cloud around the carbon-carbon triple bond allows testing the selectivity of the weakly polar-negative and weakly polar-positive sol-gels As Figure 3 illustrates PABA passes through the polar sol-gel and is enhanced by either the silver or gold particles At 1 mgml the concentration of neutral PABA is ca 20 times that of the ionized form (pKa = 48) For electropositive silver the PABA anion is expected to interact through the carboxylate group and the associated vibrational modes are expected to dominate the spectrum Conversely for electronegative gold either form of PABA is expected to interact through the amine group The clear differences in our spectra support this expectation Furthermore bands at 840 and 1405 cm-1 assigned to a COO- bend and stretch respectively are significantly more intense for silver than gold Additional bands at 1140 and 1195 cm-1 are assigned to CH bending modes while bands at 1450 1500 and 1605 cm-1 are assigned to ring vibrational modes A very similar SER spectrum for PABA on a silver-coated alumina substrate has previously been reported with similar assignments19 For the gold-doped sol-gel new bands appear at 690 1355 and 1585 cm-1 The first band is assigned to a ring-H bending mode the second band to a ring-N- stretching mode and the third band to a possible NH2 scissors mode or ring mode The second band is not observed in the normal Raman spectrum but infrared bands occur at this frequency for aromatic ring-secondary amine stretching modes The scissors mode occurs at this frequency in Raman spectra for several chemicals but is absent in the PABA Raman spectrum Alternatively this mode may be the1600 cm-1 ring mode that has been shifted by the gold interaction Again a very similar SER spectrum of PABA has been reported but surprisingly using silver (colloids)2021 not gold as the enhancement medium These researchers also assumed the primary interaction of PABA with silver was through the carboxylate anion and made assignments accordingly For example they assigned the 1359 cm-1 to a COO- stretch not to the amine group as we have They also favor the ring stretching mode assignment for the 1582 cm-1 band Finally it should be said that other researchers have argued that the most dominant band in the SER spectra at 1450 cm-1 a ring vibration mode suggests that PABA lies flat on the surface and the π-orbitals dominate the surface interaction22
Figure 3 SER spectra of A) PABA using polar-negative and B) polar-positive sol-gels and C) PA using weakly polar-negative and D) weakly polar-positive sol-gels PABA is 1 mgmL PA is 1 vv Spectral conditions 75 mw 1064 nm 100 scans (15 min) 8 cm-1 resolution Non-polar PA passes through the non-polar sol-gels and is also enhanced by both metals The spectra are easily understood For electropositive silver PA interacts through the cylindrical triple bond π electron cloud and a -CequivC- doublet occurs near 2000 cm-1 The interaction is reasonably strong since this band appears at 2112 cm-1 in the normal Raman spectrum For electronegative gold this interaction is unlikely and only very weak bands occur near 2000 cm-1 The remaining bands are at 1000 cm-1 1200 cm-1 doublet and 1595 cm-1 all appear in the normal Raman spectra at virtually the same frequencies and are assigned to the symmetric ring-breathing mode CH bending modes and the trigonal ring-breathing mode respectively The polarnon-polar selectivity of the polar-negative and weakly polar-negative sol-gels was tested by adding a 11 molar mixture of PABA and PA The selective enhancement is quite good (Figure 4) The spectrum obtained using the polar sol-gel represents 78 PABA and 22 PA while the spectrum obtained using the weakly polar sol-gel represents 9 PABA and 91 PA The band peak intensities at 2000 cm-1 for PA and 1450 cm-1 for PABA were used for these calculations and are expanded in Figure 4 for clarity
C A
Wavenumber (∆cm-1) Wavenumber (∆cm-1)
2NH COOH
D B
C CH
Proc SPIE Vol 4577
170
Figure 4 SERS of 11 MM of PABA and PA in A) polar-negative and C) weakly polar-negative sol-gels The lower traces compare the pure chemicals B) 1 mgml PABA in polar-negative sol-gel and D) 1 PA in weakly polar-negative sol-gel while the insets magnify the minority species for clarity (x5 in A and x10 in B) Spectral conditions as in Figure 3 Following this development of selective sol-gels that maintained SER activity we measured cyanide and MPA (Figure 5) Not surprisingly the best sensitivity for both hydrolysis products was obtained using the polar-negative sol-gel The interaction of the cyanide anion with the silver surface is sufficient to shift the CequivN stretch observed at 2080 cm-1 in the normal Raman spectrum to 2145 cm-1 in the surface-enhanced Raman spectrum Furthermore the band is substantially broadened This anion has been extensively studied by electrolytic SERS and this shift and broadening have been attributed to the formation of a tetrahedral Ag(CN)3
2- surface structure23 Figure 5 Surface-enhanced (upper traces) and normal Raman spectra (bottom traces) of A) CN- and B) MPA in silver-doped TMOS SERS conditions as in Figure 3 and 1 mgmL Note MPA yields two distinct spectra for neutral (top) and acidic pH (middle) The normal Raman spectra employed pure powders 500 scans and 900 mW of 1064 nm SER measurements of MPA with the polar-negative sol-gel yielded two unique spectral signatures that depended on solution pH (Figure 5) For more neutral solutions the P-C stretch of MPA at 762 cm-1 dominates and the CH2 stretch at 2922 cm-1 appears The SN is sufficiently high that the anti-Stokes Raman shift at -762 cm-1 is observed For deprotonated MPA an oxygen-surface mode appears at 325 cm-1 (as well as its anti-Stokes complement) suggesting a strong interaction This results in substantial enhancement of the P-O-C mode at 1051 cm-1 (upper trace) Others report that this mode dominates the infrared spectra of nerve agents measured in water8 Comparison of the two spectra suggests the following molecule-to-surface orientations The appearance of the oxygen-surface and P-O-C modes in the upper spectrum of Figure 5B indicates that the tetrahedral molecule interacts with the silver surface through the deprotonated oxygen and is oriented end-on The
A B
Wavenumber (∆cm-1)
C A
Wavenumber (∆cm-1) Wavenumber (∆cm-1)
D B
Wavenumber (∆cm-1)
Proc SPIE Vol 4577
171
dominance of the P-C and the CH2 stretches and the disappearance of the P-O-C mode in the upper spectrum suggest the molecule is oriented with the methyl-hydroxyl-hydroxyl face towards the surface However considerably more research must be performed to verify these points Both chemical forms were measured as a function of concentration to determine the onset of monolayer coverage and estimate expected detection limits (Figure 6) Below monolayer coverage the signal to concentration dependence should be linear and the SN of any spectral measurement in this range can be used to predict the detection limit In the spectra presented here the peak height was used as the signal while the noise as root-mean-squared (RMS) was measured between 4400-4600 cm-1 Since noise is distributed evenly throughout the spectrum when transformed this region was used since it does not have any contributions from signals or baseline offsets Figure 6 shows a series of spectra for MPA along with a plot of the 1050 cm-1 band intensity (the noise was identical for this series) as a function of sample concentration A clear discontinuity appears in the vicinity of 01 mgmL (19 ppm) indicating the onset of monolayer coverage A detection limit defined as a SN of 3 was calculated for the 01 and 005gmL samples at 24x10-4 and 25x10-4 gL respectively A more modest detection limit of 101x10-4 gL was obtained using the 760 cm-1 band in the second series of concentration measurements These detection limits correspond to 46 and 190 ppb for a 3-min scan and 75 mW of incident laser power
Figure 6 A) Concentration dependence of MPA SERS measured in silver-doped TMOS) B) Concentrations are 001 005 01 05 1 gL (188 94 188 94 188 ppm) I760 series (bull) and I1050 series (∆)
Enhancement factors for cyanide and methyl phosphonic acid can be determined by estimating the number of molecules contributing to the surface-enhanced and normal Raman spectra The enhancement factor EF can be defined by the following equation
EF = (ISERSINR)bull(MNRMSERS) bull(PNRPSERS) bull(TNRTSERS)12
where I is the spectral band intensity M is the sample mass P is the incident laser power and T is the measurement time (or number of scans) for the two measurements For the normal Raman spectra a cylindrical scattering volume is assumed based on the laser area (28x10-7m2 6x10-4m diameter spot) and the penetration depth (1x10-3 m)24 The density of KCN and MPA as powders were measured at 0572 and 0516 gcm3 indicating that 16x10-4 and 144x10-4 g produced the normal Raman signals in Figure 5 respectively The number of molecules contributing to the SER spectra are those on the silver particles embedded in the sol-gel The total silver surface area can be determined from the average particle size concentration and the scattering volume Previous scanning electron micrographs showed the average silver particle size to be 40 nm in diameter (335x10-23m3)17 The silver concentration is 012M based on the reactant molar concentrations and dilution factors And the scattering volume is 76x10-
11m3 again based on a cylindrical scattering volume defined by a laser area of 28x10-7m2 and a sol-gel thickness of 27x10-
4m This volume contains 123x10-6g of silver equivalent to 35x109 silver particles with a collective surface area of 18x10-
5m2 However it may reasonably be assumed that at least half of the silver surface is in contact with the sol-gel matrix and unavailable for analyte interaction If we assume monolayer coverage and that each CN molecule occupies 15x10-20m2 then approximately 62x1014 molecules or 27x10-8g of CN contribute to the SER spectrum (20x10-19m2 46x1013 molecules 74x10-9g for MPA) Accordingly the EF for cyanide equals 48x104 ((180599) bull(16x10-427x10-8) bull(90075) bull(500100)12) The EF for MPA is considerably higher at 87x106 ((60326) bull(144x10-474x10-9) bull(90075) bull(500200)12)
0
20
40
60
80
100
120
140
0 02 04 06 08 1 12[MPA] (mgmL)
I (76
0)
0
100
200
300
400
500
600
I (10
50)
Wavenumber (∆cm-1)
A B
Proc SPIE Vol 4577
172
4 CONCLUSIONS Here we present for the first time surface-enhanced Raman spectra of Tabun and Sarin obtained using silver-doped sol-gels However the inferior enhancement suggested employing SER-active sol-gels varying in composition to improve both detection limits and selectivity of the target analytes To this end we successfully demonstrated the capabilities of four sol-gels that select for 1) polar-positive 2) polar-negative 3) weakly polar-positive and 4) weakly polar-negative chemical species p-aminobenzoic acid was used to show that silver could be used to attract polar-negative chemicals or functional groups (carboxylate anion) and that gold could be used to attract polar-positive chemicals (amine) while a mixture of p-aminobenzoic acid and phenylacetylene was used to show that tetramethyl orthosilicate preferentially solvates polar chemicals and that tetraethyl orthosilicate preferentially solvates weakly polar chemicals This increased sample control was applied to cyanide and methyl phosphonic acid two hydrolysis products of chemical warfare agents Exceptional results were obtained for methyl phosphonic acid allowing measurement of 1x10-2 gL for a 3-min scan and 75 mW of incident laser power with an estimated detection limit of 245x10-4 gL and an enhancement factor of 87x106 However this detection limit is 76 times less sensitive than required for the JSAWM (32x10-6gL for the G-agents) Finally we note that the measurements performed here employed an FT-Raman spectrometer This instrumentation was chosen over dispersive Raman instrumentation for the high wavelength accuracy afforded by the HeNe reference laser (Connes Advantage25) which would allow reliable spectral subtraction matching of observed spectra to stored library spectra and confident use of chemometric approaches Such data analysis is likely to be required to identify the chemical agents as well as distinguish hydrolysis products However this instrumentation which employs 1064 nm excitation and InGaAs detection sacrifices sensitivity We believe that the measurement sensitivity can be improved by at least two orders of magnitude by using shorter laser excitation wavelengths (eg 532 nm) This would provide more efficient Raman scattering (fourth power dependence on laser excitation wavelength) more efficient generation of plasmon modes and allow using more efficient detector material (Si vs InGaAs) These modifications are underway
5 ACKNOWLEDGEMENTS The authors would like to thank Dr R Yin and J Jensen of the US Army for supporting this work (Contract Number DAAD13-01-C-0019) They would also like to thank Advanced Fuel Research for making their laboratory facilities available
6 REFERENCES 1 Tu Anthony ldquoOverview of Sarin Terrorist Incidents in Japan in 1994 and 1995rdquo 6th CBW Protection Symposium
Stockholm Sweden 10-15 May 1998 2 Jensen JL NC Wong W R Loerop SPIE 4775-03 (2001) 3 JSAWM Requirements at wwwsbccomapgeaarmymilRDAecbcrtPRODSERJSAWMjsawmhtml 4 ldquoThe Chemical Weapons Convention Redefines Analytical Challengerdquo Analytical Chemistry News amp Features June 1
397A (1998) 5 Johnston RL Hoefler CM Fargo JC and Moberley B ldquoThe Defense Nuclear Agencyrsquos ChemicalBiochemical
Weapons Agreements Verification Technology Research Development Test and Evaluation Program and its Requirements for On-Site Analysisrdquo AT-ONSITE 5-8 (1994)
6 Black RM Clarke RJ Read RW and Reid MT ldquoApplication of gas chromatography-mass spectrometry and gas chromatography-tandem mass spectrometry to the analysis of chemical warfare samples found to contain residues of the nerve agent sarin sulphur mustard and their degradation productsrdquo J Chromatography 662 301-321 (1994)
7 Hoffland LD Piffath RJ Bouck JBrdquoSpectral signatures of chemical agents and simulantsrdquo Optical Engineering 24 982-984 (1985)
8 Braue EHJ Pannella MGrdquoCIRCLE CELL FT-IR Analysis of Chemical Warfare Agents in Aqueous Solutionsrdquo Applied Spectroscopy 44 1513-1520 (1990)
9 Tseng C-H Mann CK and Vickers TJ ldquoDetermination of Organics on Metal Surfaces by Raman Spectroscopyrdquo Applied Spectroscopy 47 1767-1771 (1993)
10 Staff Reporter ldquoUS Army Tests Lidar to Detect Biological Toxinsrdquo Photonic Spectra pg 50 December 1998
Proc SPIE Vol 4577
Proc SPIE Vol 4577
173
11 Christesen SD Raman cross sections of chemical agents and simulants Applied Spectroscopy 42 318-321 (1988) 12 Alak AM and Vo-Dinh T ldquoSERS of Organophosphorous Chemical Agentsrdquo Analytical Chemistry 59 2149-2153
(1987) 13 Norrod KL Sudnik LM Rousell D and Rowlen KL ldquoQuantitative Comparison of Five SERS Substrates
Sensitivity and Detection Limitrdquo Applied Spectroscopy 51 994-1001 (1997) 14 Lee Y and Farquharson S ldquoSERS Sample Vials Based on Sol-Gel Process for Trace Pesticide Analysisrdquo SPIE 4206
140-146 (2000) 15 Farquharson S and Lee Y ldquoTrace Drug Analysis by Surface-Enhanced Raman Spectroscopyrdquo SPIE 4200-16 (2000) 16 Lee Y Farquharson S and Rainey P M Surface-Enhanced Raman Sensor for Trace Chemical Detection in Water
SPIE 3857 76-84 (1999) 17 Lee Y Farquharson S Kwong H and Shahriari M ldquoSurface-Enhanced Raman Sensor for Surface-Enhanced Raman
Spectroscopyrdquo SPIE 3537 252-260 (1998) 18 Farquharson S Smith W Carangelo R C and Brouillette C ldquoIndustrial Raman Providing Easy Immediate Cost
Effective Chemical Analysis Anywhererdquo SPIE 3859 14-23 (1999) 19 Narayanan VA JM Bello JD Stokes and T Vo-Dinh Analusis 19 307-310 (1991) 20 Laserna JJ E L Torres and JD Winefordner Analytica Chemica Acta 469-480 (1987) 21 Torres EL and JD Winefordner Analytical Chemistry 59 1626-1632 (1987) 22 Suh JS DP DiLella M Moskovits J Phys Chem 87 1540-1544 (1983) 23 Benner RE R Dornhaus R Chang and BL Laube Correlations in the Raman spectra of cyanide complexes adsorbed
at silver electrodes with voltammograms Surface Science 101 341 (1980) 24 Chase D B and JF Rabolt Fourier Transform Raman Spectroscopy Acad Press Ch1 p 131 (1994) 25 Connes J Rev Opt Theor Instrum 40 45 (1961)
APPLIED SPECTROSCOPY 351
focusing the 488 nm laser beam 2 mm deep into thebulk of the crystals to avoid contributions from potentialdepletion layers As shown in Fig 1b the dependence ofthe Raman shift on the degree of deuteration is almostperfectly linear and ts very well with D 5 22684cmR1 24526 where D is the degree of deuteration (in )and R is the spectral mean of the PO4 vibration in cm21A linear correlation coef cient of 0998 indicates an ex-cellent linear dependence of the Raman peak shift withdegree of deuteration This result shows that the shift ofthe PO 4 peak is simply caused by the linear increase inatomic mass due to isotope substitution which decreasesthe length of hydrogen-like bonds
This excellent linear dependence allows us to map thepro le of the DH exchange layer at the surface of DKDPcrystals by acquiring Raman spectra and determining theposition of the PO 4 peak for various depths This methodis preferable over other methods such as determining thestrength of the OD vibration directly (eg at 715 cm21)because the position of the most intense peak in the Ra-man spectrum can be measured more precisely than theintensity of some of the weakest peaks in the spectrumThis is demonstrated in Fig 2 where depth-dependentRaman spectra (Fig 2a) and the resulting exchange layerpro les for two DKDP crystals are shown (Fig 2b) Thespectra in Fig 2a were obtained from a depth scan of aDKDP crystal with 75 degree of deuteration in thebulk grown at 45 8C The spectra start out as DKDP with30 deuteration close to the surface and approach thebulk DKDP spectrum within a few micrometers of depthThe fact that the relative degree of deuteration does notextend to 0 D is due to the limited depth resolution ofthe Raman microprobe which averages over 4 mm indepth Figure 2b depicts the resulting DH exchange layerpro les for this and a second crystal grown at 63 8Crespectively Both crystals had the same exposure to am-bient conditions and their main difference is the temper-ature at which they were grown The different exchangelayer pro les indicate that crystals grown at differenttemperatures have differen t proton conductiv ities 13
which leads to a difference in their rate of deuteriumdepletion The parameters controlling this behavior arecurrently the objective of a detailed study the results ofwhich will be reported elsewhere
CONCLUSION
In conclusion we have shown that the shift of the to-tally symmetric PO 4 stretch mode in the Raman spectrumof DKDP crystals scales linearly with degree of deuter-ation This allows us to correlate Raman peak positionsto deuteration levels in these crystals We have presenteda new technique to determine DH diffusion pro les inDKDP frequency conversion crystals based on micro-Ra-man spectroscopy This technique is fast inexpensiveand works under various environmental conditionswhich will allow us to better understand and control deu-terium depletion in DKDP crystals
ACKNOWLEDGMENTS
We would like to thank M Runkel for rst discovering DKDP crack-ing R Floyd for providing DKDP crystals and L Chase and A Burn-ham for their support and helpful discussions This work was performedunder the auspices of the US Department of Energy by the University
of California Lawrence Livermore National Laboratory under contract W-7405-Eng-48
1 J J De Yoreo A K Burnham and P K Whitman Int Mat Rev47 113 (2002)
2 C E Barker R A Sacks B M Van Wonterghern J A Caird JR Murray J H Campbell K Kyle R B Ehrlich and N DNielsen Proc SPIE-Int Soc Opt Eng 2633 501 (1995)
3 T Suratwala paper to be published4 Cleveland Crystals Inc httpwwwclevelandcrystalscom
KDPshtmltable5 E A Popova I T Savatinova and I A Velichko Sov Phys Solid
State 12 1543 (1971)6 I P Kaminow R C C Leite and S P S Porto J Phys Chem
Solids 26 2085 (1965)7 J A Subramony B J Marquardt J W Macklin and B Kahr
Chem Mat 11 1312 (1999)8 H Tanaka M Tokunaga and I Tatsuzaki Solid State Commun
49 153 (1984)9 R J Nelmes G M Meyer and J E Tibballs J Phys C 15 59
(1982)10 M A Yakshin D W Kim Y S Kim Y Y Broslavets O E
Sidoryuk and S Goldstein Laser Physics 7 941 (1997)11 I Takenaga Y Tominaga S Endo and M Kobayashi Solid State
Commun 84 931 (1992)12 C Krenn personal communication13 M Sharon and A K Kalia J Solid State Chem 21 171 (1977)
Rapid Dipicolinic Acid Extractionfrom Bacillus Spores Detectedby Surface-EnhancedRaman Spectroscopy
STUART FARQUHARSON ALAND GIFT PAUL MAKSYMIUK andFRANK E INSCOREReal-Time Analyzers Inc East Hartford Connecticut06108
Index Headings Dipicolinic acid Bacillus spores Anthrax Surface-enhanced Raman spectroscopy
INTRODUCTION
The anxiety caused by the distribution of anthrax en-dospores through the US postal system in October 2001was exacerbated by the long time required for positiveidenti cation of the Bacillus anthracis spores and the un-known extent of their distribution Since that time manymethods capable of rapid eld analysis have been inves-tigated to augment or replace the laboratory method ofgrowing microorganisms in culture media which takesdays to perform12 Prominent among these approachesare polymerase chain reactions (PCR)3 immunoassaysand detection of calcium dipicolinate as a biochemicalsignature PCR employs primers to separate organism-
Received 10 October 2003 accepted 14 November 2003 Author to whom correspondence should be sent
352 Volume 58 Number 3 2004
speci c nucleic acid sequences (eg capsular protein en-coding gene for Bacillus anthracis)4 and polymerases toamplify the segment until it is detectable Recently am-pli cation times have been substantially reduced andcomplete analysis can now be performed in an hour orless Immunoassay methods are also being developed thatuse competitive binding of the bioagent (as an antigen)and its labeled conjugate for a limited number of anti-bodies Although analyses can be performed in under 30minutes a well-de ned anthrax antigen has not yet beenidenti ed5ndash7 and consequently the false-positive rate isunacceptably high8
A number of other methods are being developed witha focus on the detection of calcium dipicolinate (CaDPA)and its derivatives as a B anthracis signature This is sobecause only spore-forming bacteria contain CaDPA andthe most common potentially interfering spores such aspollen and mold spores do not Relatively fast methodshave been developed to chemically extract CaDPA andthen detect it directly by uorescence9 or indirectly byluminescence1011 In the latter case hot dodecylamine(DDA) has been used to extract dipicolinic acid (DPA)and terbium has been utilized to form a highly lumines-cent DPA complex11 Although measurements have beenperformed in as little as ve minutes it was found thatas many as three concentration-dependent complexes canform each with different lifetimes This coupled withthe fact that the Tb31 cation produces the same lumines-cence spectrum makes determinations of low spore con-centrations problematic
It has been long known that Raman spectra of Bacillispores are dominated by bands associated with CaDPA12
and that these spectra may provide a suitable anthrax sig-nature at the genus level13 Since that time considerableimprovements in Raman instrumentation have led to lab-oratory measurements of single Bacilli spores14 and to eld measurements of spores captured from a mail-sort-ing system15 However the single spore measurementsrequired complex instrumentation that is not ruggedwhile the eld measurements required milligrams of sam-ple Furthermore the Raman spectra of both measure-ments contained uorescence contributions that would in-crease uncertainty in quanti cation
In related research we demonstrated that nanogramquantities of DPA could be detected by uorescence-freesurface-enhanced Raman spectroscopy (SERS)16 We alsodemonstrated that microliter volumes of chemicals canbe detected by SERS using metal-doped sol-gel-packedglass capillaries17 Towards the goal of developing a rap-id eld SERS-based anthrax spore detector we havecombined our previous research and we now report thatDPA can be extracted from a 10 mg B cereus spore sam-ple using DDA in 1 minute and can be detected by SERSin an additional 1 minute
EXPERIMENTAL
Dipicolinic acid (26-pyridinedicarboxylic acid DPA)and dodecylamine (DDA) were used as received fromSigma-Aldrich (Milwaukee WI) Lyophilized B cereusspores prepared according to the literature13 were sup-plied by the University of Rhode Island and used as re-ceived Multiple particles approximately 01 mm3 each
were separated and weighed at 5 to 15 mg representing05 to 15 million spores The sample masses were con-sistent with a previous determination of spore density at0081 gmL that indicated a high degree of entrained air
All chemicals used to prepare the silver-doped sol-gelcoated capillaries were also obtained and used as receivedfrom Sigma-Aldrich According to previously publishedprocedures17 two precursor solutions were preparedmixed and then drawn into 1-mm-diameter glass capil-laries The silver amine precursor consisted of a 51 vvratio of 1 N AgNO 3 to 28 NH3OH while the alkoxideprecursor consisted of a 21 vv ratio of methanol to te-tramethyl orthosilicate The alkoxide precursors weremixed with silver amine precursor in an 81 vv ratioApproximately 015 mL was drawn into the capillarycoating a 15-mm length After sol-gel formation the in-corporated silver ions were reduced with dilute sodiumborohydride which was followed by a water wash to re-move residual reducing agent
A 100 mL drop of a 50 mM DDA solution in ethanolpre-heated to 78 8C was added to each of the B cereusparticles to digest the spore coat After 1 minute the re-sultant solution was drawn into a SER-active capillarythat was immediately xed horizontally to an XY posi-tioning stage (Conix Research Spring eld OR) just in-side the focal point of an f 07 aspheric lens The lensfocused the beam into the sample and collected the scat-tered radiation back along the same axis A dichroic lter(Omega Optical Brattleborough VT) was used to re ectthe excitation laser to the lens and pass the Raman scat-tered radiation collected by the lens An f 2 achromatwas used to collimate the laser beam exiting a 200-mm-core-diameter source ber optic while a second f 2 ach-romat was used to focus the scattered radiation into a 365mm ber optic (Spectran Avon CT) A short-pass lterwas placed in the excitation beam path to block the sil-icon Raman scattering generated in the source ber fromre ecting off sampling optics and reaching the detectorA long-pass lter was placed in the collection beam pathto block the sample Rayleigh scattering from reachingthe detector A 785 nm diode laser (Process InstrumentsInc model 785-600 Salt Lake City UT) was used todeliver 100 to 150 mW of power to the sample A Fouriertransform Raman spectrometer (Real-Time Analyzersmodel IRA-785 East Hartford CT) and a silicon photo-avalanche detector (Perkin Elmer model C30902S Stam-ford CT) were used to acquire the SER spectra
RESULTS AND DISCUSSION
As an initial experiment the SER spectrum of 1 gLof DPA in water was measured using the newly devel-oped silver-doped sol-gel-coated capillaries (Fig 1A) Atthis concentration a high signal-to-noise ratio (SN) isobtained in 1 min In fact a reasonable spectrum is ob-tained in the same time frame for 1 mgL (Fig 1B) TheSER spectra are reasonably similar to the normal Raman(NR) spectrum obtained for a saturated solution of DPAin 1 N KOH (Fig 1C) and the following band shifts areobserved (NR to SER) 647 to 657 cm21 817 to 815cm21 998 to 1008 cm21 1384 to 1382 cm21 1434 to1428 cm21 and 1569 to 1567 cm21 Many of these bandshave been previously assigned1213 such as 998 cm21 to
APPLIED SPECTROSCOPY 353
FIG 1 SERS of DPA in water using silver-doped sol-gel-coated glasscapillary for (A) 1 gL and (B) 1 mgL (C ) NR of saturated DPA in 1N KOH in a glass capillary Spectral conditions (A) and (B) 150 mWof 785 nm 1-min acquisition time (C) 450 mW of 785 nm 5-minacquisition time both 8 cm 21 resolution
FIG 2 SERS of DPA extracted from 10 mg B cereus particle using100 mL of 50 mM hot DDA acquired in (A) 1 minute and (B) 2 seconds(C ) Attempted SERS of 50 mM hot DDA in ethanol using silver-dopedsol-gel-coated glass capillary acquired in 1 min Spectral conditions150 mW of 785 nm 8 cm21 resolution
the symmetric ring stretch 1384 cm21 to the OndashCndashOsymmetric stretch 1428 cm21 to the symmetric ring CndashH bend and 1569 cm21 to the asymmetric OndashCndashOstretch
The rst B cereus samples consisted of 2 mg of sporesin 2 mL of 5 mM hot DDA The samples were main-tained at 78 8C for 40 min and while hot approximately10 mL was drawn into a SER-active capillary Since spec-tra of DPA were obtained for these initial samples small-er spore masses higher DDA concentrations and shorterheating periods were examined In due course it wasfound that 10 mg of spores could be digested by 100 mLof 50 mM hot DDA in one minute and detected (Fig2A) In fact the signal was suf ciently intense that it canbe observed in as little as two seconds (Fig 2B) Theamount of DPA that was extracted was estimated to bebetween 5 and 10 mgL by comparing the signal intensityof the 1008 cm21 band to that measured for DPA in waterThis is consistent with previous research that found thatthe majority of the DPA is extracted from spores usingDDA11 and that B cereus spores contain approximately10 DPA by weight18 The SN of 127 for the 1008 cm21
band in the 1-minute SER spectrum suggests a limit ofdetection of approximately 250 ng of B cereus sporesbased on a SN of 3 Finally it should be noted that DDAdid not produce a detectable SER spectrum as shown inFig 2C
CONCLUSION
We have demonstrated that by combining rapid extrac-tion of dipicolinic acid from Bacillus cereus spores withchemical identi cation by surface-enhanced Raman spec-troscopy as little as 10 mg of spores can be detected Infact the entire measurement from the time of adding hotdodecylamine to the spores to the time when the dipi-colinic acid SER spectrum is acquired and analyzedcould be performed in less than two minutes The abilityof this method to distinguish between spore-forming bac-teria such as Bacillus anthracis and non-DPA containingpowders could help prevent costly shutdowns associated
with the appearance of suspicious material or intentionalmailing of common substances as an anthrax hoax Thismethod could also prove useful in detecting the locationof anthrax endospores in mail distribution facilities if an-other veri ed attack should occur
Research continues to fully characterize the surface-enhanced Raman spectroscopy signal intensities as afunction of sample concentration and to explore otherextractants that do not require the use of elevated tem-perature
ACKNOWLEDGMENTS
The authors are grateful for the support of the National Science Foun-dation (DMI-0296116 and DMI-0215819) and the US Army(DAAD13-02-C-0015 Joint Service Agent Water Monitor program)The authors are indebted to Chetan Shende for preparing the sol-gelcapillaries The authors also thank James Gillespie Nicholas Fell andAugustus Fountain for providing important background informationMark Farquharson for laboratory support and Professor Jay Sperry ofthe University of Rhode Island for supplying B cereus spores
1 V A Pasechnik C C Shone and P Hambleton Bioseparations 3267 (1993)
2 P J Jackson M E Hugh-Jones D M Adair G Green K K HillC R Kuske L M Grinberg F A Abramova and P Keim ProcNatl Acad Sci USA 95 1224 (1998)
3 B R Glick and J J Pasternak Molecular Biology Principles andApplications of Recombinant DNA (ASM Press Washington DC1994)
4 C A Bell J R Uhl T L Had eld J C David R F Meyer TF Smith and F R Cockerill III J Clin Microbiol 40 2897(2002)
5 D L Gatto-Menking H Yu J G Bruno M T Goode M Millerand A W Zulich Biosens Bioelectron 10 501 (1995)
6 J J Quinlan and P M Foegeding J Rapid Methods AutomationMicrobiol 6 1 (1998)
7 A A Hindle and E A H Hall Analyst (Cambridge UK) 1241599 (1999)
8 M S Ascher US Department of Health amp Human Services(httpwwwhhsgovophppresentationsAscherdoc)
9 R Nudelman B V Bronk and S Efrima Appl Spectrosc 54 445(2000)
10 D L Rosen C Sharpless and L B McBrown Anal Chem 691082 (1997)
354 Volume 58 Number 3 2004
11 P M Pellegrino N F Fell Jr and J B Gillespie Anal ChimActa 455 167 (2002)
12 W H Woodruff T G Spiro and C Gilvarg Biochem BiophysRes Commun 58 197 (1974)
13 E Ghiamati R S Manoharan W H Nelson and J F SperryAppl Spectrosc 46 357 (1992)
14 A P Esposito C E Talley T Huser C W Hollars C M Schal-dach and S M Lane Appl Spectrosc 57 868 (2003)
15 S Farquharson L Grigely V Khitrov W W Smith J F Sperryand G Fenerty J Raman Spectrosc paper accep ted (2003)
16 S Farquharson W W Smith S Elliott and J F Sperry SPIE-IntSoc Opt Eng 3855 110 (1999)
17 S Farquharson and P Maksymiuk Appl Spectrosc 57 479(2003)
18 F W Janssen A J Lund and L E Anderson Science (Washing-ton DC) 127 26 (1958)
SPIE -2003-5269 117
pH dependence of methyl phosphonic acid dipicolinic acid and cyanide by surface-enhanced Raman spectroscopy
Stuart Farquharson Alan Gift Paul Maksymiuk Frank Inscore and Wayne Smith
Real-Time Analyzers 87 Church Street East Hartford CT 06108
ABSTRACT US and Coalition forces fighting terrorism in Afghanistan and Iraq must consider a wide range of attack scenarios in addition to car bombings Among these is the intentional poisoning of water supplies to obstruct military operations To counter such attacks the military is developing portable analyzers that can identify and quantify potential chemical agents in water supplies at microgram per liter concentrations within 10 minutes To aid this effort we have been investigating the value of a surface-enhanced Raman spectroscopy based portable analyzer In particular we have been developing silver-doped sol-gels to generate SER spectra of chemical agents and their hydrolysis products Here we present SER spectra of methyl phosphonic acid and cyanide as a function of pH an important factor affecting quantitation measurements which to our knowledge has not been examined In addition dipicolinic acid a chemical signature associated with anthrax-causing spores is also presented Keywords Chemical warfare agents agent detection agent hydrolysis SERS Raman spectroscopy homeland security
1 INTRODUCTION In the past decade the Unites States and its allies have been challenged by a different kind of warfare exemplified by the terrorist attacks of September 11 2001 Suicide bombings and the use of chemical agents are the norm and military personnel must consider a wide range of attack scenarios Among these is the intentional poisoning of water supplies to obstruct military operations in Afghanistan and Iraq Currently colorimetric paper is used to detect agents on-site while gas chromatography combined with mass spectrometry (GCMS) is used in mobile support laboratories However both methods have severe drawbacks The paper changes color in response to contact with many chemicals besides CWAs causing a high incidence of false positives1 while GCMS although very chemically specific requires hours to perform and constant re-calibration234 Military operations would be greatly aided by a portable analyzer that can identify and quantify potential chemical agents at concentrations that impact safety This includes the analysis of drinking water supplies distribution and storage systems To meet this goal the Department of Defense has been investigating numerous approaches under the auspices of the Joint Service Agent Water Monitor (JSAWM) program5 The JSAWM must be capable of identifying the chemical agents with no false-positives and quantifying the agents at microgram per liter concentrations within 10 minutes678 Recently we and others have been investigating the ability of surface-enhanced Raman spectroscopy (SERS) to measure chemical agents9-12 bioagents13-17 and their hydrolysis products in water SERS employs the interaction of surface plasmon modes of metal particles with target analytes to increase scattering efficiency by as much as 1 million times18 In 1987 the potential of this approach to measure CWAs was demonstrated by measuring a series of pesticides19 Several of these organophosphonates have chemical structures similar to CWAs in particular P=O functional groups In our studies we have been employing silver-doped sol-gels to promote the SER effect The porous silica network of the sol-gel matrix offers a unique environment for immobilizing and stabilizing SER-active metal particles20-23 The sol-gels have been coated on the internal walls of standard glass vials to allow rapid SER analysis of small sample quantities without preparation We have measured over 100 chemicals with enhancements of 104 to 106 demonstrated reversible measurements in a flowing system reproducible measurements from vial-to-vial and measurements in multiple solvents including water20-23 Previously we used these vials to perform preliminary measurements of cyanide (CN) methylphosphonic acid (MPA) and dipicolinic acid (DPA) MPA is a hydrolysis product of the nerve agents (eg sarin Reaction 1) and may be a valuable indicator of nerve agent usage particularly since the alkyl methylphosphonic acids are relatively more stable than their corresponding parent complexes24 DPA is
SPIE -2003-5269 118
a chemical signature of spore forming bacteria such as Bacillus anthracis And in light of the inability to rapidly detect the anthrax spores distributed through the US mail in October 2001 a number of methods are being developed to extract and analyze this signature Reaction 1 Stepwise hydrolysis of Sarin to form hydrofluoric acid (HF) isopropyl methylphosphonic acid (IMPA) then methyl phosphonic acid (MPA) and 2-propanol In our previous SERS investigations MPA and DPA were measured at 50 and 100 mgL respectively In both cases limits of detection (LOD) were estimated at 100 microgL providing encouragement in that SERS may satisfy the needs of the JSAWM Since it has been shown that pH can substantially influence the intensity of SER bands25 which would clearly influence quantitative analysis we undertook the present study to determine the severity of these effects for cyanide methyl phosphonic acid and dipicolinic acid Furthermore we previously observed a band at 1050 cm-1 for MPA6 possibly due to an anion formed at basic pH Here we investigate the source of this spectral anomaly
2 EXPERIMENTAL All chemicals including potassium cyanide methyl phosphonic acid dipicolinic acid and those used to prepare the silver-doped and gold-doped sol-gel coated vials were obtained and used as received from Sigma-Aldrich All samples were prepared in HPLC grade water (Fischer Scientific Fair Lawn NJ) for SERS measurements The pH of these samples was adjusted using dilute nitric acid or potassium hydroxide and verified using a pH electrode (Corning Inc Corning NY) that had been calibrated with pH 400 700 and 1000 buffered standards from Fischer Scientific Once prepared the samples were transferred into the silver-doped sol-gel vials (Real-Time Analyzers Inc East Hartford CT) The vials were coated in a manner similar to that previously reported by adding ammonium hydroxide to a solution of silver nitrate tetramethyl orthosilicate (TMOS) and methanol20 Gold-vials were coated by adding nitric acid to a solution of gold tetrachloride TMOS and methanol The two precursor solutions were prepared mixed and transferred to 2-mL glass vials dried and heated After sol-gel formation the incorporated metal ions were reduced with dilute sodium borohydride (1mgmL) which was followed by a water wash to remove residual reducing agent After the resultant analyte solution was introduced the SER-active vial was immediately fixed horizontally to an XY positioning stage (Conix Research Springfield OR) just inside the focal point of an f07 aspheric lens The lens focused the beam into the sample and collected the scattered radiation back along the same axis A dichroic filter (Omega Optical Brattleborough VT) was used to reflect the excitation laser to the lens and pass the Raman scattered radiation collected by the lens An f2 achromat was used to collimate the laser beam exiting a 200 microm core diameter source fiber optic while a second f2 achromat was used to focus the scattered radiation into a 365 microm fiber optic (Spectran Avon CT) A short pass filter was placed in the excitation beam path to block the silicon Raman scattering generated in the source fiber from reflecting off sampling optics and reaching the detector A long pass filter was placed in the collection beam path to block the sample Rayleigh scattering from reaching the detector A 785 nm diode laser (Process Instruments Inc model 785-600 Salt Lake City UT) was used to deliver 100 to 150 mW of power to the sample A Fourier transform Raman spectrometer (Real-Time Analyzers model IRA-785 East Hartford CT) and a silicon photo-avalanche detector (Perkin Elmer model C30902S Stamford CT) were used to acquire the SER spectra
3 RESULTS AND DISCUSSION In a previous study of MPA6 aimed at developing a concentration calibration curve and determining limits of detection (LOD) we observed an anomaly at 1050 cm-1 Since it was found that the band intensity changed as a function of concentration the band must be associated with a sample parameter Two possible parameters photon flux and pH are examined here The first parameter was investigated by irradiating a 1mgmL MPA sample in a SER-active vial with laser powers of 200 mW and above and monitoring spectral changes It was immediately found that the 1050 cm-1 band
2O+ H HF + +OH
OH3H C
OP
OF
CH
CH3
3
3H C
OCP
OOH
CH
CH3
3
3H C
O
CPHO
CH
CH
3
3
C
Sarin IMPA MPA 2-propanol
SPIE -2003-5269 119
appeared and grew as a function of time and that the higher the laser power the faster the growth Figure 1 shows the growth of the 1050 cm-1 band over the course of 30 minutes when using 150 mW of 785 nm excitation while Figure 2A shows that the growth can be fit with a first order exponential equation namely I1050 = 03+05e-013t Furthermore the 760 cm-1 band could be fit with a first order decay equation with an identical exponential rate constant ie I760 = 08-
08e-013t The rates represent classical first order kinetics and their correspondence allows one to conclude that MPA is being transformed one-for-one into a photo-generated product At this time the photoproduct has not been positively identified but phosphonic (phosphorous) acid and phosphonate are likely candidates since the symmetric P(OH)3 stretch occurs at ~1050cm-1 Our studies show that a reduction in laser power to 100 mW at the sample for MPA essentially eliminates this degradation process This laser power or lower was used for further measurements
Methyl phosphonic acid is a diprotic acid that stepwise dissociates into two anions MPA- and MPA= according to the following reactions26
MPA MPA- + H+ pKa1 = 212 Reaction 2
MPA- MPA= + H+ pKa2 = 729 Reaction 3 The relative concentrations of MPA MPA- and MPA= can be determined at any pH by expressing [MPA] and [MPA=] in terms of [MPA-] using Reactions 2 and 3 and summing all three to equal the total starting concentration here 2 mgmL (0021M MW = 9602) viz
[MPA] + [MPA-] + [MPA=] = 0021M Equation 1 substituting from Reactions 1 and 2
([H+][MPA-])K1a + [MPA-] + (K2a[MPA-])[H+] = 0021M Equation 2 rearranging [MPA-] = 0021M(1+[H+]K1a + K2a[H+]) Equation 3 The relative concentrations of MPA MPA- and MPA= as a function of pH are shown in Figure 3 It is worth noting that near neutral pH both MPA- and MPA= will be present To confirm that the SER signal followed this pH dependence a starting solution consisting of 20 mg of MPA in 10 mL HPLC grade water was prepared and brought to pH of 20 using dilute nitric acid From this solution 2 mL were added to a SER-active vial and the SER spectrum recorded At this pH a peak at 760 cm-1 was barely discernable The 2 mL solution was returned to the starting solution and the pH was re-measured to correct for any changes that the silver-doped sol-gel vials might cause In most cases the change was less than 02 pH units and the pH is reported as the before and after average Next the pH of the
Figure 2 A) Exponential growth of 1050 cm-1 band and B) exponential decay of 760 cm-1 band for spectral series in Figure 1
Figure 1 Growth of 1050 cm-1 band as a function of time due to exposure to 150 mW of 785 nm Spectra are 5 sec each collected every 100-sec from 0 to 30-min
B
570 770 970 1170 1370Raman Shift (cm-1)
Arbit
rary
Unit
s
MPA Photodegradation
600 800 1000 1200 1400 Wavenumber (cm-1)
A
0 10 20 30 0 10 20 30 time (min) time (min)
Ram
an In
tens
ity (r
elat
ive)
30
min
0
SPIE -2003-5269 120
starting solution was adjusted to 325 using dilute KOH Again 2 mL were added to a vial and the SER spectrum recorded At this pH a reasonably strong 760 cm-1 band was observed This process was repeated as spectra were recorded at pHs of 70 74 75 79 85 and 100 A total of 1 mL of KOH was added diluting the total concentration by 10 Next the pH of the starting solution was made acidic by adding dilute nitric acid dropwise This time spectra were recorded at pHs of 72 69 64 and 37 Figure 4 shows the SER spectra for representative pHs (spectra were left out to simplify the figure) while Figure 3 shows the 760 cm-1 peak intensities as a function of pH (The band intensities were adjusted to compensate for dilution effects caused by the addition of HNO3 and KOH then normalized to 0021 M for the most intense band observed at pH 37) It is clear from Figure 3 that the 760 cm-1 band follows the MPA- concentration as a function of pH and must be assigned to this anion No bands were observed that corresponded to MPA or MPA= The lack of an MPA SER spectrum may be due to the absence of an attraction between the neutral analyte and the electropositive silver surface The same reasoning suggests that a strong SER spectrum should be observed for MPA= but it is not and a satisfactory explanation has not been found
0000
0005
0010
0015
0020
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14pH
Con
cent
ratio
n [M
]
pK1 = 212 pK2 = 729
MPA- MPA=
MPA
Figure 4 SER spectra of 002M MPA as a function of pH Conditions 100 mW of 785 nm 36 scans (1 min) 8 cm-1 recorded 2 min after sample introduction pH 19 and 20 not apparent on this scale pH 69 and 74 near identical to 70 and 75 and not shown for clarity
00
02
04
06
08
10
12
14
16
18
0 5 10 15 20 25 30Measurement Number
Ram
an In
tens
ity (7
60 c
m-1
)
Figure 3 Concentration dependence of MPA MPA- and MPA= as a function of pH for a 002M sample Intensity of 760 cm-1
band from Figure 6 as a function of pH ( for increasing basic adjustment for increasing acidic adjustment error was measured at ~10 for pH 64)
Figure 5 SER spectra of 002M MPA at pH 64 measured around a vial at three heights (9 points per height) Conditions as in Figure 4 but 10-sec scans
Since these measurements involved the removal and replacement of the SER-active vial in the sample holder to remove and add sample variation in the intensity as a function of vial position was minimized by illuminating the exact same height along the vial wall But this does not account for variability of the SERS response of the sol-gel coating around the vial To analyze this effect a vial containing MPA at pH 64 was rotated at ~ 40o
intervals at the original height and 18rdquo above and below this value Figure 5 shows the intensity of the 760 cm-1
band for the 27 positions It was found that the average value was 137plusmn014 an RSD of 10 overall and 5 for each height An error bar is included in Figure 4 for the pH 64 measurement
SPIE -2003-5269 121
With the above analyses in mind a preliminary investigation of the SER spectral response for MPA (as MPA-) as a function of concentration was performed A single vial was used for these measurements beginning with 1 mgL followed by measurements of 10 100 and 1000 mgL In all cases the pH was ~7 and 3 positions around the vial were measured per concentration Since the 760 cm-1 band was not observed for concentrations of 1 or 10 mgL using 100 mW of 785 nm the laser power at the sample was raised to 200 mW beginning with the 10 mgL concentration Photo-degradation was largely avoided (and not observed) by exposing the sample for only 33 seconds per spectral acquisition Representative spectra for 10 100 and 1000 mgL are shown in Figure 6 while a plot of the 760 cm-1 band intensity as a function of concentration is shown in Figure 7 These values were also used to estimate limits of detection based on the signal-to-noise ratio (SN) of the 760 cm-1 band using 100 mW of 785 nm laser excitation and a 10-min acquisition time As summarized in the Figure 7 inset the lower the measured concentration the lower the predicted LOD Even if as estimated 210 microgL could be measured using the silver-doped sol-gel vials an improvement of a factor of 70 is still required to meet JSAWM goals of measuring 3 microgL in 10-minutes Similar to MPA DPA is a diprotic acid (pKa1 = 216 and pKa2 = 692) and variations in pH will effect the relative concentrations of DPA DPA- and DPA= and possibly the SER spectra and band intensities This could prove significant if an acid or base is used to denature anthrax spores with the goal of extracting and analyzing DPA The MPA pH study described above was mimicked for DPA except that the starting solution consisted of 20 mg of DPA in 20 mL HPLC grade water (60x10-3M MW = 1671) The initial solution had a pH of 245 which was made basic by dilute KOH to pHs of 355 433 487 559 1069 and 1166 SER spectra were recorded at each pH using 100 mW of 785 nm and a 44-sec acquisition time Next one drop of concentrated nitric acid was used to remake the solution acidic at a pH of 200 Again sequential pH measurements were performed at 383 510 735 and 822 The solution pH was made acidic a third time but to pH 219 171 then 135 Throughout this process no more than 20 drops of acid or base were added and therefore the concentration was diluted by no more than 10 Most of the spectral bands showed a minor decrease in intensity as a function of increasing pH values However the bands at 525 and 795 cm-1 showed the most dramatic changes which occurred at acid pH Figure 8 shows the SER spectra of DPA for the spectral region and pH range of interest The identity of the DPA species was determined by plotting the normalized peak intensities with the lowest value set to 0 and the highest to 0006 M as a function of pH and overlaying these values on a plot of the relative concentrations for DPA DPA- and DPA= as previously done for MPA (Figure 9) As can be seen the 525 cm-1 band clearly corresponds to DPA The correspondence of the 795 cm-1 band to this species is less clear as the band retains intensity until pH of 55 This can be attributed to contribution to the overlapping band at 810 cm-1 which does not change as a function of pH The fact that most bands are observed at all pHs suggest that the primary interaction with silver is through the ring nitrogen This is supported by the fact that the most intense band occurs at 1008 cm-1 attributed to a symmetric ring breathing mode and that this interaction has been characterized for pyridine in numerous papers27
0
1
2
3
4
5
6
7
0 200 400 600 800 1000 1200
MPA Concentration (mgL)
760
Ban
d In
tens
ity
Figure 6 SER spectra of MPA in water at A) 1000 B) 100 and C) 10 mgL Conditions pH of 7 silver-doped sol-gel coated vial 200 mW of 785 nm 33-sec 8 cm-1 resolution
Figure 7 Plot of SER intensity of 760 cm-1 band of MPA as a function of concentration using 200 mW of 785 nm Inset table includes average intensity LOD standard deviation and percent deviation for each concentration but for 100 mW and 10 min
conc (mgL) LOD ave stddev dev10 021 002 1021100 072 010 14301000 312 040 1280
A
B C
SPIE -2003-5269 122
The ability to detect dipicolinic acid (DPA) by SERS is demonstrated in Figure 10 Even at 1 mgL the primary bands are visible The intensity of the dominant ring breathing mode observed at 1008 cm-1 for DPA in water (pH = 25 to 55) is plotted as a function of concentration in Figure 11 These values were also used to estimate limits of detection based on the SN of the 1008 cm-1 band using 100 mW of 785 nm laser excitation and a 10-min acquisition time Again the lower the measured concentration the lower the predicted LOD (see Figure 11 inset) and detection of 160 microgL is possible Gastrointestinal anthrax requires significant more spores than inhalation anthrax28 and a limit of detection might be placed at 1 million spores in 1 liter of water or 10 microgL Since each spore contains ~10 CaDPA by weight29 a goal for DPA might be 1 microgL indicating that the present measurements must be improved by nearly two orders of magnitude Due to the increased hazards of handling HCN gas KCN salt was used for these experiments Nevertheless all sample preparations were performed in a chemical hood KCN completely dissolves in water but its conjugate acid HCN is formed and has a Ka of 615x10-1030 viz
HCN CN- + H+ pKa = 921 Reaction 4
Figure 8 SER spectra of 1 mgmL DPA as a function of pH Conditions 100 mW 785 nm 100 scans (44-sec)
Figure 9 Concentration dependence of DPA DPA- and DPA= as a function of pH for a 0006M sample Intensity of 525 () and 795 (diams) cm-1 bands from Fig 8 as a function of pH
pH
135 171 219 383
0
0001
0002
0003
0004
0005
0006
0007
0 2 4 6 8 10 12 14pH
Con
cent
ratio
n [M
]
DPADPA-DPA=795525
DPA DPA=DPA-
pK2 = 692pK1 = 216
795 525
0
05
1
15
2
25
0 200 400 600 800 1000 1200
DPA Concentration (mgL)
1008
Ban
d In
tens
ity
conc (mgL) lod-10min-100mw1 017
10 016100 103
1000 355
Figure 10 SER spectra of DPA in water at A) 1000 B) 100 C) 10 and D) 1 mgL Conditions pH of 25-55 silver-doped sol-gel coated vial 175 mW of 785 nm 1-min 8 cm-1 D) has been multiplied by x10 to make bands visible
A
B C
D
Figure 11 Plot of SER intensity of 1008 cm-1 band of DPA as a function of concentration using 175 mW of 785 nm Inset table includes LOD in mgL for each concentration but for 100 mW and 10 min
SPIE -2003-5269 123
Consequently the cyanide concentration must be determined for each initial KCN concentration Specifically the samples prepared with concentrations of 01 1 10 100 and 1000 mgL of KCN produced CN- concentrations of 63x10-3 033 69 89 and 964 mgL at pHs of 816 90 967 102 and 107 respectively The pH dependence for the HCN and CN- concentrations are shown in Figure 12 Thus as the amount of KCN added to the solution decreases so does the pH of the solution (becomes less basic) and according to Reaction 4 the relative amount of CN- to HCN also decreases For example in the preparation of a 01 mgL solution of KCN the pH is shifted from 7 for pure water to only 816 and only 63 of the starting material becomes CN- or 63x10-3 mgL In comparison for a solution of 1000 mgL the pH is shifted from 7 to 107 and 96 of the starting material becomes CN- This is significant since the cyanide ion is better able to adsorb onto the silver particles and become SERS active SER spectra of 10 100 and 1000 mgL of KCN samples are shown in Figure 13 and a corresponding plot of the 2100 cm-1 band intensity as a function of concentration is shown in Figure 14
000102030405060708091011
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14pH
Con
cent
ratio
n [m
gm
L]
pKa = 921
CN -HCN
The SER spectra of cyanide are dominated by the single CN stretch at 2100 cm-1 which occurs in normal Raman spectra of solutions at 2080 cm-1 However a low frequency mode occurs at 135 cm-1 due to a Ag-CN stretch (not shown) indicative of a strong surface interaction It is also observed that as the concentration decreases the CN stretch shifts to 2140 cm-1 This shift has been attribute to the formation of a tetrahedral Ag(CN)3
2- surface structure31 as well
Figure 14 Concentration dependence of KCN SERS measured under conditions in Fig 11 Concentrations are 1 01 and 001 mgml Intensities are measured for the CN stretch at 2100 cm-1 Inset table includes LOD in mgL for each concentration in Figs 13 and 15 but for 100 mW and 10 min
Figure 13 SER spectra of KCN in water at A) 1000 B) 100 and C) 10 mgL Conditions pHs of 107 102 and 97 silver-doped sol-gel coated vial 100 mW of 785 nm 1-min 8 cm-1 C) has been multiplied by x10 to make band visible
Figure 12 Concentration dependence of HCN and CN- as a function of pH for a 1 mgmL sample Calculated intensity of 2100 () cm-1 band for a 1 mgml sample at pHs of 816 90 967 102 and 107
A
B
C
0
20
40
60
80
100
120
140
0 200 400 600 800 1000 1200CN Concentration (mgL)
2100
Ban
d In
tens
ity
Figure 15 SER spectra of KCN in water at A) 10 B) 1 and C) 01 mgL Conditions pHs of 97 12 and 12 gold-doped sol-gel coated vial 100 mW of 785 nm 1-min 8 cm-1 C) has been multiplied by x10 to make band visible
A
B
C
conc (mgL) lod-10min-100mw Condition01 001 Au-pH 121 007 Au-pH 12
69 003 Au-pH 9769 016 Ag-pH 9789 022 Ag-pH 102946 113 Ag-pH 107
SPIE -2003-5269 124
as to CN adsorbed to two different surface sites32 Alternatively the 2140 cm-1 band could be attributed to HCN since this species dominates at lower concentrations However it is unlikely that this species would be attracted to the electropositive silver surface Further both peaks should be present at pHs between 85 and 105 but this is not observed It has also been suggested that at concentrations near and above monolayer coverage the CN- species is forced to adsorb end-on due to crowding and at lower concentrations the molecule can reorient to lie flat33 This suggests that the 2100 and 2140 cm-1 bands correspond to the end-on and flat orientations respectively As Figures 13 and 14 show the intensity of the CN stretch for the 89 mgL sample is nearly as intense as the 964 mgmL band This suggests that the Raman signal for the flat orientation is more enhanced However more extensive measurements are required to verify this point Since resent research has suggested that cyanide may be more effectively detected on gold measurements of KCN solutions were also performed using gold-doped sol-gel vials Preliminary measurements are shown in Figure 15 for samples prepared from 01 1 and 10 mgL KCN Since the pHs are 816 90 967 the resultant CN- concentrations are 63x10-3 033 and 69 mgL Initially only the highest concentration was observed and the signal intensity was significantly better than the equivalent concentration measured using silver In an effort to shift Reaction 4 to the left transforming HCN to CN- (Le Chatelierrsquos principle) KOH was added to the lower concentration samples producing solutions with pH 12 This effectively forces all of the cyanide in solution to be CN- or 01 and 10 mgL respectively More importantly the CN stretch is now observed in the SER spectra The band appears at 2125 cm-1 as has been previously reported for gold12 As calculated for MPA and DPA LODs can be estimated from this data For the three concentrations of cyanide on silver the LODs are 016 to 11 mgL for 100 mW of 785 nm laser excitation and a 10-min acquisition time For gold pH adjusted the LODs are10 to 70 microgL an improvement of more than 10 times silver Nevertheless either substrate is sufficient to meet the JSAWM goals of measuring 3 mgL in 10-minutes as the requirements form cyanide are much less stringent than the nerve agents
4 CONCLUSIONS Here we examined the surface-enhanced Raman spectral response for methyl phosphonic acid dipicolinic acid and cyanide as a function of pH It was determined that the most prominent peak at 760 cm-1 reaches a maximum intensity between pH 3 and 7 and corresponds to the MPA- species Neither the MPA nor MPA= species appear to generate a SER spectrum and consequently no spectra were observed below pH 2 or above pH 8 In this study we also found that higher laser powers could cause photodegradation of MPA signified by the exponential growth of a band at 1050 cm-1 which is tentatively assigned to phosphorous acid Unlike MPA DPA was observed at all pHs This is attributed to the dominant interaction of the pyridine functional group with silver Minor spectral changes were observed at acid pHs and were assigned to neutral DPA Like MPA SER spectra of cyanide were pH dependent No spectra were observed for the HCN species while CN- was best observed at pHs more basic than 8 Preliminary concentration studies for the three analytes allowed estimating limits of detection for MPA DPA and CN using 100 mW of 785 nm and a 10-min acquisition time of 210 165 and 70 microgL respectively Although the latter value suggests that cyanide can be measured at sufficiently low concentrations to meet JSAWM goal improvements by 100 to 200 times are required for MPA and DPA It is clear from this study that pH of the sample is important and must be taken into account when developing concentration calibration curves as well as fieldable analyzers Future work will include tailoring the sol-gel with specific functional groups to dictate sample pH and thereby optimize sensitivity
ACKNOWLEDGMENTS The authors are grateful for the support of the US Army (DAAD13-02-C-0015 Joint Service Agent Water Monitor program) The authors would also like to thank Dr Steve Christensen of the US Army SBCCOM for helpful discussions and Mr Chetan Shende of Real-Time Analyzers for assistance in development of the gold-doped sol-gels
REFERENCES 1 Erickson B Analytical Chemistry News amp Features June 1 397A (1998)
SPIE -2003-5269 125
2 Johnston RL Hoefler CM Fargo JC and Moberley B AT-ONSITE 5-8 (1994) 3 Black RM RJ Clarke RW Read and MT Reid J Chromatography 662 301-321 (1994) 4 USA Army Center for Health Promotion and Preventive Medicine (USACHPPM) Technical Guideline 230A 5 Jensen JL NC Wong W R Loerop SPIE 4775-03 (2001) 6 Hoenig SL Handbook of Chemical Warfare and Terrorism Greenwood Press Wesport CT (2002) 7 Munro NB SS Talmage GD Griffin LC Waters AP Watson JF King and V Hauschild Env Health
Persp 107 933-974 (1999) 8 Holstege CP Kirk M Sidell FR Crit Care Clin13 923-42 (1997) 9 Farquharson S P Maksymiuk K Ong and S Christesen SPIE 4577 166-173 (2001) 10 Lee Y and S Farquharson SPIE 4378 21-26 (2001) 11 Spencer KM J Sylvia S Clauson and J Janni SPIE 4577 158-165 (2001) 12 Tessier P S Christesen K Ong E Clemente A Lenhoff E Kaler and O Velev Applied Spectroscopy 56
1524-1530 (2002) 13 Farquharson S WW Smith S Elliott and JF Sperry SPIE 3533 207-214 (1998) 14 Farquharson S WW Smith S Elliott and JF Sperry SPIE 3855110-116 (1999) 15 Farquharson S WW Smith YH Lee S Elliott and JF Sperry SPIE 4575 62-72 (2002) 16 Guzelian AA J Sylvia J Janni S Clauson and KM Spencer SPIE 4577 182-192 (2001) 17 Shende C F Inscore A Gift P Maksymiuk and S Farquharson in press 18 Weaver MJ S Farquharson and MA Tadayyoni J Chem Phys 82 4867-4874 (1985) 19 Alak AM and T Vo-Dinh Analytical Chemistry 59 2149-2153 (1987) 20 Lee Y and S Farquharson SPIE 4206 140-146 (2000) 21 Farquharson S and Y Lee SPIE 4200-16 (2000) 22 Lee Y S Farquharson and P M Rainey SPIE 3857 76-84 (1999) 23 Lee Y S Farquharson H Kwong and M Shahriari SPIE 3537 252-260 (1998) 24 Wang J M Pumera G Collins and A Mulchandani Analytical Chemistry 74 6121-6125 (2002) 25 Dou X YM Jung Z-Q Cao and Y Ozaki Applied Spectroscopy 53 1440-1447 (1999) 26 Data supplied by S Christesen and K Ewing 27 Kerker M and B Thompson Eds SPIE MS 10 (1990) 28 Inglesby TV DA Henderson JG Bartlett JAMA 287 2236 (2002) 29 FW Janssen AJ Lund and LE Anderson Science 127 26 (1958) 30 Lide DR Ed Handbook of Chemistry and Physics CRC Press 77th Ed (1996-7) 31 Billmann J G Kovacs and A Otto Surf Sci 92 153 (1980) 32 Murray CA and S Bodoff Phys Rev B 32 671 (1985) 33 Kellogg D and J Pemberton J Phys Chem 91 1120 (1987)
SPIE-2003-5269 16
Chemical agent detection by surface-enhanced Raman spectroscopy
Stuart Farquharson Alan Gift Paul Maksymiuk Frank Inscore and Wayne Smith
Real-Time Analyzers 87 Church Street East Hartford CT 06108
Kevin Morrisey and Steven D Christesen US Army SBCCOM Aberdeen Proving Ground MD 21010
ABSTRACT
In the past decade the Unites States and its allies have been challenged by a different kind of warfare exemplified by the terrorist attacks of September 11 2001 Although suicide bombings are the most often used form of terror military personnel must consider a wide range of attack scenarios Among these is the intentional poisoning of water supplies to obstruct military operations in Afghanistan and Iraq To counter such attacks the military is developing portable analyzers that can identify and quantify potential chemical agents in water supplies at microgram per liter concentrations within 10 minutes To aid this effort we have been investigating the value of a surface-enhanced Raman spectroscopy based portable analyzer In particular we have been developing silver-doped sol-gels to generate SER spectra of chemical agents and their hydrolysis products Here we present SER spectra of several chemical agents measured in a generic tap water Repeat measurements were performed to establish statistical error associated with SERS obtained using the sol-gel coated vials Keywords Chemical agents chemical agent detection SERS Raman spectroscopy
1 INTRODUCTION
In the autumn of 2001 terrorism within US borders became a sobering reality While extensive efforts are being implemented to secure the homeland US and Coalition forces in Afghanistan and Iraq are constantly faced with terrorist attacks In addition to car-bombings the military has identified several non-traditional attack scenarios including poisoning of water supplies by chemical warfare agents (CWAs) To counter this threat the Department of Defense has been investigating numerous approaches to detect such attacks under the auspices of the Joint Service Agent Water Monitor (JSAWM) program1 The JSAWM must be capable of identifying the chemical agents with no false-positives and quantifying the agents at microgram per liter concentrations within 10 minutes2 This includes the analysis of drinking water supplies distribution and storage systems Currently colorimetric paper is used to detect agents on-site while gas chromatography combined with mass spectrometry (GCMS) is used in mobile support laboratories However both methods have severe drawbacks The paper changes color in response to contact with many chemicals besides CWAs causing a high incidence of false positives3 while GCMS although very chemically specific requires up to an hour to perform and regular re-calibration456 Vibrational spectroscopy has also been investigated as a potential method of rapidly detecting CWAs7-
11 as early as 1985 in the case of infrared spectroscopy8 and 1988 in the case of Raman spectroscopy11 Again however these techniques also have limitations when it comes to measuring trace poisons in water Infrared spectra would be dominated by the very strong absorption of water which would obscure absorptions by most other chemicals present Whereas Raman spectroscopy is simply not a very sensitive technique and detection limits are typically grams per liter Surface-enhanced Raman spectroscopy (SERS) offers several advantages over conventional vibrational methods and that may provide the necessary sensitivity required for detecting trace quantities of chemical agents in water SERS employs the interaction of surface plasmon modes of metal particles with target analytes to increase scattering efficiency by as much as 1 million times12 In 1987 the potential of this approach to measure CWAs was demonstrated by measuring a series of pesticides13 Several of these organophosphonates have chemical structures similar to CWAs in particular P=O functional groups In the past few years we and others have further explored the ability of SERS to detect CWAs14-17 and even bioagents 18-21 We have been employing silver-doped sol-gels to promote the SER effect
SPIE-2003-5269 17
in these studies The porous silica network of the sol-gel matrix offers a unique environment for immobilizing and stabilizing SER-active metal particles22-25 The silver-doped sol-gels have been coated on the internal walls of standard glass vials to allow rapid SER analysis of small sample quantities without preparation We have measured over 100 chemicals with enhancements of 104 to 106 demonstrated reversible measurements in a flowing system reproducible measurements from vial-to-vial and measurements in multiple solvents including water21-25 Previously we used these vials to perform preliminary measurements of cyanide (CN) and methylphosphonic acid (MPA) Most of the nerve agents form MPA during hydrolysis while Tabun forms CN a chemical agent in its own right In another paper including in these proceedings we examined the limits of detection (LOD) for MPA by measuring a series of concentrations down to 50 mgL and estimated a limit of detection of 100 microgL26 These measurements provide encouragement in that SERS may satisfy the needs of the JSAWM To further establish the viability of SERS in particular silver-doped sol-gels here we present analysis of cyanide mustard and VX in tap water The measurements performed at the US Armyrsquos Edgewood Chemical Biological Center Aberdeen MD also included numerous repeat measurements to establish reproducibility
2 EXPERIMENTAL 2a General All chemicals including potassium cyanide 2-chloroethylethyl sulfide and those used to prepare the silver-doped and gold-doped sol-gel coated vials were obtained and used as received from Sigma-Aldrich All samples were prepared in a chemical hood using HPLC grade water unless otherwise noted (Fischer Scientific Fair Lawn NJ) for SERS measurements Once prepared the samples were transferred into the silver-doped sol-gel vials (Real-Time Analyzers Inc East Hartford CT) The vials were coated in a manner similar to that previously reported by adding ammonium hydroxide to a solution of silver nitrate tetramethyl orthosilicate (TMOS) and methanol22 The two precursor solutions were prepared mixed and transferred to 2-mL glass vials dried and heated After sol-gel formation the incorporated metal ions were reduced with dilute sodium borohydride (1mgmL) which is followed by a water wash to remove residual reducing agent After the resultant analyte solution was introduced the SER-active vial was fixed horizontally to an XY positioning stage (Conix Research Springfield OR) just inside the focal point of an f07 aspheric lens The lens focused the beam into the sample and collected the scattered radiation back along the same axis A dichroic filter (Omega Optical Brattleborough VT) was used to reflect the excitation laser to the lens and pass the Raman scattered radiation collected by the lens An f2 achromat was used to collimate the laser beam exiting a 200 microm core diameter source fiber optic while a second f2 achromat was used to focus the scattered radiation into a 365 microm fiber optic (Spectran Avon CT) A short pass filter was placed in the excitation beam path to block the silicon Raman scattering generated in the source fiber from reflecting off sampling optics and reaching the detector A long pass filter was placed in the collection beam path to block the sample Rayleigh scattering from reaching the detector A 785 nm diode laser (Process Instruments Inc model 785-600 Salt Lake City UT) was used to deliver 100 to 150 mW of power to the sample A Fourier transform Raman spectrometer (Real-Time Analyzers model IRA-785 East Hartford CT) and a silicon photo-avalanche detector (Perkin Elmer model C30902S Stamford CT) were used to acquire the SER spectra
2b Edgewood Chemical Biological Center The surface-enhanced Raman spectral measurements at the US Armyrsquos Chemical Biological center presented here were all performed on September 12 2003 To expedite measurements a plate was machined to hold up to 12 SER-active sample vials (Figure 1) The plate fit a standard XY plate reader that could be programmed Pure KCN bis-(2-chloroethyl)sulfide (distilled mustard gas HD) and ethyl S-2-diisopropylamino ethyl methylphosphonothioate (VX) were obtained on-site and used to prepare 1 gL tap water solutions in a chemical hood with appropriate safety equipment Simulated tap water was prepared by adding 10 chemicals most often found in tap water at appropriate concentrations to distilled water (Table 1) SER measurements were also performed in a chemical hood For added safety the FT-Raman instrument was placed outside the laboratory and 30 foot fiber optic and electrical cables were used to allow remote SERS measurements and plate manipulation For each experiment 1gL samples were prepared and added to 9 individual vials which were then loaded on the plate In some cases a tenth vial was included as a blank
Table 1 Generic Tap Water Composition
Compound mgL NaHCO3 100 CaSO4 27 MgSO4bull7H2O 67 NaNO3 1 Fulvic Acid 1 K2HPO4 07 KH2PO4 03 (NH4)2HSO4 001 NaCl 001 FeSO4 0001 pH 76-78
SPIE-2003-5269 18
A software program was written that allowed selecting the sequence that the vials were measured the number of positions along the length of the vials to measure (1 to 5) and the number of scans to co-add During sample analysis the program displayed the vial being analyzed the point being analyzed and the spectrum as it was being acquired Once all the data was collected a second software program was written to rapidly analyze the data The spectra collected for all the vials on a plate could be loaded at one time and then the spectra for each point could be displayed simultaneously or separately The user could then select the Raman peak to analyze in terms of peak height or area This was accomplished by selecting points on either side of the peak to define a baseline of zero The peak height or area could then be computed for all of the spectra loaded and then exported to a spreadsheet for statistical analysis
Figure 1 A) Vial Holder 6 slots to hold 2 vials each end-to-end B) Measurement Configuration Program user selects vials to measure sequence number of points per vial (1 to 5) and number of scans per point C) Spectral Acquisition Program shows spectrum being collected which vial and position D) Spectra Analysis Program user selects spectra to analyze by plate vial and point (s) as well as two wavenumbers defining the peak and the baseline to subtract The image is of 5 repeat measurements of 10 mgL KCN in generic tap water 16 sec each 100 mW of 785 nm
3 RESULTS AND DISCUSSION Raman and surface-enhanced Raman spectra were obtained for potassium cyanide bis-(2-chloroethyl)sulfide and ethyl S-2-diisopropylamino ethyl methylphosphonothioate representing three classes of chemical agents cyanides mustards and nerve agents respectively Spectra were also obtained for 2-chloroethyl ethyl sulfide (CEES) a structural analogue to HD which was included in the study to aid in assigning spectral bands KCN salt was used for cyanide experiments to avoid the increased hazards of handling HCN gas KCN completely dissolves in water forming its conjugate acid HCN according to its Ka of 615x10-1027 and at a concentration of 1 mgmL results in a pH 107 solution This is significant in that only CN- appears to interact sufficiently with silver to produce a SER spectrum and no spectral signal is observed below pH 726 Figure 2 shows the SER and normal Raman spectra for KCN The SERS spectrum of 1mgml KCN in water shows a single intense somewhat broad feature at
A B
C D
SPIE-2003-5269 19
2100 cm-1 assigned to the single CequivN stretch The band is much sharper in the normal Raman spectra of the solid KCN salt at 2074 cm-1 This band does broaden and shift to 2080 cm-1 in solution (not shown) However the observed SERS frequency is attributed to interaction with silver and low frequency mode at 135 cm-1 attributed to a Ag-CN stretch (not shown) supports this conclusion
cm-1 band to a C-S stretch but the authors concede that it is in fact more likely a C-Cl stretch28 It appears that the most intense Raman bands at 648 692 and 747 cm-1 shift to 620 660 and 720 cm-1 in the SER spectra and are tentatively assigned as above The width of these bands suggests that they overlap underlying spectral features Additional bands in the Raman spectra occur at 972 1034 1049 1263 1286 1423 1442 2865 2935 and 2960 cm-1 Corresponding bands occur in the SER spectra at 964 1015 1054 1286 1410 1447 2865 and 2935 cm-1 Most of these bands are associated with alkane modes specifically the bands at approximately 1040 cm-1 to a C-C stretch 1290 cm-1 to a CH2 in-phase twist 1440 cm-1 to a CH2 wag 2865 cm-1 to a symmetric CH2 stretch and 2965 cm-1 to an asymmetric CH2 stretch The Raman and SER spectra of sulfur mustard were measured at the Edgewood center (Figure 4) Both spectra are largely similar to CEES The C-Cl and C-S bands in the Raman spectrum of HD now occur at 640 655 700 739 and 760 cm-1 and are more resolved possibly due to the increased molecular symmetry Theoretical calculations indicate that the first three bands are due to C-Cl stretching modes and the latter two to C-S stretching modes28 Only the C-Cl bands maintain significant intensity in the SER spectra occurring at 624 and 643 cm-1 which is attributed to the
Prior to measurements of HD CEES was examined by Raman and SER spectroscopy (Figure 3) CEES also known as half-mustard is essentially identical to HD except one of the chlorine end atoms is replaced by a hydrogen atom Again although not as toxic as HD CEES is a blister agent and dilute aqueous samples were prepared using appropriate safety equipment Both the Raman and SER spectra of CEES are similar and dominated by bands between 600 and 800 cm-1 These are associated with C-Cl and C-S stretching modes which are tentatively assigned to 648 and 747 cm-1 in the Raman spectra respectively The shoulder at 630 cm-1 the overlapped band at 660 cm-1 and the strong band at 692 cm-1 could also be due to these modes or their asymmetric counterparts It is worth noting that theoretical calculations assign the 692
Figure 2 A) SER and B) NR spectra of KCN Conditions A) 1 mgml in tap water 100 mW of 785 nm at sample 1-min acquisition time B) solid 300 mW of 785 nm 5-min All spectra are 8 cm-1 resolution
A
B
Figure 3 A) SER and B) NR spectra of CEES Conditions A) 1 vv (10 mgml) in MeOH 100 mW of 785 nm 1-min acquisition time B) neat 300 mW of 785 nm 5-min
A
B
Cl-CH2-CH2-S-CH2-CH3
A
B
Figure 4 A) SER and B) NR of HD Conditions A) 1mgml in tap water B) pure both 100 mW of 785 nm 1-min
Cl-CH2-CH2-S-CH2-CH2-Cl
SPIE-2003-5269 20
expected strong interaction between chlorine and silver and adds support to the assignment of this band to a C-Cl stretch Weaker overlapping bands occur at 670 692 and 724 cm-1 the latter possibly due to C-S stretching modes Again the alkane modes are apparent in the normal Raman spectra of HD but only a broad feature at 1300 to 1450 cm-1 suggests CH2 contributions in the SER spectrum Although the observed bands in the VX spectrum have not been assigned (Figure 4) a computer generated Raman spectrum29 predicts many of the same features with surprising accuracy and are used here Two intense bands at 460 and 530 cm-1 closely match predicted bands at 463 and 546 cm-1 assigned to a CH3-P=O bend and a PO2CS wag Three highly overlapped bands occur at 694 745 and 771 cm-1 matching predicted bands at 713 730 and 760 cm-1 The first
Table 2 Measured SER peak heights for the CN stretch at 2100 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
Figure 5 A) SERS and B) NR spectra of VX Conditions A) 1 vv (10 mgml) in MeOH B) pure sample both 100 mW of 785 nm 1-min acquisition time
A
B
two have been assigned to a C-S stretch and CH2bend respectively while the latter has been attributed to either a P-C stretch or an O-C-C stretch Although the 745 cm-1 band may alternatively be assigned to a C-S stretch based on the previous measurements of CEES and HD The relatively intense bands at 890 1106 1218 1445 and 1465 cm-1 also match predicted bands at 880 1108 1216 1440 and 1464 cm-1 that are assigned to a C-C stretch CH3 rock N-C3 stretch various C-H3 bends and C-H bends respectively Both the computer generated and the measured spectra contain numerous other less intense bands One is worth mentioning A unique band appears at 370 cm-1 that is predicted at 368 cm-1 and corresponds to an O-P=O bend The surface-enhanced Raman spectrum of VX is also rich with spectral features It has the unique low frequency band at 370 cm-1 as well as a second band at 380 cm-1 that is assigned tothe S-P-O bend predicted in the normal Raman spectrum at 388 cm-1 Based on the measured and predicted normal Raman spectra the following SERS assignments are given 460 cm-1 to the CH3-P=O bend 544 cm-1 to the PO2CS wag 738 cm-1 to a C-S stretch (based on arguments above) 890 cm-1
to a C-C stretch 1101 cm-1 to a CH3 rock and 1456
cm-1 to a C-H bend The ability of SERS to measure chemical agents in water containing real-world chemical interferents was tested by using the generic tap water described in Table 1 The ability to reproduce measurements was accomplished by preparing three separate water stock solutions which were used to prepare three sample solutions each of 1 mgmL KCN in the generic tap water The samples defined as vials 1-9 were then measured at 4 points per vial in a semi-random fashion such that errors associated with stock solution preparation and errors associated with instrument drift could be identified No trends were apparent that signified such systematic errors Each spectrum collected consisted of 20 averaged scans taking 16 seconds at 8 cm-1 resolution The laser power at the sample was measured periodically during the day and it ranged from 102 to 105 mW spacccc
spot Vial 1 Vial 2 Vial 41 13754 13004 128192 13519 12692 12909
1 3 13541 12721 126394 13462 12648 12651
ave 13569 12766 12755Vial 6 Vial 3 Vial 5
1 11586 14068 10757 2 11236 14402 115122 3 11376 14546 11559
4 10894 11763 11214ave 11273 13695 11261
Vial 7 Vial8 Vial 91 11198 15804 11023
2 11243 15716 88753 3 11562 15216 11218
4 11655 15084 11077ave 11415 15455 10548
CN AVG STDEV ERR1 mgmL pk ht 12526 1560 1245
stock solution
SPIE-2003-5269 21
The CN measurements consisting of 36 data points produced an average height of 125 for the 2100 cm-1 peak with a standard deviation of 156 or 125 (Table 2) The HD and VX measurements were performed precisely the same way (Tables 3 and 4) For HD the 624 cm-1 peak was used for analysis and it had an average height of 53 with a standard deviation of 068 or 129 while for VX the 544 cm-1 peak was used for analysis and it had an average height of 1051 with a standard deviation of 308 or 293 The greater error in the VX measurements can be somewhat attributed to Vial 4 which produced lower SER signal intensities But removing this vial from the data set changes the standard deviation to 233 only a modest improvement
spot Vial 1 Vial 2 Vial 41 525 609 4682 45 675 484
1 3 527 644 5094 559 807 669
ave 51525 68375 5325Vial 6 Vial 3 Vial 5
1 521 574 575 2 536 53 4492 3 509 508 451
4 565 594 379ave 53275 5515 4635
Vial 7 Vial 8 Vial 91 457 589 413
2 583 497 4273 3 544 505 4
4 528 5 509ave 528 52275 43725
HD AVG STDEV ERR1 mgmL pk ht 530 068 1291
stock solution
spot Vial 1 Vial 2 Vial 41 1464 1034 4672 1485 99 61
1 3 1491 989 5684 1041 777 553
ave 137025 9475 5495Vial 6 Vial 3 Vial 5
1 1058 942 1293 2 697 121 965
2 3 727 1206 11124 689 1254 746
ave 79275 1153 1029Vial 7 Vial 8 Vial 9
1 1127 783 163 2 1358 812 1614
3 3 1371 908 15124 1097 875 1397
ave 123825 8445 153825
VX AVG STDEV ERR1 mgmL pk ht 1051 308 2925
stock solution
4 CONCLUSIONS In this paper we examined the ability of surface-enhanced Raman spectroscopy to reproducibly measure CN HD and VX in tap water without chemical interference Both normal and surface-enhanced Raman spectra were examined to select unique bands suitable to identify and quantify these chemical agents For SER measurements the 2100 cm-1 C-N stretch was used for CN the 624 cm-1 C-Cl stretch was used for HD and the 544 cm-1 PO2CS wag was used for VX It was determined that 1 mgmL samples of each of these chemicals measured 36 times in glass vials coated with a silver-doped sol-gel reproduced measurements with standard deviations of 125 129 and 293 It was further found that the 10 chemicals added to simulate generic tap water did not interfere with or alter the SER spectra It should be noted that the concentrations used in this study were considerably greater than those required by the JSAWM program Current work involves improving SER sensitivity and designing sampling systems with better reproducibility This includes the development of fractal silver and gold structures within the sol-gel matrix and the development of chemically selective sol-gels
ACKNOWLEDGMENTS The authors are grateful for the support of the US Army (DAAD13-02-C-0015 Joint Service Agent Water Monitor program) The authors would also like to thank Janet Jensen Ronald Crosier and Kristina Gonser for helpful discussions
Table 4 Measured SER peak heights for the VX band at 544 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
Table 3 Measured SER peak heights for the HD band at 624 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
SPIE-2003-5269 22
REFERENCES 1 Jensen JL NC Wong W R Loerop SPIE 4775-03 (2001) 2 JSAWM Requirements at wwwsbccomapgeaarmymilRDAecbcrtPRODSERJSAWMjsawmhtml 3 Erickson B ldquoThe Chemical Weapons Convention Redefines Analytical Challengerdquo Anal Chem 70 397A (1998) 4 Johnston RL Hoefler CM Fargo JC and Moberley B ldquoThe Defense Nuclear Agencyrsquos
ChemicalBiochemical Weapons Agreements Verification Technology Research Development Test and Evaluation Program and its Requirements for On-Site Analysisrdquo AT-ONSITE 5-8 (1994)
5 Black RM Clarke RJ Read RW and Reid MT ldquoApplication of GC-MS and GC-tandem MS to the analysis of chemical warfare samples found to contain residues of the nerve agent sarin sulphur mustard and their degradation productsrdquo J Chrom 662 301-321 (1994)
6 USA Army Center for Health Promotion and Preventive Medicine (USACHPPM) Technical Guideline 230A 7 Hoffland LD Piffath RJ Bouck JB rdquoSpectral signatures of chemical agents and simulantsrdquo Optical
Engineering 24 982-984 (1985) 8 Braue EHJ Pannella MGrdquoCIRCLE CELL FT-IR Analysis of Chemical Warfare Agents in Aqueous Solutionsrdquo
App Spectrosc 44 1513-1520 (1990) 9 Tseng C-H Mann CK and Vickers TJ ldquoDetermination of Organics on Metal Surfaces by Raman
Spectroscopyrdquo App Spectrosc 47 1767-1771 (1993) 10 Staff Reporter ldquoUS Army Tests Lidar to Detect Biological Toxinsrdquo Photonic Spectra pg 50 December 1998 11 Christesen SD Raman cross sections of chemical agents and simulants App Spectrosc 42 318-321 (1988) 12 Weaver MJ Farquharson S Tadayyoni MA ldquoSurface-enhancement factors for Raman scattering at silver electrodesrdquo J Chem Phys 82 4867-4874 (1985) 13 Alak AM and Vo-Dinh T ldquoSERS of Organophosphorous Chemical Agentsrdquo Anal Chem 59 2149-2153 (1987) 14 Lee Y Farquharson S ldquoRapid chemical agent identification by SERSrdquo SPIE 4378 21-26 (2001) 15 Farquharson S Maksymiuk P Ong K Christesen S ldquoChemical agent identification by surface-enhanced Raman spectroscopyrdquo SPIE 4577 166-173 (2001) 16 Spencer KM Sylvia J Clauson S and Janni J ldquoSurface Enhanced Raman as a Water Monitor for Warfare
Agents in Waterrdquo SPIE 4577 158-165 (2001) 17 Tessier P Christesen S Ong K Clemente E Lenhoff A Kaler E Velev O ldquoOn-line spectroscopic characterization of sodium cyanide with nanostructured gold surface-enhanced Raman spectroscopy substratesrdquo
App Spectrosc 56 1524-1530 (2002) 18 Farquharson S W W Smith S Elliott and J F Sperry Rapid biological agent identification by surface- enhanced Raman spectroscopy SPIE 3855110-116 (1999) 19 Farquharson S WW Smith YH Lee S Elliott and J F Sperry Detection of biological signatures A
comparison of electrolytic and metal-doped sol-gel surface-enhanced Raman media SPIE 4575 62-72 (2002) 20 Guzelian AA Sylvia J Janni J Clauson S and Spencer KM ldquoSERS of whole cell bacteria and trace levels of biological moleculesrdquo SPIE 4577 182-192 (2001) 21 Shende C Inscore F Gift A Maksymiuk P Farquharson S ldquoRapid dipicolinic acid extraction from Bacillus spores detected by surface-enhanced Raman spectroscopyrdquo App Spectrosc 58 accepted 22 Lee Y and Farquharson S ldquoSERS Sample Vials Based on Sol-Gel Process for Trace Pesticide Analysisrdquo SPIE
4206 140-146 (2000) 23 Farquharson S and Lee Y ldquoTrace Drug Analysis by SERSrdquo SPIE 4200-16 (2000) 24 Lee Y Farquharson S and Rainey P M Surface-Enhanced Raman Sensor for Trace Chemical Detection in
Water SPIE 3857 76-84 (1999) 25 Lee Y Farquharson S Kwong H and Shahriari M ldquoSol-Gel Chemical Sensor for Surface-Enhanced Raman
Spectroscopyrdquo SPIE 3537 252-260 (1998) 26 Farquharson S A Gift P Maksymiuk F Inscore and W Smith ldquopH dependence of methyl phosphonic acid
dipicolinic acid and cyanide by surface-enhanced Raman spectroscopyrdquo SPIE 5269-19 (2003) 27 Lide DR Ed Handbook of Chemistry and Physics CRC Press 77th Ed (1996-7) 28 Sosa C RJ Bartlett K KuBulat and WB Person ldquoA theoretical study of harmonic vibrational frequencies and
infrared intensities of XCH2CH2SCH2CH2X and XCH2CH2SH (= H Cl)rdquo J Phys Chem 93 577-588 (1993) 29 Hameka HF and JO Jenson ldquoComputer-generated predictions of the structure and the IR and Raman spectra of
VXrdquo ERDEC-TR-065 May 1993
- Chemagents Appendicespdf
-
- SPIE2001-4575-Bioagent-SERSapdf
-
- Detection of bioagent signatures A comparison of electrolytic and metal-doped sol-gel surface-enhanced Raman media
- Stuart Farquharson Wayne Smith and Yuan Lee
-
- Real-Time Analyzers 87 Church Street East Hartford CT 06108
-
- ABSTRACT
- 4 CONCLUSIONS
- 5 ACKNOWLEDGEMENTS
- 6 REFERENCES
-
SPIE-4378-2001
23
In an effort to demonstrate the broad capabilities of the SER-active vials to measure chemical agents spectra of a nerve agent simulant dimethyl methylphosphonate a mustard gas simulant 2-chloroethyl ethyl sulfide and hydrolysis products pinacolyl methylphosphonate and methylphosphonic acid were collected DMMP is widely used by the US Army as a chemical warfare simulant because its chemical structure volatility and water solubility are similar to those of nerve agents25 DMMP is completely miscible and stable in water at room temperature26 Figure 3 compares the SER spectrum to the normal Raman spectrum of DMMP A number of the normal Raman bands are SER-active such as the P-C stretching mode which shifts from 715 to 735 cm-1 and the C-H stretching modes at 2855 2930 2960 and 3000 cm-1 which shift slightly Surprisingly the P=O stretching mode at 1250 cm-1 virtually disappears However the most dramatic change is the appearance of an intense triplet in the SER spectrum near 1000 cm-1 The bands at 1000 cm-1 1030 cm-1 and 1075 cm-1 likely involved the P-O-C bond This is supported by the nearly identical triplets observed for the SER spectra of fonofos and fonofoxon1719 It is also worth noting that a band appears at 425 cm-1 in the SER spectrum that may be unique to DMMP and useful for identification The enhancement factor is estimated at 120000 based on the normal Raman and SER P-C band intensity taking into account the difference in sample concentrations and spectral acquisition conditions A detection limit based on a signal-to-noise ratio of 3 can be estimated at 16 ppm
10-7 10-6 10-5 10-4 10-3 10-2 10-110-2
10-1
100
101
102
Figure 1 SER spectra of A) 70 B) 35 and C) 7 micromolar p-amino benzoic acid in water Conditions 80 mW of 1064 nm laser excitation 100 averaged scans (15 min) at 8 cm-1 resolution
Figure 2 SER spectral intensity for p-aminobenzoic acid as a function of concentration using RTA SER-active vials
Concentration (M)
Figure 3 A) SER and B) normal Raman spectra of dimethyl methylphosphonate Conditions SERS as in Figure 1 normal Raman 500 mW and 200 scans
Figure 4 A) SER and B) normal Raman spectra of 2-chloroethyl ethyl sulfide Conditions as in Figure 3
Wavenumbers (∆cm-1)
CH3O-P-OCH3
=
O
_
CH3
CH3O-P-OCH3
=
O
_
CH3
Cl-CH2-CH2-S-CH2-CH3 A
B
A
B
A
B
C
Wavenumbers (∆cm-1)
Wavenumbers (∆cm-1)
500 1000 1500 2000
SPIE-4378-2001
24
2-Chloroethyl ethyl sulfide a blister agent simulant has a chemical structure similar to the mustard gas (Cl-CH2-CH2-S-CH2-CH2-Cl) with only one terminal chlorine Due to its low solubility in water CEES was dissolved in methanol for the SER measurement Again the prominent Raman modes are SER-active and even maintain relative intensity (Figure 4) The primary difference is that the SER bands appear to broaden such that the triplet near 700 cm-1 becomes a doublet and the shoulders at 2875 and 2970 cm-1 become less defined Again the latter bands are assigned to C-H stretching modes A single band at 700 cm-1 which is attributed to the C-S-C asymmetric stretch dominates the reported infrared spectrum of mustard gas12 A corresponding symmetric stretch is reported at 705 cm-1 in the Raman spectrum of mustard gas27 Here a corresponding symmetric stretch appears but as a doublet at 700 and 755 cm-1 presumably due to the loss in symmetry for CEES The band at 655cm-1 can also be confidently assigned to a C-Cl stretch The SER spectral bands at 620 and 730 cm-1 are probably due to the same modes ie C-Cl and C-S-C stretches respectively The enhancement factor for CEES was somewhat less than DMMP at approximately 62000 as is the estimated detection limit of 22 ppm The ability to rapidly detect trace quantities of chemical agents in the gas phase would be invaluable as an early warning system Although the Raman scattering cross-sections for the nerve agents suggest that remote detection by Raman-based LIDAR is unlikely16 a SER-based system for perimeter monitoring could prove successful As a preliminary measurement we prepared a 10 by volume solution of CEES in methanol exposed a SER-active vial to the equilibrium vapor phase in a sealed jar and monitored the SER spectrum as a function of time Initially the vial was removed through a transfer chamber every hour to record the SER spectrum After ten hours spectra were recorded only every ten hours As illustrated by Figure 5 the sol-gel performed as a dosimeter in that the spectra increased as a function of exposure time The most intense SER bands at 620 and 2930 cm-1 are discernable in the first few hours The spectrum after 40 hours is nearly identical to the solution phase spectrum except for a diminished intensity of the 730 cm-1 band This may be due to methanol solvation effects or surface-orientation effects Based on the relative concentrations of methanol and CEES and their partial pressures we estimate the equilibrium concentration of CEES to between 1 and 2 micromolar Although not shown this concentration could be detected in one hour As previously stated the analysis of chemical agents in water is important in identifying poisoned water It is also important to decommissioning activities in which agents are destroyed by hydrolysis (acid or base) Furthermore any analytical technique used must be capable of distinguishing between parent CWA and hydrolysis products to assess safety or effectiveness of decommissioning For example soman has a hydrolysis half-life of ~23 hours at ambient temperatures and neutral pH28 and forms hydrofluoric acid (somewhat toxic) and pinacolyl methylphosphonate (relatively non-toxic)2930 PMP further hydrolyzes to form methyl phosphonic acid and 33-dimethyl-2-butanol (both non-toxic) The structural similarities between soman PMP and MPA are expected to produce similar Raman as well as SER spectra Figure 6 compares PMP and MPA but not the highly toxic parent CWA soman As with DMMP the P-C stretch the P-O-C mode and C-H stretches are readily apparent Yet it is worth noting that the band positions are reasonably different The former two bands appear at 764 and 1042 cm-1 for MPA while they are at 788 and 1032 cm-1 for PMP More importantly a unique band at 546 cm-1 as yet unassigned appears in the PMP spectrum
Figure 6 SER spectra of A) methyl phosphonic acid and B) pinacolyl methylphosphanate (note unique band at 546 cm-1) Spectral conditions as in Figure 1
Figure 5 SER spectra of 2-chloroethyl ethyl sulfide vapor as a function of time (10 hour increments to top which is 40 hours) Bottom trace is a blank Spectral conditions as in Figure 1
A
B
Wavenumbers (∆cm-1) Wavenumbers (∆cm-1)
HO-P-OH
=
O
CH3
_
HO-P-OH
=
O
=
O
CH3
_
CH3
_HO-P-O-CH-C-CH3
=O
CH3
_
CH3
_CH3_CH3
_
HO-P-O-CH-C-CH3
=O
=O
CH3
_
CH3
_
CH3
_
CH3
_CH3_CH3
_CH3
_
SPIE-4378-2001
25
4 CONCLUSIONS We have successfully measured the SER spectra of chemical agent simulants dimethyl metylphosphonate and 2-chloroethyl ethyl sulfide and chemical agent hydrolysis products pinacolyl methylphosphonate and methylphosphonic acid using silver-doped sol-gel coated sample vials Measurements were obtained in both aqueous and gas phase The P-C stretching mode was SER-active for all four chemicals allowing identification by class Within this group each chemical contained at least one unique spectral band that could be used for identification (Table 1) Furthermore these bands do not appear to coincide with SER spectra reported for organophosphorus pesticides the most likely source of false-positives Although surface enhancement factors appear to be an order of magnitude better than those previously presented in the literature for similar chemicals17 measurement sensitivity needs to be improved by 1 to 2 orders of magnitude to provide adequate warning of chemical agent use Current research efforts to increase surface-enhancement optical collection efficiency and instrument design are being pursued to achieve the required sensitivity
Table 1 Enhancement factors detection limits and unique SER bands fro chemicals studied Agent Simulant Enhancement Detection limit Unique bands (cm-1)
Dimethyl methylphosphonate 120000 90 microM (16 ppm) 425 2-Chloroethyl ethyl sulfide 62000 60 microM (22 ppm) 620
Methylphosphonic acid 110000 3 microM (60 ppb) 764 1042 Pinacolyl methylphosphonate 150000 70 microM (14 ppm) 546 788 1032
5 ACKNOWLEDGEMENTS The authors would like to thank Drs Janet Jensen and Steven Christesen of Aberdeen Proving Ground for encouraging this work They would also like to thank Advanced Fuel Research for making their laboratory facilities available
6 REFERENCES 1 ldquoThe Chemical Weapons Convention ndash A Guided Tour the Organization for the Prohibition of Chemical Weaponsrdquo at
httpwwwopcwnlguidehtm 2 Robinson JP and J Goldblat Chemical Warfare In The Iraq-Iran War Stockholm International Peace Research
Institute Fact Sheet at httpprojectssiprisecbwresearchfactsheet-1984html (1984) 3 ldquoChemistry of GB (Sarin)rdquo at httpwwwmitretekorgmissionenvenechemicalagentssarinhtml 4 Tu Anthony ldquoOverview of Sarin Terrorist Incidents in Japan in 1994 and 1995rdquo 6th CBW Protection Symposium
Stockholm Sweden 10-15 May 1998 5 Staff Reporter ldquoGoing out with a bangrdquo Newsweek June 28 1999 6 See UNSCOM reports in httpwwwunorgdeptsunscom (1999) 7 Treven T Saddamrsquos Secrets Harper Collins (1999) 8 ldquoDecaying Sarin-filled Rockets Spark Fearsrdquo Janersquos Defense Weekly 25(20)3 (1996) 9 ldquoThe Chemical Weapons Convention Redefines Analytical Challengerdquo Analytical Chemistry News amp Features June 1
397A (1998) 10 Johnston RL Hoefler CM Fargo JC and Moberley B ldquoThe Defense Nuclear Agencyrsquos ChemicalBiochemical
Weapons Agreements Verification Technology Research Development Test and Evaluation Program and its Requirements for On-Site Analysisrdquo AT-ONSITE 5-8 (1994)
11 Black RM Clarke RJ Read RW and Reid MT ldquoApplication of gas chromatography-mass spectrometry and gas chromatography-tandem mass spectrometry to the analysis of chemical warfare samples found to contain residues of the nerve agent sarin sulphur mustard and their degradation productsrdquo J Chromatography 662 301-321 (1994)
12 Hoffland LD Piffath RJ Bouck JBrdquoSpectral signatures of chemical agents and simulantsrdquo Optical Engineering 24 982-984 (1985)
13 Braue EHJ Pannella MGrdquoCIRCLE CELL FT-IR Analysis of Chemical Warfare Agents in Aqueous Solutionsrdquo Applied Spectroscopy 44 1513-1520 (1990)
SPIE-4378-2001
26
14 Tseng C-H Mann CK and Vickers TJ ldquoDetermination of Organics on Metal Surfaces by Raman Spectroscopyrdquo
Applied Spectroscopy 47 1767-1771 (1993) 15 Staff Reporter ldquoUS Army Tests Lidar to Detect Biological Toxinsrdquo Photonic Spectra pg 50 December 1998 16 Christesen SD Raman cross sections of chemical agents and simulants Applied Spectroscopy 42 318-321 (1988) 17 Alak AM and Vo-Dinh T ldquoSERS of Organophosphorous Chemical Agentsrdquo Analytical Chemistry 59 2149-2153
(1987) 18 Norrod KL Sudnik LM Rousell D and Rowlen KL ldquoQuantitative Comparison of Five SERS Substrates
Sensitivity and Detection Limitrdquo Applied Spectroscopy 51 994-1001 (1997) 19 Lee Y and Farquharson S ldquoSERS Sample Vials Based on Sol-Gel Process for Trace Pesticide Analysisrdquo SPIE 4206
140-146 (2000) 20 Farquharson S and Lee Y ldquoTrace Drug Analysis by Surface-Enhanced Raman Spectroscopyrdquo SPIE 4200-16 (2000) 21 Lee Y Farquharson S and Rainey P M Surface-Enhanced Raman Sensor for Trace Chemical Detection in Water
SPIE 3857 76-84 (1999) 22 Lee Y Farquharson S Kwong H and Shahriari M ldquoSurface-Enhanced Raman Sensor for Surface-Enhanced Raman
Spectroscopyrdquo SPIE 3537 252-260 (1998) 23 see Material Safety Data Sheets for details 24 Farquharson S Smith W Carangelo R C and Brouillette C ldquoIndustrial Raman Providing Easy Immediate Cost
Effective Chemical Analysis Anywhererdquo SPIE 3859 14-23 (1999) 25 Bennett S Bane J Benford P and Pratt R ldquoEnvironmental Hazards of Chemical Agent Simulantsrdquo Aberdeen
Proving Ground Maryland Chemical Research and Development Center CRDC-TR-84055 (1984) 26 Mabey W and Mill T Critical Review of Hydrolysis of Organic Compounds in Water under Environmental Conditions
Journal of Physics and Chemistry Reference Data 7(2) 383-414 (1978) 27 Christesen S MacIver B Procell L Sorrick D Carabba M and Bello J ldquo Noninstrusive Analysis of Chemical Agent
Identification Sets Using a Portable Fiber-Optic Raman Spectrometerrdquo Applied Spectroscopy 53 850-855 (1999) 28 Meylan WM and Howard PH J Pharm Sci 84 83-92 (1995) 29 Jenkins A Uy O and Murray G ldquoPolymer-Based Lanthanide Luminescent Sensor for Detection of Hydrolysis Product
of the Nerve Agent Soman in Waterrdquo Analytical Chemistry 71 373-378 (1999) 30 Nassar A Lucas S and Hoffland L ldquoDetermination of Chemical Warfare Agent Degradation Products at Low-Part-
per-Billion Levels in Aqueous Samples and Sub-Part-per-Million Levels in Soils Using Capillary Electrophoresisrdquo Analytical Chemistry 71 1285-1292 (1999)
SPIE 2001-4575
62
Detection of bioagent signatures A comparison of electrolytic and metal-doped sol-gel surface-enhanced Raman media
Stuart Farquharson Wayne Smith and Yuan Lee
Real-Time Analyzers 87 Church Street East Hartford CT 06108
Susan Elliott and Jay F Sperry University of Rhode Island 45 Lower College Rd Kingston RI 02881
ABSTRACT Since September 11 2001 the threat of terrorist attacks and biological warfare within US borders has become a sobering reality In an effort to aid military personnel and the public at large we have been investigating the utility of surface-enhanced Raman spectroscopy (SERS) to provide rapid identification of chemical agents directly and biological agents through their chemical signatures This approach is based on the ability of Raman spectroscopy to identify molecular structure through the abundant vibration information provided in spectra and the ability of SERS to detect extremely low concentrations (eg part-per-billion) through the enhancement of Raman scattering by six orders of magnitude or more Towards the goal of developing a portable analyzer we have been studying the ability of two SER media to obtain continuous (ie reversible) and quantitative (ie reproducible) measurements Here we compare measurements of nucleic acid-bases adenosine monophosphate and ribonucleic acid extracted from Escherichia coli Bacillus subtilis and Staphylococcus aureus obtained by electrolytic SERS and metal-doped sol-gel SERS The capabilities of these SER media are summarized in terms of rapid detection of B anthracis and dipicolinic acid Keywords bioagent detection SERS RNA analysis bacterial analysis Raman spectroscopy
1 INTRODUCTION The recent distribution of anthrax through the US postal system and the subsequent infection and death of several postal and national media employees amplifies the need for methods to rapidly detect and identify this and other chemical and biological warfare agents (BWA) The primary methods currently used immunoassays for screening and nucleic acid (NA) sequencing for positive identification of BWAs (bacteria protozoa and viruses) have serious limitations123 Immunoassay methods employ competitive binding of the bioagent (as an antigen) and its labeled (eg fluorescence) conjugate for a limited number of antibodies Although this analysis method is fast and semi-quantitative other chemicals may compete for the antibodies interfere with the enzymatic reaction or interfere with the measurement (eg it fluoresces) resulting in a high number of false positive responses1 Furthermore the antibodies denature due to moisture and heat limiting shelf life and require sterile often refrigerated storage Positive identification of a BWA can be accomplished by sequencing deoxyribonucleic acid or ribonucleic acid (DNA and RNA)23 This requires enumeration of a nucleic acid sequence through polymerase chain reactions (PCR) or multiplication of the microorganism through culture growth to provide sufficient quantities of DNA or RNA for analysis Unfortunately PCR and culture growth require from several hours to several days23 Consequently a wide variety of technologies have been investigated for rapid identification of BWAs The Department of Defense is actively monitoring 200 such technologies4 This includes traditional methods such as gas chromatographic separation coupled with ion mobility spectrometry detection5 to exotic methods based on nature such as monitoring toxin induced color changes in fish scales6 Although all of these techniques have achieved varying degrees of success none are yet capable of detecting and identifying BWAs in 10 minutes or less Towards this goal we have been investigating the ability of SERS to detect sub-nanogram quantities of DNA or RNA (eliminating enumeration) determine relative NA base concentrations and identify BWA taxonomy
To whom correspondences should be addresses e-mailfarqureal-time-analyzerscom wwwreal-time-analyzerscom
SPIE 2001-4575
63
Raman spectroscopy has a rich history of investigating biochemical and biological processes7 Some of the earliest laser-Raman studies demonstrated that the five NA bases adenine (A) cytosine (C) guanine (G) thymine (T in DNA) and uracil (U in RNA) yielded distinct spectra with several bands suitable for identification and quantification8 Furthermore these studies included exceptional spectra of both DNA and RNA for which the NA bases as well as several phosphate bands were easily identified9 However since the Raman effect is very inefficient (very low conversion of incident radiation to inelastically scattered Raman radiation) these samples had to be highly concentrated Fortunately two phenomena exist that can increase the generation of Raman photons by six orders of magnitude or more known as the resonance Raman and surface-enhanced Raman effects1011 Resonance Raman scattering occurs when the laser excitation wavelength is in resonance with an electronic transition of a molecule (within the absorption envelope)10 Excitation at ultraviolet wavelengths has been used to obtain resonance Raman spectra of amino acids and nucleic acids in whole bacteria1213 For example excitation at 242 nm has been used to maximize the nucleic acid spectral band intensities and minimize the amino acids band intensities A peak at 1530 cm-1 was found to be proportional to the amount of the NA base cytosine while a peak at 1485 cm-1 was proportional to the combined amount of the NA bases adenine and guanine This quantitative behavior has been used to define an A+TG+C base-pair ratio and provide a level of bacterial identification as taxonomic markers13 In recent years SERS has also been used to analyze bacterial cell components14 including amino acids15 lipids16 nucleic acids151718 and the adenine derivatives192021 SERS has proven to be one of the most sensitive methods for trace chemical analysis through the detection of single molecules2223 including DNA (dye labeled 17-mer)24 Since its discovery in 197425 the mechanism responsible for the large increase in scattering efficiency has been the subject of considerable research2627 Briefly incident laser photons couple to free conducting electrons within a metal which confined by the particle surface collectively cause the electron cloud to resonate2628 These surface plasmons are known as the physical component of the SER effect These surface plasmons can transfer energy to the molecular vibrational modes of molecules through interactions with the molecular electron orbitals2629 This interaction is known as the chemical component of the SER effect This perturbation of the molecular polarizability generates surface-enhanced Raman photons26 A number of methods have been developed to produce surfaces or solutions containing one of these metals with optimum roughness or diameter to promote SERS30 These methods include preparation of activated electrodes in electrolytic cells 1131 activated silver and gold colloid reagents32 and metal coated substrates333435 Selecting a SER-active medium for chemical and biological agent detection requires consideration of the method of deployment and hence the method of sampling Chemical aerosols or airborne bacteria will require a collection device to concentrate and continuously present the sample to the SERS medium Poisoned water supplies will also require a flow through device for continuous monitoring or a grab-sample device for periodic analysis And contaminated surfaces will require a grab-sample extractive device A SERS-based device used for continuous monitoring (air or water) must be reversible and reproducible if quantitative measurements are desired while a SERS-based device used for periodic sampling (water or surfaces) must be reproducible Both reversible and reproducible measurements have been performed using electrolytic SERS (E-SERS)36 But this requires a three-electrode sample cell and an electrolyte of known concentration to perform the necessary oxidation-reduction cycles (ORCs) to re-activate the electrode surface with new uncontaminated sites from one measurement to the next Colloids are severely limited in that continuous measurements would require a continuous supply of colloids For periodic measurements vials of colloids one per measurement could be used However aggregate size and consequently SER intensity change with sample conditions (especially pH) and quantitative reproducible measurements are unlikely Substrates appear to have the greatest potential and designs range from silver evaporated on titania particles34 to periodic gold pyramids evaporated between polystyrene beads35 Most substrates require concentrating the sample on the surface through drying to obtain the largest signal enhancements in effect making the measurements irreproducible and irreversible However successful measurements using flow systems have been obtained with glass posts but manufacturing costs appear prohibitive In an effort to overcome these limitations we have developed metal-doped sol-gels to provide SERS measurements that are reproducible reversible and quantitative and yet not restricted to specific environments such as electrolytes solvents or evaporated surfaces3738 The porous silica network of the sol-gel offers a unique environment for stabilizing SER active metal particles and the high surface area increases the interaction between the analyte and metal particles The sol-gel can be coated on the end of fiber optics or on the internal walls of a glass flow tube for continuous measurements or standard glass sample vials for periodic measurements Previously we measured 100 mgL methylphosphonic acid (the primary hydrolysis product of nerve agents) in water with an estimated detection limit of 05 mgL (100 parts-per-billion) We have also
SPIE 2001-4575
64
demonstrated reversible and reproducible measurements of p-aminobenzoic acid (PABA) in a flow through system Here we investigate the ability of the sol-gel SERS (SG-SERS) to measure the NA bases adenosine monophosphate and RNA extracted from E coli B subtilis and S aureus The measurements are compared to those obtained by E-SERS
2 EXPERIMENTAL The inorganic chemicals and solvents used to prepare samples were spectroscopic grade and purchased from Aldrich (Milwaukee WI) Fisher (Pittsburgh PA) or Pfaltz amp Bauer (Waterbury CT) The nucleic acid bases and dipicolinic acid were purchased from Sigma (St Louis MO) Normal Raman samples were measured to establish enhancement factors In each case 1cm3 of sample was placed into a 1x1 cm glass cuvette weighed and measured Unpacked densities were typically 6-7 gcm3 For all SER measurements including RNA samples were prepared as ~01mgmL (see Figure captions for exact concentrations) in 01M KCl and buffered to a pH of 92 with Na2B4O7bullH2O Adenine pH dependence measurements used pH buffer solutions at 4 (potassium acid phthalate) 69 (potassium phosphate monobasicsodium phosphate dibasic) 92 (Na2B4O7bullH2O) and 104 (tris-hydroxymethyl amino methane) Escherichia coli Bacillus subtilis and Staphylococcus aureus cultures (250ml per 1000mL Erlenmeyer flask) were grown overnight in a Trypticase soy broth (TSB) medium containing 1 glucose in a shaking water bath at 37 oC The bacteria were harvested by centrifugation for 10 minutes at 8000 rpm in a GSA rotor at 5degC then washed once in 085 saline The gram-positive bacteria were concentrated to 20 ml and passed through a French pressure cell twice at 15000 psi to break open the cells RNA was extracted according to Protocol 44139 to ensure pristine samples for initial measurements Since this method takes approximately 4 hours a streamlined method was developed For vegetative bacteria the specimen was boiled for 30 sec in 1 ml of distilled water to lyse the cells and release the RNA For bacterial spores the specimen was first incubated in 1 ml of saline solution containing 02 mg lysozyme and phosphate-buffered to pH of 624 for 1 hr at 37 oC This solution was then boiled for 2-3 minutes in 4X loading buffer to release the RNA For both specimens RNA STAT-60TM was added to the supernatant which was centrifuged at 12000 g for 5 minutes to precipitate the ~15 water-soluble proteins This procedure allowed extracting RNA for SER analysis in ~ 10 minutes Electrophoresis shows high purity while the existence of chemicals that could interfere with the SER measurements is still under investigation The electrolytic sample cell has been described previously36 Briefly a three electrode design is incorporated into a Plexiglas structure containing a well for the reference electrode (a saturated calomel electrode Cole Parmer Vernon Hills IL) and a 5mL sample well containing the silver working electrode and platinum wire counter electrode (05 mm wire Alfa Ward Hill MA) A channel connecting the two wells contained a 2 mm diameter semi-porous membrane (10-20 micron pore Ace Glass) The silver electrode was made from a 3 mm length of 2 mm diameter silver wire (Alfa) soldered to a copper wire lead encased in a 4 mm diameter Pyrex tube A cap containing the silver electrode platinum wire and nitrogen purge and vent lines fixed the silver electrode surface 1 mm from a 1 mm thick glass plate attached to the bottom of the sample well The potentiostat used to control the three electrodes was built in-house and has been described in detail elsewhere36 A multifuntional analog digital and timing inputoutput interface card (DAQCard-1200 National Instruments) is used to both drive the electrolytic cell as well as read the current generated in the cell A LabVIEW software program is used set the oxidation potential reduction potential SER measurement potential hold times and sweep rates The amount of charge passed was plotted as a cyclic voltammogram For all spectra presented five oxidation-reduction cycles (ORCs) stepping from -03 VSCE to 03 VSCE and back to -3 VSCE at 50 mVsec were used The SG-SER measurements were accomplished by simply placing the identical samples prepared above into Simple SERS Sample VialsTM (RTA) These 2-mL glass vials are internally coated with ~ 01 micron thick silver-doped sol-gel A prototype Fourier transform Raman spectrometer (RTA) was used for these measurements40 The system consisted of a NdYAG laser (Brimrose or Spectra Physics) for excitation at 1064 nm an interferometer built by On-Line Technologies (OLT East Hartford CT) for frequency separation an uncooled InGaAs detector for signal detection (RTA) and an Intel 400 MHz Pentium II based laptop computer (Dell Round Rock TX) for interferometric control data acquisition (OLT) and analysis (LabVIEW by National Instruments Austin TX) Additional components included a Notch filter (Kaiser Ann Arbor MI) and interferometer entrance and exit optics (Edmund Scientific Barrington NJ) Fiber optics were used to deliver the excitation beam to the sample and the scattered radiation to the interferometer (2 meter lengths of 200 and 365 micron core diameter respectively Spectran Avon CT) A second Notch filter (Kaiser) was used as a beam splitter to direct the excitation beam along the same axis as the collected radiation A microscope object (20x magnification 04 numeric aperture Newport Irvine CA) was used to focus the beam into the sample and to collect the scattered radiation back along the same axis In this co-axial backscattering arrangement the excitation beam passed through the glass plate onto the silver
SPIE 2001-4575
65
electrode surface for E-SERS through the vial glass wall and into the silver-doped sol-gel film for SG-SERS or through the glass wall of the cuvette and into the solid sample for normal Raman spectroscopy All E-SERS and normal Raman spectra were obtained with 750 mW of laser power at the sample while all SG-SERS spectra were obtained with 75 mW of laser power at the system Incident powers above 200 mW in some cases degraded the sol-gel
3 RESULTS AND DISCUSSION The generation of surface-enhanced Raman scattering at electrode surfaces has been extensively researched and the optimum sample conditions are well developed2729 Several researches incorporated electrodes into flowing systems and demonstrated that continuous monitoring of chemicals is possible18 These successes suggested investigated the capability of E-SERS to measure the NA bases and RNA The E-SERS measurements also provided a benchmark to compare and evaluate SG-SERS measurements The molecular structure of adenine (as well as the other base pairs) which includes an aromatic nitrogen-containing heterocycle is ideally suited to interact with the surface plasmons and contribute substantially to the chemical component of the SER effect1119 Even with excitation at 1064 nm a 3-minute scan of 18x10-5M adenine yields high signal-to-noise (SN) E-SER spectra and all of the bands are revealed with clarity (Figure 1 Table 1) Spectra of similar quality were obtained by SG-SERS and the principal spectral bands are easily observed The identical 18x10-5M adenine sample was measured in the same 3-minute time frame but with 110th the laser power The lower power appears to reduce the SN The amount of adenine responsible for the SER spectra as well as enhancement factors for the two SER media can be determined The molecules producing the E-SERS spectrum are those on the electrode surface within the illumination area of the laser (The solution concentration only determines the number of molecules available to adsorb to the electrode surface) For the current experiments the laser illuminates an area of 28x10-7m2 or 56 x10-7m2 if we assume the ORCs increase the surface area by a factor of two Furthermore if we assume monolayer coverage on the electrode and each 3x5 angstrom molecule (lying flat) occupies 15x10-19m2 then there are ca 4x1012 molecules contributing to the Raman scattering This is ca twice the number of molecules measured at electrode surfaces using either differential capacitance-potential curve measurements or rapid linear sweep voltammetry (eg 3x1018 moleculesm2 for pyridine and pyrazine)29 Thus the adenine spectrum in Figure 1 is due to 87x10-10g (6x10-12 moles) A detection limit defined as a SN of 3 can also be calculated The SN for a 3-minute scan is 844 for the 735 cm-1 band suggesting a mass detection limit of 3x10-12g (2x10-14 moles) This is consistent with previous estimates for adenine by others of 25 x10-14 moles1530 However sub-monolayer concentrations must be measured to verify this The root-mean-squared (RMS) noise is measured between 4400-4600 cm-1 Since noise is distributed evenly throughout the spectrum when transformed this region does not have any
Wavenumbers (∆cm-1) Figure 1 A) Normal Raman spectrum of pure adenine powder B) E-SERS and C) SG-SERS of 18x10-5M adenine at pH 92 All spectra 8 cm-1 resolution 200 scans (3 min) and 1064 nm excitation A) and B) 750 mW C) 75 mW B) measurement potential of 11VSCE
A
B
C
A
B
C SG-SERS
pH 10
D
725
735
735
pH 4
Wavenumbers (∆cm-1) Figure 2 A) and C) E-SERS and B) and D) SG-SERS of adenine at A) and B) pH 104 and C) and D) pH 40 Note consistent appearance of bands at 1270 and 1375 cm-1 as the pH is changed to 10 for both SER media E-SERS used 750 mW SG-SERS used 75 mW of 1064 nm excitation
SPIE 2001-4575
66
contributions from signals or baseline offsets The measurement error is given as SplusmnRMS and for adenine this equals 234 The number of molecules contributing to the SG-SERS are those on the silver particles that are embedded in the sol-gel The total silver surface area can be determined from the average particle size (40 nm diameter) concentration (073 by weight based on molar conc and measured sol-gel density) and the scattering volume (a cylinder defined by the laser area 28x10-7m2 and sol-gel thickness10-4m) The 61x109 silver particles in this volume have a collective area of 31x10-5m2 However it may reasonably be assumed that at least half of the silver surface is in contact with the sol-gel matrix and unavailable for analyte interaction Then approximately 10x1014 molecules or 22x10-8g of adenine contribute to the SG-SER spectrum The slightly lower SN of 207 suggests a mass detection limit of 32x10-10g Determination of the enhancement factors for the two SER media requires estimating the number of adenine molecules contributing to the normal Raman spectrum Here a cylindrical scattering volume is assumed again based on the laser area (28x10-7m2) and the penetration depth (1x10-3 m)41 The density of the sample was measured at 064 gcm3 indicating that 18x10-4g (13x10-6 moles) of adenine produced the normal Raman signal The enhancement factor EF is defined by the following equation
EF = (ISERSINR)bull(MNRMSERS) bull(PNRPSERS) bull(TNRTSERS)12
where I is the spectral band intensity (here 735 cm-1) M is the sample mass P is the incident laser power and T is the measurement time (or number of scans) For the E-SERS measurement the enhancement factor is 22x105 (01780184) bull (18x10-487x10-10)) while the SG-SERS enhancement factor is 10x105 (0160184) bull(18x10-422x10-8) bull(75075) bull(315)12) The lower enhancement for the SG-SERS may be real or the available surface of the silver embedded in the sol-gel may have been overestimated In addition to enhancing the Raman scattering efficiency to an extent similar to E-SERS the SG-SER medium also yields an identical shift of the adenine ring-breathing mode from 725 cm-1 in the normal Raman to 735 cm-1 Furthermore in the course of optimizing the E-SERS sample conditions it was found that pH influenced the adenine interaction with the silver surface (Figure 2) In particular the relative band intensities of the pyrimidine ring skeletal vibrations at 1270 and 1375 cm-1 and the imidazol ring skeletal vibration at1335 cm-1 change At pH 4 adenine is protonated presumably the imidazol ring since the band at 1335 cm-1 increases in intensity while the pyrimidine bands are virtually absent Conversely at pH 10 the imidazol band decreases in intensity while the pyrimidine bands appear It is worth noting that the ring-breathing mode at 735 cm-1 changes little between pH 4 and 10 suggesting that the skeletal changes are more a function of molecule-plasmon interactions than reorientation of the molecule on the surface Measurements of the identical pH series of adenine samples by SG-SERS yielded virtually identical spectral changes This suggests that the sol-gel does not influence the measurement This is critical to reproducing measurements and performing quantitative analysis Next the remaining NA bases were measured by both E-SERS and SG-SERS and compared Previously we examined the optimum pH and electrode potentials for E-SERS measurements to determine if a common pH could be used that yielded good sensitivity for all the bases and if variations in potential could be used to provide an added degree of selectivity between the bases Primarily it was found that high quality spectra were obtained between pH 7 and 95 and that cytosine and uracil were best enhanced at potentials positive of the potential-of-zero charge (pzc ca 065VSCE for Ag) guanine and thymine near the pzc and adenine negative of the pzc In all cases the ring-breathing modes were the most intense and in general could be used to identify the NA bases (Figure 3 Table 1) Specifically adenine has an intense band at 735 cm-1 cytosine at 797 cm-1 guanine at 653 cm-1 thymine at 784 cm-1 and uracil at 800 cm-1 The adenine cytosine guanine and thymine bands are sufficiently separated that their contributions to DNA should be determinable Although adenine and guanine contributions to RNA should also be determinable cytosine and uracil are highly overlapped and unfortunately share the same potential dependence Alternate unique bands at 1183 cm-1 for cytosine and 1275 cm-1 for uracil might be suitable for calculating contributions The SG-SER spectra of the remaining NA bases faithfully reproduced the E-SER spectra In particular the primary identifying bands occur at virtually the same wavenumbers (see Table 1) However the spectra for both cytosine and thymine contain an intense band at ca 1040 cm-1 Initially this was attributed to the pH buffer but samples prepared without either the buffer or the 01M KCl electrolyte yielded identical spectra containing this band In fact the E-SER and SG-SER spectra of thymine are virtually identical except for this band Also the SG-SERS of guanine contains an intense band at 1551 cm-1 that is not observed in the E-SER spectrum This band may be due to a moderately intense band at 1553cm-1 in the normal Raman spectrum that is SG-SER active It was also found that the SG-SERS of cytosine was considerably better than the E-SERS while uracil showed the opposite relationship It is also worth noting that all of the SG-SERS were obtained with 110th the laser power Most importantly the primary ring-breathing modes in the SG-SER spectra are sufficiently intense and unique to be used in determining contributions to DNA and RNA as outlined above
SPIE 2001-4575
67
Table 1 Comparison of E-SER and SG-SER Spectral Band Positions for the NA Bases and Adenosine Monophosphate
Adenine Cytosine Guanine Thymine Uracil AMP E-SERS SG-SERS E-SERS SG-SERS E-SERS SG-SERS E-SERS SG-SERS E-SERS SG-SERS E-SERS SG-SERS1647 1638 1634 1656 1655 1630 1587 1585 1510 1580 1551 1539 1456 1456 1465 1460 1462 1480 1453 1459 1394 1398 1425 1431 1435 1399 1404 1392 1374 1375 1373 1383 1370 1335 1332 1311 1307 1333 1331 1353 1348 1331 1329 1265 1273 1280 1292 1278 1276 1275 1279 1271 1183 1195 1222 1232 1221 1219 1204 1205 1180 1144 1097 1033 1029 1038 1040 1035 1051 1037 1041 1035 963 963 957 1001 1000 961 944 884 819 817 859 866 735 737 797 799 784 782 800 800 72738 742 630 630 653 664 667 684 603 602 590 611 466 561
Bands unique to SG-SERS The next chemical to be analyzed by both E-SERS and SG-SERS was adenosine monophosphate (AMP) The E-SER spectrum yields bands due to the adenine chemical functionality at 727 961 1233 1279 1331 1381 and 1486 cm-1 In addition phosphate bands are observed at 860 1097 1453 1587 and 1705 cm-1 (Figure 5) Other researchers have noted that the ribose component does not appear to contribute to the spectrum19 The AMP spectrum also changes as a function of potential As the electrode is swept more positive (here from -09 to -03VSCE) the phosphate bands at 860 1097 1453 and 1587 cm-1 increase in intensity compared to the adenine bands while a band at 1705 cm-1 appears The adenine bands at 1233 1381 and 1486 cm-1 virtually disappear These potential dependent spectral changes are consistent with earlier studies that show that phosphate is attracted to silver at potentials positive of the pzc but repelled at potentials negative of the pzc19
Wavenumbers (∆cm-1) Figure 3 E-SERS of A) 21x10-3M cytosine at -03VSCE 1000 scans B) ~10x10-5M guanine at -06VSCE 500 scans C) 23x10-3M thymine at -06VSCE 500 scans and D) 12x10-3M uracil at -093VSCE 500 scans All spectra at pH 92 750 mW 1064 nm at 8 cm-1
Wavenumbers (∆cm-1) Figure 4 SG-SERS of A) 21x10-3M cytosine 200 scans B) ~10x10-5M guanine 200 scans C) 23x10-3M thymine 200 scans and D) 12x10-3M uracil 500 scans All spectra at pH 92 75 mW 1064 nm at 8 cm-1
A
B
C
A
B
C
D D
SPIE 2001-4575
68
The SG-SER spectrum of AMP is considerably different The adenine bands virtually disappear except for the two primary bands which shift to742 and 1329 cm-1 While the phosphate band at 1459 cm-1 has gained considerable intensity In addition two new intense bands appear at 684 and 1539 cm-1 as well as a moderately intense band at 1180 cm-1 The SG-SER spectrum has greater similarity to the E-SER spectrum at -03VSCE and suggests that the silver particles embedded in the sol-gel behave as if at a potential positive of the pzc RNA samples extracted from E coli B subtilis and S aureus were next examined by both E-SERS and SG-SERS E-SER spectra of these samples yielded quality spectra in 10 minutes in which all of the major features can be identified (Figures 6 and 7) This includes guanine at 650 cm-1 adenine at 791 cm-1 cytosine and uracil combining at 790 cm-1 and phosphate at 1100 1335 (in combination with adenine and guanine) 1465 and 1570 cm-1 Surprisingly adenine which demonstrated the greatest surface-enhanced Raman effect does not dominate the ring-breathing mode portion of the spectrum The intensities of the other base-pairs bands are of the same order of magnitude This suggests that when the base-pairs are linked together as in RNA they are enhanced in concert In fact the relative intensities are very similar to a normal Raman spectrum of E coli RNA which shows the combined cytosine and uracil band at ca twice the intensity of the adenine band and ca four times the intensity of the guanine band Unfortunately this means that the independent enhancement factors for the NA bases can not be used to estimate relative concentrations For example the relative 791 and 734 cm-1 bands for B subtilis would indicate that the cytosine andor uracil concentration was at least 20 times the adenine concentration whereas each of the four RNA bases are known to contribute 15-35 Nevertheless it is worth noting that the three RNA samples yield different relative band intensities that were reproduced in numerous measurements Although the relative concentrations of the NA bases for these samples have not been determined these differences can be quantified If it is assumed that the 650 cm-1 band represents 25 guanine the 791 cm-1 band represents 25 adenine and the 790 cm-1 50 cytosine plus uracil in the E coli RNA spectrum then the relative concentrations can be estimated for the other RNA samples To aid this calculation the three spectra were normalized to the phosphate band at 1100 cm-1 which has been shown to correlate to the total phosphate concentration and can be used as an internal standard In addition a simple baseline correction was applied (Figure 7) This yields 15 adenine 30 guanine and 55 cytosine plus uracil for B subtilis RNA and 18 adenine 25 guanine and 57 cytosine plus uracil for S aureus RNA The average SN of these measurements was 26 with an average error of 8 of the value (SplusmnN) It is also worth noting that the three RNA spectra show a marked shift in a band near 825 cm-1 This band is assigned to the symmetric stretch of the O-P-O ester linkage9 The band appears at 815 cm-1 for S aureus shifting to 820 cm-1 for B subtilis and 830 cm-1 for E coli Others have used the normal Raman intensity of the band at 815 cm-1 as a direct indication of the amount of A-class helix present while the intensity of the band at 830 cm-1 has been used as a direct indication of the amount of B-class helix present However the latter is more associated with DNA than RNA
Wavenumbers (∆cm-1) Figure 5 E-SER spectra of 020 mgmL adenosine monophosphate at A) -03 and B) -09VSCE and C) SG-SER spectra Conditions sample in 01M KCl buffered to pH 92 A) and B) 750 mW C) 75 mW of 1064 64 scans (1-min) at 8 cm-1
Wavenumbers (∆cm-1) Figure 6 E-SERS of 01 mgmL RNA from E coli 02 mgmL RNA from B subtilis and 02 mgmL RNA from S aureus Conditions 01M KCl pH 92 -03VSCE 750 mW of 1064 nm 640 scans (10 min) at 8 cm-1
AMP RNA
E coli A
B
C
B subtilis
S aureus
SPIE 2001-4575
69
SG-SER spectra of reasonable quality were also obtained for E coli and B subtilis especially the latter (Figure 8) However the spectra differ substantially from the E-SERS of the same samples Both SG-SER spectra are dominated by adenine at 735 cm-1 and a band at 1030 cm-1 Although unassigned the latter does appear in the RNA E-SER spectra Bands at 1105 and 1565 cm-1 are likely due to phosphate while bands at 1320 and 1470 cm-1 are less confidently assigned to phosphate They are significantly less intense and somewhat shifted from their SG-SERS counterparts (1335 and 1455 cm-1) A band at 670 cm-1 may be due to guanine which was observed at 664 cm-1 for SG-SERS of the pure sample However the SG-SER spectrum of AMP also had an intense 667 cm-1 band A number of other bands occur at 890 1070 1165 1245 1290 1420 1505 cm-1 and remain unassigned The SG-SER spectra are somewhat disappointing in that only adenine and guanine contributions can be positively identified This limits the ability to determine relative NA base concentrations and distinguish bacterial RNA However several of the unassigned bands may be due to the bases (eg 1030 and 1420 cm-1 due to cytosine) Further experiments will be required to clarify this point
Ecoli
B subtilis
S aureas
P G
A C+U
OPO
A B
Wavenumbers (∆cm-1) Wavenumbers (∆cm-1) Figure 7 SER spectra of RNA from A) B subtilis with contributions indicated and B) E coli B subtilis and S aureus with baseline correction and peak positions used to calculate contributions indicated G = guanine A = adenine C+U = cytosine plus uracil P = phosphate (backbone) OPO = phosphate ester linkage (A- vs B-class helix)
Wavenumbers (∆cm-1) Wavenumbers (∆cm-1) Figure 9 E-SER (-03VSCE) and SG-SER spectra of RNA from A) E coli and B) B subtilis Sample conditions as in Figure 6 E-SER spectra at 750 mW SG-SERS at 75 mW
A B E-SERS
SG-SERS
E-SERS
SG-SERS
SPIE 2001-4575
70
A final comparison was made between the two SER methods by measuring dipicolinic acid (DPA) This chemical may be invaluable as a test for spore forming bacteria specifically B anthracis 50 to 90 of B anthracis sporilates During spore formation dipicolinic acid is synthesized and once completed 10-15 of the dry spore weight is composed of the Ca2+ complex located in the spore core42 Heating in water can be used to initiate germination at which point the exosporium breaks and releases the Ca dipicolinate which becomes dipicolinic acid in water The structure of this chemical strongly
4 CONCLUSIONS Towards the goal of developing a method and technology to rapidly detect and identify bioagents we have been investigating surface-enhanced Raman spectroscopy as a tool to measure relative concentrations of nucleic acid bases in RNA and determine bioagent taxonomy Initially we investigated E-SERS since this method has been extensively researched and the optimum sample conditions are well developed However this method requires a three-electrode sample cell and electrolyte solution Incorporation of an electrolytic cell into sample systems used to analyze for BWA as aerosols in water or on surfaces can be designed using flow injection analysis technologies but cross-contamination and plugging of sample lines seems inevitable For this reason we also investigated metal-doped sol-gels as a SER-active medium Previous studies have shown this material to be active in all solvents particularly water capable of continuous measurements in flowing systems and reproducible (quantitative) between coated sample vials Here we compared SG-SER spectra to traditional E-SER spectra of the nucleic acid base pairs adenosine monophosphate and RNA High quality spectra of adenine cytosine guanine thymine and uracil were obtained by both E-SERS and SG-SERS Both methods yielded very similar spectra for the NA bases including a pH dependent study of adenine Enhancement factors and detection limits for adenine were determined as 2x105 and 16x10-11g and 1x105 and 12x10-10g for E-SERS and SG-SERS respectively (normalized to 75 mW and 10-min acquisition time) Fifty percent of the silver particle surface area in the sol-gel matrix was assumed covered by adenine which may have been overestimated yielding a lower EF and higher detection limit It should also be realized that each E-SER spectrum required several attempts to optimize the measurement conditions (pH electrode potential etc) While each SG-SER spectrum involved no sample preparation and often represents the first and only attempt to make the measurement Quality spectra of RNA extracted from Escherichia coli Bacillus subtilis and Staphylococcus aureus were obtained by E-SERS that were easily interpreted Bands due to adenine guanine cytosine plus uracil and phosphate were identified The SER band intensity of the NA bases in the RNA samples were of the same order of magnitude suggesting that their interaction with the silver surface is concerted as is their Raman enhancement Interestingly the relative SER band
Wavenumbers (∆cm-1) Figure 10 A) Raman spectrum of solid dipicolinic acid B) Ax20 C) electrolytic SERS of 6x10-3 M dipicolinic acid in 01 M KCl at a potential of +07VSCE and pH of 4 and D) sol-gel SERS of 6x10-3 M dipicolinic acid Conditions for A and C as in Figure 1 C) 100 mW of 1064 nm 50 scans 8 cm-1
A B (Ax20)
C
D
suggested that it would be SER active However the E-SER spectrum was unstable and varied considerably as a function of potential A consistent spectrum was obtained at +06VSCE (Figure 10) This potential is not recommended for measurement because the surface is actively dissolving in solution The SG-SER spectrum was considerably more stable of higher quality and easily reproduced Bands at 660 825 1010 1390 1430 1570 1590 and 3075 cm-1 were observed Enhancement factors were determined for the two media using the symmetric ring stretching mode at 995 and 1010 cm-1 for the normal Raman and SER spectra respectively E-SERS yielded an EF of 5x103 while SG-SERS yielded an EF of 2x105 for DPA The SN of the latter suggests a detection limit of 20x10-10g (based on adenine coverage 75 mW and 10-min) The differences in SER activity for these two media may be attributed to the combined electrolytic potential of the solution chemical and metal15 Again the E-SERS suggests that the SG-SERS is at a potential positive of the pzc While the instability in the E-SERS may also be associated with surface interactions of two carboxylic acid groups of dipicolinic acid during the ORCs
DPA
SPIE 2001-4575
71
intensities for RNA extracted from E coli are very similar to those measured by normal Raman spectroscopy Although the relative percent that each of the NA bases contributed to each RNA sample was not determined reproducible band intensities allowed noting the following trends The percent adenine decreases while the combined percent cytosine and guanine increase for both B subtilis and S aureus compared to E coli Quality spectra were also obtained for the RNA samples by SG-SERS but only a few bands were readily identified Calculations of NA base concentrations by SG-SERS will require further research In light of recent events we summarize the capabilities of these SERS media in terms of rapid detection of B anthracis and dipicolinic acid However these capabilities must be qualified First and foremost the level to which SERS can distinguish bacteria or viruses has not yet been determined Development of a database of both DNA and RNA base concentrations for BWAs and common bacteria to establish the level of taxonomic identification is ongoing Second rapid collection of aerosol water or surface samples is being addressed by others who report trapping particles on filters from 100 liters of air per minute Third although not presented here we have developed methods to extract RNA or DNA from cells and spores for SER analysis within 10 minutes Finally we assume a detection limit of 3600 spores per 100 liters of air is required although a 50 lethal dosage of anthrax has not been established With these qualifications a mass detection limit for RNA using SERS is estimated as follows A single measurement is performed in ca 20 minutes (140 liters collected in 14 min RNA extracted in 8 min spectral acquisition and analysis in 10 min) The average human breaths 7 liters per minute therefore the analyzer must at the very minimum detect 5000 spores in 140 liters of air One spore is approximately 2x10-18m3 (1x1x2 microm) and if a density of 075 gcm3 is assumed this corresponds to a mass of 15x10-12g Each spore contains 4-12 RNA or 12x10-13g RNA for 8 If we assume 23 of the RNA can be isolated for analysis during lysis then the proposed instrument must be able to detect 4x10-10g RNA from 5000 spores per 70 liters of air within 10 minutes As noted above the mass detection limits for adenine were estimated at 16x10-11g and 12x10-10g for E-SERS and SG-SERS respectively Although these detection limits suggests that RNA from 5000 spores is detectable with the current instrumentation it is highly likely that only a portion of an RNA segment (eg 120-nucleotide 5S rRNA) is in contact with the metal surface and will contribute to the SER effect The SN for the RNA spectra were 110th of the average SN for the four individual RNA bases suggesting a 10 contribution Furthermore effective taxonomy will likely require knowing the NA base concentrations to 1 of the value (eg 25plusmn025) Again the average measurement error for the bases is 12 These values suggest that the E-SERS is within a factor of 4 of the required detection limit whereas the SG-SERS detection limit must be improved by 25 times The same arguments can be applied to the detection of dipicolinic acid If we assume a spore releases 10 by weight DPA during germination then the proposed instrument must be able to detect 75x10-10g DPA from 5000 spores per 70 liters of air within 10 minutes The detection limit for SG-SERS was estimated at 20x10-10g and suggest that the vials are suitable to perform a rapid screen for anthrax A series of concentration dependent measurements are currently being performed to verify this assertion Finally we note that the measurements performed here employed an FT-Raman spectrometer This instrumentation was chosen over dispersive Raman instrumentation for the high wavelength accuracy afforded by the HeNe reference laser (Connes Advantage43) which would allow reliable spectral subtraction matching of observed spectra to stored library spectra and confident use of chemometric approaches Such data analysis is likely to be required to enhance BWA identification However this instrumentation sacrifices sensitivity due to a less efficient detector (InGaAs vs Si) less efficient Raman scattering and less efficient generation of plasmon modes Substantial improvements can be obtained using visible excitation and Si detection and these measurements are underway
5 ACKNOWLEDGEMENTS The authors are grateful to Drs D Cookmeyer and S Tove of the US Army Research Office (Contract Number DAAH04-96-C-0078) for their interest and support of this research The authors would also like top acknowledge Dr R Yin and J Jensen for supporting development of the metal-doped sol-gels (Contract Number DAAD13-01-C-0019) They also thank Dr Wilfred H Nelson for assistance in spectral interpretations
SPIE 2001-4575
72
6 REFERENCES 1 Roberts WL and Rainey PM Clin Chem 39 1872-1877 (1993) 2 Pasechnik VA CC Shone and P Hambleton Bioseperations 3 267-283 (1993) 3 Jackson PJ ME Hugh-Jones DM Adair G Green KK Hill CR Kuske LM Grinberg FA Abramova and P
Keim Proc Natl Acad Sci 95 1224-1229 (1998) 4 Jensen JL NC Wong W R Loerop SPIE 4775-03 (2001) 5 Snyder A Pet al SPIE 3853-15 (1999) 6 Danosky T R and McFadden P N in press (1997) 7 Woodruff WH Farquharson S Science 201 831-833 (1978) 8 Lord RC and Thomas GJJr Spectrochemica Acta 23A 2551-2591 (1967) 9 Thomas GJJr Biochim Biophys Acta 213 417-423 (1970) 10 Placzek G Handbuch der Radiologie 2 EMarx ed Akademische Verlagagescellschatt Liepzig 1934 UCRL
Trans No 526 (1959) 11 Jeanmaire DL and RP Van Duyne J Electroanalytical Chem 84 1-20 (1977) 12 Chada S Manoharan R Moenne-Loccoz P Nelson WH Peticolas WL and Sperry JF Applied Spectroscopy
47 38-43 (1993) 13 Manoharan R Ghiamati E Chada S Nelson WH and Sperry JF Applied Spectroscopy 47 2145-2150 (1993) 14 Todd EA Morris MD Applied Spectroscopy 48 545-548 (1994) 15 Wentrup-Byrne E Sarinas S and Fredericks PM Applied Spectroscopy 47 1192-1197 (1993) 16 Weldon MK VR Zhelyaskov and MD Morris Applied Spectroscopy 52 265-269 (1998) 17 Kneipp K and J Fleming J Mol Structure 145 173-179 (1986) 18 Pothier NJ and Force RK Applied Spectroscopy 46 147-151 (1992) 19 Ervin KM E Koglin JM Sequaris P Valenta and HW Nurnberg J Electroanal Chem 114 179-194 (1980) 20 Kim SK TH Joo SW Suh and MS Kim J Raman Spectrosc 17 381-386 (1986) 21 Pothier NJ and Force RK Analytical Chemistry 62 678-680 (1990) 22 Kneipp K Y Wang RR Dasari and MS Feld Applied Spectroscopy 49 780-784 (1995) 23 Nie S and Emory SR Science 275 1102 (1997) 24 Graham D WE Smith AMT Linacre CH Munro ND Watson and PC White Analytical Chemistry 69 4703-
4707 (1997) 25 Fleischmann M PJ Hendra and AJ McQuillan Chem Phys Lett 26 163-166 (1974) 26 Pettinger B J Chemical Phys 85 7442-7451 (1986) 27 Surface-Enhanced Raman Scattering Section Four Theory SPIE MS 10 M Kerker and B Thompson Eds (1990) 28 Wang D-S and Kerker M SPIE (M Kerker and B Thompson Eds) MS 10 417-429 (1990) 29 Weaver MJ Farquharson S Tadayyoni MA J Chem Phys 82 4867-4874 (1985) 30 Norrod KL Sudnik LM Rousell D and Rowlen KL Applied Spectroscopy 51 994-1001 (1997) 31 Farquharson S Weaver WJ Lay PA Magnuson RH and Taube H J Am Chem Soc 105 3350-3351 (1983) 32 Lee PC and Meisel D ldquoAdsorption and Surface-Enhanced Raman of Dyes on Silver and gold Solsrdquo J Phys Chem
86 3391-3395 (1982) 33 Li Y-S and Wang Y Applied Spectroscopy 46 142-146 (1992) 34 Bello JM DL Stokes and T Vo-Dinh Analytical Chemistry 61 1779-1783 (1989) 35 van Duyne RP J C Hulteen D A Treichel M T Smith M L Duval and T R Jensen J Phys Chem B1033854-
3863 (1999) 36 Farquharson S and W W Smith W H Nelson and J F Sperry SPIE 3533-27 207-214 (1998) 37 Lee Y H W Smith S Farquharson H C Kwon M R Shahriari and P M Rainey SPIE 3537 252-260 (1998) 38 Lee Y-H S Farquharson and P M Rainey SPIE 3857 76-84 (1999) 39 Current Protocols in Molecular Biology Wiley Interscience1003-1006 (1987) 40 Farquharson S Smith W Carangelo R C and Brouillette C SPIE 3859 14-23 (1999) 41 Chase D B and JF Rabolt Fourier Transform Raman Spectroscopy Acad Press Ch1 p 131 (1994) 42 Brock TD MT Madigan JM Martinko and J Parker Biology of Microorganisms 7th Ed Prentice Hall p 76-80
(1994) 43 Connes J Rev Opt Theor Instrum 40 45 (1961)
166
Chemical agent identification by surface-enhanced Raman spectroscopy
Stuart Farquharson and Paul Maksymiuk
Real-Time Analyzers 87 Church Street East Hartford CT 06108
Kate Ong and Steven D Christesen US Army SBCCOM Aberdeen Proving Ground MD 21010
ABSTRACT The recent distribution of anthrax through the US postal system and the subsequent infection and death of several postal and national media employees amplifies the need for methods to rapidly detect identify and quantify this and other chemical and biological warfare agents The US military has also identified water supplies as a likely method of warfare agent deployment and is funding the development of a Joint Service Agent Water Monitor (JSAWM) In an effort to aid military personnel and the public at large we are developing a portable analyzer capable of identifying and quantifying chemical agents rapidly either on-demand or continuously The approach is based on the ability of Raman spectroscopy to identify molecular structure through the abundant vibration information provided in spectra and the ability of surface-enhanced Raman spectroscopy (SERS) to detect extremely low concentrations (eg part-per-billion) through the enhancement of Raman scattering by six orders of magnitude or more A key element to the analyzer design is a new SER active medium that is capable of quantitative reversible measurements The medium consists of silver or gold nanoparticles incorporated into a sol-gel matrix The porous silica network offers a unique environment for stabilizing SER active metals and the high surface area increases the interaction between the analyte and metal particles Here we present the use of new sol-gels that also selectively enhance chemicals based on polarity and charge Base-line measurements of chemical agents and their hydrolysis products are presented and compared to the JSAWM goal of 30 micrograms per liter detection Keywords Chemical warfare agent hydrolysis product SERS Raman spectroscopy sol-gel nanoparticle
1 INTRODUCTION Since September 11 2001 the threat of terrorist attacks and biological warfare within US borders has become a sobering reality The simplicity in manufacturing ease of deployment and the relatively low cost of chemical warfare agents (CWAs) raises public concern that they may also be used by terrorists Indeed terrorists released sarin (GB) in the Tokyo subway in 19951 Countering terrorism and terrorist attacks requires recognizing likely deployment scenarios and having the required technology to rapidly detect the deployment event One method of deployment has been long identified by the US military distribution through water supplies To counter this threat the Department of Defense is funding or monitoring the capabilities of 200 technologies with the goal of developing a Joint Service Agent Water Monitor (JSAWM) that is field portable2 The JSAWM must be capable of identifying the chemical agents with no false-positives and quantifying the agents at microgram per liter concentrations within 10 minutes (Table 1)3 This includes the analysis of drinking water supplies distribution and storage systems as well as potable water supplies The earliest technologies developed for CWA detection were based on electrochemical ionization or colorimetric analysis (eg phosgene tape) Although these analyzers were easy to use they were not generally agent specific and suffered from false-positives4 More traditional laboratory methods have also been investigated and in particular combined gas chromatography and mass spectrometry (GCMS) has been very successful at eliminating false-positives56 However GCMS requires extraction repeated calibration and long analysis times (typically 20 to 60 minutes)6 making it labor intensive and less than desirable for field use More rapid analysis of agents in the solid liquid and gas phase has been To whom correspondence should be addressed emailfarqureal-time-analyzerscom
Vibrational Spectroscopy-based Sensor Systems Steven D Christesen Arthur J Sedlacek III EditorsProceedings of SPIE Vol 4557 (2002) copy 2002 SPIE middot0277-786X02$1500
167
Table 1 Chemical Agent Structures Hydrolysis Half-lives and JSAWM Thresholds Agent Short-Hand Chemical Structure Hydrolysis
Half-Life JSAWM
Thresholds
Sarin (GB) F-[O=P-CH3]-O-CH(CH3)2 213 hours 32 microgL Soman (GD) F-[O=P-CH3]-O-CH(CH3)-(C-(CH3)3) 23 hours 32 microgL Tabun (GA) (CH3)2-N-[O=P-CN]-O-C2H5 41 hours 32 microgL VX C2H5O-[O=P-CH3]-S-(CH2)2-N-(CH(CH2)2)2 821 hours 32 microgL EA2192 HO-[O=P-CH3]-S-(CH2)2-N-(CH(CH2)2)2 gt9 years 32 microgL Mustard (H) ClCH2CH2-S-CH2CH2Cl encapsulates 47 microgL Lewisite (L) ClCH=CH-As-Cl2 rapid 27 microgL HCN HCN rapid 20 mgL BZ C7NH12-O-[C=O]-COH(C6H5)2 23 microgL T-2 Toxin 87 microgL at pH 7 to 75 and 20 to 25 oC
demonstrated by vibrational spectroscopy7-10 Hoffland et al7 reported infrared absorbance spectra and absolute Raman cross sections for several chemical agents while Christesen measured Raman cross sections for sarin tabun mustard gas and VX (ethyl S-2-diisopropylamino ethyl methylphosphonothioate)11 Again however these techniques also have limitations Raman spectroscopy is simply not a very sensitive technique and detection limits are typically 01 (1000 ppm) While infrared spectroscopy would have limited value in analyzing poisoned water since the very strong infrared absorption of water would obscure most other chemicals present Nevertheless efforts to overcome these limitations have been demonstrated Braue and Pannella8 quantified the G-series nerve agents (tabun sarin and soman) in terms of infrared attenuated total reflectance using a circle-cell And Alak and Vo-Dinh demonstrated the possibility of surface-enhanced Raman spectroscopy (SERS) to identify CWAs by measuring several organophosphonates that simulate the nerve agents12 However quantitative measurements have not been demonstrated for the SER-active material used (silver coated on alumina particles) or other SER-active media13 Recently we developed silver-doped sol-gels to promote the SER effect14-17 The porous silica network of the sol-gel matrix offers a unique environment for stabilizing SER-active metal particles and the sol-gel provides a high surface area that effectively increases the number of molecules observed within the Raman scattering volume The silver-doped sol-gels have been coated on the internal walls of standard glass vials to allow rapid SER analysis of small sample quantities (lt 01 mL) without preparation We have used p-aminobenzoic acid (PABA) as a test chemical to demonstrate surface enhancements greater than 106 reversible measurements in a flowing system reproducible measurements from vial-to-vial and batch-to-batch and measurements in multiple solvents including water14-17 Recently we used these vials to measure Tabun (GB) and Sarin and several hydrolysis products pinacolyl methyl phosphonate (PMP from Soman) and methyl phosphonic acid (MPA from all G-agents Figure 1) Although a number of unique vibrational bands are observed (eg C-N stretch doublet and P-C stretch) the G-agents were only observed for 5 concentrations and all spectra required baseline corrections Figure 1 Surface enhanced Raman spectra of ~5 vv A) Tabun and B) Sarin C) 1 vv PMP and D) 10 ppm MPA using sol-gel sample vials 785 nm excitation 1-min scan and CCD detection Performed at Aberdeen Proving Ground
C A
Wavenumber (∆cm-1) Wavenumber (∆cm-1)
D B
790
545 C-N
2135 2190 P-C 770
760 1290
Proc SPIE Vol 4577
168
Nevertheless MPA was readily observed for a 10 ppm sample with an estimated detection limit of 04 ppm (based on a signal-to-noise ratio of 3 for the 760 cm-1 band intensity) This measurement provides encouragement in that SERS may satisfy the needs of a JSAWM Furthermore MPA is also a hydrolysis product of VX and V-gas and EA2192 (Figure 2) and may prove a valuable indicator of agent usage Figure 2 Hydrolysis of Sarin to form hydrofluoric acid (HF) methylphosphonic acid 1-methylethyl ester (MPAMME) methyl phosphonic acid (MPA) and 2-propanol With this initial albeit modest success we began analyzing chemicals with various sol-gel compositions that we have been developing Here we describe four sol-gel compositions that select for 1) polar-positive 2) polar-negative 3) weakly polar-positive and 4) weakly polar-negative chemical species The ability of these sol-gels to select and enhance Raman scattering is described for several test chemicals and MPA
2 EXPERIMENTAL The chemicals analyzed as well as all chemicals used to prepare the metal-doped sol-gels were obtained at their purest commercially available grade from Aldrich (Milwaukee WI) The sol-gel designed to select for polar-negative species was prepared from a silver amine complex tetramethyl orthosilicate (TMOS) and methanol After mixing 02 mL of the sol-gel solution was transferred into a glass vial (2 mL) dried and heated The incorporated silver ions were then reduced using dilute sodium borohydride The vials were washed and dried prior to the addition of a sample solution In a similar manner the sol-gel designed to select for polar-positive species was prepared from a gold salt TMOS and methanol The sol-gel designed to select for weakly polar-negative species was prepared from a silver amine complex tetraethyl orthosilicate (TEOS) and methanol And the last sol-gel designed to select for weakly polar-positive species was prepared from a gold salt TEOS and methanol All samples were prepared in a chemical hood and transferred into plain or SER-active vials for analysis Normal Raman spectral measurements employed 1-mL pure samples that were placed in a 1-cm3 cuvette and weighed This yielded a powder density that allowed accurate calculation of molecules in the optical collection field SERS measurements employed 1-mg sample per mL water concentrations unless otherwise stated Once prepared a 01 mL sample was placed into one of the four selective sample vials which in turn was placed into the sample compartment of a Raman spectrometer for analysis A prototype Fourier transform Raman spectrometer (RTA) was used for these measurements18 The system consisted of a NdYAG laser (Brimrose) for excitation at 1064 nm an interferometer built by On-Line Technologies (OLT East Hartford CT) for frequency separation an uncooled InGaAs detector for signal detection (RTA) and an Intel 400 MHz Pentium II based laptop computer (Dell Round Rock TX) for interferometric control data acquisition (OLT) and analysis (LabVIEW by National Instruments Austin TX) Additional components included a Notch filter (Kaiser Ann Arbor MI) and interferometer entrance and exit optics (Edmund Scientific Barrington NJ) Fiber optics were used to deliver the excitation beam to the sample and the scattered radiation to the interferometer (1 meter lengths of 200 and 365 micron core diameter respectively Spectran Avon CT) A second Notch filter (Kaiser) was used as a beam splitter to direct the excitation beam along the same axis as the collected radiation A microscope object (20x04 Newport Irvine CA) was used to focus the beam into the sample and to collect the scattered radiation back along the same axis In this co-axial backscattering arrangement the excitation beam was passed through the outside of a glass vial and focused onto the silver-doped sol-gel film (01-03 mm thickness) containing the sample
3 RESULTS AND DISCUSSION p-aminobenzoic acid (PABA) and phenyl acetylene (PA) and were used to refine the selectivity and SER-activity of the four different metal-doped sol-gels PABA is a popular chemical used to evaluate the performance of SER-active media Here the polar end groups can be used to test selectivity of the polar-negative and polar-positive sol-gels PA is essentially non-
2O+ H HF + +OH
OH3H C
OP
OF
CH
CH3
3
3H C
OCP
OOH
CH
CH3
3
3H C
O
CPHO
CH
CH
3
3
C
Sarin MPAMME MPA 2-propanol
Proc SPIE Vol 4577
169
polar but a high electron density in the cylindrical π cloud around the carbon-carbon triple bond allows testing the selectivity of the weakly polar-negative and weakly polar-positive sol-gels As Figure 3 illustrates PABA passes through the polar sol-gel and is enhanced by either the silver or gold particles At 1 mgml the concentration of neutral PABA is ca 20 times that of the ionized form (pKa = 48) For electropositive silver the PABA anion is expected to interact through the carboxylate group and the associated vibrational modes are expected to dominate the spectrum Conversely for electronegative gold either form of PABA is expected to interact through the amine group The clear differences in our spectra support this expectation Furthermore bands at 840 and 1405 cm-1 assigned to a COO- bend and stretch respectively are significantly more intense for silver than gold Additional bands at 1140 and 1195 cm-1 are assigned to CH bending modes while bands at 1450 1500 and 1605 cm-1 are assigned to ring vibrational modes A very similar SER spectrum for PABA on a silver-coated alumina substrate has previously been reported with similar assignments19 For the gold-doped sol-gel new bands appear at 690 1355 and 1585 cm-1 The first band is assigned to a ring-H bending mode the second band to a ring-N- stretching mode and the third band to a possible NH2 scissors mode or ring mode The second band is not observed in the normal Raman spectrum but infrared bands occur at this frequency for aromatic ring-secondary amine stretching modes The scissors mode occurs at this frequency in Raman spectra for several chemicals but is absent in the PABA Raman spectrum Alternatively this mode may be the1600 cm-1 ring mode that has been shifted by the gold interaction Again a very similar SER spectrum of PABA has been reported but surprisingly using silver (colloids)2021 not gold as the enhancement medium These researchers also assumed the primary interaction of PABA with silver was through the carboxylate anion and made assignments accordingly For example they assigned the 1359 cm-1 to a COO- stretch not to the amine group as we have They also favor the ring stretching mode assignment for the 1582 cm-1 band Finally it should be said that other researchers have argued that the most dominant band in the SER spectra at 1450 cm-1 a ring vibration mode suggests that PABA lies flat on the surface and the π-orbitals dominate the surface interaction22
Figure 3 SER spectra of A) PABA using polar-negative and B) polar-positive sol-gels and C) PA using weakly polar-negative and D) weakly polar-positive sol-gels PABA is 1 mgmL PA is 1 vv Spectral conditions 75 mw 1064 nm 100 scans (15 min) 8 cm-1 resolution Non-polar PA passes through the non-polar sol-gels and is also enhanced by both metals The spectra are easily understood For electropositive silver PA interacts through the cylindrical triple bond π electron cloud and a -CequivC- doublet occurs near 2000 cm-1 The interaction is reasonably strong since this band appears at 2112 cm-1 in the normal Raman spectrum For electronegative gold this interaction is unlikely and only very weak bands occur near 2000 cm-1 The remaining bands are at 1000 cm-1 1200 cm-1 doublet and 1595 cm-1 all appear in the normal Raman spectra at virtually the same frequencies and are assigned to the symmetric ring-breathing mode CH bending modes and the trigonal ring-breathing mode respectively The polarnon-polar selectivity of the polar-negative and weakly polar-negative sol-gels was tested by adding a 11 molar mixture of PABA and PA The selective enhancement is quite good (Figure 4) The spectrum obtained using the polar sol-gel represents 78 PABA and 22 PA while the spectrum obtained using the weakly polar sol-gel represents 9 PABA and 91 PA The band peak intensities at 2000 cm-1 for PA and 1450 cm-1 for PABA were used for these calculations and are expanded in Figure 4 for clarity
C A
Wavenumber (∆cm-1) Wavenumber (∆cm-1)
2NH COOH
D B
C CH
Proc SPIE Vol 4577
170
Figure 4 SERS of 11 MM of PABA and PA in A) polar-negative and C) weakly polar-negative sol-gels The lower traces compare the pure chemicals B) 1 mgml PABA in polar-negative sol-gel and D) 1 PA in weakly polar-negative sol-gel while the insets magnify the minority species for clarity (x5 in A and x10 in B) Spectral conditions as in Figure 3 Following this development of selective sol-gels that maintained SER activity we measured cyanide and MPA (Figure 5) Not surprisingly the best sensitivity for both hydrolysis products was obtained using the polar-negative sol-gel The interaction of the cyanide anion with the silver surface is sufficient to shift the CequivN stretch observed at 2080 cm-1 in the normal Raman spectrum to 2145 cm-1 in the surface-enhanced Raman spectrum Furthermore the band is substantially broadened This anion has been extensively studied by electrolytic SERS and this shift and broadening have been attributed to the formation of a tetrahedral Ag(CN)3
2- surface structure23 Figure 5 Surface-enhanced (upper traces) and normal Raman spectra (bottom traces) of A) CN- and B) MPA in silver-doped TMOS SERS conditions as in Figure 3 and 1 mgmL Note MPA yields two distinct spectra for neutral (top) and acidic pH (middle) The normal Raman spectra employed pure powders 500 scans and 900 mW of 1064 nm SER measurements of MPA with the polar-negative sol-gel yielded two unique spectral signatures that depended on solution pH (Figure 5) For more neutral solutions the P-C stretch of MPA at 762 cm-1 dominates and the CH2 stretch at 2922 cm-1 appears The SN is sufficiently high that the anti-Stokes Raman shift at -762 cm-1 is observed For deprotonated MPA an oxygen-surface mode appears at 325 cm-1 (as well as its anti-Stokes complement) suggesting a strong interaction This results in substantial enhancement of the P-O-C mode at 1051 cm-1 (upper trace) Others report that this mode dominates the infrared spectra of nerve agents measured in water8 Comparison of the two spectra suggests the following molecule-to-surface orientations The appearance of the oxygen-surface and P-O-C modes in the upper spectrum of Figure 5B indicates that the tetrahedral molecule interacts with the silver surface through the deprotonated oxygen and is oriented end-on The
A B
Wavenumber (∆cm-1)
C A
Wavenumber (∆cm-1) Wavenumber (∆cm-1)
D B
Wavenumber (∆cm-1)
Proc SPIE Vol 4577
171
dominance of the P-C and the CH2 stretches and the disappearance of the P-O-C mode in the upper spectrum suggest the molecule is oriented with the methyl-hydroxyl-hydroxyl face towards the surface However considerably more research must be performed to verify these points Both chemical forms were measured as a function of concentration to determine the onset of monolayer coverage and estimate expected detection limits (Figure 6) Below monolayer coverage the signal to concentration dependence should be linear and the SN of any spectral measurement in this range can be used to predict the detection limit In the spectra presented here the peak height was used as the signal while the noise as root-mean-squared (RMS) was measured between 4400-4600 cm-1 Since noise is distributed evenly throughout the spectrum when transformed this region was used since it does not have any contributions from signals or baseline offsets Figure 6 shows a series of spectra for MPA along with a plot of the 1050 cm-1 band intensity (the noise was identical for this series) as a function of sample concentration A clear discontinuity appears in the vicinity of 01 mgmL (19 ppm) indicating the onset of monolayer coverage A detection limit defined as a SN of 3 was calculated for the 01 and 005gmL samples at 24x10-4 and 25x10-4 gL respectively A more modest detection limit of 101x10-4 gL was obtained using the 760 cm-1 band in the second series of concentration measurements These detection limits correspond to 46 and 190 ppb for a 3-min scan and 75 mW of incident laser power
Figure 6 A) Concentration dependence of MPA SERS measured in silver-doped TMOS) B) Concentrations are 001 005 01 05 1 gL (188 94 188 94 188 ppm) I760 series (bull) and I1050 series (∆)
Enhancement factors for cyanide and methyl phosphonic acid can be determined by estimating the number of molecules contributing to the surface-enhanced and normal Raman spectra The enhancement factor EF can be defined by the following equation
EF = (ISERSINR)bull(MNRMSERS) bull(PNRPSERS) bull(TNRTSERS)12
where I is the spectral band intensity M is the sample mass P is the incident laser power and T is the measurement time (or number of scans) for the two measurements For the normal Raman spectra a cylindrical scattering volume is assumed based on the laser area (28x10-7m2 6x10-4m diameter spot) and the penetration depth (1x10-3 m)24 The density of KCN and MPA as powders were measured at 0572 and 0516 gcm3 indicating that 16x10-4 and 144x10-4 g produced the normal Raman signals in Figure 5 respectively The number of molecules contributing to the SER spectra are those on the silver particles embedded in the sol-gel The total silver surface area can be determined from the average particle size concentration and the scattering volume Previous scanning electron micrographs showed the average silver particle size to be 40 nm in diameter (335x10-23m3)17 The silver concentration is 012M based on the reactant molar concentrations and dilution factors And the scattering volume is 76x10-
11m3 again based on a cylindrical scattering volume defined by a laser area of 28x10-7m2 and a sol-gel thickness of 27x10-
4m This volume contains 123x10-6g of silver equivalent to 35x109 silver particles with a collective surface area of 18x10-
5m2 However it may reasonably be assumed that at least half of the silver surface is in contact with the sol-gel matrix and unavailable for analyte interaction If we assume monolayer coverage and that each CN molecule occupies 15x10-20m2 then approximately 62x1014 molecules or 27x10-8g of CN contribute to the SER spectrum (20x10-19m2 46x1013 molecules 74x10-9g for MPA) Accordingly the EF for cyanide equals 48x104 ((180599) bull(16x10-427x10-8) bull(90075) bull(500100)12) The EF for MPA is considerably higher at 87x106 ((60326) bull(144x10-474x10-9) bull(90075) bull(500200)12)
0
20
40
60
80
100
120
140
0 02 04 06 08 1 12[MPA] (mgmL)
I (76
0)
0
100
200
300
400
500
600
I (10
50)
Wavenumber (∆cm-1)
A B
Proc SPIE Vol 4577
172
4 CONCLUSIONS Here we present for the first time surface-enhanced Raman spectra of Tabun and Sarin obtained using silver-doped sol-gels However the inferior enhancement suggested employing SER-active sol-gels varying in composition to improve both detection limits and selectivity of the target analytes To this end we successfully demonstrated the capabilities of four sol-gels that select for 1) polar-positive 2) polar-negative 3) weakly polar-positive and 4) weakly polar-negative chemical species p-aminobenzoic acid was used to show that silver could be used to attract polar-negative chemicals or functional groups (carboxylate anion) and that gold could be used to attract polar-positive chemicals (amine) while a mixture of p-aminobenzoic acid and phenylacetylene was used to show that tetramethyl orthosilicate preferentially solvates polar chemicals and that tetraethyl orthosilicate preferentially solvates weakly polar chemicals This increased sample control was applied to cyanide and methyl phosphonic acid two hydrolysis products of chemical warfare agents Exceptional results were obtained for methyl phosphonic acid allowing measurement of 1x10-2 gL for a 3-min scan and 75 mW of incident laser power with an estimated detection limit of 245x10-4 gL and an enhancement factor of 87x106 However this detection limit is 76 times less sensitive than required for the JSAWM (32x10-6gL for the G-agents) Finally we note that the measurements performed here employed an FT-Raman spectrometer This instrumentation was chosen over dispersive Raman instrumentation for the high wavelength accuracy afforded by the HeNe reference laser (Connes Advantage25) which would allow reliable spectral subtraction matching of observed spectra to stored library spectra and confident use of chemometric approaches Such data analysis is likely to be required to identify the chemical agents as well as distinguish hydrolysis products However this instrumentation which employs 1064 nm excitation and InGaAs detection sacrifices sensitivity We believe that the measurement sensitivity can be improved by at least two orders of magnitude by using shorter laser excitation wavelengths (eg 532 nm) This would provide more efficient Raman scattering (fourth power dependence on laser excitation wavelength) more efficient generation of plasmon modes and allow using more efficient detector material (Si vs InGaAs) These modifications are underway
5 ACKNOWLEDGEMENTS The authors would like to thank Dr R Yin and J Jensen of the US Army for supporting this work (Contract Number DAAD13-01-C-0019) They would also like to thank Advanced Fuel Research for making their laboratory facilities available
6 REFERENCES 1 Tu Anthony ldquoOverview of Sarin Terrorist Incidents in Japan in 1994 and 1995rdquo 6th CBW Protection Symposium
Stockholm Sweden 10-15 May 1998 2 Jensen JL NC Wong W R Loerop SPIE 4775-03 (2001) 3 JSAWM Requirements at wwwsbccomapgeaarmymilRDAecbcrtPRODSERJSAWMjsawmhtml 4 ldquoThe Chemical Weapons Convention Redefines Analytical Challengerdquo Analytical Chemistry News amp Features June 1
397A (1998) 5 Johnston RL Hoefler CM Fargo JC and Moberley B ldquoThe Defense Nuclear Agencyrsquos ChemicalBiochemical
Weapons Agreements Verification Technology Research Development Test and Evaluation Program and its Requirements for On-Site Analysisrdquo AT-ONSITE 5-8 (1994)
6 Black RM Clarke RJ Read RW and Reid MT ldquoApplication of gas chromatography-mass spectrometry and gas chromatography-tandem mass spectrometry to the analysis of chemical warfare samples found to contain residues of the nerve agent sarin sulphur mustard and their degradation productsrdquo J Chromatography 662 301-321 (1994)
7 Hoffland LD Piffath RJ Bouck JBrdquoSpectral signatures of chemical agents and simulantsrdquo Optical Engineering 24 982-984 (1985)
8 Braue EHJ Pannella MGrdquoCIRCLE CELL FT-IR Analysis of Chemical Warfare Agents in Aqueous Solutionsrdquo Applied Spectroscopy 44 1513-1520 (1990)
9 Tseng C-H Mann CK and Vickers TJ ldquoDetermination of Organics on Metal Surfaces by Raman Spectroscopyrdquo Applied Spectroscopy 47 1767-1771 (1993)
10 Staff Reporter ldquoUS Army Tests Lidar to Detect Biological Toxinsrdquo Photonic Spectra pg 50 December 1998
Proc SPIE Vol 4577
Proc SPIE Vol 4577
173
11 Christesen SD Raman cross sections of chemical agents and simulants Applied Spectroscopy 42 318-321 (1988) 12 Alak AM and Vo-Dinh T ldquoSERS of Organophosphorous Chemical Agentsrdquo Analytical Chemistry 59 2149-2153
(1987) 13 Norrod KL Sudnik LM Rousell D and Rowlen KL ldquoQuantitative Comparison of Five SERS Substrates
Sensitivity and Detection Limitrdquo Applied Spectroscopy 51 994-1001 (1997) 14 Lee Y and Farquharson S ldquoSERS Sample Vials Based on Sol-Gel Process for Trace Pesticide Analysisrdquo SPIE 4206
140-146 (2000) 15 Farquharson S and Lee Y ldquoTrace Drug Analysis by Surface-Enhanced Raman Spectroscopyrdquo SPIE 4200-16 (2000) 16 Lee Y Farquharson S and Rainey P M Surface-Enhanced Raman Sensor for Trace Chemical Detection in Water
SPIE 3857 76-84 (1999) 17 Lee Y Farquharson S Kwong H and Shahriari M ldquoSurface-Enhanced Raman Sensor for Surface-Enhanced Raman
Spectroscopyrdquo SPIE 3537 252-260 (1998) 18 Farquharson S Smith W Carangelo R C and Brouillette C ldquoIndustrial Raman Providing Easy Immediate Cost
Effective Chemical Analysis Anywhererdquo SPIE 3859 14-23 (1999) 19 Narayanan VA JM Bello JD Stokes and T Vo-Dinh Analusis 19 307-310 (1991) 20 Laserna JJ E L Torres and JD Winefordner Analytica Chemica Acta 469-480 (1987) 21 Torres EL and JD Winefordner Analytical Chemistry 59 1626-1632 (1987) 22 Suh JS DP DiLella M Moskovits J Phys Chem 87 1540-1544 (1983) 23 Benner RE R Dornhaus R Chang and BL Laube Correlations in the Raman spectra of cyanide complexes adsorbed
at silver electrodes with voltammograms Surface Science 101 341 (1980) 24 Chase D B and JF Rabolt Fourier Transform Raman Spectroscopy Acad Press Ch1 p 131 (1994) 25 Connes J Rev Opt Theor Instrum 40 45 (1961)
APPLIED SPECTROSCOPY 351
focusing the 488 nm laser beam 2 mm deep into thebulk of the crystals to avoid contributions from potentialdepletion layers As shown in Fig 1b the dependence ofthe Raman shift on the degree of deuteration is almostperfectly linear and ts very well with D 5 22684cmR1 24526 where D is the degree of deuteration (in )and R is the spectral mean of the PO4 vibration in cm21A linear correlation coef cient of 0998 indicates an ex-cellent linear dependence of the Raman peak shift withdegree of deuteration This result shows that the shift ofthe PO 4 peak is simply caused by the linear increase inatomic mass due to isotope substitution which decreasesthe length of hydrogen-like bonds
This excellent linear dependence allows us to map thepro le of the DH exchange layer at the surface of DKDPcrystals by acquiring Raman spectra and determining theposition of the PO 4 peak for various depths This methodis preferable over other methods such as determining thestrength of the OD vibration directly (eg at 715 cm21)because the position of the most intense peak in the Ra-man spectrum can be measured more precisely than theintensity of some of the weakest peaks in the spectrumThis is demonstrated in Fig 2 where depth-dependentRaman spectra (Fig 2a) and the resulting exchange layerpro les for two DKDP crystals are shown (Fig 2b) Thespectra in Fig 2a were obtained from a depth scan of aDKDP crystal with 75 degree of deuteration in thebulk grown at 45 8C The spectra start out as DKDP with30 deuteration close to the surface and approach thebulk DKDP spectrum within a few micrometers of depthThe fact that the relative degree of deuteration does notextend to 0 D is due to the limited depth resolution ofthe Raman microprobe which averages over 4 mm indepth Figure 2b depicts the resulting DH exchange layerpro les for this and a second crystal grown at 63 8Crespectively Both crystals had the same exposure to am-bient conditions and their main difference is the temper-ature at which they were grown The different exchangelayer pro les indicate that crystals grown at differenttemperatures have differen t proton conductiv ities 13
which leads to a difference in their rate of deuteriumdepletion The parameters controlling this behavior arecurrently the objective of a detailed study the results ofwhich will be reported elsewhere
CONCLUSION
In conclusion we have shown that the shift of the to-tally symmetric PO 4 stretch mode in the Raman spectrumof DKDP crystals scales linearly with degree of deuter-ation This allows us to correlate Raman peak positionsto deuteration levels in these crystals We have presenteda new technique to determine DH diffusion pro les inDKDP frequency conversion crystals based on micro-Ra-man spectroscopy This technique is fast inexpensiveand works under various environmental conditionswhich will allow us to better understand and control deu-terium depletion in DKDP crystals
ACKNOWLEDGMENTS
We would like to thank M Runkel for rst discovering DKDP crack-ing R Floyd for providing DKDP crystals and L Chase and A Burn-ham for their support and helpful discussions This work was performedunder the auspices of the US Department of Energy by the University
of California Lawrence Livermore National Laboratory under contract W-7405-Eng-48
1 J J De Yoreo A K Burnham and P K Whitman Int Mat Rev47 113 (2002)
2 C E Barker R A Sacks B M Van Wonterghern J A Caird JR Murray J H Campbell K Kyle R B Ehrlich and N DNielsen Proc SPIE-Int Soc Opt Eng 2633 501 (1995)
3 T Suratwala paper to be published4 Cleveland Crystals Inc httpwwwclevelandcrystalscom
KDPshtmltable5 E A Popova I T Savatinova and I A Velichko Sov Phys Solid
State 12 1543 (1971)6 I P Kaminow R C C Leite and S P S Porto J Phys Chem
Solids 26 2085 (1965)7 J A Subramony B J Marquardt J W Macklin and B Kahr
Chem Mat 11 1312 (1999)8 H Tanaka M Tokunaga and I Tatsuzaki Solid State Commun
49 153 (1984)9 R J Nelmes G M Meyer and J E Tibballs J Phys C 15 59
(1982)10 M A Yakshin D W Kim Y S Kim Y Y Broslavets O E
Sidoryuk and S Goldstein Laser Physics 7 941 (1997)11 I Takenaga Y Tominaga S Endo and M Kobayashi Solid State
Commun 84 931 (1992)12 C Krenn personal communication13 M Sharon and A K Kalia J Solid State Chem 21 171 (1977)
Rapid Dipicolinic Acid Extractionfrom Bacillus Spores Detectedby Surface-EnhancedRaman Spectroscopy
STUART FARQUHARSON ALAND GIFT PAUL MAKSYMIUK andFRANK E INSCOREReal-Time Analyzers Inc East Hartford Connecticut06108
Index Headings Dipicolinic acid Bacillus spores Anthrax Surface-enhanced Raman spectroscopy
INTRODUCTION
The anxiety caused by the distribution of anthrax en-dospores through the US postal system in October 2001was exacerbated by the long time required for positiveidenti cation of the Bacillus anthracis spores and the un-known extent of their distribution Since that time manymethods capable of rapid eld analysis have been inves-tigated to augment or replace the laboratory method ofgrowing microorganisms in culture media which takesdays to perform12 Prominent among these approachesare polymerase chain reactions (PCR)3 immunoassaysand detection of calcium dipicolinate as a biochemicalsignature PCR employs primers to separate organism-
Received 10 October 2003 accepted 14 November 2003 Author to whom correspondence should be sent
352 Volume 58 Number 3 2004
speci c nucleic acid sequences (eg capsular protein en-coding gene for Bacillus anthracis)4 and polymerases toamplify the segment until it is detectable Recently am-pli cation times have been substantially reduced andcomplete analysis can now be performed in an hour orless Immunoassay methods are also being developed thatuse competitive binding of the bioagent (as an antigen)and its labeled conjugate for a limited number of anti-bodies Although analyses can be performed in under 30minutes a well-de ned anthrax antigen has not yet beenidenti ed5ndash7 and consequently the false-positive rate isunacceptably high8
A number of other methods are being developed witha focus on the detection of calcium dipicolinate (CaDPA)and its derivatives as a B anthracis signature This is sobecause only spore-forming bacteria contain CaDPA andthe most common potentially interfering spores such aspollen and mold spores do not Relatively fast methodshave been developed to chemically extract CaDPA andthen detect it directly by uorescence9 or indirectly byluminescence1011 In the latter case hot dodecylamine(DDA) has been used to extract dipicolinic acid (DPA)and terbium has been utilized to form a highly lumines-cent DPA complex11 Although measurements have beenperformed in as little as ve minutes it was found thatas many as three concentration-dependent complexes canform each with different lifetimes This coupled withthe fact that the Tb31 cation produces the same lumines-cence spectrum makes determinations of low spore con-centrations problematic
It has been long known that Raman spectra of Bacillispores are dominated by bands associated with CaDPA12
and that these spectra may provide a suitable anthrax sig-nature at the genus level13 Since that time considerableimprovements in Raman instrumentation have led to lab-oratory measurements of single Bacilli spores14 and to eld measurements of spores captured from a mail-sort-ing system15 However the single spore measurementsrequired complex instrumentation that is not ruggedwhile the eld measurements required milligrams of sam-ple Furthermore the Raman spectra of both measure-ments contained uorescence contributions that would in-crease uncertainty in quanti cation
In related research we demonstrated that nanogramquantities of DPA could be detected by uorescence-freesurface-enhanced Raman spectroscopy (SERS)16 We alsodemonstrated that microliter volumes of chemicals canbe detected by SERS using metal-doped sol-gel-packedglass capillaries17 Towards the goal of developing a rap-id eld SERS-based anthrax spore detector we havecombined our previous research and we now report thatDPA can be extracted from a 10 mg B cereus spore sam-ple using DDA in 1 minute and can be detected by SERSin an additional 1 minute
EXPERIMENTAL
Dipicolinic acid (26-pyridinedicarboxylic acid DPA)and dodecylamine (DDA) were used as received fromSigma-Aldrich (Milwaukee WI) Lyophilized B cereusspores prepared according to the literature13 were sup-plied by the University of Rhode Island and used as re-ceived Multiple particles approximately 01 mm3 each
were separated and weighed at 5 to 15 mg representing05 to 15 million spores The sample masses were con-sistent with a previous determination of spore density at0081 gmL that indicated a high degree of entrained air
All chemicals used to prepare the silver-doped sol-gelcoated capillaries were also obtained and used as receivedfrom Sigma-Aldrich According to previously publishedprocedures17 two precursor solutions were preparedmixed and then drawn into 1-mm-diameter glass capil-laries The silver amine precursor consisted of a 51 vvratio of 1 N AgNO 3 to 28 NH3OH while the alkoxideprecursor consisted of a 21 vv ratio of methanol to te-tramethyl orthosilicate The alkoxide precursors weremixed with silver amine precursor in an 81 vv ratioApproximately 015 mL was drawn into the capillarycoating a 15-mm length After sol-gel formation the in-corporated silver ions were reduced with dilute sodiumborohydride which was followed by a water wash to re-move residual reducing agent
A 100 mL drop of a 50 mM DDA solution in ethanolpre-heated to 78 8C was added to each of the B cereusparticles to digest the spore coat After 1 minute the re-sultant solution was drawn into a SER-active capillarythat was immediately xed horizontally to an XY posi-tioning stage (Conix Research Spring eld OR) just in-side the focal point of an f 07 aspheric lens The lensfocused the beam into the sample and collected the scat-tered radiation back along the same axis A dichroic lter(Omega Optical Brattleborough VT) was used to re ectthe excitation laser to the lens and pass the Raman scat-tered radiation collected by the lens An f 2 achromatwas used to collimate the laser beam exiting a 200-mm-core-diameter source ber optic while a second f 2 ach-romat was used to focus the scattered radiation into a 365mm ber optic (Spectran Avon CT) A short-pass lterwas placed in the excitation beam path to block the sil-icon Raman scattering generated in the source ber fromre ecting off sampling optics and reaching the detectorA long-pass lter was placed in the collection beam pathto block the sample Rayleigh scattering from reachingthe detector A 785 nm diode laser (Process InstrumentsInc model 785-600 Salt Lake City UT) was used todeliver 100 to 150 mW of power to the sample A Fouriertransform Raman spectrometer (Real-Time Analyzersmodel IRA-785 East Hartford CT) and a silicon photo-avalanche detector (Perkin Elmer model C30902S Stam-ford CT) were used to acquire the SER spectra
RESULTS AND DISCUSSION
As an initial experiment the SER spectrum of 1 gLof DPA in water was measured using the newly devel-oped silver-doped sol-gel-coated capillaries (Fig 1A) Atthis concentration a high signal-to-noise ratio (SN) isobtained in 1 min In fact a reasonable spectrum is ob-tained in the same time frame for 1 mgL (Fig 1B) TheSER spectra are reasonably similar to the normal Raman(NR) spectrum obtained for a saturated solution of DPAin 1 N KOH (Fig 1C) and the following band shifts areobserved (NR to SER) 647 to 657 cm21 817 to 815cm21 998 to 1008 cm21 1384 to 1382 cm21 1434 to1428 cm21 and 1569 to 1567 cm21 Many of these bandshave been previously assigned1213 such as 998 cm21 to
APPLIED SPECTROSCOPY 353
FIG 1 SERS of DPA in water using silver-doped sol-gel-coated glasscapillary for (A) 1 gL and (B) 1 mgL (C ) NR of saturated DPA in 1N KOH in a glass capillary Spectral conditions (A) and (B) 150 mWof 785 nm 1-min acquisition time (C) 450 mW of 785 nm 5-minacquisition time both 8 cm 21 resolution
FIG 2 SERS of DPA extracted from 10 mg B cereus particle using100 mL of 50 mM hot DDA acquired in (A) 1 minute and (B) 2 seconds(C ) Attempted SERS of 50 mM hot DDA in ethanol using silver-dopedsol-gel-coated glass capillary acquired in 1 min Spectral conditions150 mW of 785 nm 8 cm21 resolution
the symmetric ring stretch 1384 cm21 to the OndashCndashOsymmetric stretch 1428 cm21 to the symmetric ring CndashH bend and 1569 cm21 to the asymmetric OndashCndashOstretch
The rst B cereus samples consisted of 2 mg of sporesin 2 mL of 5 mM hot DDA The samples were main-tained at 78 8C for 40 min and while hot approximately10 mL was drawn into a SER-active capillary Since spec-tra of DPA were obtained for these initial samples small-er spore masses higher DDA concentrations and shorterheating periods were examined In due course it wasfound that 10 mg of spores could be digested by 100 mLof 50 mM hot DDA in one minute and detected (Fig2A) In fact the signal was suf ciently intense that it canbe observed in as little as two seconds (Fig 2B) Theamount of DPA that was extracted was estimated to bebetween 5 and 10 mgL by comparing the signal intensityof the 1008 cm21 band to that measured for DPA in waterThis is consistent with previous research that found thatthe majority of the DPA is extracted from spores usingDDA11 and that B cereus spores contain approximately10 DPA by weight18 The SN of 127 for the 1008 cm21
band in the 1-minute SER spectrum suggests a limit ofdetection of approximately 250 ng of B cereus sporesbased on a SN of 3 Finally it should be noted that DDAdid not produce a detectable SER spectrum as shown inFig 2C
CONCLUSION
We have demonstrated that by combining rapid extrac-tion of dipicolinic acid from Bacillus cereus spores withchemical identi cation by surface-enhanced Raman spec-troscopy as little as 10 mg of spores can be detected Infact the entire measurement from the time of adding hotdodecylamine to the spores to the time when the dipi-colinic acid SER spectrum is acquired and analyzedcould be performed in less than two minutes The abilityof this method to distinguish between spore-forming bac-teria such as Bacillus anthracis and non-DPA containingpowders could help prevent costly shutdowns associated
with the appearance of suspicious material or intentionalmailing of common substances as an anthrax hoax Thismethod could also prove useful in detecting the locationof anthrax endospores in mail distribution facilities if an-other veri ed attack should occur
Research continues to fully characterize the surface-enhanced Raman spectroscopy signal intensities as afunction of sample concentration and to explore otherextractants that do not require the use of elevated tem-perature
ACKNOWLEDGMENTS
The authors are grateful for the support of the National Science Foun-dation (DMI-0296116 and DMI-0215819) and the US Army(DAAD13-02-C-0015 Joint Service Agent Water Monitor program)The authors are indebted to Chetan Shende for preparing the sol-gelcapillaries The authors also thank James Gillespie Nicholas Fell andAugustus Fountain for providing important background informationMark Farquharson for laboratory support and Professor Jay Sperry ofthe University of Rhode Island for supplying B cereus spores
1 V A Pasechnik C C Shone and P Hambleton Bioseparations 3267 (1993)
2 P J Jackson M E Hugh-Jones D M Adair G Green K K HillC R Kuske L M Grinberg F A Abramova and P Keim ProcNatl Acad Sci USA 95 1224 (1998)
3 B R Glick and J J Pasternak Molecular Biology Principles andApplications of Recombinant DNA (ASM Press Washington DC1994)
4 C A Bell J R Uhl T L Had eld J C David R F Meyer TF Smith and F R Cockerill III J Clin Microbiol 40 2897(2002)
5 D L Gatto-Menking H Yu J G Bruno M T Goode M Millerand A W Zulich Biosens Bioelectron 10 501 (1995)
6 J J Quinlan and P M Foegeding J Rapid Methods AutomationMicrobiol 6 1 (1998)
7 A A Hindle and E A H Hall Analyst (Cambridge UK) 1241599 (1999)
8 M S Ascher US Department of Health amp Human Services(httpwwwhhsgovophppresentationsAscherdoc)
9 R Nudelman B V Bronk and S Efrima Appl Spectrosc 54 445(2000)
10 D L Rosen C Sharpless and L B McBrown Anal Chem 691082 (1997)
354 Volume 58 Number 3 2004
11 P M Pellegrino N F Fell Jr and J B Gillespie Anal ChimActa 455 167 (2002)
12 W H Woodruff T G Spiro and C Gilvarg Biochem BiophysRes Commun 58 197 (1974)
13 E Ghiamati R S Manoharan W H Nelson and J F SperryAppl Spectrosc 46 357 (1992)
14 A P Esposito C E Talley T Huser C W Hollars C M Schal-dach and S M Lane Appl Spectrosc 57 868 (2003)
15 S Farquharson L Grigely V Khitrov W W Smith J F Sperryand G Fenerty J Raman Spectrosc paper accep ted (2003)
16 S Farquharson W W Smith S Elliott and J F Sperry SPIE-IntSoc Opt Eng 3855 110 (1999)
17 S Farquharson and P Maksymiuk Appl Spectrosc 57 479(2003)
18 F W Janssen A J Lund and L E Anderson Science (Washing-ton DC) 127 26 (1958)
SPIE -2003-5269 117
pH dependence of methyl phosphonic acid dipicolinic acid and cyanide by surface-enhanced Raman spectroscopy
Stuart Farquharson Alan Gift Paul Maksymiuk Frank Inscore and Wayne Smith
Real-Time Analyzers 87 Church Street East Hartford CT 06108
ABSTRACT US and Coalition forces fighting terrorism in Afghanistan and Iraq must consider a wide range of attack scenarios in addition to car bombings Among these is the intentional poisoning of water supplies to obstruct military operations To counter such attacks the military is developing portable analyzers that can identify and quantify potential chemical agents in water supplies at microgram per liter concentrations within 10 minutes To aid this effort we have been investigating the value of a surface-enhanced Raman spectroscopy based portable analyzer In particular we have been developing silver-doped sol-gels to generate SER spectra of chemical agents and their hydrolysis products Here we present SER spectra of methyl phosphonic acid and cyanide as a function of pH an important factor affecting quantitation measurements which to our knowledge has not been examined In addition dipicolinic acid a chemical signature associated with anthrax-causing spores is also presented Keywords Chemical warfare agents agent detection agent hydrolysis SERS Raman spectroscopy homeland security
1 INTRODUCTION In the past decade the Unites States and its allies have been challenged by a different kind of warfare exemplified by the terrorist attacks of September 11 2001 Suicide bombings and the use of chemical agents are the norm and military personnel must consider a wide range of attack scenarios Among these is the intentional poisoning of water supplies to obstruct military operations in Afghanistan and Iraq Currently colorimetric paper is used to detect agents on-site while gas chromatography combined with mass spectrometry (GCMS) is used in mobile support laboratories However both methods have severe drawbacks The paper changes color in response to contact with many chemicals besides CWAs causing a high incidence of false positives1 while GCMS although very chemically specific requires hours to perform and constant re-calibration234 Military operations would be greatly aided by a portable analyzer that can identify and quantify potential chemical agents at concentrations that impact safety This includes the analysis of drinking water supplies distribution and storage systems To meet this goal the Department of Defense has been investigating numerous approaches under the auspices of the Joint Service Agent Water Monitor (JSAWM) program5 The JSAWM must be capable of identifying the chemical agents with no false-positives and quantifying the agents at microgram per liter concentrations within 10 minutes678 Recently we and others have been investigating the ability of surface-enhanced Raman spectroscopy (SERS) to measure chemical agents9-12 bioagents13-17 and their hydrolysis products in water SERS employs the interaction of surface plasmon modes of metal particles with target analytes to increase scattering efficiency by as much as 1 million times18 In 1987 the potential of this approach to measure CWAs was demonstrated by measuring a series of pesticides19 Several of these organophosphonates have chemical structures similar to CWAs in particular P=O functional groups In our studies we have been employing silver-doped sol-gels to promote the SER effect The porous silica network of the sol-gel matrix offers a unique environment for immobilizing and stabilizing SER-active metal particles20-23 The sol-gels have been coated on the internal walls of standard glass vials to allow rapid SER analysis of small sample quantities without preparation We have measured over 100 chemicals with enhancements of 104 to 106 demonstrated reversible measurements in a flowing system reproducible measurements from vial-to-vial and measurements in multiple solvents including water20-23 Previously we used these vials to perform preliminary measurements of cyanide (CN) methylphosphonic acid (MPA) and dipicolinic acid (DPA) MPA is a hydrolysis product of the nerve agents (eg sarin Reaction 1) and may be a valuable indicator of nerve agent usage particularly since the alkyl methylphosphonic acids are relatively more stable than their corresponding parent complexes24 DPA is
SPIE -2003-5269 118
a chemical signature of spore forming bacteria such as Bacillus anthracis And in light of the inability to rapidly detect the anthrax spores distributed through the US mail in October 2001 a number of methods are being developed to extract and analyze this signature Reaction 1 Stepwise hydrolysis of Sarin to form hydrofluoric acid (HF) isopropyl methylphosphonic acid (IMPA) then methyl phosphonic acid (MPA) and 2-propanol In our previous SERS investigations MPA and DPA were measured at 50 and 100 mgL respectively In both cases limits of detection (LOD) were estimated at 100 microgL providing encouragement in that SERS may satisfy the needs of the JSAWM Since it has been shown that pH can substantially influence the intensity of SER bands25 which would clearly influence quantitative analysis we undertook the present study to determine the severity of these effects for cyanide methyl phosphonic acid and dipicolinic acid Furthermore we previously observed a band at 1050 cm-1 for MPA6 possibly due to an anion formed at basic pH Here we investigate the source of this spectral anomaly
2 EXPERIMENTAL All chemicals including potassium cyanide methyl phosphonic acid dipicolinic acid and those used to prepare the silver-doped and gold-doped sol-gel coated vials were obtained and used as received from Sigma-Aldrich All samples were prepared in HPLC grade water (Fischer Scientific Fair Lawn NJ) for SERS measurements The pH of these samples was adjusted using dilute nitric acid or potassium hydroxide and verified using a pH electrode (Corning Inc Corning NY) that had been calibrated with pH 400 700 and 1000 buffered standards from Fischer Scientific Once prepared the samples were transferred into the silver-doped sol-gel vials (Real-Time Analyzers Inc East Hartford CT) The vials were coated in a manner similar to that previously reported by adding ammonium hydroxide to a solution of silver nitrate tetramethyl orthosilicate (TMOS) and methanol20 Gold-vials were coated by adding nitric acid to a solution of gold tetrachloride TMOS and methanol The two precursor solutions were prepared mixed and transferred to 2-mL glass vials dried and heated After sol-gel formation the incorporated metal ions were reduced with dilute sodium borohydride (1mgmL) which was followed by a water wash to remove residual reducing agent After the resultant analyte solution was introduced the SER-active vial was immediately fixed horizontally to an XY positioning stage (Conix Research Springfield OR) just inside the focal point of an f07 aspheric lens The lens focused the beam into the sample and collected the scattered radiation back along the same axis A dichroic filter (Omega Optical Brattleborough VT) was used to reflect the excitation laser to the lens and pass the Raman scattered radiation collected by the lens An f2 achromat was used to collimate the laser beam exiting a 200 microm core diameter source fiber optic while a second f2 achromat was used to focus the scattered radiation into a 365 microm fiber optic (Spectran Avon CT) A short pass filter was placed in the excitation beam path to block the silicon Raman scattering generated in the source fiber from reflecting off sampling optics and reaching the detector A long pass filter was placed in the collection beam path to block the sample Rayleigh scattering from reaching the detector A 785 nm diode laser (Process Instruments Inc model 785-600 Salt Lake City UT) was used to deliver 100 to 150 mW of power to the sample A Fourier transform Raman spectrometer (Real-Time Analyzers model IRA-785 East Hartford CT) and a silicon photo-avalanche detector (Perkin Elmer model C30902S Stamford CT) were used to acquire the SER spectra
3 RESULTS AND DISCUSSION In a previous study of MPA6 aimed at developing a concentration calibration curve and determining limits of detection (LOD) we observed an anomaly at 1050 cm-1 Since it was found that the band intensity changed as a function of concentration the band must be associated with a sample parameter Two possible parameters photon flux and pH are examined here The first parameter was investigated by irradiating a 1mgmL MPA sample in a SER-active vial with laser powers of 200 mW and above and monitoring spectral changes It was immediately found that the 1050 cm-1 band
2O+ H HF + +OH
OH3H C
OP
OF
CH
CH3
3
3H C
OCP
OOH
CH
CH3
3
3H C
O
CPHO
CH
CH
3
3
C
Sarin IMPA MPA 2-propanol
SPIE -2003-5269 119
appeared and grew as a function of time and that the higher the laser power the faster the growth Figure 1 shows the growth of the 1050 cm-1 band over the course of 30 minutes when using 150 mW of 785 nm excitation while Figure 2A shows that the growth can be fit with a first order exponential equation namely I1050 = 03+05e-013t Furthermore the 760 cm-1 band could be fit with a first order decay equation with an identical exponential rate constant ie I760 = 08-
08e-013t The rates represent classical first order kinetics and their correspondence allows one to conclude that MPA is being transformed one-for-one into a photo-generated product At this time the photoproduct has not been positively identified but phosphonic (phosphorous) acid and phosphonate are likely candidates since the symmetric P(OH)3 stretch occurs at ~1050cm-1 Our studies show that a reduction in laser power to 100 mW at the sample for MPA essentially eliminates this degradation process This laser power or lower was used for further measurements
Methyl phosphonic acid is a diprotic acid that stepwise dissociates into two anions MPA- and MPA= according to the following reactions26
MPA MPA- + H+ pKa1 = 212 Reaction 2
MPA- MPA= + H+ pKa2 = 729 Reaction 3 The relative concentrations of MPA MPA- and MPA= can be determined at any pH by expressing [MPA] and [MPA=] in terms of [MPA-] using Reactions 2 and 3 and summing all three to equal the total starting concentration here 2 mgmL (0021M MW = 9602) viz
[MPA] + [MPA-] + [MPA=] = 0021M Equation 1 substituting from Reactions 1 and 2
([H+][MPA-])K1a + [MPA-] + (K2a[MPA-])[H+] = 0021M Equation 2 rearranging [MPA-] = 0021M(1+[H+]K1a + K2a[H+]) Equation 3 The relative concentrations of MPA MPA- and MPA= as a function of pH are shown in Figure 3 It is worth noting that near neutral pH both MPA- and MPA= will be present To confirm that the SER signal followed this pH dependence a starting solution consisting of 20 mg of MPA in 10 mL HPLC grade water was prepared and brought to pH of 20 using dilute nitric acid From this solution 2 mL were added to a SER-active vial and the SER spectrum recorded At this pH a peak at 760 cm-1 was barely discernable The 2 mL solution was returned to the starting solution and the pH was re-measured to correct for any changes that the silver-doped sol-gel vials might cause In most cases the change was less than 02 pH units and the pH is reported as the before and after average Next the pH of the
Figure 2 A) Exponential growth of 1050 cm-1 band and B) exponential decay of 760 cm-1 band for spectral series in Figure 1
Figure 1 Growth of 1050 cm-1 band as a function of time due to exposure to 150 mW of 785 nm Spectra are 5 sec each collected every 100-sec from 0 to 30-min
B
570 770 970 1170 1370Raman Shift (cm-1)
Arbit
rary
Unit
s
MPA Photodegradation
600 800 1000 1200 1400 Wavenumber (cm-1)
A
0 10 20 30 0 10 20 30 time (min) time (min)
Ram
an In
tens
ity (r
elat
ive)
30
min
0
SPIE -2003-5269 120
starting solution was adjusted to 325 using dilute KOH Again 2 mL were added to a vial and the SER spectrum recorded At this pH a reasonably strong 760 cm-1 band was observed This process was repeated as spectra were recorded at pHs of 70 74 75 79 85 and 100 A total of 1 mL of KOH was added diluting the total concentration by 10 Next the pH of the starting solution was made acidic by adding dilute nitric acid dropwise This time spectra were recorded at pHs of 72 69 64 and 37 Figure 4 shows the SER spectra for representative pHs (spectra were left out to simplify the figure) while Figure 3 shows the 760 cm-1 peak intensities as a function of pH (The band intensities were adjusted to compensate for dilution effects caused by the addition of HNO3 and KOH then normalized to 0021 M for the most intense band observed at pH 37) It is clear from Figure 3 that the 760 cm-1 band follows the MPA- concentration as a function of pH and must be assigned to this anion No bands were observed that corresponded to MPA or MPA= The lack of an MPA SER spectrum may be due to the absence of an attraction between the neutral analyte and the electropositive silver surface The same reasoning suggests that a strong SER spectrum should be observed for MPA= but it is not and a satisfactory explanation has not been found
0000
0005
0010
0015
0020
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14pH
Con
cent
ratio
n [M
]
pK1 = 212 pK2 = 729
MPA- MPA=
MPA
Figure 4 SER spectra of 002M MPA as a function of pH Conditions 100 mW of 785 nm 36 scans (1 min) 8 cm-1 recorded 2 min after sample introduction pH 19 and 20 not apparent on this scale pH 69 and 74 near identical to 70 and 75 and not shown for clarity
00
02
04
06
08
10
12
14
16
18
0 5 10 15 20 25 30Measurement Number
Ram
an In
tens
ity (7
60 c
m-1
)
Figure 3 Concentration dependence of MPA MPA- and MPA= as a function of pH for a 002M sample Intensity of 760 cm-1
band from Figure 6 as a function of pH ( for increasing basic adjustment for increasing acidic adjustment error was measured at ~10 for pH 64)
Figure 5 SER spectra of 002M MPA at pH 64 measured around a vial at three heights (9 points per height) Conditions as in Figure 4 but 10-sec scans
Since these measurements involved the removal and replacement of the SER-active vial in the sample holder to remove and add sample variation in the intensity as a function of vial position was minimized by illuminating the exact same height along the vial wall But this does not account for variability of the SERS response of the sol-gel coating around the vial To analyze this effect a vial containing MPA at pH 64 was rotated at ~ 40o
intervals at the original height and 18rdquo above and below this value Figure 5 shows the intensity of the 760 cm-1
band for the 27 positions It was found that the average value was 137plusmn014 an RSD of 10 overall and 5 for each height An error bar is included in Figure 4 for the pH 64 measurement
SPIE -2003-5269 121
With the above analyses in mind a preliminary investigation of the SER spectral response for MPA (as MPA-) as a function of concentration was performed A single vial was used for these measurements beginning with 1 mgL followed by measurements of 10 100 and 1000 mgL In all cases the pH was ~7 and 3 positions around the vial were measured per concentration Since the 760 cm-1 band was not observed for concentrations of 1 or 10 mgL using 100 mW of 785 nm the laser power at the sample was raised to 200 mW beginning with the 10 mgL concentration Photo-degradation was largely avoided (and not observed) by exposing the sample for only 33 seconds per spectral acquisition Representative spectra for 10 100 and 1000 mgL are shown in Figure 6 while a plot of the 760 cm-1 band intensity as a function of concentration is shown in Figure 7 These values were also used to estimate limits of detection based on the signal-to-noise ratio (SN) of the 760 cm-1 band using 100 mW of 785 nm laser excitation and a 10-min acquisition time As summarized in the Figure 7 inset the lower the measured concentration the lower the predicted LOD Even if as estimated 210 microgL could be measured using the silver-doped sol-gel vials an improvement of a factor of 70 is still required to meet JSAWM goals of measuring 3 microgL in 10-minutes Similar to MPA DPA is a diprotic acid (pKa1 = 216 and pKa2 = 692) and variations in pH will effect the relative concentrations of DPA DPA- and DPA= and possibly the SER spectra and band intensities This could prove significant if an acid or base is used to denature anthrax spores with the goal of extracting and analyzing DPA The MPA pH study described above was mimicked for DPA except that the starting solution consisted of 20 mg of DPA in 20 mL HPLC grade water (60x10-3M MW = 1671) The initial solution had a pH of 245 which was made basic by dilute KOH to pHs of 355 433 487 559 1069 and 1166 SER spectra were recorded at each pH using 100 mW of 785 nm and a 44-sec acquisition time Next one drop of concentrated nitric acid was used to remake the solution acidic at a pH of 200 Again sequential pH measurements were performed at 383 510 735 and 822 The solution pH was made acidic a third time but to pH 219 171 then 135 Throughout this process no more than 20 drops of acid or base were added and therefore the concentration was diluted by no more than 10 Most of the spectral bands showed a minor decrease in intensity as a function of increasing pH values However the bands at 525 and 795 cm-1 showed the most dramatic changes which occurred at acid pH Figure 8 shows the SER spectra of DPA for the spectral region and pH range of interest The identity of the DPA species was determined by plotting the normalized peak intensities with the lowest value set to 0 and the highest to 0006 M as a function of pH and overlaying these values on a plot of the relative concentrations for DPA DPA- and DPA= as previously done for MPA (Figure 9) As can be seen the 525 cm-1 band clearly corresponds to DPA The correspondence of the 795 cm-1 band to this species is less clear as the band retains intensity until pH of 55 This can be attributed to contribution to the overlapping band at 810 cm-1 which does not change as a function of pH The fact that most bands are observed at all pHs suggest that the primary interaction with silver is through the ring nitrogen This is supported by the fact that the most intense band occurs at 1008 cm-1 attributed to a symmetric ring breathing mode and that this interaction has been characterized for pyridine in numerous papers27
0
1
2
3
4
5
6
7
0 200 400 600 800 1000 1200
MPA Concentration (mgL)
760
Ban
d In
tens
ity
Figure 6 SER spectra of MPA in water at A) 1000 B) 100 and C) 10 mgL Conditions pH of 7 silver-doped sol-gel coated vial 200 mW of 785 nm 33-sec 8 cm-1 resolution
Figure 7 Plot of SER intensity of 760 cm-1 band of MPA as a function of concentration using 200 mW of 785 nm Inset table includes average intensity LOD standard deviation and percent deviation for each concentration but for 100 mW and 10 min
conc (mgL) LOD ave stddev dev10 021 002 1021100 072 010 14301000 312 040 1280
A
B C
SPIE -2003-5269 122
The ability to detect dipicolinic acid (DPA) by SERS is demonstrated in Figure 10 Even at 1 mgL the primary bands are visible The intensity of the dominant ring breathing mode observed at 1008 cm-1 for DPA in water (pH = 25 to 55) is plotted as a function of concentration in Figure 11 These values were also used to estimate limits of detection based on the SN of the 1008 cm-1 band using 100 mW of 785 nm laser excitation and a 10-min acquisition time Again the lower the measured concentration the lower the predicted LOD (see Figure 11 inset) and detection of 160 microgL is possible Gastrointestinal anthrax requires significant more spores than inhalation anthrax28 and a limit of detection might be placed at 1 million spores in 1 liter of water or 10 microgL Since each spore contains ~10 CaDPA by weight29 a goal for DPA might be 1 microgL indicating that the present measurements must be improved by nearly two orders of magnitude Due to the increased hazards of handling HCN gas KCN salt was used for these experiments Nevertheless all sample preparations were performed in a chemical hood KCN completely dissolves in water but its conjugate acid HCN is formed and has a Ka of 615x10-1030 viz
HCN CN- + H+ pKa = 921 Reaction 4
Figure 8 SER spectra of 1 mgmL DPA as a function of pH Conditions 100 mW 785 nm 100 scans (44-sec)
Figure 9 Concentration dependence of DPA DPA- and DPA= as a function of pH for a 0006M sample Intensity of 525 () and 795 (diams) cm-1 bands from Fig 8 as a function of pH
pH
135 171 219 383
0
0001
0002
0003
0004
0005
0006
0007
0 2 4 6 8 10 12 14pH
Con
cent
ratio
n [M
]
DPADPA-DPA=795525
DPA DPA=DPA-
pK2 = 692pK1 = 216
795 525
0
05
1
15
2
25
0 200 400 600 800 1000 1200
DPA Concentration (mgL)
1008
Ban
d In
tens
ity
conc (mgL) lod-10min-100mw1 017
10 016100 103
1000 355
Figure 10 SER spectra of DPA in water at A) 1000 B) 100 C) 10 and D) 1 mgL Conditions pH of 25-55 silver-doped sol-gel coated vial 175 mW of 785 nm 1-min 8 cm-1 D) has been multiplied by x10 to make bands visible
A
B C
D
Figure 11 Plot of SER intensity of 1008 cm-1 band of DPA as a function of concentration using 175 mW of 785 nm Inset table includes LOD in mgL for each concentration but for 100 mW and 10 min
SPIE -2003-5269 123
Consequently the cyanide concentration must be determined for each initial KCN concentration Specifically the samples prepared with concentrations of 01 1 10 100 and 1000 mgL of KCN produced CN- concentrations of 63x10-3 033 69 89 and 964 mgL at pHs of 816 90 967 102 and 107 respectively The pH dependence for the HCN and CN- concentrations are shown in Figure 12 Thus as the amount of KCN added to the solution decreases so does the pH of the solution (becomes less basic) and according to Reaction 4 the relative amount of CN- to HCN also decreases For example in the preparation of a 01 mgL solution of KCN the pH is shifted from 7 for pure water to only 816 and only 63 of the starting material becomes CN- or 63x10-3 mgL In comparison for a solution of 1000 mgL the pH is shifted from 7 to 107 and 96 of the starting material becomes CN- This is significant since the cyanide ion is better able to adsorb onto the silver particles and become SERS active SER spectra of 10 100 and 1000 mgL of KCN samples are shown in Figure 13 and a corresponding plot of the 2100 cm-1 band intensity as a function of concentration is shown in Figure 14
000102030405060708091011
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14pH
Con
cent
ratio
n [m
gm
L]
pKa = 921
CN -HCN
The SER spectra of cyanide are dominated by the single CN stretch at 2100 cm-1 which occurs in normal Raman spectra of solutions at 2080 cm-1 However a low frequency mode occurs at 135 cm-1 due to a Ag-CN stretch (not shown) indicative of a strong surface interaction It is also observed that as the concentration decreases the CN stretch shifts to 2140 cm-1 This shift has been attribute to the formation of a tetrahedral Ag(CN)3
2- surface structure31 as well
Figure 14 Concentration dependence of KCN SERS measured under conditions in Fig 11 Concentrations are 1 01 and 001 mgml Intensities are measured for the CN stretch at 2100 cm-1 Inset table includes LOD in mgL for each concentration in Figs 13 and 15 but for 100 mW and 10 min
Figure 13 SER spectra of KCN in water at A) 1000 B) 100 and C) 10 mgL Conditions pHs of 107 102 and 97 silver-doped sol-gel coated vial 100 mW of 785 nm 1-min 8 cm-1 C) has been multiplied by x10 to make band visible
Figure 12 Concentration dependence of HCN and CN- as a function of pH for a 1 mgmL sample Calculated intensity of 2100 () cm-1 band for a 1 mgml sample at pHs of 816 90 967 102 and 107
A
B
C
0
20
40
60
80
100
120
140
0 200 400 600 800 1000 1200CN Concentration (mgL)
2100
Ban
d In
tens
ity
Figure 15 SER spectra of KCN in water at A) 10 B) 1 and C) 01 mgL Conditions pHs of 97 12 and 12 gold-doped sol-gel coated vial 100 mW of 785 nm 1-min 8 cm-1 C) has been multiplied by x10 to make band visible
A
B
C
conc (mgL) lod-10min-100mw Condition01 001 Au-pH 121 007 Au-pH 12
69 003 Au-pH 9769 016 Ag-pH 9789 022 Ag-pH 102946 113 Ag-pH 107
SPIE -2003-5269 124
as to CN adsorbed to two different surface sites32 Alternatively the 2140 cm-1 band could be attributed to HCN since this species dominates at lower concentrations However it is unlikely that this species would be attracted to the electropositive silver surface Further both peaks should be present at pHs between 85 and 105 but this is not observed It has also been suggested that at concentrations near and above monolayer coverage the CN- species is forced to adsorb end-on due to crowding and at lower concentrations the molecule can reorient to lie flat33 This suggests that the 2100 and 2140 cm-1 bands correspond to the end-on and flat orientations respectively As Figures 13 and 14 show the intensity of the CN stretch for the 89 mgL sample is nearly as intense as the 964 mgmL band This suggests that the Raman signal for the flat orientation is more enhanced However more extensive measurements are required to verify this point Since resent research has suggested that cyanide may be more effectively detected on gold measurements of KCN solutions were also performed using gold-doped sol-gel vials Preliminary measurements are shown in Figure 15 for samples prepared from 01 1 and 10 mgL KCN Since the pHs are 816 90 967 the resultant CN- concentrations are 63x10-3 033 and 69 mgL Initially only the highest concentration was observed and the signal intensity was significantly better than the equivalent concentration measured using silver In an effort to shift Reaction 4 to the left transforming HCN to CN- (Le Chatelierrsquos principle) KOH was added to the lower concentration samples producing solutions with pH 12 This effectively forces all of the cyanide in solution to be CN- or 01 and 10 mgL respectively More importantly the CN stretch is now observed in the SER spectra The band appears at 2125 cm-1 as has been previously reported for gold12 As calculated for MPA and DPA LODs can be estimated from this data For the three concentrations of cyanide on silver the LODs are 016 to 11 mgL for 100 mW of 785 nm laser excitation and a 10-min acquisition time For gold pH adjusted the LODs are10 to 70 microgL an improvement of more than 10 times silver Nevertheless either substrate is sufficient to meet the JSAWM goals of measuring 3 mgL in 10-minutes as the requirements form cyanide are much less stringent than the nerve agents
4 CONCLUSIONS Here we examined the surface-enhanced Raman spectral response for methyl phosphonic acid dipicolinic acid and cyanide as a function of pH It was determined that the most prominent peak at 760 cm-1 reaches a maximum intensity between pH 3 and 7 and corresponds to the MPA- species Neither the MPA nor MPA= species appear to generate a SER spectrum and consequently no spectra were observed below pH 2 or above pH 8 In this study we also found that higher laser powers could cause photodegradation of MPA signified by the exponential growth of a band at 1050 cm-1 which is tentatively assigned to phosphorous acid Unlike MPA DPA was observed at all pHs This is attributed to the dominant interaction of the pyridine functional group with silver Minor spectral changes were observed at acid pHs and were assigned to neutral DPA Like MPA SER spectra of cyanide were pH dependent No spectra were observed for the HCN species while CN- was best observed at pHs more basic than 8 Preliminary concentration studies for the three analytes allowed estimating limits of detection for MPA DPA and CN using 100 mW of 785 nm and a 10-min acquisition time of 210 165 and 70 microgL respectively Although the latter value suggests that cyanide can be measured at sufficiently low concentrations to meet JSAWM goal improvements by 100 to 200 times are required for MPA and DPA It is clear from this study that pH of the sample is important and must be taken into account when developing concentration calibration curves as well as fieldable analyzers Future work will include tailoring the sol-gel with specific functional groups to dictate sample pH and thereby optimize sensitivity
ACKNOWLEDGMENTS The authors are grateful for the support of the US Army (DAAD13-02-C-0015 Joint Service Agent Water Monitor program) The authors would also like to thank Dr Steve Christensen of the US Army SBCCOM for helpful discussions and Mr Chetan Shende of Real-Time Analyzers for assistance in development of the gold-doped sol-gels
REFERENCES 1 Erickson B Analytical Chemistry News amp Features June 1 397A (1998)
SPIE -2003-5269 125
2 Johnston RL Hoefler CM Fargo JC and Moberley B AT-ONSITE 5-8 (1994) 3 Black RM RJ Clarke RW Read and MT Reid J Chromatography 662 301-321 (1994) 4 USA Army Center for Health Promotion and Preventive Medicine (USACHPPM) Technical Guideline 230A 5 Jensen JL NC Wong W R Loerop SPIE 4775-03 (2001) 6 Hoenig SL Handbook of Chemical Warfare and Terrorism Greenwood Press Wesport CT (2002) 7 Munro NB SS Talmage GD Griffin LC Waters AP Watson JF King and V Hauschild Env Health
Persp 107 933-974 (1999) 8 Holstege CP Kirk M Sidell FR Crit Care Clin13 923-42 (1997) 9 Farquharson S P Maksymiuk K Ong and S Christesen SPIE 4577 166-173 (2001) 10 Lee Y and S Farquharson SPIE 4378 21-26 (2001) 11 Spencer KM J Sylvia S Clauson and J Janni SPIE 4577 158-165 (2001) 12 Tessier P S Christesen K Ong E Clemente A Lenhoff E Kaler and O Velev Applied Spectroscopy 56
1524-1530 (2002) 13 Farquharson S WW Smith S Elliott and JF Sperry SPIE 3533 207-214 (1998) 14 Farquharson S WW Smith S Elliott and JF Sperry SPIE 3855110-116 (1999) 15 Farquharson S WW Smith YH Lee S Elliott and JF Sperry SPIE 4575 62-72 (2002) 16 Guzelian AA J Sylvia J Janni S Clauson and KM Spencer SPIE 4577 182-192 (2001) 17 Shende C F Inscore A Gift P Maksymiuk and S Farquharson in press 18 Weaver MJ S Farquharson and MA Tadayyoni J Chem Phys 82 4867-4874 (1985) 19 Alak AM and T Vo-Dinh Analytical Chemistry 59 2149-2153 (1987) 20 Lee Y and S Farquharson SPIE 4206 140-146 (2000) 21 Farquharson S and Y Lee SPIE 4200-16 (2000) 22 Lee Y S Farquharson and P M Rainey SPIE 3857 76-84 (1999) 23 Lee Y S Farquharson H Kwong and M Shahriari SPIE 3537 252-260 (1998) 24 Wang J M Pumera G Collins and A Mulchandani Analytical Chemistry 74 6121-6125 (2002) 25 Dou X YM Jung Z-Q Cao and Y Ozaki Applied Spectroscopy 53 1440-1447 (1999) 26 Data supplied by S Christesen and K Ewing 27 Kerker M and B Thompson Eds SPIE MS 10 (1990) 28 Inglesby TV DA Henderson JG Bartlett JAMA 287 2236 (2002) 29 FW Janssen AJ Lund and LE Anderson Science 127 26 (1958) 30 Lide DR Ed Handbook of Chemistry and Physics CRC Press 77th Ed (1996-7) 31 Billmann J G Kovacs and A Otto Surf Sci 92 153 (1980) 32 Murray CA and S Bodoff Phys Rev B 32 671 (1985) 33 Kellogg D and J Pemberton J Phys Chem 91 1120 (1987)
SPIE-2003-5269 16
Chemical agent detection by surface-enhanced Raman spectroscopy
Stuart Farquharson Alan Gift Paul Maksymiuk Frank Inscore and Wayne Smith
Real-Time Analyzers 87 Church Street East Hartford CT 06108
Kevin Morrisey and Steven D Christesen US Army SBCCOM Aberdeen Proving Ground MD 21010
ABSTRACT
In the past decade the Unites States and its allies have been challenged by a different kind of warfare exemplified by the terrorist attacks of September 11 2001 Although suicide bombings are the most often used form of terror military personnel must consider a wide range of attack scenarios Among these is the intentional poisoning of water supplies to obstruct military operations in Afghanistan and Iraq To counter such attacks the military is developing portable analyzers that can identify and quantify potential chemical agents in water supplies at microgram per liter concentrations within 10 minutes To aid this effort we have been investigating the value of a surface-enhanced Raman spectroscopy based portable analyzer In particular we have been developing silver-doped sol-gels to generate SER spectra of chemical agents and their hydrolysis products Here we present SER spectra of several chemical agents measured in a generic tap water Repeat measurements were performed to establish statistical error associated with SERS obtained using the sol-gel coated vials Keywords Chemical agents chemical agent detection SERS Raman spectroscopy
1 INTRODUCTION
In the autumn of 2001 terrorism within US borders became a sobering reality While extensive efforts are being implemented to secure the homeland US and Coalition forces in Afghanistan and Iraq are constantly faced with terrorist attacks In addition to car-bombings the military has identified several non-traditional attack scenarios including poisoning of water supplies by chemical warfare agents (CWAs) To counter this threat the Department of Defense has been investigating numerous approaches to detect such attacks under the auspices of the Joint Service Agent Water Monitor (JSAWM) program1 The JSAWM must be capable of identifying the chemical agents with no false-positives and quantifying the agents at microgram per liter concentrations within 10 minutes2 This includes the analysis of drinking water supplies distribution and storage systems Currently colorimetric paper is used to detect agents on-site while gas chromatography combined with mass spectrometry (GCMS) is used in mobile support laboratories However both methods have severe drawbacks The paper changes color in response to contact with many chemicals besides CWAs causing a high incidence of false positives3 while GCMS although very chemically specific requires up to an hour to perform and regular re-calibration456 Vibrational spectroscopy has also been investigated as a potential method of rapidly detecting CWAs7-
11 as early as 1985 in the case of infrared spectroscopy8 and 1988 in the case of Raman spectroscopy11 Again however these techniques also have limitations when it comes to measuring trace poisons in water Infrared spectra would be dominated by the very strong absorption of water which would obscure absorptions by most other chemicals present Whereas Raman spectroscopy is simply not a very sensitive technique and detection limits are typically grams per liter Surface-enhanced Raman spectroscopy (SERS) offers several advantages over conventional vibrational methods and that may provide the necessary sensitivity required for detecting trace quantities of chemical agents in water SERS employs the interaction of surface plasmon modes of metal particles with target analytes to increase scattering efficiency by as much as 1 million times12 In 1987 the potential of this approach to measure CWAs was demonstrated by measuring a series of pesticides13 Several of these organophosphonates have chemical structures similar to CWAs in particular P=O functional groups In the past few years we and others have further explored the ability of SERS to detect CWAs14-17 and even bioagents 18-21 We have been employing silver-doped sol-gels to promote the SER effect
SPIE-2003-5269 17
in these studies The porous silica network of the sol-gel matrix offers a unique environment for immobilizing and stabilizing SER-active metal particles22-25 The silver-doped sol-gels have been coated on the internal walls of standard glass vials to allow rapid SER analysis of small sample quantities without preparation We have measured over 100 chemicals with enhancements of 104 to 106 demonstrated reversible measurements in a flowing system reproducible measurements from vial-to-vial and measurements in multiple solvents including water21-25 Previously we used these vials to perform preliminary measurements of cyanide (CN) and methylphosphonic acid (MPA) Most of the nerve agents form MPA during hydrolysis while Tabun forms CN a chemical agent in its own right In another paper including in these proceedings we examined the limits of detection (LOD) for MPA by measuring a series of concentrations down to 50 mgL and estimated a limit of detection of 100 microgL26 These measurements provide encouragement in that SERS may satisfy the needs of the JSAWM To further establish the viability of SERS in particular silver-doped sol-gels here we present analysis of cyanide mustard and VX in tap water The measurements performed at the US Armyrsquos Edgewood Chemical Biological Center Aberdeen MD also included numerous repeat measurements to establish reproducibility
2 EXPERIMENTAL 2a General All chemicals including potassium cyanide 2-chloroethylethyl sulfide and those used to prepare the silver-doped and gold-doped sol-gel coated vials were obtained and used as received from Sigma-Aldrich All samples were prepared in a chemical hood using HPLC grade water unless otherwise noted (Fischer Scientific Fair Lawn NJ) for SERS measurements Once prepared the samples were transferred into the silver-doped sol-gel vials (Real-Time Analyzers Inc East Hartford CT) The vials were coated in a manner similar to that previously reported by adding ammonium hydroxide to a solution of silver nitrate tetramethyl orthosilicate (TMOS) and methanol22 The two precursor solutions were prepared mixed and transferred to 2-mL glass vials dried and heated After sol-gel formation the incorporated metal ions were reduced with dilute sodium borohydride (1mgmL) which is followed by a water wash to remove residual reducing agent After the resultant analyte solution was introduced the SER-active vial was fixed horizontally to an XY positioning stage (Conix Research Springfield OR) just inside the focal point of an f07 aspheric lens The lens focused the beam into the sample and collected the scattered radiation back along the same axis A dichroic filter (Omega Optical Brattleborough VT) was used to reflect the excitation laser to the lens and pass the Raman scattered radiation collected by the lens An f2 achromat was used to collimate the laser beam exiting a 200 microm core diameter source fiber optic while a second f2 achromat was used to focus the scattered radiation into a 365 microm fiber optic (Spectran Avon CT) A short pass filter was placed in the excitation beam path to block the silicon Raman scattering generated in the source fiber from reflecting off sampling optics and reaching the detector A long pass filter was placed in the collection beam path to block the sample Rayleigh scattering from reaching the detector A 785 nm diode laser (Process Instruments Inc model 785-600 Salt Lake City UT) was used to deliver 100 to 150 mW of power to the sample A Fourier transform Raman spectrometer (Real-Time Analyzers model IRA-785 East Hartford CT) and a silicon photo-avalanche detector (Perkin Elmer model C30902S Stamford CT) were used to acquire the SER spectra
2b Edgewood Chemical Biological Center The surface-enhanced Raman spectral measurements at the US Armyrsquos Chemical Biological center presented here were all performed on September 12 2003 To expedite measurements a plate was machined to hold up to 12 SER-active sample vials (Figure 1) The plate fit a standard XY plate reader that could be programmed Pure KCN bis-(2-chloroethyl)sulfide (distilled mustard gas HD) and ethyl S-2-diisopropylamino ethyl methylphosphonothioate (VX) were obtained on-site and used to prepare 1 gL tap water solutions in a chemical hood with appropriate safety equipment Simulated tap water was prepared by adding 10 chemicals most often found in tap water at appropriate concentrations to distilled water (Table 1) SER measurements were also performed in a chemical hood For added safety the FT-Raman instrument was placed outside the laboratory and 30 foot fiber optic and electrical cables were used to allow remote SERS measurements and plate manipulation For each experiment 1gL samples were prepared and added to 9 individual vials which were then loaded on the plate In some cases a tenth vial was included as a blank
Table 1 Generic Tap Water Composition
Compound mgL NaHCO3 100 CaSO4 27 MgSO4bull7H2O 67 NaNO3 1 Fulvic Acid 1 K2HPO4 07 KH2PO4 03 (NH4)2HSO4 001 NaCl 001 FeSO4 0001 pH 76-78
SPIE-2003-5269 18
A software program was written that allowed selecting the sequence that the vials were measured the number of positions along the length of the vials to measure (1 to 5) and the number of scans to co-add During sample analysis the program displayed the vial being analyzed the point being analyzed and the spectrum as it was being acquired Once all the data was collected a second software program was written to rapidly analyze the data The spectra collected for all the vials on a plate could be loaded at one time and then the spectra for each point could be displayed simultaneously or separately The user could then select the Raman peak to analyze in terms of peak height or area This was accomplished by selecting points on either side of the peak to define a baseline of zero The peak height or area could then be computed for all of the spectra loaded and then exported to a spreadsheet for statistical analysis
Figure 1 A) Vial Holder 6 slots to hold 2 vials each end-to-end B) Measurement Configuration Program user selects vials to measure sequence number of points per vial (1 to 5) and number of scans per point C) Spectral Acquisition Program shows spectrum being collected which vial and position D) Spectra Analysis Program user selects spectra to analyze by plate vial and point (s) as well as two wavenumbers defining the peak and the baseline to subtract The image is of 5 repeat measurements of 10 mgL KCN in generic tap water 16 sec each 100 mW of 785 nm
3 RESULTS AND DISCUSSION Raman and surface-enhanced Raman spectra were obtained for potassium cyanide bis-(2-chloroethyl)sulfide and ethyl S-2-diisopropylamino ethyl methylphosphonothioate representing three classes of chemical agents cyanides mustards and nerve agents respectively Spectra were also obtained for 2-chloroethyl ethyl sulfide (CEES) a structural analogue to HD which was included in the study to aid in assigning spectral bands KCN salt was used for cyanide experiments to avoid the increased hazards of handling HCN gas KCN completely dissolves in water forming its conjugate acid HCN according to its Ka of 615x10-1027 and at a concentration of 1 mgmL results in a pH 107 solution This is significant in that only CN- appears to interact sufficiently with silver to produce a SER spectrum and no spectral signal is observed below pH 726 Figure 2 shows the SER and normal Raman spectra for KCN The SERS spectrum of 1mgml KCN in water shows a single intense somewhat broad feature at
A B
C D
SPIE-2003-5269 19
2100 cm-1 assigned to the single CequivN stretch The band is much sharper in the normal Raman spectra of the solid KCN salt at 2074 cm-1 This band does broaden and shift to 2080 cm-1 in solution (not shown) However the observed SERS frequency is attributed to interaction with silver and low frequency mode at 135 cm-1 attributed to a Ag-CN stretch (not shown) supports this conclusion
cm-1 band to a C-S stretch but the authors concede that it is in fact more likely a C-Cl stretch28 It appears that the most intense Raman bands at 648 692 and 747 cm-1 shift to 620 660 and 720 cm-1 in the SER spectra and are tentatively assigned as above The width of these bands suggests that they overlap underlying spectral features Additional bands in the Raman spectra occur at 972 1034 1049 1263 1286 1423 1442 2865 2935 and 2960 cm-1 Corresponding bands occur in the SER spectra at 964 1015 1054 1286 1410 1447 2865 and 2935 cm-1 Most of these bands are associated with alkane modes specifically the bands at approximately 1040 cm-1 to a C-C stretch 1290 cm-1 to a CH2 in-phase twist 1440 cm-1 to a CH2 wag 2865 cm-1 to a symmetric CH2 stretch and 2965 cm-1 to an asymmetric CH2 stretch The Raman and SER spectra of sulfur mustard were measured at the Edgewood center (Figure 4) Both spectra are largely similar to CEES The C-Cl and C-S bands in the Raman spectrum of HD now occur at 640 655 700 739 and 760 cm-1 and are more resolved possibly due to the increased molecular symmetry Theoretical calculations indicate that the first three bands are due to C-Cl stretching modes and the latter two to C-S stretching modes28 Only the C-Cl bands maintain significant intensity in the SER spectra occurring at 624 and 643 cm-1 which is attributed to the
Prior to measurements of HD CEES was examined by Raman and SER spectroscopy (Figure 3) CEES also known as half-mustard is essentially identical to HD except one of the chlorine end atoms is replaced by a hydrogen atom Again although not as toxic as HD CEES is a blister agent and dilute aqueous samples were prepared using appropriate safety equipment Both the Raman and SER spectra of CEES are similar and dominated by bands between 600 and 800 cm-1 These are associated with C-Cl and C-S stretching modes which are tentatively assigned to 648 and 747 cm-1 in the Raman spectra respectively The shoulder at 630 cm-1 the overlapped band at 660 cm-1 and the strong band at 692 cm-1 could also be due to these modes or their asymmetric counterparts It is worth noting that theoretical calculations assign the 692
Figure 2 A) SER and B) NR spectra of KCN Conditions A) 1 mgml in tap water 100 mW of 785 nm at sample 1-min acquisition time B) solid 300 mW of 785 nm 5-min All spectra are 8 cm-1 resolution
A
B
Figure 3 A) SER and B) NR spectra of CEES Conditions A) 1 vv (10 mgml) in MeOH 100 mW of 785 nm 1-min acquisition time B) neat 300 mW of 785 nm 5-min
A
B
Cl-CH2-CH2-S-CH2-CH3
A
B
Figure 4 A) SER and B) NR of HD Conditions A) 1mgml in tap water B) pure both 100 mW of 785 nm 1-min
Cl-CH2-CH2-S-CH2-CH2-Cl
SPIE-2003-5269 20
expected strong interaction between chlorine and silver and adds support to the assignment of this band to a C-Cl stretch Weaker overlapping bands occur at 670 692 and 724 cm-1 the latter possibly due to C-S stretching modes Again the alkane modes are apparent in the normal Raman spectra of HD but only a broad feature at 1300 to 1450 cm-1 suggests CH2 contributions in the SER spectrum Although the observed bands in the VX spectrum have not been assigned (Figure 4) a computer generated Raman spectrum29 predicts many of the same features with surprising accuracy and are used here Two intense bands at 460 and 530 cm-1 closely match predicted bands at 463 and 546 cm-1 assigned to a CH3-P=O bend and a PO2CS wag Three highly overlapped bands occur at 694 745 and 771 cm-1 matching predicted bands at 713 730 and 760 cm-1 The first
Table 2 Measured SER peak heights for the CN stretch at 2100 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
Figure 5 A) SERS and B) NR spectra of VX Conditions A) 1 vv (10 mgml) in MeOH B) pure sample both 100 mW of 785 nm 1-min acquisition time
A
B
two have been assigned to a C-S stretch and CH2bend respectively while the latter has been attributed to either a P-C stretch or an O-C-C stretch Although the 745 cm-1 band may alternatively be assigned to a C-S stretch based on the previous measurements of CEES and HD The relatively intense bands at 890 1106 1218 1445 and 1465 cm-1 also match predicted bands at 880 1108 1216 1440 and 1464 cm-1 that are assigned to a C-C stretch CH3 rock N-C3 stretch various C-H3 bends and C-H bends respectively Both the computer generated and the measured spectra contain numerous other less intense bands One is worth mentioning A unique band appears at 370 cm-1 that is predicted at 368 cm-1 and corresponds to an O-P=O bend The surface-enhanced Raman spectrum of VX is also rich with spectral features It has the unique low frequency band at 370 cm-1 as well as a second band at 380 cm-1 that is assigned tothe S-P-O bend predicted in the normal Raman spectrum at 388 cm-1 Based on the measured and predicted normal Raman spectra the following SERS assignments are given 460 cm-1 to the CH3-P=O bend 544 cm-1 to the PO2CS wag 738 cm-1 to a C-S stretch (based on arguments above) 890 cm-1
to a C-C stretch 1101 cm-1 to a CH3 rock and 1456
cm-1 to a C-H bend The ability of SERS to measure chemical agents in water containing real-world chemical interferents was tested by using the generic tap water described in Table 1 The ability to reproduce measurements was accomplished by preparing three separate water stock solutions which were used to prepare three sample solutions each of 1 mgmL KCN in the generic tap water The samples defined as vials 1-9 were then measured at 4 points per vial in a semi-random fashion such that errors associated with stock solution preparation and errors associated with instrument drift could be identified No trends were apparent that signified such systematic errors Each spectrum collected consisted of 20 averaged scans taking 16 seconds at 8 cm-1 resolution The laser power at the sample was measured periodically during the day and it ranged from 102 to 105 mW spacccc
spot Vial 1 Vial 2 Vial 41 13754 13004 128192 13519 12692 12909
1 3 13541 12721 126394 13462 12648 12651
ave 13569 12766 12755Vial 6 Vial 3 Vial 5
1 11586 14068 10757 2 11236 14402 115122 3 11376 14546 11559
4 10894 11763 11214ave 11273 13695 11261
Vial 7 Vial8 Vial 91 11198 15804 11023
2 11243 15716 88753 3 11562 15216 11218
4 11655 15084 11077ave 11415 15455 10548
CN AVG STDEV ERR1 mgmL pk ht 12526 1560 1245
stock solution
SPIE-2003-5269 21
The CN measurements consisting of 36 data points produced an average height of 125 for the 2100 cm-1 peak with a standard deviation of 156 or 125 (Table 2) The HD and VX measurements were performed precisely the same way (Tables 3 and 4) For HD the 624 cm-1 peak was used for analysis and it had an average height of 53 with a standard deviation of 068 or 129 while for VX the 544 cm-1 peak was used for analysis and it had an average height of 1051 with a standard deviation of 308 or 293 The greater error in the VX measurements can be somewhat attributed to Vial 4 which produced lower SER signal intensities But removing this vial from the data set changes the standard deviation to 233 only a modest improvement
spot Vial 1 Vial 2 Vial 41 525 609 4682 45 675 484
1 3 527 644 5094 559 807 669
ave 51525 68375 5325Vial 6 Vial 3 Vial 5
1 521 574 575 2 536 53 4492 3 509 508 451
4 565 594 379ave 53275 5515 4635
Vial 7 Vial 8 Vial 91 457 589 413
2 583 497 4273 3 544 505 4
4 528 5 509ave 528 52275 43725
HD AVG STDEV ERR1 mgmL pk ht 530 068 1291
stock solution
spot Vial 1 Vial 2 Vial 41 1464 1034 4672 1485 99 61
1 3 1491 989 5684 1041 777 553
ave 137025 9475 5495Vial 6 Vial 3 Vial 5
1 1058 942 1293 2 697 121 965
2 3 727 1206 11124 689 1254 746
ave 79275 1153 1029Vial 7 Vial 8 Vial 9
1 1127 783 163 2 1358 812 1614
3 3 1371 908 15124 1097 875 1397
ave 123825 8445 153825
VX AVG STDEV ERR1 mgmL pk ht 1051 308 2925
stock solution
4 CONCLUSIONS In this paper we examined the ability of surface-enhanced Raman spectroscopy to reproducibly measure CN HD and VX in tap water without chemical interference Both normal and surface-enhanced Raman spectra were examined to select unique bands suitable to identify and quantify these chemical agents For SER measurements the 2100 cm-1 C-N stretch was used for CN the 624 cm-1 C-Cl stretch was used for HD and the 544 cm-1 PO2CS wag was used for VX It was determined that 1 mgmL samples of each of these chemicals measured 36 times in glass vials coated with a silver-doped sol-gel reproduced measurements with standard deviations of 125 129 and 293 It was further found that the 10 chemicals added to simulate generic tap water did not interfere with or alter the SER spectra It should be noted that the concentrations used in this study were considerably greater than those required by the JSAWM program Current work involves improving SER sensitivity and designing sampling systems with better reproducibility This includes the development of fractal silver and gold structures within the sol-gel matrix and the development of chemically selective sol-gels
ACKNOWLEDGMENTS The authors are grateful for the support of the US Army (DAAD13-02-C-0015 Joint Service Agent Water Monitor program) The authors would also like to thank Janet Jensen Ronald Crosier and Kristina Gonser for helpful discussions
Table 4 Measured SER peak heights for the VX band at 544 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
Table 3 Measured SER peak heights for the HD band at 624 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
SPIE-2003-5269 22
REFERENCES 1 Jensen JL NC Wong W R Loerop SPIE 4775-03 (2001) 2 JSAWM Requirements at wwwsbccomapgeaarmymilRDAecbcrtPRODSERJSAWMjsawmhtml 3 Erickson B ldquoThe Chemical Weapons Convention Redefines Analytical Challengerdquo Anal Chem 70 397A (1998) 4 Johnston RL Hoefler CM Fargo JC and Moberley B ldquoThe Defense Nuclear Agencyrsquos
ChemicalBiochemical Weapons Agreements Verification Technology Research Development Test and Evaluation Program and its Requirements for On-Site Analysisrdquo AT-ONSITE 5-8 (1994)
5 Black RM Clarke RJ Read RW and Reid MT ldquoApplication of GC-MS and GC-tandem MS to the analysis of chemical warfare samples found to contain residues of the nerve agent sarin sulphur mustard and their degradation productsrdquo J Chrom 662 301-321 (1994)
6 USA Army Center for Health Promotion and Preventive Medicine (USACHPPM) Technical Guideline 230A 7 Hoffland LD Piffath RJ Bouck JB rdquoSpectral signatures of chemical agents and simulantsrdquo Optical
Engineering 24 982-984 (1985) 8 Braue EHJ Pannella MGrdquoCIRCLE CELL FT-IR Analysis of Chemical Warfare Agents in Aqueous Solutionsrdquo
App Spectrosc 44 1513-1520 (1990) 9 Tseng C-H Mann CK and Vickers TJ ldquoDetermination of Organics on Metal Surfaces by Raman
Spectroscopyrdquo App Spectrosc 47 1767-1771 (1993) 10 Staff Reporter ldquoUS Army Tests Lidar to Detect Biological Toxinsrdquo Photonic Spectra pg 50 December 1998 11 Christesen SD Raman cross sections of chemical agents and simulants App Spectrosc 42 318-321 (1988) 12 Weaver MJ Farquharson S Tadayyoni MA ldquoSurface-enhancement factors for Raman scattering at silver electrodesrdquo J Chem Phys 82 4867-4874 (1985) 13 Alak AM and Vo-Dinh T ldquoSERS of Organophosphorous Chemical Agentsrdquo Anal Chem 59 2149-2153 (1987) 14 Lee Y Farquharson S ldquoRapid chemical agent identification by SERSrdquo SPIE 4378 21-26 (2001) 15 Farquharson S Maksymiuk P Ong K Christesen S ldquoChemical agent identification by surface-enhanced Raman spectroscopyrdquo SPIE 4577 166-173 (2001) 16 Spencer KM Sylvia J Clauson S and Janni J ldquoSurface Enhanced Raman as a Water Monitor for Warfare
Agents in Waterrdquo SPIE 4577 158-165 (2001) 17 Tessier P Christesen S Ong K Clemente E Lenhoff A Kaler E Velev O ldquoOn-line spectroscopic characterization of sodium cyanide with nanostructured gold surface-enhanced Raman spectroscopy substratesrdquo
App Spectrosc 56 1524-1530 (2002) 18 Farquharson S W W Smith S Elliott and J F Sperry Rapid biological agent identification by surface- enhanced Raman spectroscopy SPIE 3855110-116 (1999) 19 Farquharson S WW Smith YH Lee S Elliott and J F Sperry Detection of biological signatures A
comparison of electrolytic and metal-doped sol-gel surface-enhanced Raman media SPIE 4575 62-72 (2002) 20 Guzelian AA Sylvia J Janni J Clauson S and Spencer KM ldquoSERS of whole cell bacteria and trace levels of biological moleculesrdquo SPIE 4577 182-192 (2001) 21 Shende C Inscore F Gift A Maksymiuk P Farquharson S ldquoRapid dipicolinic acid extraction from Bacillus spores detected by surface-enhanced Raman spectroscopyrdquo App Spectrosc 58 accepted 22 Lee Y and Farquharson S ldquoSERS Sample Vials Based on Sol-Gel Process for Trace Pesticide Analysisrdquo SPIE
4206 140-146 (2000) 23 Farquharson S and Lee Y ldquoTrace Drug Analysis by SERSrdquo SPIE 4200-16 (2000) 24 Lee Y Farquharson S and Rainey P M Surface-Enhanced Raman Sensor for Trace Chemical Detection in
Water SPIE 3857 76-84 (1999) 25 Lee Y Farquharson S Kwong H and Shahriari M ldquoSol-Gel Chemical Sensor for Surface-Enhanced Raman
Spectroscopyrdquo SPIE 3537 252-260 (1998) 26 Farquharson S A Gift P Maksymiuk F Inscore and W Smith ldquopH dependence of methyl phosphonic acid
dipicolinic acid and cyanide by surface-enhanced Raman spectroscopyrdquo SPIE 5269-19 (2003) 27 Lide DR Ed Handbook of Chemistry and Physics CRC Press 77th Ed (1996-7) 28 Sosa C RJ Bartlett K KuBulat and WB Person ldquoA theoretical study of harmonic vibrational frequencies and
infrared intensities of XCH2CH2SCH2CH2X and XCH2CH2SH (= H Cl)rdquo J Phys Chem 93 577-588 (1993) 29 Hameka HF and JO Jenson ldquoComputer-generated predictions of the structure and the IR and Raman spectra of
VXrdquo ERDEC-TR-065 May 1993
- Chemagents Appendicespdf
-
- SPIE2001-4575-Bioagent-SERSapdf
-
- Detection of bioagent signatures A comparison of electrolytic and metal-doped sol-gel surface-enhanced Raman media
- Stuart Farquharson Wayne Smith and Yuan Lee
-
- Real-Time Analyzers 87 Church Street East Hartford CT 06108
-
- ABSTRACT
- 4 CONCLUSIONS
- 5 ACKNOWLEDGEMENTS
- 6 REFERENCES
-
SPIE-4378-2001
24
2-Chloroethyl ethyl sulfide a blister agent simulant has a chemical structure similar to the mustard gas (Cl-CH2-CH2-S-CH2-CH2-Cl) with only one terminal chlorine Due to its low solubility in water CEES was dissolved in methanol for the SER measurement Again the prominent Raman modes are SER-active and even maintain relative intensity (Figure 4) The primary difference is that the SER bands appear to broaden such that the triplet near 700 cm-1 becomes a doublet and the shoulders at 2875 and 2970 cm-1 become less defined Again the latter bands are assigned to C-H stretching modes A single band at 700 cm-1 which is attributed to the C-S-C asymmetric stretch dominates the reported infrared spectrum of mustard gas12 A corresponding symmetric stretch is reported at 705 cm-1 in the Raman spectrum of mustard gas27 Here a corresponding symmetric stretch appears but as a doublet at 700 and 755 cm-1 presumably due to the loss in symmetry for CEES The band at 655cm-1 can also be confidently assigned to a C-Cl stretch The SER spectral bands at 620 and 730 cm-1 are probably due to the same modes ie C-Cl and C-S-C stretches respectively The enhancement factor for CEES was somewhat less than DMMP at approximately 62000 as is the estimated detection limit of 22 ppm The ability to rapidly detect trace quantities of chemical agents in the gas phase would be invaluable as an early warning system Although the Raman scattering cross-sections for the nerve agents suggest that remote detection by Raman-based LIDAR is unlikely16 a SER-based system for perimeter monitoring could prove successful As a preliminary measurement we prepared a 10 by volume solution of CEES in methanol exposed a SER-active vial to the equilibrium vapor phase in a sealed jar and monitored the SER spectrum as a function of time Initially the vial was removed through a transfer chamber every hour to record the SER spectrum After ten hours spectra were recorded only every ten hours As illustrated by Figure 5 the sol-gel performed as a dosimeter in that the spectra increased as a function of exposure time The most intense SER bands at 620 and 2930 cm-1 are discernable in the first few hours The spectrum after 40 hours is nearly identical to the solution phase spectrum except for a diminished intensity of the 730 cm-1 band This may be due to methanol solvation effects or surface-orientation effects Based on the relative concentrations of methanol and CEES and their partial pressures we estimate the equilibrium concentration of CEES to between 1 and 2 micromolar Although not shown this concentration could be detected in one hour As previously stated the analysis of chemical agents in water is important in identifying poisoned water It is also important to decommissioning activities in which agents are destroyed by hydrolysis (acid or base) Furthermore any analytical technique used must be capable of distinguishing between parent CWA and hydrolysis products to assess safety or effectiveness of decommissioning For example soman has a hydrolysis half-life of ~23 hours at ambient temperatures and neutral pH28 and forms hydrofluoric acid (somewhat toxic) and pinacolyl methylphosphonate (relatively non-toxic)2930 PMP further hydrolyzes to form methyl phosphonic acid and 33-dimethyl-2-butanol (both non-toxic) The structural similarities between soman PMP and MPA are expected to produce similar Raman as well as SER spectra Figure 6 compares PMP and MPA but not the highly toxic parent CWA soman As with DMMP the P-C stretch the P-O-C mode and C-H stretches are readily apparent Yet it is worth noting that the band positions are reasonably different The former two bands appear at 764 and 1042 cm-1 for MPA while they are at 788 and 1032 cm-1 for PMP More importantly a unique band at 546 cm-1 as yet unassigned appears in the PMP spectrum
Figure 6 SER spectra of A) methyl phosphonic acid and B) pinacolyl methylphosphanate (note unique band at 546 cm-1) Spectral conditions as in Figure 1
Figure 5 SER spectra of 2-chloroethyl ethyl sulfide vapor as a function of time (10 hour increments to top which is 40 hours) Bottom trace is a blank Spectral conditions as in Figure 1
A
B
Wavenumbers (∆cm-1) Wavenumbers (∆cm-1)
HO-P-OH
=
O
CH3
_
HO-P-OH
=
O
=
O
CH3
_
CH3
_HO-P-O-CH-C-CH3
=O
CH3
_
CH3
_CH3_CH3
_
HO-P-O-CH-C-CH3
=O
=O
CH3
_
CH3
_
CH3
_
CH3
_CH3_CH3
_CH3
_
SPIE-4378-2001
25
4 CONCLUSIONS We have successfully measured the SER spectra of chemical agent simulants dimethyl metylphosphonate and 2-chloroethyl ethyl sulfide and chemical agent hydrolysis products pinacolyl methylphosphonate and methylphosphonic acid using silver-doped sol-gel coated sample vials Measurements were obtained in both aqueous and gas phase The P-C stretching mode was SER-active for all four chemicals allowing identification by class Within this group each chemical contained at least one unique spectral band that could be used for identification (Table 1) Furthermore these bands do not appear to coincide with SER spectra reported for organophosphorus pesticides the most likely source of false-positives Although surface enhancement factors appear to be an order of magnitude better than those previously presented in the literature for similar chemicals17 measurement sensitivity needs to be improved by 1 to 2 orders of magnitude to provide adequate warning of chemical agent use Current research efforts to increase surface-enhancement optical collection efficiency and instrument design are being pursued to achieve the required sensitivity
Table 1 Enhancement factors detection limits and unique SER bands fro chemicals studied Agent Simulant Enhancement Detection limit Unique bands (cm-1)
Dimethyl methylphosphonate 120000 90 microM (16 ppm) 425 2-Chloroethyl ethyl sulfide 62000 60 microM (22 ppm) 620
Methylphosphonic acid 110000 3 microM (60 ppb) 764 1042 Pinacolyl methylphosphonate 150000 70 microM (14 ppm) 546 788 1032
5 ACKNOWLEDGEMENTS The authors would like to thank Drs Janet Jensen and Steven Christesen of Aberdeen Proving Ground for encouraging this work They would also like to thank Advanced Fuel Research for making their laboratory facilities available
6 REFERENCES 1 ldquoThe Chemical Weapons Convention ndash A Guided Tour the Organization for the Prohibition of Chemical Weaponsrdquo at
httpwwwopcwnlguidehtm 2 Robinson JP and J Goldblat Chemical Warfare In The Iraq-Iran War Stockholm International Peace Research
Institute Fact Sheet at httpprojectssiprisecbwresearchfactsheet-1984html (1984) 3 ldquoChemistry of GB (Sarin)rdquo at httpwwwmitretekorgmissionenvenechemicalagentssarinhtml 4 Tu Anthony ldquoOverview of Sarin Terrorist Incidents in Japan in 1994 and 1995rdquo 6th CBW Protection Symposium
Stockholm Sweden 10-15 May 1998 5 Staff Reporter ldquoGoing out with a bangrdquo Newsweek June 28 1999 6 See UNSCOM reports in httpwwwunorgdeptsunscom (1999) 7 Treven T Saddamrsquos Secrets Harper Collins (1999) 8 ldquoDecaying Sarin-filled Rockets Spark Fearsrdquo Janersquos Defense Weekly 25(20)3 (1996) 9 ldquoThe Chemical Weapons Convention Redefines Analytical Challengerdquo Analytical Chemistry News amp Features June 1
397A (1998) 10 Johnston RL Hoefler CM Fargo JC and Moberley B ldquoThe Defense Nuclear Agencyrsquos ChemicalBiochemical
Weapons Agreements Verification Technology Research Development Test and Evaluation Program and its Requirements for On-Site Analysisrdquo AT-ONSITE 5-8 (1994)
11 Black RM Clarke RJ Read RW and Reid MT ldquoApplication of gas chromatography-mass spectrometry and gas chromatography-tandem mass spectrometry to the analysis of chemical warfare samples found to contain residues of the nerve agent sarin sulphur mustard and their degradation productsrdquo J Chromatography 662 301-321 (1994)
12 Hoffland LD Piffath RJ Bouck JBrdquoSpectral signatures of chemical agents and simulantsrdquo Optical Engineering 24 982-984 (1985)
13 Braue EHJ Pannella MGrdquoCIRCLE CELL FT-IR Analysis of Chemical Warfare Agents in Aqueous Solutionsrdquo Applied Spectroscopy 44 1513-1520 (1990)
SPIE-4378-2001
26
14 Tseng C-H Mann CK and Vickers TJ ldquoDetermination of Organics on Metal Surfaces by Raman Spectroscopyrdquo
Applied Spectroscopy 47 1767-1771 (1993) 15 Staff Reporter ldquoUS Army Tests Lidar to Detect Biological Toxinsrdquo Photonic Spectra pg 50 December 1998 16 Christesen SD Raman cross sections of chemical agents and simulants Applied Spectroscopy 42 318-321 (1988) 17 Alak AM and Vo-Dinh T ldquoSERS of Organophosphorous Chemical Agentsrdquo Analytical Chemistry 59 2149-2153
(1987) 18 Norrod KL Sudnik LM Rousell D and Rowlen KL ldquoQuantitative Comparison of Five SERS Substrates
Sensitivity and Detection Limitrdquo Applied Spectroscopy 51 994-1001 (1997) 19 Lee Y and Farquharson S ldquoSERS Sample Vials Based on Sol-Gel Process for Trace Pesticide Analysisrdquo SPIE 4206
140-146 (2000) 20 Farquharson S and Lee Y ldquoTrace Drug Analysis by Surface-Enhanced Raman Spectroscopyrdquo SPIE 4200-16 (2000) 21 Lee Y Farquharson S and Rainey P M Surface-Enhanced Raman Sensor for Trace Chemical Detection in Water
SPIE 3857 76-84 (1999) 22 Lee Y Farquharson S Kwong H and Shahriari M ldquoSurface-Enhanced Raman Sensor for Surface-Enhanced Raman
Spectroscopyrdquo SPIE 3537 252-260 (1998) 23 see Material Safety Data Sheets for details 24 Farquharson S Smith W Carangelo R C and Brouillette C ldquoIndustrial Raman Providing Easy Immediate Cost
Effective Chemical Analysis Anywhererdquo SPIE 3859 14-23 (1999) 25 Bennett S Bane J Benford P and Pratt R ldquoEnvironmental Hazards of Chemical Agent Simulantsrdquo Aberdeen
Proving Ground Maryland Chemical Research and Development Center CRDC-TR-84055 (1984) 26 Mabey W and Mill T Critical Review of Hydrolysis of Organic Compounds in Water under Environmental Conditions
Journal of Physics and Chemistry Reference Data 7(2) 383-414 (1978) 27 Christesen S MacIver B Procell L Sorrick D Carabba M and Bello J ldquo Noninstrusive Analysis of Chemical Agent
Identification Sets Using a Portable Fiber-Optic Raman Spectrometerrdquo Applied Spectroscopy 53 850-855 (1999) 28 Meylan WM and Howard PH J Pharm Sci 84 83-92 (1995) 29 Jenkins A Uy O and Murray G ldquoPolymer-Based Lanthanide Luminescent Sensor for Detection of Hydrolysis Product
of the Nerve Agent Soman in Waterrdquo Analytical Chemistry 71 373-378 (1999) 30 Nassar A Lucas S and Hoffland L ldquoDetermination of Chemical Warfare Agent Degradation Products at Low-Part-
per-Billion Levels in Aqueous Samples and Sub-Part-per-Million Levels in Soils Using Capillary Electrophoresisrdquo Analytical Chemistry 71 1285-1292 (1999)
SPIE 2001-4575
62
Detection of bioagent signatures A comparison of electrolytic and metal-doped sol-gel surface-enhanced Raman media
Stuart Farquharson Wayne Smith and Yuan Lee
Real-Time Analyzers 87 Church Street East Hartford CT 06108
Susan Elliott and Jay F Sperry University of Rhode Island 45 Lower College Rd Kingston RI 02881
ABSTRACT Since September 11 2001 the threat of terrorist attacks and biological warfare within US borders has become a sobering reality In an effort to aid military personnel and the public at large we have been investigating the utility of surface-enhanced Raman spectroscopy (SERS) to provide rapid identification of chemical agents directly and biological agents through their chemical signatures This approach is based on the ability of Raman spectroscopy to identify molecular structure through the abundant vibration information provided in spectra and the ability of SERS to detect extremely low concentrations (eg part-per-billion) through the enhancement of Raman scattering by six orders of magnitude or more Towards the goal of developing a portable analyzer we have been studying the ability of two SER media to obtain continuous (ie reversible) and quantitative (ie reproducible) measurements Here we compare measurements of nucleic acid-bases adenosine monophosphate and ribonucleic acid extracted from Escherichia coli Bacillus subtilis and Staphylococcus aureus obtained by electrolytic SERS and metal-doped sol-gel SERS The capabilities of these SER media are summarized in terms of rapid detection of B anthracis and dipicolinic acid Keywords bioagent detection SERS RNA analysis bacterial analysis Raman spectroscopy
1 INTRODUCTION The recent distribution of anthrax through the US postal system and the subsequent infection and death of several postal and national media employees amplifies the need for methods to rapidly detect and identify this and other chemical and biological warfare agents (BWA) The primary methods currently used immunoassays for screening and nucleic acid (NA) sequencing for positive identification of BWAs (bacteria protozoa and viruses) have serious limitations123 Immunoassay methods employ competitive binding of the bioagent (as an antigen) and its labeled (eg fluorescence) conjugate for a limited number of antibodies Although this analysis method is fast and semi-quantitative other chemicals may compete for the antibodies interfere with the enzymatic reaction or interfere with the measurement (eg it fluoresces) resulting in a high number of false positive responses1 Furthermore the antibodies denature due to moisture and heat limiting shelf life and require sterile often refrigerated storage Positive identification of a BWA can be accomplished by sequencing deoxyribonucleic acid or ribonucleic acid (DNA and RNA)23 This requires enumeration of a nucleic acid sequence through polymerase chain reactions (PCR) or multiplication of the microorganism through culture growth to provide sufficient quantities of DNA or RNA for analysis Unfortunately PCR and culture growth require from several hours to several days23 Consequently a wide variety of technologies have been investigated for rapid identification of BWAs The Department of Defense is actively monitoring 200 such technologies4 This includes traditional methods such as gas chromatographic separation coupled with ion mobility spectrometry detection5 to exotic methods based on nature such as monitoring toxin induced color changes in fish scales6 Although all of these techniques have achieved varying degrees of success none are yet capable of detecting and identifying BWAs in 10 minutes or less Towards this goal we have been investigating the ability of SERS to detect sub-nanogram quantities of DNA or RNA (eliminating enumeration) determine relative NA base concentrations and identify BWA taxonomy
To whom correspondences should be addresses e-mailfarqureal-time-analyzerscom wwwreal-time-analyzerscom
SPIE 2001-4575
63
Raman spectroscopy has a rich history of investigating biochemical and biological processes7 Some of the earliest laser-Raman studies demonstrated that the five NA bases adenine (A) cytosine (C) guanine (G) thymine (T in DNA) and uracil (U in RNA) yielded distinct spectra with several bands suitable for identification and quantification8 Furthermore these studies included exceptional spectra of both DNA and RNA for which the NA bases as well as several phosphate bands were easily identified9 However since the Raman effect is very inefficient (very low conversion of incident radiation to inelastically scattered Raman radiation) these samples had to be highly concentrated Fortunately two phenomena exist that can increase the generation of Raman photons by six orders of magnitude or more known as the resonance Raman and surface-enhanced Raman effects1011 Resonance Raman scattering occurs when the laser excitation wavelength is in resonance with an electronic transition of a molecule (within the absorption envelope)10 Excitation at ultraviolet wavelengths has been used to obtain resonance Raman spectra of amino acids and nucleic acids in whole bacteria1213 For example excitation at 242 nm has been used to maximize the nucleic acid spectral band intensities and minimize the amino acids band intensities A peak at 1530 cm-1 was found to be proportional to the amount of the NA base cytosine while a peak at 1485 cm-1 was proportional to the combined amount of the NA bases adenine and guanine This quantitative behavior has been used to define an A+TG+C base-pair ratio and provide a level of bacterial identification as taxonomic markers13 In recent years SERS has also been used to analyze bacterial cell components14 including amino acids15 lipids16 nucleic acids151718 and the adenine derivatives192021 SERS has proven to be one of the most sensitive methods for trace chemical analysis through the detection of single molecules2223 including DNA (dye labeled 17-mer)24 Since its discovery in 197425 the mechanism responsible for the large increase in scattering efficiency has been the subject of considerable research2627 Briefly incident laser photons couple to free conducting electrons within a metal which confined by the particle surface collectively cause the electron cloud to resonate2628 These surface plasmons are known as the physical component of the SER effect These surface plasmons can transfer energy to the molecular vibrational modes of molecules through interactions with the molecular electron orbitals2629 This interaction is known as the chemical component of the SER effect This perturbation of the molecular polarizability generates surface-enhanced Raman photons26 A number of methods have been developed to produce surfaces or solutions containing one of these metals with optimum roughness or diameter to promote SERS30 These methods include preparation of activated electrodes in electrolytic cells 1131 activated silver and gold colloid reagents32 and metal coated substrates333435 Selecting a SER-active medium for chemical and biological agent detection requires consideration of the method of deployment and hence the method of sampling Chemical aerosols or airborne bacteria will require a collection device to concentrate and continuously present the sample to the SERS medium Poisoned water supplies will also require a flow through device for continuous monitoring or a grab-sample device for periodic analysis And contaminated surfaces will require a grab-sample extractive device A SERS-based device used for continuous monitoring (air or water) must be reversible and reproducible if quantitative measurements are desired while a SERS-based device used for periodic sampling (water or surfaces) must be reproducible Both reversible and reproducible measurements have been performed using electrolytic SERS (E-SERS)36 But this requires a three-electrode sample cell and an electrolyte of known concentration to perform the necessary oxidation-reduction cycles (ORCs) to re-activate the electrode surface with new uncontaminated sites from one measurement to the next Colloids are severely limited in that continuous measurements would require a continuous supply of colloids For periodic measurements vials of colloids one per measurement could be used However aggregate size and consequently SER intensity change with sample conditions (especially pH) and quantitative reproducible measurements are unlikely Substrates appear to have the greatest potential and designs range from silver evaporated on titania particles34 to periodic gold pyramids evaporated between polystyrene beads35 Most substrates require concentrating the sample on the surface through drying to obtain the largest signal enhancements in effect making the measurements irreproducible and irreversible However successful measurements using flow systems have been obtained with glass posts but manufacturing costs appear prohibitive In an effort to overcome these limitations we have developed metal-doped sol-gels to provide SERS measurements that are reproducible reversible and quantitative and yet not restricted to specific environments such as electrolytes solvents or evaporated surfaces3738 The porous silica network of the sol-gel offers a unique environment for stabilizing SER active metal particles and the high surface area increases the interaction between the analyte and metal particles The sol-gel can be coated on the end of fiber optics or on the internal walls of a glass flow tube for continuous measurements or standard glass sample vials for periodic measurements Previously we measured 100 mgL methylphosphonic acid (the primary hydrolysis product of nerve agents) in water with an estimated detection limit of 05 mgL (100 parts-per-billion) We have also
SPIE 2001-4575
64
demonstrated reversible and reproducible measurements of p-aminobenzoic acid (PABA) in a flow through system Here we investigate the ability of the sol-gel SERS (SG-SERS) to measure the NA bases adenosine monophosphate and RNA extracted from E coli B subtilis and S aureus The measurements are compared to those obtained by E-SERS
2 EXPERIMENTAL The inorganic chemicals and solvents used to prepare samples were spectroscopic grade and purchased from Aldrich (Milwaukee WI) Fisher (Pittsburgh PA) or Pfaltz amp Bauer (Waterbury CT) The nucleic acid bases and dipicolinic acid were purchased from Sigma (St Louis MO) Normal Raman samples were measured to establish enhancement factors In each case 1cm3 of sample was placed into a 1x1 cm glass cuvette weighed and measured Unpacked densities were typically 6-7 gcm3 For all SER measurements including RNA samples were prepared as ~01mgmL (see Figure captions for exact concentrations) in 01M KCl and buffered to a pH of 92 with Na2B4O7bullH2O Adenine pH dependence measurements used pH buffer solutions at 4 (potassium acid phthalate) 69 (potassium phosphate monobasicsodium phosphate dibasic) 92 (Na2B4O7bullH2O) and 104 (tris-hydroxymethyl amino methane) Escherichia coli Bacillus subtilis and Staphylococcus aureus cultures (250ml per 1000mL Erlenmeyer flask) were grown overnight in a Trypticase soy broth (TSB) medium containing 1 glucose in a shaking water bath at 37 oC The bacteria were harvested by centrifugation for 10 minutes at 8000 rpm in a GSA rotor at 5degC then washed once in 085 saline The gram-positive bacteria were concentrated to 20 ml and passed through a French pressure cell twice at 15000 psi to break open the cells RNA was extracted according to Protocol 44139 to ensure pristine samples for initial measurements Since this method takes approximately 4 hours a streamlined method was developed For vegetative bacteria the specimen was boiled for 30 sec in 1 ml of distilled water to lyse the cells and release the RNA For bacterial spores the specimen was first incubated in 1 ml of saline solution containing 02 mg lysozyme and phosphate-buffered to pH of 624 for 1 hr at 37 oC This solution was then boiled for 2-3 minutes in 4X loading buffer to release the RNA For both specimens RNA STAT-60TM was added to the supernatant which was centrifuged at 12000 g for 5 minutes to precipitate the ~15 water-soluble proteins This procedure allowed extracting RNA for SER analysis in ~ 10 minutes Electrophoresis shows high purity while the existence of chemicals that could interfere with the SER measurements is still under investigation The electrolytic sample cell has been described previously36 Briefly a three electrode design is incorporated into a Plexiglas structure containing a well for the reference electrode (a saturated calomel electrode Cole Parmer Vernon Hills IL) and a 5mL sample well containing the silver working electrode and platinum wire counter electrode (05 mm wire Alfa Ward Hill MA) A channel connecting the two wells contained a 2 mm diameter semi-porous membrane (10-20 micron pore Ace Glass) The silver electrode was made from a 3 mm length of 2 mm diameter silver wire (Alfa) soldered to a copper wire lead encased in a 4 mm diameter Pyrex tube A cap containing the silver electrode platinum wire and nitrogen purge and vent lines fixed the silver electrode surface 1 mm from a 1 mm thick glass plate attached to the bottom of the sample well The potentiostat used to control the three electrodes was built in-house and has been described in detail elsewhere36 A multifuntional analog digital and timing inputoutput interface card (DAQCard-1200 National Instruments) is used to both drive the electrolytic cell as well as read the current generated in the cell A LabVIEW software program is used set the oxidation potential reduction potential SER measurement potential hold times and sweep rates The amount of charge passed was plotted as a cyclic voltammogram For all spectra presented five oxidation-reduction cycles (ORCs) stepping from -03 VSCE to 03 VSCE and back to -3 VSCE at 50 mVsec were used The SG-SER measurements were accomplished by simply placing the identical samples prepared above into Simple SERS Sample VialsTM (RTA) These 2-mL glass vials are internally coated with ~ 01 micron thick silver-doped sol-gel A prototype Fourier transform Raman spectrometer (RTA) was used for these measurements40 The system consisted of a NdYAG laser (Brimrose or Spectra Physics) for excitation at 1064 nm an interferometer built by On-Line Technologies (OLT East Hartford CT) for frequency separation an uncooled InGaAs detector for signal detection (RTA) and an Intel 400 MHz Pentium II based laptop computer (Dell Round Rock TX) for interferometric control data acquisition (OLT) and analysis (LabVIEW by National Instruments Austin TX) Additional components included a Notch filter (Kaiser Ann Arbor MI) and interferometer entrance and exit optics (Edmund Scientific Barrington NJ) Fiber optics were used to deliver the excitation beam to the sample and the scattered radiation to the interferometer (2 meter lengths of 200 and 365 micron core diameter respectively Spectran Avon CT) A second Notch filter (Kaiser) was used as a beam splitter to direct the excitation beam along the same axis as the collected radiation A microscope object (20x magnification 04 numeric aperture Newport Irvine CA) was used to focus the beam into the sample and to collect the scattered radiation back along the same axis In this co-axial backscattering arrangement the excitation beam passed through the glass plate onto the silver
SPIE 2001-4575
65
electrode surface for E-SERS through the vial glass wall and into the silver-doped sol-gel film for SG-SERS or through the glass wall of the cuvette and into the solid sample for normal Raman spectroscopy All E-SERS and normal Raman spectra were obtained with 750 mW of laser power at the sample while all SG-SERS spectra were obtained with 75 mW of laser power at the system Incident powers above 200 mW in some cases degraded the sol-gel
3 RESULTS AND DISCUSSION The generation of surface-enhanced Raman scattering at electrode surfaces has been extensively researched and the optimum sample conditions are well developed2729 Several researches incorporated electrodes into flowing systems and demonstrated that continuous monitoring of chemicals is possible18 These successes suggested investigated the capability of E-SERS to measure the NA bases and RNA The E-SERS measurements also provided a benchmark to compare and evaluate SG-SERS measurements The molecular structure of adenine (as well as the other base pairs) which includes an aromatic nitrogen-containing heterocycle is ideally suited to interact with the surface plasmons and contribute substantially to the chemical component of the SER effect1119 Even with excitation at 1064 nm a 3-minute scan of 18x10-5M adenine yields high signal-to-noise (SN) E-SER spectra and all of the bands are revealed with clarity (Figure 1 Table 1) Spectra of similar quality were obtained by SG-SERS and the principal spectral bands are easily observed The identical 18x10-5M adenine sample was measured in the same 3-minute time frame but with 110th the laser power The lower power appears to reduce the SN The amount of adenine responsible for the SER spectra as well as enhancement factors for the two SER media can be determined The molecules producing the E-SERS spectrum are those on the electrode surface within the illumination area of the laser (The solution concentration only determines the number of molecules available to adsorb to the electrode surface) For the current experiments the laser illuminates an area of 28x10-7m2 or 56 x10-7m2 if we assume the ORCs increase the surface area by a factor of two Furthermore if we assume monolayer coverage on the electrode and each 3x5 angstrom molecule (lying flat) occupies 15x10-19m2 then there are ca 4x1012 molecules contributing to the Raman scattering This is ca twice the number of molecules measured at electrode surfaces using either differential capacitance-potential curve measurements or rapid linear sweep voltammetry (eg 3x1018 moleculesm2 for pyridine and pyrazine)29 Thus the adenine spectrum in Figure 1 is due to 87x10-10g (6x10-12 moles) A detection limit defined as a SN of 3 can also be calculated The SN for a 3-minute scan is 844 for the 735 cm-1 band suggesting a mass detection limit of 3x10-12g (2x10-14 moles) This is consistent with previous estimates for adenine by others of 25 x10-14 moles1530 However sub-monolayer concentrations must be measured to verify this The root-mean-squared (RMS) noise is measured between 4400-4600 cm-1 Since noise is distributed evenly throughout the spectrum when transformed this region does not have any
Wavenumbers (∆cm-1) Figure 1 A) Normal Raman spectrum of pure adenine powder B) E-SERS and C) SG-SERS of 18x10-5M adenine at pH 92 All spectra 8 cm-1 resolution 200 scans (3 min) and 1064 nm excitation A) and B) 750 mW C) 75 mW B) measurement potential of 11VSCE
A
B
C
A
B
C SG-SERS
pH 10
D
725
735
735
pH 4
Wavenumbers (∆cm-1) Figure 2 A) and C) E-SERS and B) and D) SG-SERS of adenine at A) and B) pH 104 and C) and D) pH 40 Note consistent appearance of bands at 1270 and 1375 cm-1 as the pH is changed to 10 for both SER media E-SERS used 750 mW SG-SERS used 75 mW of 1064 nm excitation
SPIE 2001-4575
66
contributions from signals or baseline offsets The measurement error is given as SplusmnRMS and for adenine this equals 234 The number of molecules contributing to the SG-SERS are those on the silver particles that are embedded in the sol-gel The total silver surface area can be determined from the average particle size (40 nm diameter) concentration (073 by weight based on molar conc and measured sol-gel density) and the scattering volume (a cylinder defined by the laser area 28x10-7m2 and sol-gel thickness10-4m) The 61x109 silver particles in this volume have a collective area of 31x10-5m2 However it may reasonably be assumed that at least half of the silver surface is in contact with the sol-gel matrix and unavailable for analyte interaction Then approximately 10x1014 molecules or 22x10-8g of adenine contribute to the SG-SER spectrum The slightly lower SN of 207 suggests a mass detection limit of 32x10-10g Determination of the enhancement factors for the two SER media requires estimating the number of adenine molecules contributing to the normal Raman spectrum Here a cylindrical scattering volume is assumed again based on the laser area (28x10-7m2) and the penetration depth (1x10-3 m)41 The density of the sample was measured at 064 gcm3 indicating that 18x10-4g (13x10-6 moles) of adenine produced the normal Raman signal The enhancement factor EF is defined by the following equation
EF = (ISERSINR)bull(MNRMSERS) bull(PNRPSERS) bull(TNRTSERS)12
where I is the spectral band intensity (here 735 cm-1) M is the sample mass P is the incident laser power and T is the measurement time (or number of scans) For the E-SERS measurement the enhancement factor is 22x105 (01780184) bull (18x10-487x10-10)) while the SG-SERS enhancement factor is 10x105 (0160184) bull(18x10-422x10-8) bull(75075) bull(315)12) The lower enhancement for the SG-SERS may be real or the available surface of the silver embedded in the sol-gel may have been overestimated In addition to enhancing the Raman scattering efficiency to an extent similar to E-SERS the SG-SER medium also yields an identical shift of the adenine ring-breathing mode from 725 cm-1 in the normal Raman to 735 cm-1 Furthermore in the course of optimizing the E-SERS sample conditions it was found that pH influenced the adenine interaction with the silver surface (Figure 2) In particular the relative band intensities of the pyrimidine ring skeletal vibrations at 1270 and 1375 cm-1 and the imidazol ring skeletal vibration at1335 cm-1 change At pH 4 adenine is protonated presumably the imidazol ring since the band at 1335 cm-1 increases in intensity while the pyrimidine bands are virtually absent Conversely at pH 10 the imidazol band decreases in intensity while the pyrimidine bands appear It is worth noting that the ring-breathing mode at 735 cm-1 changes little between pH 4 and 10 suggesting that the skeletal changes are more a function of molecule-plasmon interactions than reorientation of the molecule on the surface Measurements of the identical pH series of adenine samples by SG-SERS yielded virtually identical spectral changes This suggests that the sol-gel does not influence the measurement This is critical to reproducing measurements and performing quantitative analysis Next the remaining NA bases were measured by both E-SERS and SG-SERS and compared Previously we examined the optimum pH and electrode potentials for E-SERS measurements to determine if a common pH could be used that yielded good sensitivity for all the bases and if variations in potential could be used to provide an added degree of selectivity between the bases Primarily it was found that high quality spectra were obtained between pH 7 and 95 and that cytosine and uracil were best enhanced at potentials positive of the potential-of-zero charge (pzc ca 065VSCE for Ag) guanine and thymine near the pzc and adenine negative of the pzc In all cases the ring-breathing modes were the most intense and in general could be used to identify the NA bases (Figure 3 Table 1) Specifically adenine has an intense band at 735 cm-1 cytosine at 797 cm-1 guanine at 653 cm-1 thymine at 784 cm-1 and uracil at 800 cm-1 The adenine cytosine guanine and thymine bands are sufficiently separated that their contributions to DNA should be determinable Although adenine and guanine contributions to RNA should also be determinable cytosine and uracil are highly overlapped and unfortunately share the same potential dependence Alternate unique bands at 1183 cm-1 for cytosine and 1275 cm-1 for uracil might be suitable for calculating contributions The SG-SER spectra of the remaining NA bases faithfully reproduced the E-SER spectra In particular the primary identifying bands occur at virtually the same wavenumbers (see Table 1) However the spectra for both cytosine and thymine contain an intense band at ca 1040 cm-1 Initially this was attributed to the pH buffer but samples prepared without either the buffer or the 01M KCl electrolyte yielded identical spectra containing this band In fact the E-SER and SG-SER spectra of thymine are virtually identical except for this band Also the SG-SERS of guanine contains an intense band at 1551 cm-1 that is not observed in the E-SER spectrum This band may be due to a moderately intense band at 1553cm-1 in the normal Raman spectrum that is SG-SER active It was also found that the SG-SERS of cytosine was considerably better than the E-SERS while uracil showed the opposite relationship It is also worth noting that all of the SG-SERS were obtained with 110th the laser power Most importantly the primary ring-breathing modes in the SG-SER spectra are sufficiently intense and unique to be used in determining contributions to DNA and RNA as outlined above
SPIE 2001-4575
67
Table 1 Comparison of E-SER and SG-SER Spectral Band Positions for the NA Bases and Adenosine Monophosphate
Adenine Cytosine Guanine Thymine Uracil AMP E-SERS SG-SERS E-SERS SG-SERS E-SERS SG-SERS E-SERS SG-SERS E-SERS SG-SERS E-SERS SG-SERS1647 1638 1634 1656 1655 1630 1587 1585 1510 1580 1551 1539 1456 1456 1465 1460 1462 1480 1453 1459 1394 1398 1425 1431 1435 1399 1404 1392 1374 1375 1373 1383 1370 1335 1332 1311 1307 1333 1331 1353 1348 1331 1329 1265 1273 1280 1292 1278 1276 1275 1279 1271 1183 1195 1222 1232 1221 1219 1204 1205 1180 1144 1097 1033 1029 1038 1040 1035 1051 1037 1041 1035 963 963 957 1001 1000 961 944 884 819 817 859 866 735 737 797 799 784 782 800 800 72738 742 630 630 653 664 667 684 603 602 590 611 466 561
Bands unique to SG-SERS The next chemical to be analyzed by both E-SERS and SG-SERS was adenosine monophosphate (AMP) The E-SER spectrum yields bands due to the adenine chemical functionality at 727 961 1233 1279 1331 1381 and 1486 cm-1 In addition phosphate bands are observed at 860 1097 1453 1587 and 1705 cm-1 (Figure 5) Other researchers have noted that the ribose component does not appear to contribute to the spectrum19 The AMP spectrum also changes as a function of potential As the electrode is swept more positive (here from -09 to -03VSCE) the phosphate bands at 860 1097 1453 and 1587 cm-1 increase in intensity compared to the adenine bands while a band at 1705 cm-1 appears The adenine bands at 1233 1381 and 1486 cm-1 virtually disappear These potential dependent spectral changes are consistent with earlier studies that show that phosphate is attracted to silver at potentials positive of the pzc but repelled at potentials negative of the pzc19
Wavenumbers (∆cm-1) Figure 3 E-SERS of A) 21x10-3M cytosine at -03VSCE 1000 scans B) ~10x10-5M guanine at -06VSCE 500 scans C) 23x10-3M thymine at -06VSCE 500 scans and D) 12x10-3M uracil at -093VSCE 500 scans All spectra at pH 92 750 mW 1064 nm at 8 cm-1
Wavenumbers (∆cm-1) Figure 4 SG-SERS of A) 21x10-3M cytosine 200 scans B) ~10x10-5M guanine 200 scans C) 23x10-3M thymine 200 scans and D) 12x10-3M uracil 500 scans All spectra at pH 92 75 mW 1064 nm at 8 cm-1
A
B
C
A
B
C
D D
SPIE 2001-4575
68
The SG-SER spectrum of AMP is considerably different The adenine bands virtually disappear except for the two primary bands which shift to742 and 1329 cm-1 While the phosphate band at 1459 cm-1 has gained considerable intensity In addition two new intense bands appear at 684 and 1539 cm-1 as well as a moderately intense band at 1180 cm-1 The SG-SER spectrum has greater similarity to the E-SER spectrum at -03VSCE and suggests that the silver particles embedded in the sol-gel behave as if at a potential positive of the pzc RNA samples extracted from E coli B subtilis and S aureus were next examined by both E-SERS and SG-SERS E-SER spectra of these samples yielded quality spectra in 10 minutes in which all of the major features can be identified (Figures 6 and 7) This includes guanine at 650 cm-1 adenine at 791 cm-1 cytosine and uracil combining at 790 cm-1 and phosphate at 1100 1335 (in combination with adenine and guanine) 1465 and 1570 cm-1 Surprisingly adenine which demonstrated the greatest surface-enhanced Raman effect does not dominate the ring-breathing mode portion of the spectrum The intensities of the other base-pairs bands are of the same order of magnitude This suggests that when the base-pairs are linked together as in RNA they are enhanced in concert In fact the relative intensities are very similar to a normal Raman spectrum of E coli RNA which shows the combined cytosine and uracil band at ca twice the intensity of the adenine band and ca four times the intensity of the guanine band Unfortunately this means that the independent enhancement factors for the NA bases can not be used to estimate relative concentrations For example the relative 791 and 734 cm-1 bands for B subtilis would indicate that the cytosine andor uracil concentration was at least 20 times the adenine concentration whereas each of the four RNA bases are known to contribute 15-35 Nevertheless it is worth noting that the three RNA samples yield different relative band intensities that were reproduced in numerous measurements Although the relative concentrations of the NA bases for these samples have not been determined these differences can be quantified If it is assumed that the 650 cm-1 band represents 25 guanine the 791 cm-1 band represents 25 adenine and the 790 cm-1 50 cytosine plus uracil in the E coli RNA spectrum then the relative concentrations can be estimated for the other RNA samples To aid this calculation the three spectra were normalized to the phosphate band at 1100 cm-1 which has been shown to correlate to the total phosphate concentration and can be used as an internal standard In addition a simple baseline correction was applied (Figure 7) This yields 15 adenine 30 guanine and 55 cytosine plus uracil for B subtilis RNA and 18 adenine 25 guanine and 57 cytosine plus uracil for S aureus RNA The average SN of these measurements was 26 with an average error of 8 of the value (SplusmnN) It is also worth noting that the three RNA spectra show a marked shift in a band near 825 cm-1 This band is assigned to the symmetric stretch of the O-P-O ester linkage9 The band appears at 815 cm-1 for S aureus shifting to 820 cm-1 for B subtilis and 830 cm-1 for E coli Others have used the normal Raman intensity of the band at 815 cm-1 as a direct indication of the amount of A-class helix present while the intensity of the band at 830 cm-1 has been used as a direct indication of the amount of B-class helix present However the latter is more associated with DNA than RNA
Wavenumbers (∆cm-1) Figure 5 E-SER spectra of 020 mgmL adenosine monophosphate at A) -03 and B) -09VSCE and C) SG-SER spectra Conditions sample in 01M KCl buffered to pH 92 A) and B) 750 mW C) 75 mW of 1064 64 scans (1-min) at 8 cm-1
Wavenumbers (∆cm-1) Figure 6 E-SERS of 01 mgmL RNA from E coli 02 mgmL RNA from B subtilis and 02 mgmL RNA from S aureus Conditions 01M KCl pH 92 -03VSCE 750 mW of 1064 nm 640 scans (10 min) at 8 cm-1
AMP RNA
E coli A
B
C
B subtilis
S aureus
SPIE 2001-4575
69
SG-SER spectra of reasonable quality were also obtained for E coli and B subtilis especially the latter (Figure 8) However the spectra differ substantially from the E-SERS of the same samples Both SG-SER spectra are dominated by adenine at 735 cm-1 and a band at 1030 cm-1 Although unassigned the latter does appear in the RNA E-SER spectra Bands at 1105 and 1565 cm-1 are likely due to phosphate while bands at 1320 and 1470 cm-1 are less confidently assigned to phosphate They are significantly less intense and somewhat shifted from their SG-SERS counterparts (1335 and 1455 cm-1) A band at 670 cm-1 may be due to guanine which was observed at 664 cm-1 for SG-SERS of the pure sample However the SG-SER spectrum of AMP also had an intense 667 cm-1 band A number of other bands occur at 890 1070 1165 1245 1290 1420 1505 cm-1 and remain unassigned The SG-SER spectra are somewhat disappointing in that only adenine and guanine contributions can be positively identified This limits the ability to determine relative NA base concentrations and distinguish bacterial RNA However several of the unassigned bands may be due to the bases (eg 1030 and 1420 cm-1 due to cytosine) Further experiments will be required to clarify this point
Ecoli
B subtilis
S aureas
P G
A C+U
OPO
A B
Wavenumbers (∆cm-1) Wavenumbers (∆cm-1) Figure 7 SER spectra of RNA from A) B subtilis with contributions indicated and B) E coli B subtilis and S aureus with baseline correction and peak positions used to calculate contributions indicated G = guanine A = adenine C+U = cytosine plus uracil P = phosphate (backbone) OPO = phosphate ester linkage (A- vs B-class helix)
Wavenumbers (∆cm-1) Wavenumbers (∆cm-1) Figure 9 E-SER (-03VSCE) and SG-SER spectra of RNA from A) E coli and B) B subtilis Sample conditions as in Figure 6 E-SER spectra at 750 mW SG-SERS at 75 mW
A B E-SERS
SG-SERS
E-SERS
SG-SERS
SPIE 2001-4575
70
A final comparison was made between the two SER methods by measuring dipicolinic acid (DPA) This chemical may be invaluable as a test for spore forming bacteria specifically B anthracis 50 to 90 of B anthracis sporilates During spore formation dipicolinic acid is synthesized and once completed 10-15 of the dry spore weight is composed of the Ca2+ complex located in the spore core42 Heating in water can be used to initiate germination at which point the exosporium breaks and releases the Ca dipicolinate which becomes dipicolinic acid in water The structure of this chemical strongly
4 CONCLUSIONS Towards the goal of developing a method and technology to rapidly detect and identify bioagents we have been investigating surface-enhanced Raman spectroscopy as a tool to measure relative concentrations of nucleic acid bases in RNA and determine bioagent taxonomy Initially we investigated E-SERS since this method has been extensively researched and the optimum sample conditions are well developed However this method requires a three-electrode sample cell and electrolyte solution Incorporation of an electrolytic cell into sample systems used to analyze for BWA as aerosols in water or on surfaces can be designed using flow injection analysis technologies but cross-contamination and plugging of sample lines seems inevitable For this reason we also investigated metal-doped sol-gels as a SER-active medium Previous studies have shown this material to be active in all solvents particularly water capable of continuous measurements in flowing systems and reproducible (quantitative) between coated sample vials Here we compared SG-SER spectra to traditional E-SER spectra of the nucleic acid base pairs adenosine monophosphate and RNA High quality spectra of adenine cytosine guanine thymine and uracil were obtained by both E-SERS and SG-SERS Both methods yielded very similar spectra for the NA bases including a pH dependent study of adenine Enhancement factors and detection limits for adenine were determined as 2x105 and 16x10-11g and 1x105 and 12x10-10g for E-SERS and SG-SERS respectively (normalized to 75 mW and 10-min acquisition time) Fifty percent of the silver particle surface area in the sol-gel matrix was assumed covered by adenine which may have been overestimated yielding a lower EF and higher detection limit It should also be realized that each E-SER spectrum required several attempts to optimize the measurement conditions (pH electrode potential etc) While each SG-SER spectrum involved no sample preparation and often represents the first and only attempt to make the measurement Quality spectra of RNA extracted from Escherichia coli Bacillus subtilis and Staphylococcus aureus were obtained by E-SERS that were easily interpreted Bands due to adenine guanine cytosine plus uracil and phosphate were identified The SER band intensity of the NA bases in the RNA samples were of the same order of magnitude suggesting that their interaction with the silver surface is concerted as is their Raman enhancement Interestingly the relative SER band
Wavenumbers (∆cm-1) Figure 10 A) Raman spectrum of solid dipicolinic acid B) Ax20 C) electrolytic SERS of 6x10-3 M dipicolinic acid in 01 M KCl at a potential of +07VSCE and pH of 4 and D) sol-gel SERS of 6x10-3 M dipicolinic acid Conditions for A and C as in Figure 1 C) 100 mW of 1064 nm 50 scans 8 cm-1
A B (Ax20)
C
D
suggested that it would be SER active However the E-SER spectrum was unstable and varied considerably as a function of potential A consistent spectrum was obtained at +06VSCE (Figure 10) This potential is not recommended for measurement because the surface is actively dissolving in solution The SG-SER spectrum was considerably more stable of higher quality and easily reproduced Bands at 660 825 1010 1390 1430 1570 1590 and 3075 cm-1 were observed Enhancement factors were determined for the two media using the symmetric ring stretching mode at 995 and 1010 cm-1 for the normal Raman and SER spectra respectively E-SERS yielded an EF of 5x103 while SG-SERS yielded an EF of 2x105 for DPA The SN of the latter suggests a detection limit of 20x10-10g (based on adenine coverage 75 mW and 10-min) The differences in SER activity for these two media may be attributed to the combined electrolytic potential of the solution chemical and metal15 Again the E-SERS suggests that the SG-SERS is at a potential positive of the pzc While the instability in the E-SERS may also be associated with surface interactions of two carboxylic acid groups of dipicolinic acid during the ORCs
DPA
SPIE 2001-4575
71
intensities for RNA extracted from E coli are very similar to those measured by normal Raman spectroscopy Although the relative percent that each of the NA bases contributed to each RNA sample was not determined reproducible band intensities allowed noting the following trends The percent adenine decreases while the combined percent cytosine and guanine increase for both B subtilis and S aureus compared to E coli Quality spectra were also obtained for the RNA samples by SG-SERS but only a few bands were readily identified Calculations of NA base concentrations by SG-SERS will require further research In light of recent events we summarize the capabilities of these SERS media in terms of rapid detection of B anthracis and dipicolinic acid However these capabilities must be qualified First and foremost the level to which SERS can distinguish bacteria or viruses has not yet been determined Development of a database of both DNA and RNA base concentrations for BWAs and common bacteria to establish the level of taxonomic identification is ongoing Second rapid collection of aerosol water or surface samples is being addressed by others who report trapping particles on filters from 100 liters of air per minute Third although not presented here we have developed methods to extract RNA or DNA from cells and spores for SER analysis within 10 minutes Finally we assume a detection limit of 3600 spores per 100 liters of air is required although a 50 lethal dosage of anthrax has not been established With these qualifications a mass detection limit for RNA using SERS is estimated as follows A single measurement is performed in ca 20 minutes (140 liters collected in 14 min RNA extracted in 8 min spectral acquisition and analysis in 10 min) The average human breaths 7 liters per minute therefore the analyzer must at the very minimum detect 5000 spores in 140 liters of air One spore is approximately 2x10-18m3 (1x1x2 microm) and if a density of 075 gcm3 is assumed this corresponds to a mass of 15x10-12g Each spore contains 4-12 RNA or 12x10-13g RNA for 8 If we assume 23 of the RNA can be isolated for analysis during lysis then the proposed instrument must be able to detect 4x10-10g RNA from 5000 spores per 70 liters of air within 10 minutes As noted above the mass detection limits for adenine were estimated at 16x10-11g and 12x10-10g for E-SERS and SG-SERS respectively Although these detection limits suggests that RNA from 5000 spores is detectable with the current instrumentation it is highly likely that only a portion of an RNA segment (eg 120-nucleotide 5S rRNA) is in contact with the metal surface and will contribute to the SER effect The SN for the RNA spectra were 110th of the average SN for the four individual RNA bases suggesting a 10 contribution Furthermore effective taxonomy will likely require knowing the NA base concentrations to 1 of the value (eg 25plusmn025) Again the average measurement error for the bases is 12 These values suggest that the E-SERS is within a factor of 4 of the required detection limit whereas the SG-SERS detection limit must be improved by 25 times The same arguments can be applied to the detection of dipicolinic acid If we assume a spore releases 10 by weight DPA during germination then the proposed instrument must be able to detect 75x10-10g DPA from 5000 spores per 70 liters of air within 10 minutes The detection limit for SG-SERS was estimated at 20x10-10g and suggest that the vials are suitable to perform a rapid screen for anthrax A series of concentration dependent measurements are currently being performed to verify this assertion Finally we note that the measurements performed here employed an FT-Raman spectrometer This instrumentation was chosen over dispersive Raman instrumentation for the high wavelength accuracy afforded by the HeNe reference laser (Connes Advantage43) which would allow reliable spectral subtraction matching of observed spectra to stored library spectra and confident use of chemometric approaches Such data analysis is likely to be required to enhance BWA identification However this instrumentation sacrifices sensitivity due to a less efficient detector (InGaAs vs Si) less efficient Raman scattering and less efficient generation of plasmon modes Substantial improvements can be obtained using visible excitation and Si detection and these measurements are underway
5 ACKNOWLEDGEMENTS The authors are grateful to Drs D Cookmeyer and S Tove of the US Army Research Office (Contract Number DAAH04-96-C-0078) for their interest and support of this research The authors would also like top acknowledge Dr R Yin and J Jensen for supporting development of the metal-doped sol-gels (Contract Number DAAD13-01-C-0019) They also thank Dr Wilfred H Nelson for assistance in spectral interpretations
SPIE 2001-4575
72
6 REFERENCES 1 Roberts WL and Rainey PM Clin Chem 39 1872-1877 (1993) 2 Pasechnik VA CC Shone and P Hambleton Bioseperations 3 267-283 (1993) 3 Jackson PJ ME Hugh-Jones DM Adair G Green KK Hill CR Kuske LM Grinberg FA Abramova and P
Keim Proc Natl Acad Sci 95 1224-1229 (1998) 4 Jensen JL NC Wong W R Loerop SPIE 4775-03 (2001) 5 Snyder A Pet al SPIE 3853-15 (1999) 6 Danosky T R and McFadden P N in press (1997) 7 Woodruff WH Farquharson S Science 201 831-833 (1978) 8 Lord RC and Thomas GJJr Spectrochemica Acta 23A 2551-2591 (1967) 9 Thomas GJJr Biochim Biophys Acta 213 417-423 (1970) 10 Placzek G Handbuch der Radiologie 2 EMarx ed Akademische Verlagagescellschatt Liepzig 1934 UCRL
Trans No 526 (1959) 11 Jeanmaire DL and RP Van Duyne J Electroanalytical Chem 84 1-20 (1977) 12 Chada S Manoharan R Moenne-Loccoz P Nelson WH Peticolas WL and Sperry JF Applied Spectroscopy
47 38-43 (1993) 13 Manoharan R Ghiamati E Chada S Nelson WH and Sperry JF Applied Spectroscopy 47 2145-2150 (1993) 14 Todd EA Morris MD Applied Spectroscopy 48 545-548 (1994) 15 Wentrup-Byrne E Sarinas S and Fredericks PM Applied Spectroscopy 47 1192-1197 (1993) 16 Weldon MK VR Zhelyaskov and MD Morris Applied Spectroscopy 52 265-269 (1998) 17 Kneipp K and J Fleming J Mol Structure 145 173-179 (1986) 18 Pothier NJ and Force RK Applied Spectroscopy 46 147-151 (1992) 19 Ervin KM E Koglin JM Sequaris P Valenta and HW Nurnberg J Electroanal Chem 114 179-194 (1980) 20 Kim SK TH Joo SW Suh and MS Kim J Raman Spectrosc 17 381-386 (1986) 21 Pothier NJ and Force RK Analytical Chemistry 62 678-680 (1990) 22 Kneipp K Y Wang RR Dasari and MS Feld Applied Spectroscopy 49 780-784 (1995) 23 Nie S and Emory SR Science 275 1102 (1997) 24 Graham D WE Smith AMT Linacre CH Munro ND Watson and PC White Analytical Chemistry 69 4703-
4707 (1997) 25 Fleischmann M PJ Hendra and AJ McQuillan Chem Phys Lett 26 163-166 (1974) 26 Pettinger B J Chemical Phys 85 7442-7451 (1986) 27 Surface-Enhanced Raman Scattering Section Four Theory SPIE MS 10 M Kerker and B Thompson Eds (1990) 28 Wang D-S and Kerker M SPIE (M Kerker and B Thompson Eds) MS 10 417-429 (1990) 29 Weaver MJ Farquharson S Tadayyoni MA J Chem Phys 82 4867-4874 (1985) 30 Norrod KL Sudnik LM Rousell D and Rowlen KL Applied Spectroscopy 51 994-1001 (1997) 31 Farquharson S Weaver WJ Lay PA Magnuson RH and Taube H J Am Chem Soc 105 3350-3351 (1983) 32 Lee PC and Meisel D ldquoAdsorption and Surface-Enhanced Raman of Dyes on Silver and gold Solsrdquo J Phys Chem
86 3391-3395 (1982) 33 Li Y-S and Wang Y Applied Spectroscopy 46 142-146 (1992) 34 Bello JM DL Stokes and T Vo-Dinh Analytical Chemistry 61 1779-1783 (1989) 35 van Duyne RP J C Hulteen D A Treichel M T Smith M L Duval and T R Jensen J Phys Chem B1033854-
3863 (1999) 36 Farquharson S and W W Smith W H Nelson and J F Sperry SPIE 3533-27 207-214 (1998) 37 Lee Y H W Smith S Farquharson H C Kwon M R Shahriari and P M Rainey SPIE 3537 252-260 (1998) 38 Lee Y-H S Farquharson and P M Rainey SPIE 3857 76-84 (1999) 39 Current Protocols in Molecular Biology Wiley Interscience1003-1006 (1987) 40 Farquharson S Smith W Carangelo R C and Brouillette C SPIE 3859 14-23 (1999) 41 Chase D B and JF Rabolt Fourier Transform Raman Spectroscopy Acad Press Ch1 p 131 (1994) 42 Brock TD MT Madigan JM Martinko and J Parker Biology of Microorganisms 7th Ed Prentice Hall p 76-80
(1994) 43 Connes J Rev Opt Theor Instrum 40 45 (1961)
166
Chemical agent identification by surface-enhanced Raman spectroscopy
Stuart Farquharson and Paul Maksymiuk
Real-Time Analyzers 87 Church Street East Hartford CT 06108
Kate Ong and Steven D Christesen US Army SBCCOM Aberdeen Proving Ground MD 21010
ABSTRACT The recent distribution of anthrax through the US postal system and the subsequent infection and death of several postal and national media employees amplifies the need for methods to rapidly detect identify and quantify this and other chemical and biological warfare agents The US military has also identified water supplies as a likely method of warfare agent deployment and is funding the development of a Joint Service Agent Water Monitor (JSAWM) In an effort to aid military personnel and the public at large we are developing a portable analyzer capable of identifying and quantifying chemical agents rapidly either on-demand or continuously The approach is based on the ability of Raman spectroscopy to identify molecular structure through the abundant vibration information provided in spectra and the ability of surface-enhanced Raman spectroscopy (SERS) to detect extremely low concentrations (eg part-per-billion) through the enhancement of Raman scattering by six orders of magnitude or more A key element to the analyzer design is a new SER active medium that is capable of quantitative reversible measurements The medium consists of silver or gold nanoparticles incorporated into a sol-gel matrix The porous silica network offers a unique environment for stabilizing SER active metals and the high surface area increases the interaction between the analyte and metal particles Here we present the use of new sol-gels that also selectively enhance chemicals based on polarity and charge Base-line measurements of chemical agents and their hydrolysis products are presented and compared to the JSAWM goal of 30 micrograms per liter detection Keywords Chemical warfare agent hydrolysis product SERS Raman spectroscopy sol-gel nanoparticle
1 INTRODUCTION Since September 11 2001 the threat of terrorist attacks and biological warfare within US borders has become a sobering reality The simplicity in manufacturing ease of deployment and the relatively low cost of chemical warfare agents (CWAs) raises public concern that they may also be used by terrorists Indeed terrorists released sarin (GB) in the Tokyo subway in 19951 Countering terrorism and terrorist attacks requires recognizing likely deployment scenarios and having the required technology to rapidly detect the deployment event One method of deployment has been long identified by the US military distribution through water supplies To counter this threat the Department of Defense is funding or monitoring the capabilities of 200 technologies with the goal of developing a Joint Service Agent Water Monitor (JSAWM) that is field portable2 The JSAWM must be capable of identifying the chemical agents with no false-positives and quantifying the agents at microgram per liter concentrations within 10 minutes (Table 1)3 This includes the analysis of drinking water supplies distribution and storage systems as well as potable water supplies The earliest technologies developed for CWA detection were based on electrochemical ionization or colorimetric analysis (eg phosgene tape) Although these analyzers were easy to use they were not generally agent specific and suffered from false-positives4 More traditional laboratory methods have also been investigated and in particular combined gas chromatography and mass spectrometry (GCMS) has been very successful at eliminating false-positives56 However GCMS requires extraction repeated calibration and long analysis times (typically 20 to 60 minutes)6 making it labor intensive and less than desirable for field use More rapid analysis of agents in the solid liquid and gas phase has been To whom correspondence should be addressed emailfarqureal-time-analyzerscom
Vibrational Spectroscopy-based Sensor Systems Steven D Christesen Arthur J Sedlacek III EditorsProceedings of SPIE Vol 4557 (2002) copy 2002 SPIE middot0277-786X02$1500
167
Table 1 Chemical Agent Structures Hydrolysis Half-lives and JSAWM Thresholds Agent Short-Hand Chemical Structure Hydrolysis
Half-Life JSAWM
Thresholds
Sarin (GB) F-[O=P-CH3]-O-CH(CH3)2 213 hours 32 microgL Soman (GD) F-[O=P-CH3]-O-CH(CH3)-(C-(CH3)3) 23 hours 32 microgL Tabun (GA) (CH3)2-N-[O=P-CN]-O-C2H5 41 hours 32 microgL VX C2H5O-[O=P-CH3]-S-(CH2)2-N-(CH(CH2)2)2 821 hours 32 microgL EA2192 HO-[O=P-CH3]-S-(CH2)2-N-(CH(CH2)2)2 gt9 years 32 microgL Mustard (H) ClCH2CH2-S-CH2CH2Cl encapsulates 47 microgL Lewisite (L) ClCH=CH-As-Cl2 rapid 27 microgL HCN HCN rapid 20 mgL BZ C7NH12-O-[C=O]-COH(C6H5)2 23 microgL T-2 Toxin 87 microgL at pH 7 to 75 and 20 to 25 oC
demonstrated by vibrational spectroscopy7-10 Hoffland et al7 reported infrared absorbance spectra and absolute Raman cross sections for several chemical agents while Christesen measured Raman cross sections for sarin tabun mustard gas and VX (ethyl S-2-diisopropylamino ethyl methylphosphonothioate)11 Again however these techniques also have limitations Raman spectroscopy is simply not a very sensitive technique and detection limits are typically 01 (1000 ppm) While infrared spectroscopy would have limited value in analyzing poisoned water since the very strong infrared absorption of water would obscure most other chemicals present Nevertheless efforts to overcome these limitations have been demonstrated Braue and Pannella8 quantified the G-series nerve agents (tabun sarin and soman) in terms of infrared attenuated total reflectance using a circle-cell And Alak and Vo-Dinh demonstrated the possibility of surface-enhanced Raman spectroscopy (SERS) to identify CWAs by measuring several organophosphonates that simulate the nerve agents12 However quantitative measurements have not been demonstrated for the SER-active material used (silver coated on alumina particles) or other SER-active media13 Recently we developed silver-doped sol-gels to promote the SER effect14-17 The porous silica network of the sol-gel matrix offers a unique environment for stabilizing SER-active metal particles and the sol-gel provides a high surface area that effectively increases the number of molecules observed within the Raman scattering volume The silver-doped sol-gels have been coated on the internal walls of standard glass vials to allow rapid SER analysis of small sample quantities (lt 01 mL) without preparation We have used p-aminobenzoic acid (PABA) as a test chemical to demonstrate surface enhancements greater than 106 reversible measurements in a flowing system reproducible measurements from vial-to-vial and batch-to-batch and measurements in multiple solvents including water14-17 Recently we used these vials to measure Tabun (GB) and Sarin and several hydrolysis products pinacolyl methyl phosphonate (PMP from Soman) and methyl phosphonic acid (MPA from all G-agents Figure 1) Although a number of unique vibrational bands are observed (eg C-N stretch doublet and P-C stretch) the G-agents were only observed for 5 concentrations and all spectra required baseline corrections Figure 1 Surface enhanced Raman spectra of ~5 vv A) Tabun and B) Sarin C) 1 vv PMP and D) 10 ppm MPA using sol-gel sample vials 785 nm excitation 1-min scan and CCD detection Performed at Aberdeen Proving Ground
C A
Wavenumber (∆cm-1) Wavenumber (∆cm-1)
D B
790
545 C-N
2135 2190 P-C 770
760 1290
Proc SPIE Vol 4577
168
Nevertheless MPA was readily observed for a 10 ppm sample with an estimated detection limit of 04 ppm (based on a signal-to-noise ratio of 3 for the 760 cm-1 band intensity) This measurement provides encouragement in that SERS may satisfy the needs of a JSAWM Furthermore MPA is also a hydrolysis product of VX and V-gas and EA2192 (Figure 2) and may prove a valuable indicator of agent usage Figure 2 Hydrolysis of Sarin to form hydrofluoric acid (HF) methylphosphonic acid 1-methylethyl ester (MPAMME) methyl phosphonic acid (MPA) and 2-propanol With this initial albeit modest success we began analyzing chemicals with various sol-gel compositions that we have been developing Here we describe four sol-gel compositions that select for 1) polar-positive 2) polar-negative 3) weakly polar-positive and 4) weakly polar-negative chemical species The ability of these sol-gels to select and enhance Raman scattering is described for several test chemicals and MPA
2 EXPERIMENTAL The chemicals analyzed as well as all chemicals used to prepare the metal-doped sol-gels were obtained at their purest commercially available grade from Aldrich (Milwaukee WI) The sol-gel designed to select for polar-negative species was prepared from a silver amine complex tetramethyl orthosilicate (TMOS) and methanol After mixing 02 mL of the sol-gel solution was transferred into a glass vial (2 mL) dried and heated The incorporated silver ions were then reduced using dilute sodium borohydride The vials were washed and dried prior to the addition of a sample solution In a similar manner the sol-gel designed to select for polar-positive species was prepared from a gold salt TMOS and methanol The sol-gel designed to select for weakly polar-negative species was prepared from a silver amine complex tetraethyl orthosilicate (TEOS) and methanol And the last sol-gel designed to select for weakly polar-positive species was prepared from a gold salt TEOS and methanol All samples were prepared in a chemical hood and transferred into plain or SER-active vials for analysis Normal Raman spectral measurements employed 1-mL pure samples that were placed in a 1-cm3 cuvette and weighed This yielded a powder density that allowed accurate calculation of molecules in the optical collection field SERS measurements employed 1-mg sample per mL water concentrations unless otherwise stated Once prepared a 01 mL sample was placed into one of the four selective sample vials which in turn was placed into the sample compartment of a Raman spectrometer for analysis A prototype Fourier transform Raman spectrometer (RTA) was used for these measurements18 The system consisted of a NdYAG laser (Brimrose) for excitation at 1064 nm an interferometer built by On-Line Technologies (OLT East Hartford CT) for frequency separation an uncooled InGaAs detector for signal detection (RTA) and an Intel 400 MHz Pentium II based laptop computer (Dell Round Rock TX) for interferometric control data acquisition (OLT) and analysis (LabVIEW by National Instruments Austin TX) Additional components included a Notch filter (Kaiser Ann Arbor MI) and interferometer entrance and exit optics (Edmund Scientific Barrington NJ) Fiber optics were used to deliver the excitation beam to the sample and the scattered radiation to the interferometer (1 meter lengths of 200 and 365 micron core diameter respectively Spectran Avon CT) A second Notch filter (Kaiser) was used as a beam splitter to direct the excitation beam along the same axis as the collected radiation A microscope object (20x04 Newport Irvine CA) was used to focus the beam into the sample and to collect the scattered radiation back along the same axis In this co-axial backscattering arrangement the excitation beam was passed through the outside of a glass vial and focused onto the silver-doped sol-gel film (01-03 mm thickness) containing the sample
3 RESULTS AND DISCUSSION p-aminobenzoic acid (PABA) and phenyl acetylene (PA) and were used to refine the selectivity and SER-activity of the four different metal-doped sol-gels PABA is a popular chemical used to evaluate the performance of SER-active media Here the polar end groups can be used to test selectivity of the polar-negative and polar-positive sol-gels PA is essentially non-
2O+ H HF + +OH
OH3H C
OP
OF
CH
CH3
3
3H C
OCP
OOH
CH
CH3
3
3H C
O
CPHO
CH
CH
3
3
C
Sarin MPAMME MPA 2-propanol
Proc SPIE Vol 4577
169
polar but a high electron density in the cylindrical π cloud around the carbon-carbon triple bond allows testing the selectivity of the weakly polar-negative and weakly polar-positive sol-gels As Figure 3 illustrates PABA passes through the polar sol-gel and is enhanced by either the silver or gold particles At 1 mgml the concentration of neutral PABA is ca 20 times that of the ionized form (pKa = 48) For electropositive silver the PABA anion is expected to interact through the carboxylate group and the associated vibrational modes are expected to dominate the spectrum Conversely for electronegative gold either form of PABA is expected to interact through the amine group The clear differences in our spectra support this expectation Furthermore bands at 840 and 1405 cm-1 assigned to a COO- bend and stretch respectively are significantly more intense for silver than gold Additional bands at 1140 and 1195 cm-1 are assigned to CH bending modes while bands at 1450 1500 and 1605 cm-1 are assigned to ring vibrational modes A very similar SER spectrum for PABA on a silver-coated alumina substrate has previously been reported with similar assignments19 For the gold-doped sol-gel new bands appear at 690 1355 and 1585 cm-1 The first band is assigned to a ring-H bending mode the second band to a ring-N- stretching mode and the third band to a possible NH2 scissors mode or ring mode The second band is not observed in the normal Raman spectrum but infrared bands occur at this frequency for aromatic ring-secondary amine stretching modes The scissors mode occurs at this frequency in Raman spectra for several chemicals but is absent in the PABA Raman spectrum Alternatively this mode may be the1600 cm-1 ring mode that has been shifted by the gold interaction Again a very similar SER spectrum of PABA has been reported but surprisingly using silver (colloids)2021 not gold as the enhancement medium These researchers also assumed the primary interaction of PABA with silver was through the carboxylate anion and made assignments accordingly For example they assigned the 1359 cm-1 to a COO- stretch not to the amine group as we have They also favor the ring stretching mode assignment for the 1582 cm-1 band Finally it should be said that other researchers have argued that the most dominant band in the SER spectra at 1450 cm-1 a ring vibration mode suggests that PABA lies flat on the surface and the π-orbitals dominate the surface interaction22
Figure 3 SER spectra of A) PABA using polar-negative and B) polar-positive sol-gels and C) PA using weakly polar-negative and D) weakly polar-positive sol-gels PABA is 1 mgmL PA is 1 vv Spectral conditions 75 mw 1064 nm 100 scans (15 min) 8 cm-1 resolution Non-polar PA passes through the non-polar sol-gels and is also enhanced by both metals The spectra are easily understood For electropositive silver PA interacts through the cylindrical triple bond π electron cloud and a -CequivC- doublet occurs near 2000 cm-1 The interaction is reasonably strong since this band appears at 2112 cm-1 in the normal Raman spectrum For electronegative gold this interaction is unlikely and only very weak bands occur near 2000 cm-1 The remaining bands are at 1000 cm-1 1200 cm-1 doublet and 1595 cm-1 all appear in the normal Raman spectra at virtually the same frequencies and are assigned to the symmetric ring-breathing mode CH bending modes and the trigonal ring-breathing mode respectively The polarnon-polar selectivity of the polar-negative and weakly polar-negative sol-gels was tested by adding a 11 molar mixture of PABA and PA The selective enhancement is quite good (Figure 4) The spectrum obtained using the polar sol-gel represents 78 PABA and 22 PA while the spectrum obtained using the weakly polar sol-gel represents 9 PABA and 91 PA The band peak intensities at 2000 cm-1 for PA and 1450 cm-1 for PABA were used for these calculations and are expanded in Figure 4 for clarity
C A
Wavenumber (∆cm-1) Wavenumber (∆cm-1)
2NH COOH
D B
C CH
Proc SPIE Vol 4577
170
Figure 4 SERS of 11 MM of PABA and PA in A) polar-negative and C) weakly polar-negative sol-gels The lower traces compare the pure chemicals B) 1 mgml PABA in polar-negative sol-gel and D) 1 PA in weakly polar-negative sol-gel while the insets magnify the minority species for clarity (x5 in A and x10 in B) Spectral conditions as in Figure 3 Following this development of selective sol-gels that maintained SER activity we measured cyanide and MPA (Figure 5) Not surprisingly the best sensitivity for both hydrolysis products was obtained using the polar-negative sol-gel The interaction of the cyanide anion with the silver surface is sufficient to shift the CequivN stretch observed at 2080 cm-1 in the normal Raman spectrum to 2145 cm-1 in the surface-enhanced Raman spectrum Furthermore the band is substantially broadened This anion has been extensively studied by electrolytic SERS and this shift and broadening have been attributed to the formation of a tetrahedral Ag(CN)3
2- surface structure23 Figure 5 Surface-enhanced (upper traces) and normal Raman spectra (bottom traces) of A) CN- and B) MPA in silver-doped TMOS SERS conditions as in Figure 3 and 1 mgmL Note MPA yields two distinct spectra for neutral (top) and acidic pH (middle) The normal Raman spectra employed pure powders 500 scans and 900 mW of 1064 nm SER measurements of MPA with the polar-negative sol-gel yielded two unique spectral signatures that depended on solution pH (Figure 5) For more neutral solutions the P-C stretch of MPA at 762 cm-1 dominates and the CH2 stretch at 2922 cm-1 appears The SN is sufficiently high that the anti-Stokes Raman shift at -762 cm-1 is observed For deprotonated MPA an oxygen-surface mode appears at 325 cm-1 (as well as its anti-Stokes complement) suggesting a strong interaction This results in substantial enhancement of the P-O-C mode at 1051 cm-1 (upper trace) Others report that this mode dominates the infrared spectra of nerve agents measured in water8 Comparison of the two spectra suggests the following molecule-to-surface orientations The appearance of the oxygen-surface and P-O-C modes in the upper spectrum of Figure 5B indicates that the tetrahedral molecule interacts with the silver surface through the deprotonated oxygen and is oriented end-on The
A B
Wavenumber (∆cm-1)
C A
Wavenumber (∆cm-1) Wavenumber (∆cm-1)
D B
Wavenumber (∆cm-1)
Proc SPIE Vol 4577
171
dominance of the P-C and the CH2 stretches and the disappearance of the P-O-C mode in the upper spectrum suggest the molecule is oriented with the methyl-hydroxyl-hydroxyl face towards the surface However considerably more research must be performed to verify these points Both chemical forms were measured as a function of concentration to determine the onset of monolayer coverage and estimate expected detection limits (Figure 6) Below monolayer coverage the signal to concentration dependence should be linear and the SN of any spectral measurement in this range can be used to predict the detection limit In the spectra presented here the peak height was used as the signal while the noise as root-mean-squared (RMS) was measured between 4400-4600 cm-1 Since noise is distributed evenly throughout the spectrum when transformed this region was used since it does not have any contributions from signals or baseline offsets Figure 6 shows a series of spectra for MPA along with a plot of the 1050 cm-1 band intensity (the noise was identical for this series) as a function of sample concentration A clear discontinuity appears in the vicinity of 01 mgmL (19 ppm) indicating the onset of monolayer coverage A detection limit defined as a SN of 3 was calculated for the 01 and 005gmL samples at 24x10-4 and 25x10-4 gL respectively A more modest detection limit of 101x10-4 gL was obtained using the 760 cm-1 band in the second series of concentration measurements These detection limits correspond to 46 and 190 ppb for a 3-min scan and 75 mW of incident laser power
Figure 6 A) Concentration dependence of MPA SERS measured in silver-doped TMOS) B) Concentrations are 001 005 01 05 1 gL (188 94 188 94 188 ppm) I760 series (bull) and I1050 series (∆)
Enhancement factors for cyanide and methyl phosphonic acid can be determined by estimating the number of molecules contributing to the surface-enhanced and normal Raman spectra The enhancement factor EF can be defined by the following equation
EF = (ISERSINR)bull(MNRMSERS) bull(PNRPSERS) bull(TNRTSERS)12
where I is the spectral band intensity M is the sample mass P is the incident laser power and T is the measurement time (or number of scans) for the two measurements For the normal Raman spectra a cylindrical scattering volume is assumed based on the laser area (28x10-7m2 6x10-4m diameter spot) and the penetration depth (1x10-3 m)24 The density of KCN and MPA as powders were measured at 0572 and 0516 gcm3 indicating that 16x10-4 and 144x10-4 g produced the normal Raman signals in Figure 5 respectively The number of molecules contributing to the SER spectra are those on the silver particles embedded in the sol-gel The total silver surface area can be determined from the average particle size concentration and the scattering volume Previous scanning electron micrographs showed the average silver particle size to be 40 nm in diameter (335x10-23m3)17 The silver concentration is 012M based on the reactant molar concentrations and dilution factors And the scattering volume is 76x10-
11m3 again based on a cylindrical scattering volume defined by a laser area of 28x10-7m2 and a sol-gel thickness of 27x10-
4m This volume contains 123x10-6g of silver equivalent to 35x109 silver particles with a collective surface area of 18x10-
5m2 However it may reasonably be assumed that at least half of the silver surface is in contact with the sol-gel matrix and unavailable for analyte interaction If we assume monolayer coverage and that each CN molecule occupies 15x10-20m2 then approximately 62x1014 molecules or 27x10-8g of CN contribute to the SER spectrum (20x10-19m2 46x1013 molecules 74x10-9g for MPA) Accordingly the EF for cyanide equals 48x104 ((180599) bull(16x10-427x10-8) bull(90075) bull(500100)12) The EF for MPA is considerably higher at 87x106 ((60326) bull(144x10-474x10-9) bull(90075) bull(500200)12)
0
20
40
60
80
100
120
140
0 02 04 06 08 1 12[MPA] (mgmL)
I (76
0)
0
100
200
300
400
500
600
I (10
50)
Wavenumber (∆cm-1)
A B
Proc SPIE Vol 4577
172
4 CONCLUSIONS Here we present for the first time surface-enhanced Raman spectra of Tabun and Sarin obtained using silver-doped sol-gels However the inferior enhancement suggested employing SER-active sol-gels varying in composition to improve both detection limits and selectivity of the target analytes To this end we successfully demonstrated the capabilities of four sol-gels that select for 1) polar-positive 2) polar-negative 3) weakly polar-positive and 4) weakly polar-negative chemical species p-aminobenzoic acid was used to show that silver could be used to attract polar-negative chemicals or functional groups (carboxylate anion) and that gold could be used to attract polar-positive chemicals (amine) while a mixture of p-aminobenzoic acid and phenylacetylene was used to show that tetramethyl orthosilicate preferentially solvates polar chemicals and that tetraethyl orthosilicate preferentially solvates weakly polar chemicals This increased sample control was applied to cyanide and methyl phosphonic acid two hydrolysis products of chemical warfare agents Exceptional results were obtained for methyl phosphonic acid allowing measurement of 1x10-2 gL for a 3-min scan and 75 mW of incident laser power with an estimated detection limit of 245x10-4 gL and an enhancement factor of 87x106 However this detection limit is 76 times less sensitive than required for the JSAWM (32x10-6gL for the G-agents) Finally we note that the measurements performed here employed an FT-Raman spectrometer This instrumentation was chosen over dispersive Raman instrumentation for the high wavelength accuracy afforded by the HeNe reference laser (Connes Advantage25) which would allow reliable spectral subtraction matching of observed spectra to stored library spectra and confident use of chemometric approaches Such data analysis is likely to be required to identify the chemical agents as well as distinguish hydrolysis products However this instrumentation which employs 1064 nm excitation and InGaAs detection sacrifices sensitivity We believe that the measurement sensitivity can be improved by at least two orders of magnitude by using shorter laser excitation wavelengths (eg 532 nm) This would provide more efficient Raman scattering (fourth power dependence on laser excitation wavelength) more efficient generation of plasmon modes and allow using more efficient detector material (Si vs InGaAs) These modifications are underway
5 ACKNOWLEDGEMENTS The authors would like to thank Dr R Yin and J Jensen of the US Army for supporting this work (Contract Number DAAD13-01-C-0019) They would also like to thank Advanced Fuel Research for making their laboratory facilities available
6 REFERENCES 1 Tu Anthony ldquoOverview of Sarin Terrorist Incidents in Japan in 1994 and 1995rdquo 6th CBW Protection Symposium
Stockholm Sweden 10-15 May 1998 2 Jensen JL NC Wong W R Loerop SPIE 4775-03 (2001) 3 JSAWM Requirements at wwwsbccomapgeaarmymilRDAecbcrtPRODSERJSAWMjsawmhtml 4 ldquoThe Chemical Weapons Convention Redefines Analytical Challengerdquo Analytical Chemistry News amp Features June 1
397A (1998) 5 Johnston RL Hoefler CM Fargo JC and Moberley B ldquoThe Defense Nuclear Agencyrsquos ChemicalBiochemical
Weapons Agreements Verification Technology Research Development Test and Evaluation Program and its Requirements for On-Site Analysisrdquo AT-ONSITE 5-8 (1994)
6 Black RM Clarke RJ Read RW and Reid MT ldquoApplication of gas chromatography-mass spectrometry and gas chromatography-tandem mass spectrometry to the analysis of chemical warfare samples found to contain residues of the nerve agent sarin sulphur mustard and their degradation productsrdquo J Chromatography 662 301-321 (1994)
7 Hoffland LD Piffath RJ Bouck JBrdquoSpectral signatures of chemical agents and simulantsrdquo Optical Engineering 24 982-984 (1985)
8 Braue EHJ Pannella MGrdquoCIRCLE CELL FT-IR Analysis of Chemical Warfare Agents in Aqueous Solutionsrdquo Applied Spectroscopy 44 1513-1520 (1990)
9 Tseng C-H Mann CK and Vickers TJ ldquoDetermination of Organics on Metal Surfaces by Raman Spectroscopyrdquo Applied Spectroscopy 47 1767-1771 (1993)
10 Staff Reporter ldquoUS Army Tests Lidar to Detect Biological Toxinsrdquo Photonic Spectra pg 50 December 1998
Proc SPIE Vol 4577
Proc SPIE Vol 4577
173
11 Christesen SD Raman cross sections of chemical agents and simulants Applied Spectroscopy 42 318-321 (1988) 12 Alak AM and Vo-Dinh T ldquoSERS of Organophosphorous Chemical Agentsrdquo Analytical Chemistry 59 2149-2153
(1987) 13 Norrod KL Sudnik LM Rousell D and Rowlen KL ldquoQuantitative Comparison of Five SERS Substrates
Sensitivity and Detection Limitrdquo Applied Spectroscopy 51 994-1001 (1997) 14 Lee Y and Farquharson S ldquoSERS Sample Vials Based on Sol-Gel Process for Trace Pesticide Analysisrdquo SPIE 4206
140-146 (2000) 15 Farquharson S and Lee Y ldquoTrace Drug Analysis by Surface-Enhanced Raman Spectroscopyrdquo SPIE 4200-16 (2000) 16 Lee Y Farquharson S and Rainey P M Surface-Enhanced Raman Sensor for Trace Chemical Detection in Water
SPIE 3857 76-84 (1999) 17 Lee Y Farquharson S Kwong H and Shahriari M ldquoSurface-Enhanced Raman Sensor for Surface-Enhanced Raman
Spectroscopyrdquo SPIE 3537 252-260 (1998) 18 Farquharson S Smith W Carangelo R C and Brouillette C ldquoIndustrial Raman Providing Easy Immediate Cost
Effective Chemical Analysis Anywhererdquo SPIE 3859 14-23 (1999) 19 Narayanan VA JM Bello JD Stokes and T Vo-Dinh Analusis 19 307-310 (1991) 20 Laserna JJ E L Torres and JD Winefordner Analytica Chemica Acta 469-480 (1987) 21 Torres EL and JD Winefordner Analytical Chemistry 59 1626-1632 (1987) 22 Suh JS DP DiLella M Moskovits J Phys Chem 87 1540-1544 (1983) 23 Benner RE R Dornhaus R Chang and BL Laube Correlations in the Raman spectra of cyanide complexes adsorbed
at silver electrodes with voltammograms Surface Science 101 341 (1980) 24 Chase D B and JF Rabolt Fourier Transform Raman Spectroscopy Acad Press Ch1 p 131 (1994) 25 Connes J Rev Opt Theor Instrum 40 45 (1961)
APPLIED SPECTROSCOPY 351
focusing the 488 nm laser beam 2 mm deep into thebulk of the crystals to avoid contributions from potentialdepletion layers As shown in Fig 1b the dependence ofthe Raman shift on the degree of deuteration is almostperfectly linear and ts very well with D 5 22684cmR1 24526 where D is the degree of deuteration (in )and R is the spectral mean of the PO4 vibration in cm21A linear correlation coef cient of 0998 indicates an ex-cellent linear dependence of the Raman peak shift withdegree of deuteration This result shows that the shift ofthe PO 4 peak is simply caused by the linear increase inatomic mass due to isotope substitution which decreasesthe length of hydrogen-like bonds
This excellent linear dependence allows us to map thepro le of the DH exchange layer at the surface of DKDPcrystals by acquiring Raman spectra and determining theposition of the PO 4 peak for various depths This methodis preferable over other methods such as determining thestrength of the OD vibration directly (eg at 715 cm21)because the position of the most intense peak in the Ra-man spectrum can be measured more precisely than theintensity of some of the weakest peaks in the spectrumThis is demonstrated in Fig 2 where depth-dependentRaman spectra (Fig 2a) and the resulting exchange layerpro les for two DKDP crystals are shown (Fig 2b) Thespectra in Fig 2a were obtained from a depth scan of aDKDP crystal with 75 degree of deuteration in thebulk grown at 45 8C The spectra start out as DKDP with30 deuteration close to the surface and approach thebulk DKDP spectrum within a few micrometers of depthThe fact that the relative degree of deuteration does notextend to 0 D is due to the limited depth resolution ofthe Raman microprobe which averages over 4 mm indepth Figure 2b depicts the resulting DH exchange layerpro les for this and a second crystal grown at 63 8Crespectively Both crystals had the same exposure to am-bient conditions and their main difference is the temper-ature at which they were grown The different exchangelayer pro les indicate that crystals grown at differenttemperatures have differen t proton conductiv ities 13
which leads to a difference in their rate of deuteriumdepletion The parameters controlling this behavior arecurrently the objective of a detailed study the results ofwhich will be reported elsewhere
CONCLUSION
In conclusion we have shown that the shift of the to-tally symmetric PO 4 stretch mode in the Raman spectrumof DKDP crystals scales linearly with degree of deuter-ation This allows us to correlate Raman peak positionsto deuteration levels in these crystals We have presenteda new technique to determine DH diffusion pro les inDKDP frequency conversion crystals based on micro-Ra-man spectroscopy This technique is fast inexpensiveand works under various environmental conditionswhich will allow us to better understand and control deu-terium depletion in DKDP crystals
ACKNOWLEDGMENTS
We would like to thank M Runkel for rst discovering DKDP crack-ing R Floyd for providing DKDP crystals and L Chase and A Burn-ham for their support and helpful discussions This work was performedunder the auspices of the US Department of Energy by the University
of California Lawrence Livermore National Laboratory under contract W-7405-Eng-48
1 J J De Yoreo A K Burnham and P K Whitman Int Mat Rev47 113 (2002)
2 C E Barker R A Sacks B M Van Wonterghern J A Caird JR Murray J H Campbell K Kyle R B Ehrlich and N DNielsen Proc SPIE-Int Soc Opt Eng 2633 501 (1995)
3 T Suratwala paper to be published4 Cleveland Crystals Inc httpwwwclevelandcrystalscom
KDPshtmltable5 E A Popova I T Savatinova and I A Velichko Sov Phys Solid
State 12 1543 (1971)6 I P Kaminow R C C Leite and S P S Porto J Phys Chem
Solids 26 2085 (1965)7 J A Subramony B J Marquardt J W Macklin and B Kahr
Chem Mat 11 1312 (1999)8 H Tanaka M Tokunaga and I Tatsuzaki Solid State Commun
49 153 (1984)9 R J Nelmes G M Meyer and J E Tibballs J Phys C 15 59
(1982)10 M A Yakshin D W Kim Y S Kim Y Y Broslavets O E
Sidoryuk and S Goldstein Laser Physics 7 941 (1997)11 I Takenaga Y Tominaga S Endo and M Kobayashi Solid State
Commun 84 931 (1992)12 C Krenn personal communication13 M Sharon and A K Kalia J Solid State Chem 21 171 (1977)
Rapid Dipicolinic Acid Extractionfrom Bacillus Spores Detectedby Surface-EnhancedRaman Spectroscopy
STUART FARQUHARSON ALAND GIFT PAUL MAKSYMIUK andFRANK E INSCOREReal-Time Analyzers Inc East Hartford Connecticut06108
Index Headings Dipicolinic acid Bacillus spores Anthrax Surface-enhanced Raman spectroscopy
INTRODUCTION
The anxiety caused by the distribution of anthrax en-dospores through the US postal system in October 2001was exacerbated by the long time required for positiveidenti cation of the Bacillus anthracis spores and the un-known extent of their distribution Since that time manymethods capable of rapid eld analysis have been inves-tigated to augment or replace the laboratory method ofgrowing microorganisms in culture media which takesdays to perform12 Prominent among these approachesare polymerase chain reactions (PCR)3 immunoassaysand detection of calcium dipicolinate as a biochemicalsignature PCR employs primers to separate organism-
Received 10 October 2003 accepted 14 November 2003 Author to whom correspondence should be sent
352 Volume 58 Number 3 2004
speci c nucleic acid sequences (eg capsular protein en-coding gene for Bacillus anthracis)4 and polymerases toamplify the segment until it is detectable Recently am-pli cation times have been substantially reduced andcomplete analysis can now be performed in an hour orless Immunoassay methods are also being developed thatuse competitive binding of the bioagent (as an antigen)and its labeled conjugate for a limited number of anti-bodies Although analyses can be performed in under 30minutes a well-de ned anthrax antigen has not yet beenidenti ed5ndash7 and consequently the false-positive rate isunacceptably high8
A number of other methods are being developed witha focus on the detection of calcium dipicolinate (CaDPA)and its derivatives as a B anthracis signature This is sobecause only spore-forming bacteria contain CaDPA andthe most common potentially interfering spores such aspollen and mold spores do not Relatively fast methodshave been developed to chemically extract CaDPA andthen detect it directly by uorescence9 or indirectly byluminescence1011 In the latter case hot dodecylamine(DDA) has been used to extract dipicolinic acid (DPA)and terbium has been utilized to form a highly lumines-cent DPA complex11 Although measurements have beenperformed in as little as ve minutes it was found thatas many as three concentration-dependent complexes canform each with different lifetimes This coupled withthe fact that the Tb31 cation produces the same lumines-cence spectrum makes determinations of low spore con-centrations problematic
It has been long known that Raman spectra of Bacillispores are dominated by bands associated with CaDPA12
and that these spectra may provide a suitable anthrax sig-nature at the genus level13 Since that time considerableimprovements in Raman instrumentation have led to lab-oratory measurements of single Bacilli spores14 and to eld measurements of spores captured from a mail-sort-ing system15 However the single spore measurementsrequired complex instrumentation that is not ruggedwhile the eld measurements required milligrams of sam-ple Furthermore the Raman spectra of both measure-ments contained uorescence contributions that would in-crease uncertainty in quanti cation
In related research we demonstrated that nanogramquantities of DPA could be detected by uorescence-freesurface-enhanced Raman spectroscopy (SERS)16 We alsodemonstrated that microliter volumes of chemicals canbe detected by SERS using metal-doped sol-gel-packedglass capillaries17 Towards the goal of developing a rap-id eld SERS-based anthrax spore detector we havecombined our previous research and we now report thatDPA can be extracted from a 10 mg B cereus spore sam-ple using DDA in 1 minute and can be detected by SERSin an additional 1 minute
EXPERIMENTAL
Dipicolinic acid (26-pyridinedicarboxylic acid DPA)and dodecylamine (DDA) were used as received fromSigma-Aldrich (Milwaukee WI) Lyophilized B cereusspores prepared according to the literature13 were sup-plied by the University of Rhode Island and used as re-ceived Multiple particles approximately 01 mm3 each
were separated and weighed at 5 to 15 mg representing05 to 15 million spores The sample masses were con-sistent with a previous determination of spore density at0081 gmL that indicated a high degree of entrained air
All chemicals used to prepare the silver-doped sol-gelcoated capillaries were also obtained and used as receivedfrom Sigma-Aldrich According to previously publishedprocedures17 two precursor solutions were preparedmixed and then drawn into 1-mm-diameter glass capil-laries The silver amine precursor consisted of a 51 vvratio of 1 N AgNO 3 to 28 NH3OH while the alkoxideprecursor consisted of a 21 vv ratio of methanol to te-tramethyl orthosilicate The alkoxide precursors weremixed with silver amine precursor in an 81 vv ratioApproximately 015 mL was drawn into the capillarycoating a 15-mm length After sol-gel formation the in-corporated silver ions were reduced with dilute sodiumborohydride which was followed by a water wash to re-move residual reducing agent
A 100 mL drop of a 50 mM DDA solution in ethanolpre-heated to 78 8C was added to each of the B cereusparticles to digest the spore coat After 1 minute the re-sultant solution was drawn into a SER-active capillarythat was immediately xed horizontally to an XY posi-tioning stage (Conix Research Spring eld OR) just in-side the focal point of an f 07 aspheric lens The lensfocused the beam into the sample and collected the scat-tered radiation back along the same axis A dichroic lter(Omega Optical Brattleborough VT) was used to re ectthe excitation laser to the lens and pass the Raman scat-tered radiation collected by the lens An f 2 achromatwas used to collimate the laser beam exiting a 200-mm-core-diameter source ber optic while a second f 2 ach-romat was used to focus the scattered radiation into a 365mm ber optic (Spectran Avon CT) A short-pass lterwas placed in the excitation beam path to block the sil-icon Raman scattering generated in the source ber fromre ecting off sampling optics and reaching the detectorA long-pass lter was placed in the collection beam pathto block the sample Rayleigh scattering from reachingthe detector A 785 nm diode laser (Process InstrumentsInc model 785-600 Salt Lake City UT) was used todeliver 100 to 150 mW of power to the sample A Fouriertransform Raman spectrometer (Real-Time Analyzersmodel IRA-785 East Hartford CT) and a silicon photo-avalanche detector (Perkin Elmer model C30902S Stam-ford CT) were used to acquire the SER spectra
RESULTS AND DISCUSSION
As an initial experiment the SER spectrum of 1 gLof DPA in water was measured using the newly devel-oped silver-doped sol-gel-coated capillaries (Fig 1A) Atthis concentration a high signal-to-noise ratio (SN) isobtained in 1 min In fact a reasonable spectrum is ob-tained in the same time frame for 1 mgL (Fig 1B) TheSER spectra are reasonably similar to the normal Raman(NR) spectrum obtained for a saturated solution of DPAin 1 N KOH (Fig 1C) and the following band shifts areobserved (NR to SER) 647 to 657 cm21 817 to 815cm21 998 to 1008 cm21 1384 to 1382 cm21 1434 to1428 cm21 and 1569 to 1567 cm21 Many of these bandshave been previously assigned1213 such as 998 cm21 to
APPLIED SPECTROSCOPY 353
FIG 1 SERS of DPA in water using silver-doped sol-gel-coated glasscapillary for (A) 1 gL and (B) 1 mgL (C ) NR of saturated DPA in 1N KOH in a glass capillary Spectral conditions (A) and (B) 150 mWof 785 nm 1-min acquisition time (C) 450 mW of 785 nm 5-minacquisition time both 8 cm 21 resolution
FIG 2 SERS of DPA extracted from 10 mg B cereus particle using100 mL of 50 mM hot DDA acquired in (A) 1 minute and (B) 2 seconds(C ) Attempted SERS of 50 mM hot DDA in ethanol using silver-dopedsol-gel-coated glass capillary acquired in 1 min Spectral conditions150 mW of 785 nm 8 cm21 resolution
the symmetric ring stretch 1384 cm21 to the OndashCndashOsymmetric stretch 1428 cm21 to the symmetric ring CndashH bend and 1569 cm21 to the asymmetric OndashCndashOstretch
The rst B cereus samples consisted of 2 mg of sporesin 2 mL of 5 mM hot DDA The samples were main-tained at 78 8C for 40 min and while hot approximately10 mL was drawn into a SER-active capillary Since spec-tra of DPA were obtained for these initial samples small-er spore masses higher DDA concentrations and shorterheating periods were examined In due course it wasfound that 10 mg of spores could be digested by 100 mLof 50 mM hot DDA in one minute and detected (Fig2A) In fact the signal was suf ciently intense that it canbe observed in as little as two seconds (Fig 2B) Theamount of DPA that was extracted was estimated to bebetween 5 and 10 mgL by comparing the signal intensityof the 1008 cm21 band to that measured for DPA in waterThis is consistent with previous research that found thatthe majority of the DPA is extracted from spores usingDDA11 and that B cereus spores contain approximately10 DPA by weight18 The SN of 127 for the 1008 cm21
band in the 1-minute SER spectrum suggests a limit ofdetection of approximately 250 ng of B cereus sporesbased on a SN of 3 Finally it should be noted that DDAdid not produce a detectable SER spectrum as shown inFig 2C
CONCLUSION
We have demonstrated that by combining rapid extrac-tion of dipicolinic acid from Bacillus cereus spores withchemical identi cation by surface-enhanced Raman spec-troscopy as little as 10 mg of spores can be detected Infact the entire measurement from the time of adding hotdodecylamine to the spores to the time when the dipi-colinic acid SER spectrum is acquired and analyzedcould be performed in less than two minutes The abilityof this method to distinguish between spore-forming bac-teria such as Bacillus anthracis and non-DPA containingpowders could help prevent costly shutdowns associated
with the appearance of suspicious material or intentionalmailing of common substances as an anthrax hoax Thismethod could also prove useful in detecting the locationof anthrax endospores in mail distribution facilities if an-other veri ed attack should occur
Research continues to fully characterize the surface-enhanced Raman spectroscopy signal intensities as afunction of sample concentration and to explore otherextractants that do not require the use of elevated tem-perature
ACKNOWLEDGMENTS
The authors are grateful for the support of the National Science Foun-dation (DMI-0296116 and DMI-0215819) and the US Army(DAAD13-02-C-0015 Joint Service Agent Water Monitor program)The authors are indebted to Chetan Shende for preparing the sol-gelcapillaries The authors also thank James Gillespie Nicholas Fell andAugustus Fountain for providing important background informationMark Farquharson for laboratory support and Professor Jay Sperry ofthe University of Rhode Island for supplying B cereus spores
1 V A Pasechnik C C Shone and P Hambleton Bioseparations 3267 (1993)
2 P J Jackson M E Hugh-Jones D M Adair G Green K K HillC R Kuske L M Grinberg F A Abramova and P Keim ProcNatl Acad Sci USA 95 1224 (1998)
3 B R Glick and J J Pasternak Molecular Biology Principles andApplications of Recombinant DNA (ASM Press Washington DC1994)
4 C A Bell J R Uhl T L Had eld J C David R F Meyer TF Smith and F R Cockerill III J Clin Microbiol 40 2897(2002)
5 D L Gatto-Menking H Yu J G Bruno M T Goode M Millerand A W Zulich Biosens Bioelectron 10 501 (1995)
6 J J Quinlan and P M Foegeding J Rapid Methods AutomationMicrobiol 6 1 (1998)
7 A A Hindle and E A H Hall Analyst (Cambridge UK) 1241599 (1999)
8 M S Ascher US Department of Health amp Human Services(httpwwwhhsgovophppresentationsAscherdoc)
9 R Nudelman B V Bronk and S Efrima Appl Spectrosc 54 445(2000)
10 D L Rosen C Sharpless and L B McBrown Anal Chem 691082 (1997)
354 Volume 58 Number 3 2004
11 P M Pellegrino N F Fell Jr and J B Gillespie Anal ChimActa 455 167 (2002)
12 W H Woodruff T G Spiro and C Gilvarg Biochem BiophysRes Commun 58 197 (1974)
13 E Ghiamati R S Manoharan W H Nelson and J F SperryAppl Spectrosc 46 357 (1992)
14 A P Esposito C E Talley T Huser C W Hollars C M Schal-dach and S M Lane Appl Spectrosc 57 868 (2003)
15 S Farquharson L Grigely V Khitrov W W Smith J F Sperryand G Fenerty J Raman Spectrosc paper accep ted (2003)
16 S Farquharson W W Smith S Elliott and J F Sperry SPIE-IntSoc Opt Eng 3855 110 (1999)
17 S Farquharson and P Maksymiuk Appl Spectrosc 57 479(2003)
18 F W Janssen A J Lund and L E Anderson Science (Washing-ton DC) 127 26 (1958)
SPIE -2003-5269 117
pH dependence of methyl phosphonic acid dipicolinic acid and cyanide by surface-enhanced Raman spectroscopy
Stuart Farquharson Alan Gift Paul Maksymiuk Frank Inscore and Wayne Smith
Real-Time Analyzers 87 Church Street East Hartford CT 06108
ABSTRACT US and Coalition forces fighting terrorism in Afghanistan and Iraq must consider a wide range of attack scenarios in addition to car bombings Among these is the intentional poisoning of water supplies to obstruct military operations To counter such attacks the military is developing portable analyzers that can identify and quantify potential chemical agents in water supplies at microgram per liter concentrations within 10 minutes To aid this effort we have been investigating the value of a surface-enhanced Raman spectroscopy based portable analyzer In particular we have been developing silver-doped sol-gels to generate SER spectra of chemical agents and their hydrolysis products Here we present SER spectra of methyl phosphonic acid and cyanide as a function of pH an important factor affecting quantitation measurements which to our knowledge has not been examined In addition dipicolinic acid a chemical signature associated with anthrax-causing spores is also presented Keywords Chemical warfare agents agent detection agent hydrolysis SERS Raman spectroscopy homeland security
1 INTRODUCTION In the past decade the Unites States and its allies have been challenged by a different kind of warfare exemplified by the terrorist attacks of September 11 2001 Suicide bombings and the use of chemical agents are the norm and military personnel must consider a wide range of attack scenarios Among these is the intentional poisoning of water supplies to obstruct military operations in Afghanistan and Iraq Currently colorimetric paper is used to detect agents on-site while gas chromatography combined with mass spectrometry (GCMS) is used in mobile support laboratories However both methods have severe drawbacks The paper changes color in response to contact with many chemicals besides CWAs causing a high incidence of false positives1 while GCMS although very chemically specific requires hours to perform and constant re-calibration234 Military operations would be greatly aided by a portable analyzer that can identify and quantify potential chemical agents at concentrations that impact safety This includes the analysis of drinking water supplies distribution and storage systems To meet this goal the Department of Defense has been investigating numerous approaches under the auspices of the Joint Service Agent Water Monitor (JSAWM) program5 The JSAWM must be capable of identifying the chemical agents with no false-positives and quantifying the agents at microgram per liter concentrations within 10 minutes678 Recently we and others have been investigating the ability of surface-enhanced Raman spectroscopy (SERS) to measure chemical agents9-12 bioagents13-17 and their hydrolysis products in water SERS employs the interaction of surface plasmon modes of metal particles with target analytes to increase scattering efficiency by as much as 1 million times18 In 1987 the potential of this approach to measure CWAs was demonstrated by measuring a series of pesticides19 Several of these organophosphonates have chemical structures similar to CWAs in particular P=O functional groups In our studies we have been employing silver-doped sol-gels to promote the SER effect The porous silica network of the sol-gel matrix offers a unique environment for immobilizing and stabilizing SER-active metal particles20-23 The sol-gels have been coated on the internal walls of standard glass vials to allow rapid SER analysis of small sample quantities without preparation We have measured over 100 chemicals with enhancements of 104 to 106 demonstrated reversible measurements in a flowing system reproducible measurements from vial-to-vial and measurements in multiple solvents including water20-23 Previously we used these vials to perform preliminary measurements of cyanide (CN) methylphosphonic acid (MPA) and dipicolinic acid (DPA) MPA is a hydrolysis product of the nerve agents (eg sarin Reaction 1) and may be a valuable indicator of nerve agent usage particularly since the alkyl methylphosphonic acids are relatively more stable than their corresponding parent complexes24 DPA is
SPIE -2003-5269 118
a chemical signature of spore forming bacteria such as Bacillus anthracis And in light of the inability to rapidly detect the anthrax spores distributed through the US mail in October 2001 a number of methods are being developed to extract and analyze this signature Reaction 1 Stepwise hydrolysis of Sarin to form hydrofluoric acid (HF) isopropyl methylphosphonic acid (IMPA) then methyl phosphonic acid (MPA) and 2-propanol In our previous SERS investigations MPA and DPA were measured at 50 and 100 mgL respectively In both cases limits of detection (LOD) were estimated at 100 microgL providing encouragement in that SERS may satisfy the needs of the JSAWM Since it has been shown that pH can substantially influence the intensity of SER bands25 which would clearly influence quantitative analysis we undertook the present study to determine the severity of these effects for cyanide methyl phosphonic acid and dipicolinic acid Furthermore we previously observed a band at 1050 cm-1 for MPA6 possibly due to an anion formed at basic pH Here we investigate the source of this spectral anomaly
2 EXPERIMENTAL All chemicals including potassium cyanide methyl phosphonic acid dipicolinic acid and those used to prepare the silver-doped and gold-doped sol-gel coated vials were obtained and used as received from Sigma-Aldrich All samples were prepared in HPLC grade water (Fischer Scientific Fair Lawn NJ) for SERS measurements The pH of these samples was adjusted using dilute nitric acid or potassium hydroxide and verified using a pH electrode (Corning Inc Corning NY) that had been calibrated with pH 400 700 and 1000 buffered standards from Fischer Scientific Once prepared the samples were transferred into the silver-doped sol-gel vials (Real-Time Analyzers Inc East Hartford CT) The vials were coated in a manner similar to that previously reported by adding ammonium hydroxide to a solution of silver nitrate tetramethyl orthosilicate (TMOS) and methanol20 Gold-vials were coated by adding nitric acid to a solution of gold tetrachloride TMOS and methanol The two precursor solutions were prepared mixed and transferred to 2-mL glass vials dried and heated After sol-gel formation the incorporated metal ions were reduced with dilute sodium borohydride (1mgmL) which was followed by a water wash to remove residual reducing agent After the resultant analyte solution was introduced the SER-active vial was immediately fixed horizontally to an XY positioning stage (Conix Research Springfield OR) just inside the focal point of an f07 aspheric lens The lens focused the beam into the sample and collected the scattered radiation back along the same axis A dichroic filter (Omega Optical Brattleborough VT) was used to reflect the excitation laser to the lens and pass the Raman scattered radiation collected by the lens An f2 achromat was used to collimate the laser beam exiting a 200 microm core diameter source fiber optic while a second f2 achromat was used to focus the scattered radiation into a 365 microm fiber optic (Spectran Avon CT) A short pass filter was placed in the excitation beam path to block the silicon Raman scattering generated in the source fiber from reflecting off sampling optics and reaching the detector A long pass filter was placed in the collection beam path to block the sample Rayleigh scattering from reaching the detector A 785 nm diode laser (Process Instruments Inc model 785-600 Salt Lake City UT) was used to deliver 100 to 150 mW of power to the sample A Fourier transform Raman spectrometer (Real-Time Analyzers model IRA-785 East Hartford CT) and a silicon photo-avalanche detector (Perkin Elmer model C30902S Stamford CT) were used to acquire the SER spectra
3 RESULTS AND DISCUSSION In a previous study of MPA6 aimed at developing a concentration calibration curve and determining limits of detection (LOD) we observed an anomaly at 1050 cm-1 Since it was found that the band intensity changed as a function of concentration the band must be associated with a sample parameter Two possible parameters photon flux and pH are examined here The first parameter was investigated by irradiating a 1mgmL MPA sample in a SER-active vial with laser powers of 200 mW and above and monitoring spectral changes It was immediately found that the 1050 cm-1 band
2O+ H HF + +OH
OH3H C
OP
OF
CH
CH3
3
3H C
OCP
OOH
CH
CH3
3
3H C
O
CPHO
CH
CH
3
3
C
Sarin IMPA MPA 2-propanol
SPIE -2003-5269 119
appeared and grew as a function of time and that the higher the laser power the faster the growth Figure 1 shows the growth of the 1050 cm-1 band over the course of 30 minutes when using 150 mW of 785 nm excitation while Figure 2A shows that the growth can be fit with a first order exponential equation namely I1050 = 03+05e-013t Furthermore the 760 cm-1 band could be fit with a first order decay equation with an identical exponential rate constant ie I760 = 08-
08e-013t The rates represent classical first order kinetics and their correspondence allows one to conclude that MPA is being transformed one-for-one into a photo-generated product At this time the photoproduct has not been positively identified but phosphonic (phosphorous) acid and phosphonate are likely candidates since the symmetric P(OH)3 stretch occurs at ~1050cm-1 Our studies show that a reduction in laser power to 100 mW at the sample for MPA essentially eliminates this degradation process This laser power or lower was used for further measurements
Methyl phosphonic acid is a diprotic acid that stepwise dissociates into two anions MPA- and MPA= according to the following reactions26
MPA MPA- + H+ pKa1 = 212 Reaction 2
MPA- MPA= + H+ pKa2 = 729 Reaction 3 The relative concentrations of MPA MPA- and MPA= can be determined at any pH by expressing [MPA] and [MPA=] in terms of [MPA-] using Reactions 2 and 3 and summing all three to equal the total starting concentration here 2 mgmL (0021M MW = 9602) viz
[MPA] + [MPA-] + [MPA=] = 0021M Equation 1 substituting from Reactions 1 and 2
([H+][MPA-])K1a + [MPA-] + (K2a[MPA-])[H+] = 0021M Equation 2 rearranging [MPA-] = 0021M(1+[H+]K1a + K2a[H+]) Equation 3 The relative concentrations of MPA MPA- and MPA= as a function of pH are shown in Figure 3 It is worth noting that near neutral pH both MPA- and MPA= will be present To confirm that the SER signal followed this pH dependence a starting solution consisting of 20 mg of MPA in 10 mL HPLC grade water was prepared and brought to pH of 20 using dilute nitric acid From this solution 2 mL were added to a SER-active vial and the SER spectrum recorded At this pH a peak at 760 cm-1 was barely discernable The 2 mL solution was returned to the starting solution and the pH was re-measured to correct for any changes that the silver-doped sol-gel vials might cause In most cases the change was less than 02 pH units and the pH is reported as the before and after average Next the pH of the
Figure 2 A) Exponential growth of 1050 cm-1 band and B) exponential decay of 760 cm-1 band for spectral series in Figure 1
Figure 1 Growth of 1050 cm-1 band as a function of time due to exposure to 150 mW of 785 nm Spectra are 5 sec each collected every 100-sec from 0 to 30-min
B
570 770 970 1170 1370Raman Shift (cm-1)
Arbit
rary
Unit
s
MPA Photodegradation
600 800 1000 1200 1400 Wavenumber (cm-1)
A
0 10 20 30 0 10 20 30 time (min) time (min)
Ram
an In
tens
ity (r
elat
ive)
30
min
0
SPIE -2003-5269 120
starting solution was adjusted to 325 using dilute KOH Again 2 mL were added to a vial and the SER spectrum recorded At this pH a reasonably strong 760 cm-1 band was observed This process was repeated as spectra were recorded at pHs of 70 74 75 79 85 and 100 A total of 1 mL of KOH was added diluting the total concentration by 10 Next the pH of the starting solution was made acidic by adding dilute nitric acid dropwise This time spectra were recorded at pHs of 72 69 64 and 37 Figure 4 shows the SER spectra for representative pHs (spectra were left out to simplify the figure) while Figure 3 shows the 760 cm-1 peak intensities as a function of pH (The band intensities were adjusted to compensate for dilution effects caused by the addition of HNO3 and KOH then normalized to 0021 M for the most intense band observed at pH 37) It is clear from Figure 3 that the 760 cm-1 band follows the MPA- concentration as a function of pH and must be assigned to this anion No bands were observed that corresponded to MPA or MPA= The lack of an MPA SER spectrum may be due to the absence of an attraction between the neutral analyte and the electropositive silver surface The same reasoning suggests that a strong SER spectrum should be observed for MPA= but it is not and a satisfactory explanation has not been found
0000
0005
0010
0015
0020
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14pH
Con
cent
ratio
n [M
]
pK1 = 212 pK2 = 729
MPA- MPA=
MPA
Figure 4 SER spectra of 002M MPA as a function of pH Conditions 100 mW of 785 nm 36 scans (1 min) 8 cm-1 recorded 2 min after sample introduction pH 19 and 20 not apparent on this scale pH 69 and 74 near identical to 70 and 75 and not shown for clarity
00
02
04
06
08
10
12
14
16
18
0 5 10 15 20 25 30Measurement Number
Ram
an In
tens
ity (7
60 c
m-1
)
Figure 3 Concentration dependence of MPA MPA- and MPA= as a function of pH for a 002M sample Intensity of 760 cm-1
band from Figure 6 as a function of pH ( for increasing basic adjustment for increasing acidic adjustment error was measured at ~10 for pH 64)
Figure 5 SER spectra of 002M MPA at pH 64 measured around a vial at three heights (9 points per height) Conditions as in Figure 4 but 10-sec scans
Since these measurements involved the removal and replacement of the SER-active vial in the sample holder to remove and add sample variation in the intensity as a function of vial position was minimized by illuminating the exact same height along the vial wall But this does not account for variability of the SERS response of the sol-gel coating around the vial To analyze this effect a vial containing MPA at pH 64 was rotated at ~ 40o
intervals at the original height and 18rdquo above and below this value Figure 5 shows the intensity of the 760 cm-1
band for the 27 positions It was found that the average value was 137plusmn014 an RSD of 10 overall and 5 for each height An error bar is included in Figure 4 for the pH 64 measurement
SPIE -2003-5269 121
With the above analyses in mind a preliminary investigation of the SER spectral response for MPA (as MPA-) as a function of concentration was performed A single vial was used for these measurements beginning with 1 mgL followed by measurements of 10 100 and 1000 mgL In all cases the pH was ~7 and 3 positions around the vial were measured per concentration Since the 760 cm-1 band was not observed for concentrations of 1 or 10 mgL using 100 mW of 785 nm the laser power at the sample was raised to 200 mW beginning with the 10 mgL concentration Photo-degradation was largely avoided (and not observed) by exposing the sample for only 33 seconds per spectral acquisition Representative spectra for 10 100 and 1000 mgL are shown in Figure 6 while a plot of the 760 cm-1 band intensity as a function of concentration is shown in Figure 7 These values were also used to estimate limits of detection based on the signal-to-noise ratio (SN) of the 760 cm-1 band using 100 mW of 785 nm laser excitation and a 10-min acquisition time As summarized in the Figure 7 inset the lower the measured concentration the lower the predicted LOD Even if as estimated 210 microgL could be measured using the silver-doped sol-gel vials an improvement of a factor of 70 is still required to meet JSAWM goals of measuring 3 microgL in 10-minutes Similar to MPA DPA is a diprotic acid (pKa1 = 216 and pKa2 = 692) and variations in pH will effect the relative concentrations of DPA DPA- and DPA= and possibly the SER spectra and band intensities This could prove significant if an acid or base is used to denature anthrax spores with the goal of extracting and analyzing DPA The MPA pH study described above was mimicked for DPA except that the starting solution consisted of 20 mg of DPA in 20 mL HPLC grade water (60x10-3M MW = 1671) The initial solution had a pH of 245 which was made basic by dilute KOH to pHs of 355 433 487 559 1069 and 1166 SER spectra were recorded at each pH using 100 mW of 785 nm and a 44-sec acquisition time Next one drop of concentrated nitric acid was used to remake the solution acidic at a pH of 200 Again sequential pH measurements were performed at 383 510 735 and 822 The solution pH was made acidic a third time but to pH 219 171 then 135 Throughout this process no more than 20 drops of acid or base were added and therefore the concentration was diluted by no more than 10 Most of the spectral bands showed a minor decrease in intensity as a function of increasing pH values However the bands at 525 and 795 cm-1 showed the most dramatic changes which occurred at acid pH Figure 8 shows the SER spectra of DPA for the spectral region and pH range of interest The identity of the DPA species was determined by plotting the normalized peak intensities with the lowest value set to 0 and the highest to 0006 M as a function of pH and overlaying these values on a plot of the relative concentrations for DPA DPA- and DPA= as previously done for MPA (Figure 9) As can be seen the 525 cm-1 band clearly corresponds to DPA The correspondence of the 795 cm-1 band to this species is less clear as the band retains intensity until pH of 55 This can be attributed to contribution to the overlapping band at 810 cm-1 which does not change as a function of pH The fact that most bands are observed at all pHs suggest that the primary interaction with silver is through the ring nitrogen This is supported by the fact that the most intense band occurs at 1008 cm-1 attributed to a symmetric ring breathing mode and that this interaction has been characterized for pyridine in numerous papers27
0
1
2
3
4
5
6
7
0 200 400 600 800 1000 1200
MPA Concentration (mgL)
760
Ban
d In
tens
ity
Figure 6 SER spectra of MPA in water at A) 1000 B) 100 and C) 10 mgL Conditions pH of 7 silver-doped sol-gel coated vial 200 mW of 785 nm 33-sec 8 cm-1 resolution
Figure 7 Plot of SER intensity of 760 cm-1 band of MPA as a function of concentration using 200 mW of 785 nm Inset table includes average intensity LOD standard deviation and percent deviation for each concentration but for 100 mW and 10 min
conc (mgL) LOD ave stddev dev10 021 002 1021100 072 010 14301000 312 040 1280
A
B C
SPIE -2003-5269 122
The ability to detect dipicolinic acid (DPA) by SERS is demonstrated in Figure 10 Even at 1 mgL the primary bands are visible The intensity of the dominant ring breathing mode observed at 1008 cm-1 for DPA in water (pH = 25 to 55) is plotted as a function of concentration in Figure 11 These values were also used to estimate limits of detection based on the SN of the 1008 cm-1 band using 100 mW of 785 nm laser excitation and a 10-min acquisition time Again the lower the measured concentration the lower the predicted LOD (see Figure 11 inset) and detection of 160 microgL is possible Gastrointestinal anthrax requires significant more spores than inhalation anthrax28 and a limit of detection might be placed at 1 million spores in 1 liter of water or 10 microgL Since each spore contains ~10 CaDPA by weight29 a goal for DPA might be 1 microgL indicating that the present measurements must be improved by nearly two orders of magnitude Due to the increased hazards of handling HCN gas KCN salt was used for these experiments Nevertheless all sample preparations were performed in a chemical hood KCN completely dissolves in water but its conjugate acid HCN is formed and has a Ka of 615x10-1030 viz
HCN CN- + H+ pKa = 921 Reaction 4
Figure 8 SER spectra of 1 mgmL DPA as a function of pH Conditions 100 mW 785 nm 100 scans (44-sec)
Figure 9 Concentration dependence of DPA DPA- and DPA= as a function of pH for a 0006M sample Intensity of 525 () and 795 (diams) cm-1 bands from Fig 8 as a function of pH
pH
135 171 219 383
0
0001
0002
0003
0004
0005
0006
0007
0 2 4 6 8 10 12 14pH
Con
cent
ratio
n [M
]
DPADPA-DPA=795525
DPA DPA=DPA-
pK2 = 692pK1 = 216
795 525
0
05
1
15
2
25
0 200 400 600 800 1000 1200
DPA Concentration (mgL)
1008
Ban
d In
tens
ity
conc (mgL) lod-10min-100mw1 017
10 016100 103
1000 355
Figure 10 SER spectra of DPA in water at A) 1000 B) 100 C) 10 and D) 1 mgL Conditions pH of 25-55 silver-doped sol-gel coated vial 175 mW of 785 nm 1-min 8 cm-1 D) has been multiplied by x10 to make bands visible
A
B C
D
Figure 11 Plot of SER intensity of 1008 cm-1 band of DPA as a function of concentration using 175 mW of 785 nm Inset table includes LOD in mgL for each concentration but for 100 mW and 10 min
SPIE -2003-5269 123
Consequently the cyanide concentration must be determined for each initial KCN concentration Specifically the samples prepared with concentrations of 01 1 10 100 and 1000 mgL of KCN produced CN- concentrations of 63x10-3 033 69 89 and 964 mgL at pHs of 816 90 967 102 and 107 respectively The pH dependence for the HCN and CN- concentrations are shown in Figure 12 Thus as the amount of KCN added to the solution decreases so does the pH of the solution (becomes less basic) and according to Reaction 4 the relative amount of CN- to HCN also decreases For example in the preparation of a 01 mgL solution of KCN the pH is shifted from 7 for pure water to only 816 and only 63 of the starting material becomes CN- or 63x10-3 mgL In comparison for a solution of 1000 mgL the pH is shifted from 7 to 107 and 96 of the starting material becomes CN- This is significant since the cyanide ion is better able to adsorb onto the silver particles and become SERS active SER spectra of 10 100 and 1000 mgL of KCN samples are shown in Figure 13 and a corresponding plot of the 2100 cm-1 band intensity as a function of concentration is shown in Figure 14
000102030405060708091011
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14pH
Con
cent
ratio
n [m
gm
L]
pKa = 921
CN -HCN
The SER spectra of cyanide are dominated by the single CN stretch at 2100 cm-1 which occurs in normal Raman spectra of solutions at 2080 cm-1 However a low frequency mode occurs at 135 cm-1 due to a Ag-CN stretch (not shown) indicative of a strong surface interaction It is also observed that as the concentration decreases the CN stretch shifts to 2140 cm-1 This shift has been attribute to the formation of a tetrahedral Ag(CN)3
2- surface structure31 as well
Figure 14 Concentration dependence of KCN SERS measured under conditions in Fig 11 Concentrations are 1 01 and 001 mgml Intensities are measured for the CN stretch at 2100 cm-1 Inset table includes LOD in mgL for each concentration in Figs 13 and 15 but for 100 mW and 10 min
Figure 13 SER spectra of KCN in water at A) 1000 B) 100 and C) 10 mgL Conditions pHs of 107 102 and 97 silver-doped sol-gel coated vial 100 mW of 785 nm 1-min 8 cm-1 C) has been multiplied by x10 to make band visible
Figure 12 Concentration dependence of HCN and CN- as a function of pH for a 1 mgmL sample Calculated intensity of 2100 () cm-1 band for a 1 mgml sample at pHs of 816 90 967 102 and 107
A
B
C
0
20
40
60
80
100
120
140
0 200 400 600 800 1000 1200CN Concentration (mgL)
2100
Ban
d In
tens
ity
Figure 15 SER spectra of KCN in water at A) 10 B) 1 and C) 01 mgL Conditions pHs of 97 12 and 12 gold-doped sol-gel coated vial 100 mW of 785 nm 1-min 8 cm-1 C) has been multiplied by x10 to make band visible
A
B
C
conc (mgL) lod-10min-100mw Condition01 001 Au-pH 121 007 Au-pH 12
69 003 Au-pH 9769 016 Ag-pH 9789 022 Ag-pH 102946 113 Ag-pH 107
SPIE -2003-5269 124
as to CN adsorbed to two different surface sites32 Alternatively the 2140 cm-1 band could be attributed to HCN since this species dominates at lower concentrations However it is unlikely that this species would be attracted to the electropositive silver surface Further both peaks should be present at pHs between 85 and 105 but this is not observed It has also been suggested that at concentrations near and above monolayer coverage the CN- species is forced to adsorb end-on due to crowding and at lower concentrations the molecule can reorient to lie flat33 This suggests that the 2100 and 2140 cm-1 bands correspond to the end-on and flat orientations respectively As Figures 13 and 14 show the intensity of the CN stretch for the 89 mgL sample is nearly as intense as the 964 mgmL band This suggests that the Raman signal for the flat orientation is more enhanced However more extensive measurements are required to verify this point Since resent research has suggested that cyanide may be more effectively detected on gold measurements of KCN solutions were also performed using gold-doped sol-gel vials Preliminary measurements are shown in Figure 15 for samples prepared from 01 1 and 10 mgL KCN Since the pHs are 816 90 967 the resultant CN- concentrations are 63x10-3 033 and 69 mgL Initially only the highest concentration was observed and the signal intensity was significantly better than the equivalent concentration measured using silver In an effort to shift Reaction 4 to the left transforming HCN to CN- (Le Chatelierrsquos principle) KOH was added to the lower concentration samples producing solutions with pH 12 This effectively forces all of the cyanide in solution to be CN- or 01 and 10 mgL respectively More importantly the CN stretch is now observed in the SER spectra The band appears at 2125 cm-1 as has been previously reported for gold12 As calculated for MPA and DPA LODs can be estimated from this data For the three concentrations of cyanide on silver the LODs are 016 to 11 mgL for 100 mW of 785 nm laser excitation and a 10-min acquisition time For gold pH adjusted the LODs are10 to 70 microgL an improvement of more than 10 times silver Nevertheless either substrate is sufficient to meet the JSAWM goals of measuring 3 mgL in 10-minutes as the requirements form cyanide are much less stringent than the nerve agents
4 CONCLUSIONS Here we examined the surface-enhanced Raman spectral response for methyl phosphonic acid dipicolinic acid and cyanide as a function of pH It was determined that the most prominent peak at 760 cm-1 reaches a maximum intensity between pH 3 and 7 and corresponds to the MPA- species Neither the MPA nor MPA= species appear to generate a SER spectrum and consequently no spectra were observed below pH 2 or above pH 8 In this study we also found that higher laser powers could cause photodegradation of MPA signified by the exponential growth of a band at 1050 cm-1 which is tentatively assigned to phosphorous acid Unlike MPA DPA was observed at all pHs This is attributed to the dominant interaction of the pyridine functional group with silver Minor spectral changes were observed at acid pHs and were assigned to neutral DPA Like MPA SER spectra of cyanide were pH dependent No spectra were observed for the HCN species while CN- was best observed at pHs more basic than 8 Preliminary concentration studies for the three analytes allowed estimating limits of detection for MPA DPA and CN using 100 mW of 785 nm and a 10-min acquisition time of 210 165 and 70 microgL respectively Although the latter value suggests that cyanide can be measured at sufficiently low concentrations to meet JSAWM goal improvements by 100 to 200 times are required for MPA and DPA It is clear from this study that pH of the sample is important and must be taken into account when developing concentration calibration curves as well as fieldable analyzers Future work will include tailoring the sol-gel with specific functional groups to dictate sample pH and thereby optimize sensitivity
ACKNOWLEDGMENTS The authors are grateful for the support of the US Army (DAAD13-02-C-0015 Joint Service Agent Water Monitor program) The authors would also like to thank Dr Steve Christensen of the US Army SBCCOM for helpful discussions and Mr Chetan Shende of Real-Time Analyzers for assistance in development of the gold-doped sol-gels
REFERENCES 1 Erickson B Analytical Chemistry News amp Features June 1 397A (1998)
SPIE -2003-5269 125
2 Johnston RL Hoefler CM Fargo JC and Moberley B AT-ONSITE 5-8 (1994) 3 Black RM RJ Clarke RW Read and MT Reid J Chromatography 662 301-321 (1994) 4 USA Army Center for Health Promotion and Preventive Medicine (USACHPPM) Technical Guideline 230A 5 Jensen JL NC Wong W R Loerop SPIE 4775-03 (2001) 6 Hoenig SL Handbook of Chemical Warfare and Terrorism Greenwood Press Wesport CT (2002) 7 Munro NB SS Talmage GD Griffin LC Waters AP Watson JF King and V Hauschild Env Health
Persp 107 933-974 (1999) 8 Holstege CP Kirk M Sidell FR Crit Care Clin13 923-42 (1997) 9 Farquharson S P Maksymiuk K Ong and S Christesen SPIE 4577 166-173 (2001) 10 Lee Y and S Farquharson SPIE 4378 21-26 (2001) 11 Spencer KM J Sylvia S Clauson and J Janni SPIE 4577 158-165 (2001) 12 Tessier P S Christesen K Ong E Clemente A Lenhoff E Kaler and O Velev Applied Spectroscopy 56
1524-1530 (2002) 13 Farquharson S WW Smith S Elliott and JF Sperry SPIE 3533 207-214 (1998) 14 Farquharson S WW Smith S Elliott and JF Sperry SPIE 3855110-116 (1999) 15 Farquharson S WW Smith YH Lee S Elliott and JF Sperry SPIE 4575 62-72 (2002) 16 Guzelian AA J Sylvia J Janni S Clauson and KM Spencer SPIE 4577 182-192 (2001) 17 Shende C F Inscore A Gift P Maksymiuk and S Farquharson in press 18 Weaver MJ S Farquharson and MA Tadayyoni J Chem Phys 82 4867-4874 (1985) 19 Alak AM and T Vo-Dinh Analytical Chemistry 59 2149-2153 (1987) 20 Lee Y and S Farquharson SPIE 4206 140-146 (2000) 21 Farquharson S and Y Lee SPIE 4200-16 (2000) 22 Lee Y S Farquharson and P M Rainey SPIE 3857 76-84 (1999) 23 Lee Y S Farquharson H Kwong and M Shahriari SPIE 3537 252-260 (1998) 24 Wang J M Pumera G Collins and A Mulchandani Analytical Chemistry 74 6121-6125 (2002) 25 Dou X YM Jung Z-Q Cao and Y Ozaki Applied Spectroscopy 53 1440-1447 (1999) 26 Data supplied by S Christesen and K Ewing 27 Kerker M and B Thompson Eds SPIE MS 10 (1990) 28 Inglesby TV DA Henderson JG Bartlett JAMA 287 2236 (2002) 29 FW Janssen AJ Lund and LE Anderson Science 127 26 (1958) 30 Lide DR Ed Handbook of Chemistry and Physics CRC Press 77th Ed (1996-7) 31 Billmann J G Kovacs and A Otto Surf Sci 92 153 (1980) 32 Murray CA and S Bodoff Phys Rev B 32 671 (1985) 33 Kellogg D and J Pemberton J Phys Chem 91 1120 (1987)
SPIE-2003-5269 16
Chemical agent detection by surface-enhanced Raman spectroscopy
Stuart Farquharson Alan Gift Paul Maksymiuk Frank Inscore and Wayne Smith
Real-Time Analyzers 87 Church Street East Hartford CT 06108
Kevin Morrisey and Steven D Christesen US Army SBCCOM Aberdeen Proving Ground MD 21010
ABSTRACT
In the past decade the Unites States and its allies have been challenged by a different kind of warfare exemplified by the terrorist attacks of September 11 2001 Although suicide bombings are the most often used form of terror military personnel must consider a wide range of attack scenarios Among these is the intentional poisoning of water supplies to obstruct military operations in Afghanistan and Iraq To counter such attacks the military is developing portable analyzers that can identify and quantify potential chemical agents in water supplies at microgram per liter concentrations within 10 minutes To aid this effort we have been investigating the value of a surface-enhanced Raman spectroscopy based portable analyzer In particular we have been developing silver-doped sol-gels to generate SER spectra of chemical agents and their hydrolysis products Here we present SER spectra of several chemical agents measured in a generic tap water Repeat measurements were performed to establish statistical error associated with SERS obtained using the sol-gel coated vials Keywords Chemical agents chemical agent detection SERS Raman spectroscopy
1 INTRODUCTION
In the autumn of 2001 terrorism within US borders became a sobering reality While extensive efforts are being implemented to secure the homeland US and Coalition forces in Afghanistan and Iraq are constantly faced with terrorist attacks In addition to car-bombings the military has identified several non-traditional attack scenarios including poisoning of water supplies by chemical warfare agents (CWAs) To counter this threat the Department of Defense has been investigating numerous approaches to detect such attacks under the auspices of the Joint Service Agent Water Monitor (JSAWM) program1 The JSAWM must be capable of identifying the chemical agents with no false-positives and quantifying the agents at microgram per liter concentrations within 10 minutes2 This includes the analysis of drinking water supplies distribution and storage systems Currently colorimetric paper is used to detect agents on-site while gas chromatography combined with mass spectrometry (GCMS) is used in mobile support laboratories However both methods have severe drawbacks The paper changes color in response to contact with many chemicals besides CWAs causing a high incidence of false positives3 while GCMS although very chemically specific requires up to an hour to perform and regular re-calibration456 Vibrational spectroscopy has also been investigated as a potential method of rapidly detecting CWAs7-
11 as early as 1985 in the case of infrared spectroscopy8 and 1988 in the case of Raman spectroscopy11 Again however these techniques also have limitations when it comes to measuring trace poisons in water Infrared spectra would be dominated by the very strong absorption of water which would obscure absorptions by most other chemicals present Whereas Raman spectroscopy is simply not a very sensitive technique and detection limits are typically grams per liter Surface-enhanced Raman spectroscopy (SERS) offers several advantages over conventional vibrational methods and that may provide the necessary sensitivity required for detecting trace quantities of chemical agents in water SERS employs the interaction of surface plasmon modes of metal particles with target analytes to increase scattering efficiency by as much as 1 million times12 In 1987 the potential of this approach to measure CWAs was demonstrated by measuring a series of pesticides13 Several of these organophosphonates have chemical structures similar to CWAs in particular P=O functional groups In the past few years we and others have further explored the ability of SERS to detect CWAs14-17 and even bioagents 18-21 We have been employing silver-doped sol-gels to promote the SER effect
SPIE-2003-5269 17
in these studies The porous silica network of the sol-gel matrix offers a unique environment for immobilizing and stabilizing SER-active metal particles22-25 The silver-doped sol-gels have been coated on the internal walls of standard glass vials to allow rapid SER analysis of small sample quantities without preparation We have measured over 100 chemicals with enhancements of 104 to 106 demonstrated reversible measurements in a flowing system reproducible measurements from vial-to-vial and measurements in multiple solvents including water21-25 Previously we used these vials to perform preliminary measurements of cyanide (CN) and methylphosphonic acid (MPA) Most of the nerve agents form MPA during hydrolysis while Tabun forms CN a chemical agent in its own right In another paper including in these proceedings we examined the limits of detection (LOD) for MPA by measuring a series of concentrations down to 50 mgL and estimated a limit of detection of 100 microgL26 These measurements provide encouragement in that SERS may satisfy the needs of the JSAWM To further establish the viability of SERS in particular silver-doped sol-gels here we present analysis of cyanide mustard and VX in tap water The measurements performed at the US Armyrsquos Edgewood Chemical Biological Center Aberdeen MD also included numerous repeat measurements to establish reproducibility
2 EXPERIMENTAL 2a General All chemicals including potassium cyanide 2-chloroethylethyl sulfide and those used to prepare the silver-doped and gold-doped sol-gel coated vials were obtained and used as received from Sigma-Aldrich All samples were prepared in a chemical hood using HPLC grade water unless otherwise noted (Fischer Scientific Fair Lawn NJ) for SERS measurements Once prepared the samples were transferred into the silver-doped sol-gel vials (Real-Time Analyzers Inc East Hartford CT) The vials were coated in a manner similar to that previously reported by adding ammonium hydroxide to a solution of silver nitrate tetramethyl orthosilicate (TMOS) and methanol22 The two precursor solutions were prepared mixed and transferred to 2-mL glass vials dried and heated After sol-gel formation the incorporated metal ions were reduced with dilute sodium borohydride (1mgmL) which is followed by a water wash to remove residual reducing agent After the resultant analyte solution was introduced the SER-active vial was fixed horizontally to an XY positioning stage (Conix Research Springfield OR) just inside the focal point of an f07 aspheric lens The lens focused the beam into the sample and collected the scattered radiation back along the same axis A dichroic filter (Omega Optical Brattleborough VT) was used to reflect the excitation laser to the lens and pass the Raman scattered radiation collected by the lens An f2 achromat was used to collimate the laser beam exiting a 200 microm core diameter source fiber optic while a second f2 achromat was used to focus the scattered radiation into a 365 microm fiber optic (Spectran Avon CT) A short pass filter was placed in the excitation beam path to block the silicon Raman scattering generated in the source fiber from reflecting off sampling optics and reaching the detector A long pass filter was placed in the collection beam path to block the sample Rayleigh scattering from reaching the detector A 785 nm diode laser (Process Instruments Inc model 785-600 Salt Lake City UT) was used to deliver 100 to 150 mW of power to the sample A Fourier transform Raman spectrometer (Real-Time Analyzers model IRA-785 East Hartford CT) and a silicon photo-avalanche detector (Perkin Elmer model C30902S Stamford CT) were used to acquire the SER spectra
2b Edgewood Chemical Biological Center The surface-enhanced Raman spectral measurements at the US Armyrsquos Chemical Biological center presented here were all performed on September 12 2003 To expedite measurements a plate was machined to hold up to 12 SER-active sample vials (Figure 1) The plate fit a standard XY plate reader that could be programmed Pure KCN bis-(2-chloroethyl)sulfide (distilled mustard gas HD) and ethyl S-2-diisopropylamino ethyl methylphosphonothioate (VX) were obtained on-site and used to prepare 1 gL tap water solutions in a chemical hood with appropriate safety equipment Simulated tap water was prepared by adding 10 chemicals most often found in tap water at appropriate concentrations to distilled water (Table 1) SER measurements were also performed in a chemical hood For added safety the FT-Raman instrument was placed outside the laboratory and 30 foot fiber optic and electrical cables were used to allow remote SERS measurements and plate manipulation For each experiment 1gL samples were prepared and added to 9 individual vials which were then loaded on the plate In some cases a tenth vial was included as a blank
Table 1 Generic Tap Water Composition
Compound mgL NaHCO3 100 CaSO4 27 MgSO4bull7H2O 67 NaNO3 1 Fulvic Acid 1 K2HPO4 07 KH2PO4 03 (NH4)2HSO4 001 NaCl 001 FeSO4 0001 pH 76-78
SPIE-2003-5269 18
A software program was written that allowed selecting the sequence that the vials were measured the number of positions along the length of the vials to measure (1 to 5) and the number of scans to co-add During sample analysis the program displayed the vial being analyzed the point being analyzed and the spectrum as it was being acquired Once all the data was collected a second software program was written to rapidly analyze the data The spectra collected for all the vials on a plate could be loaded at one time and then the spectra for each point could be displayed simultaneously or separately The user could then select the Raman peak to analyze in terms of peak height or area This was accomplished by selecting points on either side of the peak to define a baseline of zero The peak height or area could then be computed for all of the spectra loaded and then exported to a spreadsheet for statistical analysis
Figure 1 A) Vial Holder 6 slots to hold 2 vials each end-to-end B) Measurement Configuration Program user selects vials to measure sequence number of points per vial (1 to 5) and number of scans per point C) Spectral Acquisition Program shows spectrum being collected which vial and position D) Spectra Analysis Program user selects spectra to analyze by plate vial and point (s) as well as two wavenumbers defining the peak and the baseline to subtract The image is of 5 repeat measurements of 10 mgL KCN in generic tap water 16 sec each 100 mW of 785 nm
3 RESULTS AND DISCUSSION Raman and surface-enhanced Raman spectra were obtained for potassium cyanide bis-(2-chloroethyl)sulfide and ethyl S-2-diisopropylamino ethyl methylphosphonothioate representing three classes of chemical agents cyanides mustards and nerve agents respectively Spectra were also obtained for 2-chloroethyl ethyl sulfide (CEES) a structural analogue to HD which was included in the study to aid in assigning spectral bands KCN salt was used for cyanide experiments to avoid the increased hazards of handling HCN gas KCN completely dissolves in water forming its conjugate acid HCN according to its Ka of 615x10-1027 and at a concentration of 1 mgmL results in a pH 107 solution This is significant in that only CN- appears to interact sufficiently with silver to produce a SER spectrum and no spectral signal is observed below pH 726 Figure 2 shows the SER and normal Raman spectra for KCN The SERS spectrum of 1mgml KCN in water shows a single intense somewhat broad feature at
A B
C D
SPIE-2003-5269 19
2100 cm-1 assigned to the single CequivN stretch The band is much sharper in the normal Raman spectra of the solid KCN salt at 2074 cm-1 This band does broaden and shift to 2080 cm-1 in solution (not shown) However the observed SERS frequency is attributed to interaction with silver and low frequency mode at 135 cm-1 attributed to a Ag-CN stretch (not shown) supports this conclusion
cm-1 band to a C-S stretch but the authors concede that it is in fact more likely a C-Cl stretch28 It appears that the most intense Raman bands at 648 692 and 747 cm-1 shift to 620 660 and 720 cm-1 in the SER spectra and are tentatively assigned as above The width of these bands suggests that they overlap underlying spectral features Additional bands in the Raman spectra occur at 972 1034 1049 1263 1286 1423 1442 2865 2935 and 2960 cm-1 Corresponding bands occur in the SER spectra at 964 1015 1054 1286 1410 1447 2865 and 2935 cm-1 Most of these bands are associated with alkane modes specifically the bands at approximately 1040 cm-1 to a C-C stretch 1290 cm-1 to a CH2 in-phase twist 1440 cm-1 to a CH2 wag 2865 cm-1 to a symmetric CH2 stretch and 2965 cm-1 to an asymmetric CH2 stretch The Raman and SER spectra of sulfur mustard were measured at the Edgewood center (Figure 4) Both spectra are largely similar to CEES The C-Cl and C-S bands in the Raman spectrum of HD now occur at 640 655 700 739 and 760 cm-1 and are more resolved possibly due to the increased molecular symmetry Theoretical calculations indicate that the first three bands are due to C-Cl stretching modes and the latter two to C-S stretching modes28 Only the C-Cl bands maintain significant intensity in the SER spectra occurring at 624 and 643 cm-1 which is attributed to the
Prior to measurements of HD CEES was examined by Raman and SER spectroscopy (Figure 3) CEES also known as half-mustard is essentially identical to HD except one of the chlorine end atoms is replaced by a hydrogen atom Again although not as toxic as HD CEES is a blister agent and dilute aqueous samples were prepared using appropriate safety equipment Both the Raman and SER spectra of CEES are similar and dominated by bands between 600 and 800 cm-1 These are associated with C-Cl and C-S stretching modes which are tentatively assigned to 648 and 747 cm-1 in the Raman spectra respectively The shoulder at 630 cm-1 the overlapped band at 660 cm-1 and the strong band at 692 cm-1 could also be due to these modes or their asymmetric counterparts It is worth noting that theoretical calculations assign the 692
Figure 2 A) SER and B) NR spectra of KCN Conditions A) 1 mgml in tap water 100 mW of 785 nm at sample 1-min acquisition time B) solid 300 mW of 785 nm 5-min All spectra are 8 cm-1 resolution
A
B
Figure 3 A) SER and B) NR spectra of CEES Conditions A) 1 vv (10 mgml) in MeOH 100 mW of 785 nm 1-min acquisition time B) neat 300 mW of 785 nm 5-min
A
B
Cl-CH2-CH2-S-CH2-CH3
A
B
Figure 4 A) SER and B) NR of HD Conditions A) 1mgml in tap water B) pure both 100 mW of 785 nm 1-min
Cl-CH2-CH2-S-CH2-CH2-Cl
SPIE-2003-5269 20
expected strong interaction between chlorine and silver and adds support to the assignment of this band to a C-Cl stretch Weaker overlapping bands occur at 670 692 and 724 cm-1 the latter possibly due to C-S stretching modes Again the alkane modes are apparent in the normal Raman spectra of HD but only a broad feature at 1300 to 1450 cm-1 suggests CH2 contributions in the SER spectrum Although the observed bands in the VX spectrum have not been assigned (Figure 4) a computer generated Raman spectrum29 predicts many of the same features with surprising accuracy and are used here Two intense bands at 460 and 530 cm-1 closely match predicted bands at 463 and 546 cm-1 assigned to a CH3-P=O bend and a PO2CS wag Three highly overlapped bands occur at 694 745 and 771 cm-1 matching predicted bands at 713 730 and 760 cm-1 The first
Table 2 Measured SER peak heights for the CN stretch at 2100 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
Figure 5 A) SERS and B) NR spectra of VX Conditions A) 1 vv (10 mgml) in MeOH B) pure sample both 100 mW of 785 nm 1-min acquisition time
A
B
two have been assigned to a C-S stretch and CH2bend respectively while the latter has been attributed to either a P-C stretch or an O-C-C stretch Although the 745 cm-1 band may alternatively be assigned to a C-S stretch based on the previous measurements of CEES and HD The relatively intense bands at 890 1106 1218 1445 and 1465 cm-1 also match predicted bands at 880 1108 1216 1440 and 1464 cm-1 that are assigned to a C-C stretch CH3 rock N-C3 stretch various C-H3 bends and C-H bends respectively Both the computer generated and the measured spectra contain numerous other less intense bands One is worth mentioning A unique band appears at 370 cm-1 that is predicted at 368 cm-1 and corresponds to an O-P=O bend The surface-enhanced Raman spectrum of VX is also rich with spectral features It has the unique low frequency band at 370 cm-1 as well as a second band at 380 cm-1 that is assigned tothe S-P-O bend predicted in the normal Raman spectrum at 388 cm-1 Based on the measured and predicted normal Raman spectra the following SERS assignments are given 460 cm-1 to the CH3-P=O bend 544 cm-1 to the PO2CS wag 738 cm-1 to a C-S stretch (based on arguments above) 890 cm-1
to a C-C stretch 1101 cm-1 to a CH3 rock and 1456
cm-1 to a C-H bend The ability of SERS to measure chemical agents in water containing real-world chemical interferents was tested by using the generic tap water described in Table 1 The ability to reproduce measurements was accomplished by preparing three separate water stock solutions which were used to prepare three sample solutions each of 1 mgmL KCN in the generic tap water The samples defined as vials 1-9 were then measured at 4 points per vial in a semi-random fashion such that errors associated with stock solution preparation and errors associated with instrument drift could be identified No trends were apparent that signified such systematic errors Each spectrum collected consisted of 20 averaged scans taking 16 seconds at 8 cm-1 resolution The laser power at the sample was measured periodically during the day and it ranged from 102 to 105 mW spacccc
spot Vial 1 Vial 2 Vial 41 13754 13004 128192 13519 12692 12909
1 3 13541 12721 126394 13462 12648 12651
ave 13569 12766 12755Vial 6 Vial 3 Vial 5
1 11586 14068 10757 2 11236 14402 115122 3 11376 14546 11559
4 10894 11763 11214ave 11273 13695 11261
Vial 7 Vial8 Vial 91 11198 15804 11023
2 11243 15716 88753 3 11562 15216 11218
4 11655 15084 11077ave 11415 15455 10548
CN AVG STDEV ERR1 mgmL pk ht 12526 1560 1245
stock solution
SPIE-2003-5269 21
The CN measurements consisting of 36 data points produced an average height of 125 for the 2100 cm-1 peak with a standard deviation of 156 or 125 (Table 2) The HD and VX measurements were performed precisely the same way (Tables 3 and 4) For HD the 624 cm-1 peak was used for analysis and it had an average height of 53 with a standard deviation of 068 or 129 while for VX the 544 cm-1 peak was used for analysis and it had an average height of 1051 with a standard deviation of 308 or 293 The greater error in the VX measurements can be somewhat attributed to Vial 4 which produced lower SER signal intensities But removing this vial from the data set changes the standard deviation to 233 only a modest improvement
spot Vial 1 Vial 2 Vial 41 525 609 4682 45 675 484
1 3 527 644 5094 559 807 669
ave 51525 68375 5325Vial 6 Vial 3 Vial 5
1 521 574 575 2 536 53 4492 3 509 508 451
4 565 594 379ave 53275 5515 4635
Vial 7 Vial 8 Vial 91 457 589 413
2 583 497 4273 3 544 505 4
4 528 5 509ave 528 52275 43725
HD AVG STDEV ERR1 mgmL pk ht 530 068 1291
stock solution
spot Vial 1 Vial 2 Vial 41 1464 1034 4672 1485 99 61
1 3 1491 989 5684 1041 777 553
ave 137025 9475 5495Vial 6 Vial 3 Vial 5
1 1058 942 1293 2 697 121 965
2 3 727 1206 11124 689 1254 746
ave 79275 1153 1029Vial 7 Vial 8 Vial 9
1 1127 783 163 2 1358 812 1614
3 3 1371 908 15124 1097 875 1397
ave 123825 8445 153825
VX AVG STDEV ERR1 mgmL pk ht 1051 308 2925
stock solution
4 CONCLUSIONS In this paper we examined the ability of surface-enhanced Raman spectroscopy to reproducibly measure CN HD and VX in tap water without chemical interference Both normal and surface-enhanced Raman spectra were examined to select unique bands suitable to identify and quantify these chemical agents For SER measurements the 2100 cm-1 C-N stretch was used for CN the 624 cm-1 C-Cl stretch was used for HD and the 544 cm-1 PO2CS wag was used for VX It was determined that 1 mgmL samples of each of these chemicals measured 36 times in glass vials coated with a silver-doped sol-gel reproduced measurements with standard deviations of 125 129 and 293 It was further found that the 10 chemicals added to simulate generic tap water did not interfere with or alter the SER spectra It should be noted that the concentrations used in this study were considerably greater than those required by the JSAWM program Current work involves improving SER sensitivity and designing sampling systems with better reproducibility This includes the development of fractal silver and gold structures within the sol-gel matrix and the development of chemically selective sol-gels
ACKNOWLEDGMENTS The authors are grateful for the support of the US Army (DAAD13-02-C-0015 Joint Service Agent Water Monitor program) The authors would also like to thank Janet Jensen Ronald Crosier and Kristina Gonser for helpful discussions
Table 4 Measured SER peak heights for the VX band at 544 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
Table 3 Measured SER peak heights for the HD band at 624 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
SPIE-2003-5269 22
REFERENCES 1 Jensen JL NC Wong W R Loerop SPIE 4775-03 (2001) 2 JSAWM Requirements at wwwsbccomapgeaarmymilRDAecbcrtPRODSERJSAWMjsawmhtml 3 Erickson B ldquoThe Chemical Weapons Convention Redefines Analytical Challengerdquo Anal Chem 70 397A (1998) 4 Johnston RL Hoefler CM Fargo JC and Moberley B ldquoThe Defense Nuclear Agencyrsquos
ChemicalBiochemical Weapons Agreements Verification Technology Research Development Test and Evaluation Program and its Requirements for On-Site Analysisrdquo AT-ONSITE 5-8 (1994)
5 Black RM Clarke RJ Read RW and Reid MT ldquoApplication of GC-MS and GC-tandem MS to the analysis of chemical warfare samples found to contain residues of the nerve agent sarin sulphur mustard and their degradation productsrdquo J Chrom 662 301-321 (1994)
6 USA Army Center for Health Promotion and Preventive Medicine (USACHPPM) Technical Guideline 230A 7 Hoffland LD Piffath RJ Bouck JB rdquoSpectral signatures of chemical agents and simulantsrdquo Optical
Engineering 24 982-984 (1985) 8 Braue EHJ Pannella MGrdquoCIRCLE CELL FT-IR Analysis of Chemical Warfare Agents in Aqueous Solutionsrdquo
App Spectrosc 44 1513-1520 (1990) 9 Tseng C-H Mann CK and Vickers TJ ldquoDetermination of Organics on Metal Surfaces by Raman
Spectroscopyrdquo App Spectrosc 47 1767-1771 (1993) 10 Staff Reporter ldquoUS Army Tests Lidar to Detect Biological Toxinsrdquo Photonic Spectra pg 50 December 1998 11 Christesen SD Raman cross sections of chemical agents and simulants App Spectrosc 42 318-321 (1988) 12 Weaver MJ Farquharson S Tadayyoni MA ldquoSurface-enhancement factors for Raman scattering at silver electrodesrdquo J Chem Phys 82 4867-4874 (1985) 13 Alak AM and Vo-Dinh T ldquoSERS of Organophosphorous Chemical Agentsrdquo Anal Chem 59 2149-2153 (1987) 14 Lee Y Farquharson S ldquoRapid chemical agent identification by SERSrdquo SPIE 4378 21-26 (2001) 15 Farquharson S Maksymiuk P Ong K Christesen S ldquoChemical agent identification by surface-enhanced Raman spectroscopyrdquo SPIE 4577 166-173 (2001) 16 Spencer KM Sylvia J Clauson S and Janni J ldquoSurface Enhanced Raman as a Water Monitor for Warfare
Agents in Waterrdquo SPIE 4577 158-165 (2001) 17 Tessier P Christesen S Ong K Clemente E Lenhoff A Kaler E Velev O ldquoOn-line spectroscopic characterization of sodium cyanide with nanostructured gold surface-enhanced Raman spectroscopy substratesrdquo
App Spectrosc 56 1524-1530 (2002) 18 Farquharson S W W Smith S Elliott and J F Sperry Rapid biological agent identification by surface- enhanced Raman spectroscopy SPIE 3855110-116 (1999) 19 Farquharson S WW Smith YH Lee S Elliott and J F Sperry Detection of biological signatures A
comparison of electrolytic and metal-doped sol-gel surface-enhanced Raman media SPIE 4575 62-72 (2002) 20 Guzelian AA Sylvia J Janni J Clauson S and Spencer KM ldquoSERS of whole cell bacteria and trace levels of biological moleculesrdquo SPIE 4577 182-192 (2001) 21 Shende C Inscore F Gift A Maksymiuk P Farquharson S ldquoRapid dipicolinic acid extraction from Bacillus spores detected by surface-enhanced Raman spectroscopyrdquo App Spectrosc 58 accepted 22 Lee Y and Farquharson S ldquoSERS Sample Vials Based on Sol-Gel Process for Trace Pesticide Analysisrdquo SPIE
4206 140-146 (2000) 23 Farquharson S and Lee Y ldquoTrace Drug Analysis by SERSrdquo SPIE 4200-16 (2000) 24 Lee Y Farquharson S and Rainey P M Surface-Enhanced Raman Sensor for Trace Chemical Detection in
Water SPIE 3857 76-84 (1999) 25 Lee Y Farquharson S Kwong H and Shahriari M ldquoSol-Gel Chemical Sensor for Surface-Enhanced Raman
Spectroscopyrdquo SPIE 3537 252-260 (1998) 26 Farquharson S A Gift P Maksymiuk F Inscore and W Smith ldquopH dependence of methyl phosphonic acid
dipicolinic acid and cyanide by surface-enhanced Raman spectroscopyrdquo SPIE 5269-19 (2003) 27 Lide DR Ed Handbook of Chemistry and Physics CRC Press 77th Ed (1996-7) 28 Sosa C RJ Bartlett K KuBulat and WB Person ldquoA theoretical study of harmonic vibrational frequencies and
infrared intensities of XCH2CH2SCH2CH2X and XCH2CH2SH (= H Cl)rdquo J Phys Chem 93 577-588 (1993) 29 Hameka HF and JO Jenson ldquoComputer-generated predictions of the structure and the IR and Raman spectra of
VXrdquo ERDEC-TR-065 May 1993
- Chemagents Appendicespdf
-
- SPIE2001-4575-Bioagent-SERSapdf
-
- Detection of bioagent signatures A comparison of electrolytic and metal-doped sol-gel surface-enhanced Raman media
- Stuart Farquharson Wayne Smith and Yuan Lee
-
- Real-Time Analyzers 87 Church Street East Hartford CT 06108
-
- ABSTRACT
- 4 CONCLUSIONS
- 5 ACKNOWLEDGEMENTS
- 6 REFERENCES
-
SPIE-4378-2001
25
4 CONCLUSIONS We have successfully measured the SER spectra of chemical agent simulants dimethyl metylphosphonate and 2-chloroethyl ethyl sulfide and chemical agent hydrolysis products pinacolyl methylphosphonate and methylphosphonic acid using silver-doped sol-gel coated sample vials Measurements were obtained in both aqueous and gas phase The P-C stretching mode was SER-active for all four chemicals allowing identification by class Within this group each chemical contained at least one unique spectral band that could be used for identification (Table 1) Furthermore these bands do not appear to coincide with SER spectra reported for organophosphorus pesticides the most likely source of false-positives Although surface enhancement factors appear to be an order of magnitude better than those previously presented in the literature for similar chemicals17 measurement sensitivity needs to be improved by 1 to 2 orders of magnitude to provide adequate warning of chemical agent use Current research efforts to increase surface-enhancement optical collection efficiency and instrument design are being pursued to achieve the required sensitivity
Table 1 Enhancement factors detection limits and unique SER bands fro chemicals studied Agent Simulant Enhancement Detection limit Unique bands (cm-1)
Dimethyl methylphosphonate 120000 90 microM (16 ppm) 425 2-Chloroethyl ethyl sulfide 62000 60 microM (22 ppm) 620
Methylphosphonic acid 110000 3 microM (60 ppb) 764 1042 Pinacolyl methylphosphonate 150000 70 microM (14 ppm) 546 788 1032
5 ACKNOWLEDGEMENTS The authors would like to thank Drs Janet Jensen and Steven Christesen of Aberdeen Proving Ground for encouraging this work They would also like to thank Advanced Fuel Research for making their laboratory facilities available
6 REFERENCES 1 ldquoThe Chemical Weapons Convention ndash A Guided Tour the Organization for the Prohibition of Chemical Weaponsrdquo at
httpwwwopcwnlguidehtm 2 Robinson JP and J Goldblat Chemical Warfare In The Iraq-Iran War Stockholm International Peace Research
Institute Fact Sheet at httpprojectssiprisecbwresearchfactsheet-1984html (1984) 3 ldquoChemistry of GB (Sarin)rdquo at httpwwwmitretekorgmissionenvenechemicalagentssarinhtml 4 Tu Anthony ldquoOverview of Sarin Terrorist Incidents in Japan in 1994 and 1995rdquo 6th CBW Protection Symposium
Stockholm Sweden 10-15 May 1998 5 Staff Reporter ldquoGoing out with a bangrdquo Newsweek June 28 1999 6 See UNSCOM reports in httpwwwunorgdeptsunscom (1999) 7 Treven T Saddamrsquos Secrets Harper Collins (1999) 8 ldquoDecaying Sarin-filled Rockets Spark Fearsrdquo Janersquos Defense Weekly 25(20)3 (1996) 9 ldquoThe Chemical Weapons Convention Redefines Analytical Challengerdquo Analytical Chemistry News amp Features June 1
397A (1998) 10 Johnston RL Hoefler CM Fargo JC and Moberley B ldquoThe Defense Nuclear Agencyrsquos ChemicalBiochemical
Weapons Agreements Verification Technology Research Development Test and Evaluation Program and its Requirements for On-Site Analysisrdquo AT-ONSITE 5-8 (1994)
11 Black RM Clarke RJ Read RW and Reid MT ldquoApplication of gas chromatography-mass spectrometry and gas chromatography-tandem mass spectrometry to the analysis of chemical warfare samples found to contain residues of the nerve agent sarin sulphur mustard and their degradation productsrdquo J Chromatography 662 301-321 (1994)
12 Hoffland LD Piffath RJ Bouck JBrdquoSpectral signatures of chemical agents and simulantsrdquo Optical Engineering 24 982-984 (1985)
13 Braue EHJ Pannella MGrdquoCIRCLE CELL FT-IR Analysis of Chemical Warfare Agents in Aqueous Solutionsrdquo Applied Spectroscopy 44 1513-1520 (1990)
SPIE-4378-2001
26
14 Tseng C-H Mann CK and Vickers TJ ldquoDetermination of Organics on Metal Surfaces by Raman Spectroscopyrdquo
Applied Spectroscopy 47 1767-1771 (1993) 15 Staff Reporter ldquoUS Army Tests Lidar to Detect Biological Toxinsrdquo Photonic Spectra pg 50 December 1998 16 Christesen SD Raman cross sections of chemical agents and simulants Applied Spectroscopy 42 318-321 (1988) 17 Alak AM and Vo-Dinh T ldquoSERS of Organophosphorous Chemical Agentsrdquo Analytical Chemistry 59 2149-2153
(1987) 18 Norrod KL Sudnik LM Rousell D and Rowlen KL ldquoQuantitative Comparison of Five SERS Substrates
Sensitivity and Detection Limitrdquo Applied Spectroscopy 51 994-1001 (1997) 19 Lee Y and Farquharson S ldquoSERS Sample Vials Based on Sol-Gel Process for Trace Pesticide Analysisrdquo SPIE 4206
140-146 (2000) 20 Farquharson S and Lee Y ldquoTrace Drug Analysis by Surface-Enhanced Raman Spectroscopyrdquo SPIE 4200-16 (2000) 21 Lee Y Farquharson S and Rainey P M Surface-Enhanced Raman Sensor for Trace Chemical Detection in Water
SPIE 3857 76-84 (1999) 22 Lee Y Farquharson S Kwong H and Shahriari M ldquoSurface-Enhanced Raman Sensor for Surface-Enhanced Raman
Spectroscopyrdquo SPIE 3537 252-260 (1998) 23 see Material Safety Data Sheets for details 24 Farquharson S Smith W Carangelo R C and Brouillette C ldquoIndustrial Raman Providing Easy Immediate Cost
Effective Chemical Analysis Anywhererdquo SPIE 3859 14-23 (1999) 25 Bennett S Bane J Benford P and Pratt R ldquoEnvironmental Hazards of Chemical Agent Simulantsrdquo Aberdeen
Proving Ground Maryland Chemical Research and Development Center CRDC-TR-84055 (1984) 26 Mabey W and Mill T Critical Review of Hydrolysis of Organic Compounds in Water under Environmental Conditions
Journal of Physics and Chemistry Reference Data 7(2) 383-414 (1978) 27 Christesen S MacIver B Procell L Sorrick D Carabba M and Bello J ldquo Noninstrusive Analysis of Chemical Agent
Identification Sets Using a Portable Fiber-Optic Raman Spectrometerrdquo Applied Spectroscopy 53 850-855 (1999) 28 Meylan WM and Howard PH J Pharm Sci 84 83-92 (1995) 29 Jenkins A Uy O and Murray G ldquoPolymer-Based Lanthanide Luminescent Sensor for Detection of Hydrolysis Product
of the Nerve Agent Soman in Waterrdquo Analytical Chemistry 71 373-378 (1999) 30 Nassar A Lucas S and Hoffland L ldquoDetermination of Chemical Warfare Agent Degradation Products at Low-Part-
per-Billion Levels in Aqueous Samples and Sub-Part-per-Million Levels in Soils Using Capillary Electrophoresisrdquo Analytical Chemistry 71 1285-1292 (1999)
SPIE 2001-4575
62
Detection of bioagent signatures A comparison of electrolytic and metal-doped sol-gel surface-enhanced Raman media
Stuart Farquharson Wayne Smith and Yuan Lee
Real-Time Analyzers 87 Church Street East Hartford CT 06108
Susan Elliott and Jay F Sperry University of Rhode Island 45 Lower College Rd Kingston RI 02881
ABSTRACT Since September 11 2001 the threat of terrorist attacks and biological warfare within US borders has become a sobering reality In an effort to aid military personnel and the public at large we have been investigating the utility of surface-enhanced Raman spectroscopy (SERS) to provide rapid identification of chemical agents directly and biological agents through their chemical signatures This approach is based on the ability of Raman spectroscopy to identify molecular structure through the abundant vibration information provided in spectra and the ability of SERS to detect extremely low concentrations (eg part-per-billion) through the enhancement of Raman scattering by six orders of magnitude or more Towards the goal of developing a portable analyzer we have been studying the ability of two SER media to obtain continuous (ie reversible) and quantitative (ie reproducible) measurements Here we compare measurements of nucleic acid-bases adenosine monophosphate and ribonucleic acid extracted from Escherichia coli Bacillus subtilis and Staphylococcus aureus obtained by electrolytic SERS and metal-doped sol-gel SERS The capabilities of these SER media are summarized in terms of rapid detection of B anthracis and dipicolinic acid Keywords bioagent detection SERS RNA analysis bacterial analysis Raman spectroscopy
1 INTRODUCTION The recent distribution of anthrax through the US postal system and the subsequent infection and death of several postal and national media employees amplifies the need for methods to rapidly detect and identify this and other chemical and biological warfare agents (BWA) The primary methods currently used immunoassays for screening and nucleic acid (NA) sequencing for positive identification of BWAs (bacteria protozoa and viruses) have serious limitations123 Immunoassay methods employ competitive binding of the bioagent (as an antigen) and its labeled (eg fluorescence) conjugate for a limited number of antibodies Although this analysis method is fast and semi-quantitative other chemicals may compete for the antibodies interfere with the enzymatic reaction or interfere with the measurement (eg it fluoresces) resulting in a high number of false positive responses1 Furthermore the antibodies denature due to moisture and heat limiting shelf life and require sterile often refrigerated storage Positive identification of a BWA can be accomplished by sequencing deoxyribonucleic acid or ribonucleic acid (DNA and RNA)23 This requires enumeration of a nucleic acid sequence through polymerase chain reactions (PCR) or multiplication of the microorganism through culture growth to provide sufficient quantities of DNA or RNA for analysis Unfortunately PCR and culture growth require from several hours to several days23 Consequently a wide variety of technologies have been investigated for rapid identification of BWAs The Department of Defense is actively monitoring 200 such technologies4 This includes traditional methods such as gas chromatographic separation coupled with ion mobility spectrometry detection5 to exotic methods based on nature such as monitoring toxin induced color changes in fish scales6 Although all of these techniques have achieved varying degrees of success none are yet capable of detecting and identifying BWAs in 10 minutes or less Towards this goal we have been investigating the ability of SERS to detect sub-nanogram quantities of DNA or RNA (eliminating enumeration) determine relative NA base concentrations and identify BWA taxonomy
To whom correspondences should be addresses e-mailfarqureal-time-analyzerscom wwwreal-time-analyzerscom
SPIE 2001-4575
63
Raman spectroscopy has a rich history of investigating biochemical and biological processes7 Some of the earliest laser-Raman studies demonstrated that the five NA bases adenine (A) cytosine (C) guanine (G) thymine (T in DNA) and uracil (U in RNA) yielded distinct spectra with several bands suitable for identification and quantification8 Furthermore these studies included exceptional spectra of both DNA and RNA for which the NA bases as well as several phosphate bands were easily identified9 However since the Raman effect is very inefficient (very low conversion of incident radiation to inelastically scattered Raman radiation) these samples had to be highly concentrated Fortunately two phenomena exist that can increase the generation of Raman photons by six orders of magnitude or more known as the resonance Raman and surface-enhanced Raman effects1011 Resonance Raman scattering occurs when the laser excitation wavelength is in resonance with an electronic transition of a molecule (within the absorption envelope)10 Excitation at ultraviolet wavelengths has been used to obtain resonance Raman spectra of amino acids and nucleic acids in whole bacteria1213 For example excitation at 242 nm has been used to maximize the nucleic acid spectral band intensities and minimize the amino acids band intensities A peak at 1530 cm-1 was found to be proportional to the amount of the NA base cytosine while a peak at 1485 cm-1 was proportional to the combined amount of the NA bases adenine and guanine This quantitative behavior has been used to define an A+TG+C base-pair ratio and provide a level of bacterial identification as taxonomic markers13 In recent years SERS has also been used to analyze bacterial cell components14 including amino acids15 lipids16 nucleic acids151718 and the adenine derivatives192021 SERS has proven to be one of the most sensitive methods for trace chemical analysis through the detection of single molecules2223 including DNA (dye labeled 17-mer)24 Since its discovery in 197425 the mechanism responsible for the large increase in scattering efficiency has been the subject of considerable research2627 Briefly incident laser photons couple to free conducting electrons within a metal which confined by the particle surface collectively cause the electron cloud to resonate2628 These surface plasmons are known as the physical component of the SER effect These surface plasmons can transfer energy to the molecular vibrational modes of molecules through interactions with the molecular electron orbitals2629 This interaction is known as the chemical component of the SER effect This perturbation of the molecular polarizability generates surface-enhanced Raman photons26 A number of methods have been developed to produce surfaces or solutions containing one of these metals with optimum roughness or diameter to promote SERS30 These methods include preparation of activated electrodes in electrolytic cells 1131 activated silver and gold colloid reagents32 and metal coated substrates333435 Selecting a SER-active medium for chemical and biological agent detection requires consideration of the method of deployment and hence the method of sampling Chemical aerosols or airborne bacteria will require a collection device to concentrate and continuously present the sample to the SERS medium Poisoned water supplies will also require a flow through device for continuous monitoring or a grab-sample device for periodic analysis And contaminated surfaces will require a grab-sample extractive device A SERS-based device used for continuous monitoring (air or water) must be reversible and reproducible if quantitative measurements are desired while a SERS-based device used for periodic sampling (water or surfaces) must be reproducible Both reversible and reproducible measurements have been performed using electrolytic SERS (E-SERS)36 But this requires a three-electrode sample cell and an electrolyte of known concentration to perform the necessary oxidation-reduction cycles (ORCs) to re-activate the electrode surface with new uncontaminated sites from one measurement to the next Colloids are severely limited in that continuous measurements would require a continuous supply of colloids For periodic measurements vials of colloids one per measurement could be used However aggregate size and consequently SER intensity change with sample conditions (especially pH) and quantitative reproducible measurements are unlikely Substrates appear to have the greatest potential and designs range from silver evaporated on titania particles34 to periodic gold pyramids evaporated between polystyrene beads35 Most substrates require concentrating the sample on the surface through drying to obtain the largest signal enhancements in effect making the measurements irreproducible and irreversible However successful measurements using flow systems have been obtained with glass posts but manufacturing costs appear prohibitive In an effort to overcome these limitations we have developed metal-doped sol-gels to provide SERS measurements that are reproducible reversible and quantitative and yet not restricted to specific environments such as electrolytes solvents or evaporated surfaces3738 The porous silica network of the sol-gel offers a unique environment for stabilizing SER active metal particles and the high surface area increases the interaction between the analyte and metal particles The sol-gel can be coated on the end of fiber optics or on the internal walls of a glass flow tube for continuous measurements or standard glass sample vials for periodic measurements Previously we measured 100 mgL methylphosphonic acid (the primary hydrolysis product of nerve agents) in water with an estimated detection limit of 05 mgL (100 parts-per-billion) We have also
SPIE 2001-4575
64
demonstrated reversible and reproducible measurements of p-aminobenzoic acid (PABA) in a flow through system Here we investigate the ability of the sol-gel SERS (SG-SERS) to measure the NA bases adenosine monophosphate and RNA extracted from E coli B subtilis and S aureus The measurements are compared to those obtained by E-SERS
2 EXPERIMENTAL The inorganic chemicals and solvents used to prepare samples were spectroscopic grade and purchased from Aldrich (Milwaukee WI) Fisher (Pittsburgh PA) or Pfaltz amp Bauer (Waterbury CT) The nucleic acid bases and dipicolinic acid were purchased from Sigma (St Louis MO) Normal Raman samples were measured to establish enhancement factors In each case 1cm3 of sample was placed into a 1x1 cm glass cuvette weighed and measured Unpacked densities were typically 6-7 gcm3 For all SER measurements including RNA samples were prepared as ~01mgmL (see Figure captions for exact concentrations) in 01M KCl and buffered to a pH of 92 with Na2B4O7bullH2O Adenine pH dependence measurements used pH buffer solutions at 4 (potassium acid phthalate) 69 (potassium phosphate monobasicsodium phosphate dibasic) 92 (Na2B4O7bullH2O) and 104 (tris-hydroxymethyl amino methane) Escherichia coli Bacillus subtilis and Staphylococcus aureus cultures (250ml per 1000mL Erlenmeyer flask) were grown overnight in a Trypticase soy broth (TSB) medium containing 1 glucose in a shaking water bath at 37 oC The bacteria were harvested by centrifugation for 10 minutes at 8000 rpm in a GSA rotor at 5degC then washed once in 085 saline The gram-positive bacteria were concentrated to 20 ml and passed through a French pressure cell twice at 15000 psi to break open the cells RNA was extracted according to Protocol 44139 to ensure pristine samples for initial measurements Since this method takes approximately 4 hours a streamlined method was developed For vegetative bacteria the specimen was boiled for 30 sec in 1 ml of distilled water to lyse the cells and release the RNA For bacterial spores the specimen was first incubated in 1 ml of saline solution containing 02 mg lysozyme and phosphate-buffered to pH of 624 for 1 hr at 37 oC This solution was then boiled for 2-3 minutes in 4X loading buffer to release the RNA For both specimens RNA STAT-60TM was added to the supernatant which was centrifuged at 12000 g for 5 minutes to precipitate the ~15 water-soluble proteins This procedure allowed extracting RNA for SER analysis in ~ 10 minutes Electrophoresis shows high purity while the existence of chemicals that could interfere with the SER measurements is still under investigation The electrolytic sample cell has been described previously36 Briefly a three electrode design is incorporated into a Plexiglas structure containing a well for the reference electrode (a saturated calomel electrode Cole Parmer Vernon Hills IL) and a 5mL sample well containing the silver working electrode and platinum wire counter electrode (05 mm wire Alfa Ward Hill MA) A channel connecting the two wells contained a 2 mm diameter semi-porous membrane (10-20 micron pore Ace Glass) The silver electrode was made from a 3 mm length of 2 mm diameter silver wire (Alfa) soldered to a copper wire lead encased in a 4 mm diameter Pyrex tube A cap containing the silver electrode platinum wire and nitrogen purge and vent lines fixed the silver electrode surface 1 mm from a 1 mm thick glass plate attached to the bottom of the sample well The potentiostat used to control the three electrodes was built in-house and has been described in detail elsewhere36 A multifuntional analog digital and timing inputoutput interface card (DAQCard-1200 National Instruments) is used to both drive the electrolytic cell as well as read the current generated in the cell A LabVIEW software program is used set the oxidation potential reduction potential SER measurement potential hold times and sweep rates The amount of charge passed was plotted as a cyclic voltammogram For all spectra presented five oxidation-reduction cycles (ORCs) stepping from -03 VSCE to 03 VSCE and back to -3 VSCE at 50 mVsec were used The SG-SER measurements were accomplished by simply placing the identical samples prepared above into Simple SERS Sample VialsTM (RTA) These 2-mL glass vials are internally coated with ~ 01 micron thick silver-doped sol-gel A prototype Fourier transform Raman spectrometer (RTA) was used for these measurements40 The system consisted of a NdYAG laser (Brimrose or Spectra Physics) for excitation at 1064 nm an interferometer built by On-Line Technologies (OLT East Hartford CT) for frequency separation an uncooled InGaAs detector for signal detection (RTA) and an Intel 400 MHz Pentium II based laptop computer (Dell Round Rock TX) for interferometric control data acquisition (OLT) and analysis (LabVIEW by National Instruments Austin TX) Additional components included a Notch filter (Kaiser Ann Arbor MI) and interferometer entrance and exit optics (Edmund Scientific Barrington NJ) Fiber optics were used to deliver the excitation beam to the sample and the scattered radiation to the interferometer (2 meter lengths of 200 and 365 micron core diameter respectively Spectran Avon CT) A second Notch filter (Kaiser) was used as a beam splitter to direct the excitation beam along the same axis as the collected radiation A microscope object (20x magnification 04 numeric aperture Newport Irvine CA) was used to focus the beam into the sample and to collect the scattered radiation back along the same axis In this co-axial backscattering arrangement the excitation beam passed through the glass plate onto the silver
SPIE 2001-4575
65
electrode surface for E-SERS through the vial glass wall and into the silver-doped sol-gel film for SG-SERS or through the glass wall of the cuvette and into the solid sample for normal Raman spectroscopy All E-SERS and normal Raman spectra were obtained with 750 mW of laser power at the sample while all SG-SERS spectra were obtained with 75 mW of laser power at the system Incident powers above 200 mW in some cases degraded the sol-gel
3 RESULTS AND DISCUSSION The generation of surface-enhanced Raman scattering at electrode surfaces has been extensively researched and the optimum sample conditions are well developed2729 Several researches incorporated electrodes into flowing systems and demonstrated that continuous monitoring of chemicals is possible18 These successes suggested investigated the capability of E-SERS to measure the NA bases and RNA The E-SERS measurements also provided a benchmark to compare and evaluate SG-SERS measurements The molecular structure of adenine (as well as the other base pairs) which includes an aromatic nitrogen-containing heterocycle is ideally suited to interact with the surface plasmons and contribute substantially to the chemical component of the SER effect1119 Even with excitation at 1064 nm a 3-minute scan of 18x10-5M adenine yields high signal-to-noise (SN) E-SER spectra and all of the bands are revealed with clarity (Figure 1 Table 1) Spectra of similar quality were obtained by SG-SERS and the principal spectral bands are easily observed The identical 18x10-5M adenine sample was measured in the same 3-minute time frame but with 110th the laser power The lower power appears to reduce the SN The amount of adenine responsible for the SER spectra as well as enhancement factors for the two SER media can be determined The molecules producing the E-SERS spectrum are those on the electrode surface within the illumination area of the laser (The solution concentration only determines the number of molecules available to adsorb to the electrode surface) For the current experiments the laser illuminates an area of 28x10-7m2 or 56 x10-7m2 if we assume the ORCs increase the surface area by a factor of two Furthermore if we assume monolayer coverage on the electrode and each 3x5 angstrom molecule (lying flat) occupies 15x10-19m2 then there are ca 4x1012 molecules contributing to the Raman scattering This is ca twice the number of molecules measured at electrode surfaces using either differential capacitance-potential curve measurements or rapid linear sweep voltammetry (eg 3x1018 moleculesm2 for pyridine and pyrazine)29 Thus the adenine spectrum in Figure 1 is due to 87x10-10g (6x10-12 moles) A detection limit defined as a SN of 3 can also be calculated The SN for a 3-minute scan is 844 for the 735 cm-1 band suggesting a mass detection limit of 3x10-12g (2x10-14 moles) This is consistent with previous estimates for adenine by others of 25 x10-14 moles1530 However sub-monolayer concentrations must be measured to verify this The root-mean-squared (RMS) noise is measured between 4400-4600 cm-1 Since noise is distributed evenly throughout the spectrum when transformed this region does not have any
Wavenumbers (∆cm-1) Figure 1 A) Normal Raman spectrum of pure adenine powder B) E-SERS and C) SG-SERS of 18x10-5M adenine at pH 92 All spectra 8 cm-1 resolution 200 scans (3 min) and 1064 nm excitation A) and B) 750 mW C) 75 mW B) measurement potential of 11VSCE
A
B
C
A
B
C SG-SERS
pH 10
D
725
735
735
pH 4
Wavenumbers (∆cm-1) Figure 2 A) and C) E-SERS and B) and D) SG-SERS of adenine at A) and B) pH 104 and C) and D) pH 40 Note consistent appearance of bands at 1270 and 1375 cm-1 as the pH is changed to 10 for both SER media E-SERS used 750 mW SG-SERS used 75 mW of 1064 nm excitation
SPIE 2001-4575
66
contributions from signals or baseline offsets The measurement error is given as SplusmnRMS and for adenine this equals 234 The number of molecules contributing to the SG-SERS are those on the silver particles that are embedded in the sol-gel The total silver surface area can be determined from the average particle size (40 nm diameter) concentration (073 by weight based on molar conc and measured sol-gel density) and the scattering volume (a cylinder defined by the laser area 28x10-7m2 and sol-gel thickness10-4m) The 61x109 silver particles in this volume have a collective area of 31x10-5m2 However it may reasonably be assumed that at least half of the silver surface is in contact with the sol-gel matrix and unavailable for analyte interaction Then approximately 10x1014 molecules or 22x10-8g of adenine contribute to the SG-SER spectrum The slightly lower SN of 207 suggests a mass detection limit of 32x10-10g Determination of the enhancement factors for the two SER media requires estimating the number of adenine molecules contributing to the normal Raman spectrum Here a cylindrical scattering volume is assumed again based on the laser area (28x10-7m2) and the penetration depth (1x10-3 m)41 The density of the sample was measured at 064 gcm3 indicating that 18x10-4g (13x10-6 moles) of adenine produced the normal Raman signal The enhancement factor EF is defined by the following equation
EF = (ISERSINR)bull(MNRMSERS) bull(PNRPSERS) bull(TNRTSERS)12
where I is the spectral band intensity (here 735 cm-1) M is the sample mass P is the incident laser power and T is the measurement time (or number of scans) For the E-SERS measurement the enhancement factor is 22x105 (01780184) bull (18x10-487x10-10)) while the SG-SERS enhancement factor is 10x105 (0160184) bull(18x10-422x10-8) bull(75075) bull(315)12) The lower enhancement for the SG-SERS may be real or the available surface of the silver embedded in the sol-gel may have been overestimated In addition to enhancing the Raman scattering efficiency to an extent similar to E-SERS the SG-SER medium also yields an identical shift of the adenine ring-breathing mode from 725 cm-1 in the normal Raman to 735 cm-1 Furthermore in the course of optimizing the E-SERS sample conditions it was found that pH influenced the adenine interaction with the silver surface (Figure 2) In particular the relative band intensities of the pyrimidine ring skeletal vibrations at 1270 and 1375 cm-1 and the imidazol ring skeletal vibration at1335 cm-1 change At pH 4 adenine is protonated presumably the imidazol ring since the band at 1335 cm-1 increases in intensity while the pyrimidine bands are virtually absent Conversely at pH 10 the imidazol band decreases in intensity while the pyrimidine bands appear It is worth noting that the ring-breathing mode at 735 cm-1 changes little between pH 4 and 10 suggesting that the skeletal changes are more a function of molecule-plasmon interactions than reorientation of the molecule on the surface Measurements of the identical pH series of adenine samples by SG-SERS yielded virtually identical spectral changes This suggests that the sol-gel does not influence the measurement This is critical to reproducing measurements and performing quantitative analysis Next the remaining NA bases were measured by both E-SERS and SG-SERS and compared Previously we examined the optimum pH and electrode potentials for E-SERS measurements to determine if a common pH could be used that yielded good sensitivity for all the bases and if variations in potential could be used to provide an added degree of selectivity between the bases Primarily it was found that high quality spectra were obtained between pH 7 and 95 and that cytosine and uracil were best enhanced at potentials positive of the potential-of-zero charge (pzc ca 065VSCE for Ag) guanine and thymine near the pzc and adenine negative of the pzc In all cases the ring-breathing modes were the most intense and in general could be used to identify the NA bases (Figure 3 Table 1) Specifically adenine has an intense band at 735 cm-1 cytosine at 797 cm-1 guanine at 653 cm-1 thymine at 784 cm-1 and uracil at 800 cm-1 The adenine cytosine guanine and thymine bands are sufficiently separated that their contributions to DNA should be determinable Although adenine and guanine contributions to RNA should also be determinable cytosine and uracil are highly overlapped and unfortunately share the same potential dependence Alternate unique bands at 1183 cm-1 for cytosine and 1275 cm-1 for uracil might be suitable for calculating contributions The SG-SER spectra of the remaining NA bases faithfully reproduced the E-SER spectra In particular the primary identifying bands occur at virtually the same wavenumbers (see Table 1) However the spectra for both cytosine and thymine contain an intense band at ca 1040 cm-1 Initially this was attributed to the pH buffer but samples prepared without either the buffer or the 01M KCl electrolyte yielded identical spectra containing this band In fact the E-SER and SG-SER spectra of thymine are virtually identical except for this band Also the SG-SERS of guanine contains an intense band at 1551 cm-1 that is not observed in the E-SER spectrum This band may be due to a moderately intense band at 1553cm-1 in the normal Raman spectrum that is SG-SER active It was also found that the SG-SERS of cytosine was considerably better than the E-SERS while uracil showed the opposite relationship It is also worth noting that all of the SG-SERS were obtained with 110th the laser power Most importantly the primary ring-breathing modes in the SG-SER spectra are sufficiently intense and unique to be used in determining contributions to DNA and RNA as outlined above
SPIE 2001-4575
67
Table 1 Comparison of E-SER and SG-SER Spectral Band Positions for the NA Bases and Adenosine Monophosphate
Adenine Cytosine Guanine Thymine Uracil AMP E-SERS SG-SERS E-SERS SG-SERS E-SERS SG-SERS E-SERS SG-SERS E-SERS SG-SERS E-SERS SG-SERS1647 1638 1634 1656 1655 1630 1587 1585 1510 1580 1551 1539 1456 1456 1465 1460 1462 1480 1453 1459 1394 1398 1425 1431 1435 1399 1404 1392 1374 1375 1373 1383 1370 1335 1332 1311 1307 1333 1331 1353 1348 1331 1329 1265 1273 1280 1292 1278 1276 1275 1279 1271 1183 1195 1222 1232 1221 1219 1204 1205 1180 1144 1097 1033 1029 1038 1040 1035 1051 1037 1041 1035 963 963 957 1001 1000 961 944 884 819 817 859 866 735 737 797 799 784 782 800 800 72738 742 630 630 653 664 667 684 603 602 590 611 466 561
Bands unique to SG-SERS The next chemical to be analyzed by both E-SERS and SG-SERS was adenosine monophosphate (AMP) The E-SER spectrum yields bands due to the adenine chemical functionality at 727 961 1233 1279 1331 1381 and 1486 cm-1 In addition phosphate bands are observed at 860 1097 1453 1587 and 1705 cm-1 (Figure 5) Other researchers have noted that the ribose component does not appear to contribute to the spectrum19 The AMP spectrum also changes as a function of potential As the electrode is swept more positive (here from -09 to -03VSCE) the phosphate bands at 860 1097 1453 and 1587 cm-1 increase in intensity compared to the adenine bands while a band at 1705 cm-1 appears The adenine bands at 1233 1381 and 1486 cm-1 virtually disappear These potential dependent spectral changes are consistent with earlier studies that show that phosphate is attracted to silver at potentials positive of the pzc but repelled at potentials negative of the pzc19
Wavenumbers (∆cm-1) Figure 3 E-SERS of A) 21x10-3M cytosine at -03VSCE 1000 scans B) ~10x10-5M guanine at -06VSCE 500 scans C) 23x10-3M thymine at -06VSCE 500 scans and D) 12x10-3M uracil at -093VSCE 500 scans All spectra at pH 92 750 mW 1064 nm at 8 cm-1
Wavenumbers (∆cm-1) Figure 4 SG-SERS of A) 21x10-3M cytosine 200 scans B) ~10x10-5M guanine 200 scans C) 23x10-3M thymine 200 scans and D) 12x10-3M uracil 500 scans All spectra at pH 92 75 mW 1064 nm at 8 cm-1
A
B
C
A
B
C
D D
SPIE 2001-4575
68
The SG-SER spectrum of AMP is considerably different The adenine bands virtually disappear except for the two primary bands which shift to742 and 1329 cm-1 While the phosphate band at 1459 cm-1 has gained considerable intensity In addition two new intense bands appear at 684 and 1539 cm-1 as well as a moderately intense band at 1180 cm-1 The SG-SER spectrum has greater similarity to the E-SER spectrum at -03VSCE and suggests that the silver particles embedded in the sol-gel behave as if at a potential positive of the pzc RNA samples extracted from E coli B subtilis and S aureus were next examined by both E-SERS and SG-SERS E-SER spectra of these samples yielded quality spectra in 10 minutes in which all of the major features can be identified (Figures 6 and 7) This includes guanine at 650 cm-1 adenine at 791 cm-1 cytosine and uracil combining at 790 cm-1 and phosphate at 1100 1335 (in combination with adenine and guanine) 1465 and 1570 cm-1 Surprisingly adenine which demonstrated the greatest surface-enhanced Raman effect does not dominate the ring-breathing mode portion of the spectrum The intensities of the other base-pairs bands are of the same order of magnitude This suggests that when the base-pairs are linked together as in RNA they are enhanced in concert In fact the relative intensities are very similar to a normal Raman spectrum of E coli RNA which shows the combined cytosine and uracil band at ca twice the intensity of the adenine band and ca four times the intensity of the guanine band Unfortunately this means that the independent enhancement factors for the NA bases can not be used to estimate relative concentrations For example the relative 791 and 734 cm-1 bands for B subtilis would indicate that the cytosine andor uracil concentration was at least 20 times the adenine concentration whereas each of the four RNA bases are known to contribute 15-35 Nevertheless it is worth noting that the three RNA samples yield different relative band intensities that were reproduced in numerous measurements Although the relative concentrations of the NA bases for these samples have not been determined these differences can be quantified If it is assumed that the 650 cm-1 band represents 25 guanine the 791 cm-1 band represents 25 adenine and the 790 cm-1 50 cytosine plus uracil in the E coli RNA spectrum then the relative concentrations can be estimated for the other RNA samples To aid this calculation the three spectra were normalized to the phosphate band at 1100 cm-1 which has been shown to correlate to the total phosphate concentration and can be used as an internal standard In addition a simple baseline correction was applied (Figure 7) This yields 15 adenine 30 guanine and 55 cytosine plus uracil for B subtilis RNA and 18 adenine 25 guanine and 57 cytosine plus uracil for S aureus RNA The average SN of these measurements was 26 with an average error of 8 of the value (SplusmnN) It is also worth noting that the three RNA spectra show a marked shift in a band near 825 cm-1 This band is assigned to the symmetric stretch of the O-P-O ester linkage9 The band appears at 815 cm-1 for S aureus shifting to 820 cm-1 for B subtilis and 830 cm-1 for E coli Others have used the normal Raman intensity of the band at 815 cm-1 as a direct indication of the amount of A-class helix present while the intensity of the band at 830 cm-1 has been used as a direct indication of the amount of B-class helix present However the latter is more associated with DNA than RNA
Wavenumbers (∆cm-1) Figure 5 E-SER spectra of 020 mgmL adenosine monophosphate at A) -03 and B) -09VSCE and C) SG-SER spectra Conditions sample in 01M KCl buffered to pH 92 A) and B) 750 mW C) 75 mW of 1064 64 scans (1-min) at 8 cm-1
Wavenumbers (∆cm-1) Figure 6 E-SERS of 01 mgmL RNA from E coli 02 mgmL RNA from B subtilis and 02 mgmL RNA from S aureus Conditions 01M KCl pH 92 -03VSCE 750 mW of 1064 nm 640 scans (10 min) at 8 cm-1
AMP RNA
E coli A
B
C
B subtilis
S aureus
SPIE 2001-4575
69
SG-SER spectra of reasonable quality were also obtained for E coli and B subtilis especially the latter (Figure 8) However the spectra differ substantially from the E-SERS of the same samples Both SG-SER spectra are dominated by adenine at 735 cm-1 and a band at 1030 cm-1 Although unassigned the latter does appear in the RNA E-SER spectra Bands at 1105 and 1565 cm-1 are likely due to phosphate while bands at 1320 and 1470 cm-1 are less confidently assigned to phosphate They are significantly less intense and somewhat shifted from their SG-SERS counterparts (1335 and 1455 cm-1) A band at 670 cm-1 may be due to guanine which was observed at 664 cm-1 for SG-SERS of the pure sample However the SG-SER spectrum of AMP also had an intense 667 cm-1 band A number of other bands occur at 890 1070 1165 1245 1290 1420 1505 cm-1 and remain unassigned The SG-SER spectra are somewhat disappointing in that only adenine and guanine contributions can be positively identified This limits the ability to determine relative NA base concentrations and distinguish bacterial RNA However several of the unassigned bands may be due to the bases (eg 1030 and 1420 cm-1 due to cytosine) Further experiments will be required to clarify this point
Ecoli
B subtilis
S aureas
P G
A C+U
OPO
A B
Wavenumbers (∆cm-1) Wavenumbers (∆cm-1) Figure 7 SER spectra of RNA from A) B subtilis with contributions indicated and B) E coli B subtilis and S aureus with baseline correction and peak positions used to calculate contributions indicated G = guanine A = adenine C+U = cytosine plus uracil P = phosphate (backbone) OPO = phosphate ester linkage (A- vs B-class helix)
Wavenumbers (∆cm-1) Wavenumbers (∆cm-1) Figure 9 E-SER (-03VSCE) and SG-SER spectra of RNA from A) E coli and B) B subtilis Sample conditions as in Figure 6 E-SER spectra at 750 mW SG-SERS at 75 mW
A B E-SERS
SG-SERS
E-SERS
SG-SERS
SPIE 2001-4575
70
A final comparison was made between the two SER methods by measuring dipicolinic acid (DPA) This chemical may be invaluable as a test for spore forming bacteria specifically B anthracis 50 to 90 of B anthracis sporilates During spore formation dipicolinic acid is synthesized and once completed 10-15 of the dry spore weight is composed of the Ca2+ complex located in the spore core42 Heating in water can be used to initiate germination at which point the exosporium breaks and releases the Ca dipicolinate which becomes dipicolinic acid in water The structure of this chemical strongly
4 CONCLUSIONS Towards the goal of developing a method and technology to rapidly detect and identify bioagents we have been investigating surface-enhanced Raman spectroscopy as a tool to measure relative concentrations of nucleic acid bases in RNA and determine bioagent taxonomy Initially we investigated E-SERS since this method has been extensively researched and the optimum sample conditions are well developed However this method requires a three-electrode sample cell and electrolyte solution Incorporation of an electrolytic cell into sample systems used to analyze for BWA as aerosols in water or on surfaces can be designed using flow injection analysis technologies but cross-contamination and plugging of sample lines seems inevitable For this reason we also investigated metal-doped sol-gels as a SER-active medium Previous studies have shown this material to be active in all solvents particularly water capable of continuous measurements in flowing systems and reproducible (quantitative) between coated sample vials Here we compared SG-SER spectra to traditional E-SER spectra of the nucleic acid base pairs adenosine monophosphate and RNA High quality spectra of adenine cytosine guanine thymine and uracil were obtained by both E-SERS and SG-SERS Both methods yielded very similar spectra for the NA bases including a pH dependent study of adenine Enhancement factors and detection limits for adenine were determined as 2x105 and 16x10-11g and 1x105 and 12x10-10g for E-SERS and SG-SERS respectively (normalized to 75 mW and 10-min acquisition time) Fifty percent of the silver particle surface area in the sol-gel matrix was assumed covered by adenine which may have been overestimated yielding a lower EF and higher detection limit It should also be realized that each E-SER spectrum required several attempts to optimize the measurement conditions (pH electrode potential etc) While each SG-SER spectrum involved no sample preparation and often represents the first and only attempt to make the measurement Quality spectra of RNA extracted from Escherichia coli Bacillus subtilis and Staphylococcus aureus were obtained by E-SERS that were easily interpreted Bands due to adenine guanine cytosine plus uracil and phosphate were identified The SER band intensity of the NA bases in the RNA samples were of the same order of magnitude suggesting that their interaction with the silver surface is concerted as is their Raman enhancement Interestingly the relative SER band
Wavenumbers (∆cm-1) Figure 10 A) Raman spectrum of solid dipicolinic acid B) Ax20 C) electrolytic SERS of 6x10-3 M dipicolinic acid in 01 M KCl at a potential of +07VSCE and pH of 4 and D) sol-gel SERS of 6x10-3 M dipicolinic acid Conditions for A and C as in Figure 1 C) 100 mW of 1064 nm 50 scans 8 cm-1
A B (Ax20)
C
D
suggested that it would be SER active However the E-SER spectrum was unstable and varied considerably as a function of potential A consistent spectrum was obtained at +06VSCE (Figure 10) This potential is not recommended for measurement because the surface is actively dissolving in solution The SG-SER spectrum was considerably more stable of higher quality and easily reproduced Bands at 660 825 1010 1390 1430 1570 1590 and 3075 cm-1 were observed Enhancement factors were determined for the two media using the symmetric ring stretching mode at 995 and 1010 cm-1 for the normal Raman and SER spectra respectively E-SERS yielded an EF of 5x103 while SG-SERS yielded an EF of 2x105 for DPA The SN of the latter suggests a detection limit of 20x10-10g (based on adenine coverage 75 mW and 10-min) The differences in SER activity for these two media may be attributed to the combined electrolytic potential of the solution chemical and metal15 Again the E-SERS suggests that the SG-SERS is at a potential positive of the pzc While the instability in the E-SERS may also be associated with surface interactions of two carboxylic acid groups of dipicolinic acid during the ORCs
DPA
SPIE 2001-4575
71
intensities for RNA extracted from E coli are very similar to those measured by normal Raman spectroscopy Although the relative percent that each of the NA bases contributed to each RNA sample was not determined reproducible band intensities allowed noting the following trends The percent adenine decreases while the combined percent cytosine and guanine increase for both B subtilis and S aureus compared to E coli Quality spectra were also obtained for the RNA samples by SG-SERS but only a few bands were readily identified Calculations of NA base concentrations by SG-SERS will require further research In light of recent events we summarize the capabilities of these SERS media in terms of rapid detection of B anthracis and dipicolinic acid However these capabilities must be qualified First and foremost the level to which SERS can distinguish bacteria or viruses has not yet been determined Development of a database of both DNA and RNA base concentrations for BWAs and common bacteria to establish the level of taxonomic identification is ongoing Second rapid collection of aerosol water or surface samples is being addressed by others who report trapping particles on filters from 100 liters of air per minute Third although not presented here we have developed methods to extract RNA or DNA from cells and spores for SER analysis within 10 minutes Finally we assume a detection limit of 3600 spores per 100 liters of air is required although a 50 lethal dosage of anthrax has not been established With these qualifications a mass detection limit for RNA using SERS is estimated as follows A single measurement is performed in ca 20 minutes (140 liters collected in 14 min RNA extracted in 8 min spectral acquisition and analysis in 10 min) The average human breaths 7 liters per minute therefore the analyzer must at the very minimum detect 5000 spores in 140 liters of air One spore is approximately 2x10-18m3 (1x1x2 microm) and if a density of 075 gcm3 is assumed this corresponds to a mass of 15x10-12g Each spore contains 4-12 RNA or 12x10-13g RNA for 8 If we assume 23 of the RNA can be isolated for analysis during lysis then the proposed instrument must be able to detect 4x10-10g RNA from 5000 spores per 70 liters of air within 10 minutes As noted above the mass detection limits for adenine were estimated at 16x10-11g and 12x10-10g for E-SERS and SG-SERS respectively Although these detection limits suggests that RNA from 5000 spores is detectable with the current instrumentation it is highly likely that only a portion of an RNA segment (eg 120-nucleotide 5S rRNA) is in contact with the metal surface and will contribute to the SER effect The SN for the RNA spectra were 110th of the average SN for the four individual RNA bases suggesting a 10 contribution Furthermore effective taxonomy will likely require knowing the NA base concentrations to 1 of the value (eg 25plusmn025) Again the average measurement error for the bases is 12 These values suggest that the E-SERS is within a factor of 4 of the required detection limit whereas the SG-SERS detection limit must be improved by 25 times The same arguments can be applied to the detection of dipicolinic acid If we assume a spore releases 10 by weight DPA during germination then the proposed instrument must be able to detect 75x10-10g DPA from 5000 spores per 70 liters of air within 10 minutes The detection limit for SG-SERS was estimated at 20x10-10g and suggest that the vials are suitable to perform a rapid screen for anthrax A series of concentration dependent measurements are currently being performed to verify this assertion Finally we note that the measurements performed here employed an FT-Raman spectrometer This instrumentation was chosen over dispersive Raman instrumentation for the high wavelength accuracy afforded by the HeNe reference laser (Connes Advantage43) which would allow reliable spectral subtraction matching of observed spectra to stored library spectra and confident use of chemometric approaches Such data analysis is likely to be required to enhance BWA identification However this instrumentation sacrifices sensitivity due to a less efficient detector (InGaAs vs Si) less efficient Raman scattering and less efficient generation of plasmon modes Substantial improvements can be obtained using visible excitation and Si detection and these measurements are underway
5 ACKNOWLEDGEMENTS The authors are grateful to Drs D Cookmeyer and S Tove of the US Army Research Office (Contract Number DAAH04-96-C-0078) for their interest and support of this research The authors would also like top acknowledge Dr R Yin and J Jensen for supporting development of the metal-doped sol-gels (Contract Number DAAD13-01-C-0019) They also thank Dr Wilfred H Nelson for assistance in spectral interpretations
SPIE 2001-4575
72
6 REFERENCES 1 Roberts WL and Rainey PM Clin Chem 39 1872-1877 (1993) 2 Pasechnik VA CC Shone and P Hambleton Bioseperations 3 267-283 (1993) 3 Jackson PJ ME Hugh-Jones DM Adair G Green KK Hill CR Kuske LM Grinberg FA Abramova and P
Keim Proc Natl Acad Sci 95 1224-1229 (1998) 4 Jensen JL NC Wong W R Loerop SPIE 4775-03 (2001) 5 Snyder A Pet al SPIE 3853-15 (1999) 6 Danosky T R and McFadden P N in press (1997) 7 Woodruff WH Farquharson S Science 201 831-833 (1978) 8 Lord RC and Thomas GJJr Spectrochemica Acta 23A 2551-2591 (1967) 9 Thomas GJJr Biochim Biophys Acta 213 417-423 (1970) 10 Placzek G Handbuch der Radiologie 2 EMarx ed Akademische Verlagagescellschatt Liepzig 1934 UCRL
Trans No 526 (1959) 11 Jeanmaire DL and RP Van Duyne J Electroanalytical Chem 84 1-20 (1977) 12 Chada S Manoharan R Moenne-Loccoz P Nelson WH Peticolas WL and Sperry JF Applied Spectroscopy
47 38-43 (1993) 13 Manoharan R Ghiamati E Chada S Nelson WH and Sperry JF Applied Spectroscopy 47 2145-2150 (1993) 14 Todd EA Morris MD Applied Spectroscopy 48 545-548 (1994) 15 Wentrup-Byrne E Sarinas S and Fredericks PM Applied Spectroscopy 47 1192-1197 (1993) 16 Weldon MK VR Zhelyaskov and MD Morris Applied Spectroscopy 52 265-269 (1998) 17 Kneipp K and J Fleming J Mol Structure 145 173-179 (1986) 18 Pothier NJ and Force RK Applied Spectroscopy 46 147-151 (1992) 19 Ervin KM E Koglin JM Sequaris P Valenta and HW Nurnberg J Electroanal Chem 114 179-194 (1980) 20 Kim SK TH Joo SW Suh and MS Kim J Raman Spectrosc 17 381-386 (1986) 21 Pothier NJ and Force RK Analytical Chemistry 62 678-680 (1990) 22 Kneipp K Y Wang RR Dasari and MS Feld Applied Spectroscopy 49 780-784 (1995) 23 Nie S and Emory SR Science 275 1102 (1997) 24 Graham D WE Smith AMT Linacre CH Munro ND Watson and PC White Analytical Chemistry 69 4703-
4707 (1997) 25 Fleischmann M PJ Hendra and AJ McQuillan Chem Phys Lett 26 163-166 (1974) 26 Pettinger B J Chemical Phys 85 7442-7451 (1986) 27 Surface-Enhanced Raman Scattering Section Four Theory SPIE MS 10 M Kerker and B Thompson Eds (1990) 28 Wang D-S and Kerker M SPIE (M Kerker and B Thompson Eds) MS 10 417-429 (1990) 29 Weaver MJ Farquharson S Tadayyoni MA J Chem Phys 82 4867-4874 (1985) 30 Norrod KL Sudnik LM Rousell D and Rowlen KL Applied Spectroscopy 51 994-1001 (1997) 31 Farquharson S Weaver WJ Lay PA Magnuson RH and Taube H J Am Chem Soc 105 3350-3351 (1983) 32 Lee PC and Meisel D ldquoAdsorption and Surface-Enhanced Raman of Dyes on Silver and gold Solsrdquo J Phys Chem
86 3391-3395 (1982) 33 Li Y-S and Wang Y Applied Spectroscopy 46 142-146 (1992) 34 Bello JM DL Stokes and T Vo-Dinh Analytical Chemistry 61 1779-1783 (1989) 35 van Duyne RP J C Hulteen D A Treichel M T Smith M L Duval and T R Jensen J Phys Chem B1033854-
3863 (1999) 36 Farquharson S and W W Smith W H Nelson and J F Sperry SPIE 3533-27 207-214 (1998) 37 Lee Y H W Smith S Farquharson H C Kwon M R Shahriari and P M Rainey SPIE 3537 252-260 (1998) 38 Lee Y-H S Farquharson and P M Rainey SPIE 3857 76-84 (1999) 39 Current Protocols in Molecular Biology Wiley Interscience1003-1006 (1987) 40 Farquharson S Smith W Carangelo R C and Brouillette C SPIE 3859 14-23 (1999) 41 Chase D B and JF Rabolt Fourier Transform Raman Spectroscopy Acad Press Ch1 p 131 (1994) 42 Brock TD MT Madigan JM Martinko and J Parker Biology of Microorganisms 7th Ed Prentice Hall p 76-80
(1994) 43 Connes J Rev Opt Theor Instrum 40 45 (1961)
166
Chemical agent identification by surface-enhanced Raman spectroscopy
Stuart Farquharson and Paul Maksymiuk
Real-Time Analyzers 87 Church Street East Hartford CT 06108
Kate Ong and Steven D Christesen US Army SBCCOM Aberdeen Proving Ground MD 21010
ABSTRACT The recent distribution of anthrax through the US postal system and the subsequent infection and death of several postal and national media employees amplifies the need for methods to rapidly detect identify and quantify this and other chemical and biological warfare agents The US military has also identified water supplies as a likely method of warfare agent deployment and is funding the development of a Joint Service Agent Water Monitor (JSAWM) In an effort to aid military personnel and the public at large we are developing a portable analyzer capable of identifying and quantifying chemical agents rapidly either on-demand or continuously The approach is based on the ability of Raman spectroscopy to identify molecular structure through the abundant vibration information provided in spectra and the ability of surface-enhanced Raman spectroscopy (SERS) to detect extremely low concentrations (eg part-per-billion) through the enhancement of Raman scattering by six orders of magnitude or more A key element to the analyzer design is a new SER active medium that is capable of quantitative reversible measurements The medium consists of silver or gold nanoparticles incorporated into a sol-gel matrix The porous silica network offers a unique environment for stabilizing SER active metals and the high surface area increases the interaction between the analyte and metal particles Here we present the use of new sol-gels that also selectively enhance chemicals based on polarity and charge Base-line measurements of chemical agents and their hydrolysis products are presented and compared to the JSAWM goal of 30 micrograms per liter detection Keywords Chemical warfare agent hydrolysis product SERS Raman spectroscopy sol-gel nanoparticle
1 INTRODUCTION Since September 11 2001 the threat of terrorist attacks and biological warfare within US borders has become a sobering reality The simplicity in manufacturing ease of deployment and the relatively low cost of chemical warfare agents (CWAs) raises public concern that they may also be used by terrorists Indeed terrorists released sarin (GB) in the Tokyo subway in 19951 Countering terrorism and terrorist attacks requires recognizing likely deployment scenarios and having the required technology to rapidly detect the deployment event One method of deployment has been long identified by the US military distribution through water supplies To counter this threat the Department of Defense is funding or monitoring the capabilities of 200 technologies with the goal of developing a Joint Service Agent Water Monitor (JSAWM) that is field portable2 The JSAWM must be capable of identifying the chemical agents with no false-positives and quantifying the agents at microgram per liter concentrations within 10 minutes (Table 1)3 This includes the analysis of drinking water supplies distribution and storage systems as well as potable water supplies The earliest technologies developed for CWA detection were based on electrochemical ionization or colorimetric analysis (eg phosgene tape) Although these analyzers were easy to use they were not generally agent specific and suffered from false-positives4 More traditional laboratory methods have also been investigated and in particular combined gas chromatography and mass spectrometry (GCMS) has been very successful at eliminating false-positives56 However GCMS requires extraction repeated calibration and long analysis times (typically 20 to 60 minutes)6 making it labor intensive and less than desirable for field use More rapid analysis of agents in the solid liquid and gas phase has been To whom correspondence should be addressed emailfarqureal-time-analyzerscom
Vibrational Spectroscopy-based Sensor Systems Steven D Christesen Arthur J Sedlacek III EditorsProceedings of SPIE Vol 4557 (2002) copy 2002 SPIE middot0277-786X02$1500
167
Table 1 Chemical Agent Structures Hydrolysis Half-lives and JSAWM Thresholds Agent Short-Hand Chemical Structure Hydrolysis
Half-Life JSAWM
Thresholds
Sarin (GB) F-[O=P-CH3]-O-CH(CH3)2 213 hours 32 microgL Soman (GD) F-[O=P-CH3]-O-CH(CH3)-(C-(CH3)3) 23 hours 32 microgL Tabun (GA) (CH3)2-N-[O=P-CN]-O-C2H5 41 hours 32 microgL VX C2H5O-[O=P-CH3]-S-(CH2)2-N-(CH(CH2)2)2 821 hours 32 microgL EA2192 HO-[O=P-CH3]-S-(CH2)2-N-(CH(CH2)2)2 gt9 years 32 microgL Mustard (H) ClCH2CH2-S-CH2CH2Cl encapsulates 47 microgL Lewisite (L) ClCH=CH-As-Cl2 rapid 27 microgL HCN HCN rapid 20 mgL BZ C7NH12-O-[C=O]-COH(C6H5)2 23 microgL T-2 Toxin 87 microgL at pH 7 to 75 and 20 to 25 oC
demonstrated by vibrational spectroscopy7-10 Hoffland et al7 reported infrared absorbance spectra and absolute Raman cross sections for several chemical agents while Christesen measured Raman cross sections for sarin tabun mustard gas and VX (ethyl S-2-diisopropylamino ethyl methylphosphonothioate)11 Again however these techniques also have limitations Raman spectroscopy is simply not a very sensitive technique and detection limits are typically 01 (1000 ppm) While infrared spectroscopy would have limited value in analyzing poisoned water since the very strong infrared absorption of water would obscure most other chemicals present Nevertheless efforts to overcome these limitations have been demonstrated Braue and Pannella8 quantified the G-series nerve agents (tabun sarin and soman) in terms of infrared attenuated total reflectance using a circle-cell And Alak and Vo-Dinh demonstrated the possibility of surface-enhanced Raman spectroscopy (SERS) to identify CWAs by measuring several organophosphonates that simulate the nerve agents12 However quantitative measurements have not been demonstrated for the SER-active material used (silver coated on alumina particles) or other SER-active media13 Recently we developed silver-doped sol-gels to promote the SER effect14-17 The porous silica network of the sol-gel matrix offers a unique environment for stabilizing SER-active metal particles and the sol-gel provides a high surface area that effectively increases the number of molecules observed within the Raman scattering volume The silver-doped sol-gels have been coated on the internal walls of standard glass vials to allow rapid SER analysis of small sample quantities (lt 01 mL) without preparation We have used p-aminobenzoic acid (PABA) as a test chemical to demonstrate surface enhancements greater than 106 reversible measurements in a flowing system reproducible measurements from vial-to-vial and batch-to-batch and measurements in multiple solvents including water14-17 Recently we used these vials to measure Tabun (GB) and Sarin and several hydrolysis products pinacolyl methyl phosphonate (PMP from Soman) and methyl phosphonic acid (MPA from all G-agents Figure 1) Although a number of unique vibrational bands are observed (eg C-N stretch doublet and P-C stretch) the G-agents were only observed for 5 concentrations and all spectra required baseline corrections Figure 1 Surface enhanced Raman spectra of ~5 vv A) Tabun and B) Sarin C) 1 vv PMP and D) 10 ppm MPA using sol-gel sample vials 785 nm excitation 1-min scan and CCD detection Performed at Aberdeen Proving Ground
C A
Wavenumber (∆cm-1) Wavenumber (∆cm-1)
D B
790
545 C-N
2135 2190 P-C 770
760 1290
Proc SPIE Vol 4577
168
Nevertheless MPA was readily observed for a 10 ppm sample with an estimated detection limit of 04 ppm (based on a signal-to-noise ratio of 3 for the 760 cm-1 band intensity) This measurement provides encouragement in that SERS may satisfy the needs of a JSAWM Furthermore MPA is also a hydrolysis product of VX and V-gas and EA2192 (Figure 2) and may prove a valuable indicator of agent usage Figure 2 Hydrolysis of Sarin to form hydrofluoric acid (HF) methylphosphonic acid 1-methylethyl ester (MPAMME) methyl phosphonic acid (MPA) and 2-propanol With this initial albeit modest success we began analyzing chemicals with various sol-gel compositions that we have been developing Here we describe four sol-gel compositions that select for 1) polar-positive 2) polar-negative 3) weakly polar-positive and 4) weakly polar-negative chemical species The ability of these sol-gels to select and enhance Raman scattering is described for several test chemicals and MPA
2 EXPERIMENTAL The chemicals analyzed as well as all chemicals used to prepare the metal-doped sol-gels were obtained at their purest commercially available grade from Aldrich (Milwaukee WI) The sol-gel designed to select for polar-negative species was prepared from a silver amine complex tetramethyl orthosilicate (TMOS) and methanol After mixing 02 mL of the sol-gel solution was transferred into a glass vial (2 mL) dried and heated The incorporated silver ions were then reduced using dilute sodium borohydride The vials were washed and dried prior to the addition of a sample solution In a similar manner the sol-gel designed to select for polar-positive species was prepared from a gold salt TMOS and methanol The sol-gel designed to select for weakly polar-negative species was prepared from a silver amine complex tetraethyl orthosilicate (TEOS) and methanol And the last sol-gel designed to select for weakly polar-positive species was prepared from a gold salt TEOS and methanol All samples were prepared in a chemical hood and transferred into plain or SER-active vials for analysis Normal Raman spectral measurements employed 1-mL pure samples that were placed in a 1-cm3 cuvette and weighed This yielded a powder density that allowed accurate calculation of molecules in the optical collection field SERS measurements employed 1-mg sample per mL water concentrations unless otherwise stated Once prepared a 01 mL sample was placed into one of the four selective sample vials which in turn was placed into the sample compartment of a Raman spectrometer for analysis A prototype Fourier transform Raman spectrometer (RTA) was used for these measurements18 The system consisted of a NdYAG laser (Brimrose) for excitation at 1064 nm an interferometer built by On-Line Technologies (OLT East Hartford CT) for frequency separation an uncooled InGaAs detector for signal detection (RTA) and an Intel 400 MHz Pentium II based laptop computer (Dell Round Rock TX) for interferometric control data acquisition (OLT) and analysis (LabVIEW by National Instruments Austin TX) Additional components included a Notch filter (Kaiser Ann Arbor MI) and interferometer entrance and exit optics (Edmund Scientific Barrington NJ) Fiber optics were used to deliver the excitation beam to the sample and the scattered radiation to the interferometer (1 meter lengths of 200 and 365 micron core diameter respectively Spectran Avon CT) A second Notch filter (Kaiser) was used as a beam splitter to direct the excitation beam along the same axis as the collected radiation A microscope object (20x04 Newport Irvine CA) was used to focus the beam into the sample and to collect the scattered radiation back along the same axis In this co-axial backscattering arrangement the excitation beam was passed through the outside of a glass vial and focused onto the silver-doped sol-gel film (01-03 mm thickness) containing the sample
3 RESULTS AND DISCUSSION p-aminobenzoic acid (PABA) and phenyl acetylene (PA) and were used to refine the selectivity and SER-activity of the four different metal-doped sol-gels PABA is a popular chemical used to evaluate the performance of SER-active media Here the polar end groups can be used to test selectivity of the polar-negative and polar-positive sol-gels PA is essentially non-
2O+ H HF + +OH
OH3H C
OP
OF
CH
CH3
3
3H C
OCP
OOH
CH
CH3
3
3H C
O
CPHO
CH
CH
3
3
C
Sarin MPAMME MPA 2-propanol
Proc SPIE Vol 4577
169
polar but a high electron density in the cylindrical π cloud around the carbon-carbon triple bond allows testing the selectivity of the weakly polar-negative and weakly polar-positive sol-gels As Figure 3 illustrates PABA passes through the polar sol-gel and is enhanced by either the silver or gold particles At 1 mgml the concentration of neutral PABA is ca 20 times that of the ionized form (pKa = 48) For electropositive silver the PABA anion is expected to interact through the carboxylate group and the associated vibrational modes are expected to dominate the spectrum Conversely for electronegative gold either form of PABA is expected to interact through the amine group The clear differences in our spectra support this expectation Furthermore bands at 840 and 1405 cm-1 assigned to a COO- bend and stretch respectively are significantly more intense for silver than gold Additional bands at 1140 and 1195 cm-1 are assigned to CH bending modes while bands at 1450 1500 and 1605 cm-1 are assigned to ring vibrational modes A very similar SER spectrum for PABA on a silver-coated alumina substrate has previously been reported with similar assignments19 For the gold-doped sol-gel new bands appear at 690 1355 and 1585 cm-1 The first band is assigned to a ring-H bending mode the second band to a ring-N- stretching mode and the third band to a possible NH2 scissors mode or ring mode The second band is not observed in the normal Raman spectrum but infrared bands occur at this frequency for aromatic ring-secondary amine stretching modes The scissors mode occurs at this frequency in Raman spectra for several chemicals but is absent in the PABA Raman spectrum Alternatively this mode may be the1600 cm-1 ring mode that has been shifted by the gold interaction Again a very similar SER spectrum of PABA has been reported but surprisingly using silver (colloids)2021 not gold as the enhancement medium These researchers also assumed the primary interaction of PABA with silver was through the carboxylate anion and made assignments accordingly For example they assigned the 1359 cm-1 to a COO- stretch not to the amine group as we have They also favor the ring stretching mode assignment for the 1582 cm-1 band Finally it should be said that other researchers have argued that the most dominant band in the SER spectra at 1450 cm-1 a ring vibration mode suggests that PABA lies flat on the surface and the π-orbitals dominate the surface interaction22
Figure 3 SER spectra of A) PABA using polar-negative and B) polar-positive sol-gels and C) PA using weakly polar-negative and D) weakly polar-positive sol-gels PABA is 1 mgmL PA is 1 vv Spectral conditions 75 mw 1064 nm 100 scans (15 min) 8 cm-1 resolution Non-polar PA passes through the non-polar sol-gels and is also enhanced by both metals The spectra are easily understood For electropositive silver PA interacts through the cylindrical triple bond π electron cloud and a -CequivC- doublet occurs near 2000 cm-1 The interaction is reasonably strong since this band appears at 2112 cm-1 in the normal Raman spectrum For electronegative gold this interaction is unlikely and only very weak bands occur near 2000 cm-1 The remaining bands are at 1000 cm-1 1200 cm-1 doublet and 1595 cm-1 all appear in the normal Raman spectra at virtually the same frequencies and are assigned to the symmetric ring-breathing mode CH bending modes and the trigonal ring-breathing mode respectively The polarnon-polar selectivity of the polar-negative and weakly polar-negative sol-gels was tested by adding a 11 molar mixture of PABA and PA The selective enhancement is quite good (Figure 4) The spectrum obtained using the polar sol-gel represents 78 PABA and 22 PA while the spectrum obtained using the weakly polar sol-gel represents 9 PABA and 91 PA The band peak intensities at 2000 cm-1 for PA and 1450 cm-1 for PABA were used for these calculations and are expanded in Figure 4 for clarity
C A
Wavenumber (∆cm-1) Wavenumber (∆cm-1)
2NH COOH
D B
C CH
Proc SPIE Vol 4577
170
Figure 4 SERS of 11 MM of PABA and PA in A) polar-negative and C) weakly polar-negative sol-gels The lower traces compare the pure chemicals B) 1 mgml PABA in polar-negative sol-gel and D) 1 PA in weakly polar-negative sol-gel while the insets magnify the minority species for clarity (x5 in A and x10 in B) Spectral conditions as in Figure 3 Following this development of selective sol-gels that maintained SER activity we measured cyanide and MPA (Figure 5) Not surprisingly the best sensitivity for both hydrolysis products was obtained using the polar-negative sol-gel The interaction of the cyanide anion with the silver surface is sufficient to shift the CequivN stretch observed at 2080 cm-1 in the normal Raman spectrum to 2145 cm-1 in the surface-enhanced Raman spectrum Furthermore the band is substantially broadened This anion has been extensively studied by electrolytic SERS and this shift and broadening have been attributed to the formation of a tetrahedral Ag(CN)3
2- surface structure23 Figure 5 Surface-enhanced (upper traces) and normal Raman spectra (bottom traces) of A) CN- and B) MPA in silver-doped TMOS SERS conditions as in Figure 3 and 1 mgmL Note MPA yields two distinct spectra for neutral (top) and acidic pH (middle) The normal Raman spectra employed pure powders 500 scans and 900 mW of 1064 nm SER measurements of MPA with the polar-negative sol-gel yielded two unique spectral signatures that depended on solution pH (Figure 5) For more neutral solutions the P-C stretch of MPA at 762 cm-1 dominates and the CH2 stretch at 2922 cm-1 appears The SN is sufficiently high that the anti-Stokes Raman shift at -762 cm-1 is observed For deprotonated MPA an oxygen-surface mode appears at 325 cm-1 (as well as its anti-Stokes complement) suggesting a strong interaction This results in substantial enhancement of the P-O-C mode at 1051 cm-1 (upper trace) Others report that this mode dominates the infrared spectra of nerve agents measured in water8 Comparison of the two spectra suggests the following molecule-to-surface orientations The appearance of the oxygen-surface and P-O-C modes in the upper spectrum of Figure 5B indicates that the tetrahedral molecule interacts with the silver surface through the deprotonated oxygen and is oriented end-on The
A B
Wavenumber (∆cm-1)
C A
Wavenumber (∆cm-1) Wavenumber (∆cm-1)
D B
Wavenumber (∆cm-1)
Proc SPIE Vol 4577
171
dominance of the P-C and the CH2 stretches and the disappearance of the P-O-C mode in the upper spectrum suggest the molecule is oriented with the methyl-hydroxyl-hydroxyl face towards the surface However considerably more research must be performed to verify these points Both chemical forms were measured as a function of concentration to determine the onset of monolayer coverage and estimate expected detection limits (Figure 6) Below monolayer coverage the signal to concentration dependence should be linear and the SN of any spectral measurement in this range can be used to predict the detection limit In the spectra presented here the peak height was used as the signal while the noise as root-mean-squared (RMS) was measured between 4400-4600 cm-1 Since noise is distributed evenly throughout the spectrum when transformed this region was used since it does not have any contributions from signals or baseline offsets Figure 6 shows a series of spectra for MPA along with a plot of the 1050 cm-1 band intensity (the noise was identical for this series) as a function of sample concentration A clear discontinuity appears in the vicinity of 01 mgmL (19 ppm) indicating the onset of monolayer coverage A detection limit defined as a SN of 3 was calculated for the 01 and 005gmL samples at 24x10-4 and 25x10-4 gL respectively A more modest detection limit of 101x10-4 gL was obtained using the 760 cm-1 band in the second series of concentration measurements These detection limits correspond to 46 and 190 ppb for a 3-min scan and 75 mW of incident laser power
Figure 6 A) Concentration dependence of MPA SERS measured in silver-doped TMOS) B) Concentrations are 001 005 01 05 1 gL (188 94 188 94 188 ppm) I760 series (bull) and I1050 series (∆)
Enhancement factors for cyanide and methyl phosphonic acid can be determined by estimating the number of molecules contributing to the surface-enhanced and normal Raman spectra The enhancement factor EF can be defined by the following equation
EF = (ISERSINR)bull(MNRMSERS) bull(PNRPSERS) bull(TNRTSERS)12
where I is the spectral band intensity M is the sample mass P is the incident laser power and T is the measurement time (or number of scans) for the two measurements For the normal Raman spectra a cylindrical scattering volume is assumed based on the laser area (28x10-7m2 6x10-4m diameter spot) and the penetration depth (1x10-3 m)24 The density of KCN and MPA as powders were measured at 0572 and 0516 gcm3 indicating that 16x10-4 and 144x10-4 g produced the normal Raman signals in Figure 5 respectively The number of molecules contributing to the SER spectra are those on the silver particles embedded in the sol-gel The total silver surface area can be determined from the average particle size concentration and the scattering volume Previous scanning electron micrographs showed the average silver particle size to be 40 nm in diameter (335x10-23m3)17 The silver concentration is 012M based on the reactant molar concentrations and dilution factors And the scattering volume is 76x10-
11m3 again based on a cylindrical scattering volume defined by a laser area of 28x10-7m2 and a sol-gel thickness of 27x10-
4m This volume contains 123x10-6g of silver equivalent to 35x109 silver particles with a collective surface area of 18x10-
5m2 However it may reasonably be assumed that at least half of the silver surface is in contact with the sol-gel matrix and unavailable for analyte interaction If we assume monolayer coverage and that each CN molecule occupies 15x10-20m2 then approximately 62x1014 molecules or 27x10-8g of CN contribute to the SER spectrum (20x10-19m2 46x1013 molecules 74x10-9g for MPA) Accordingly the EF for cyanide equals 48x104 ((180599) bull(16x10-427x10-8) bull(90075) bull(500100)12) The EF for MPA is considerably higher at 87x106 ((60326) bull(144x10-474x10-9) bull(90075) bull(500200)12)
0
20
40
60
80
100
120
140
0 02 04 06 08 1 12[MPA] (mgmL)
I (76
0)
0
100
200
300
400
500
600
I (10
50)
Wavenumber (∆cm-1)
A B
Proc SPIE Vol 4577
172
4 CONCLUSIONS Here we present for the first time surface-enhanced Raman spectra of Tabun and Sarin obtained using silver-doped sol-gels However the inferior enhancement suggested employing SER-active sol-gels varying in composition to improve both detection limits and selectivity of the target analytes To this end we successfully demonstrated the capabilities of four sol-gels that select for 1) polar-positive 2) polar-negative 3) weakly polar-positive and 4) weakly polar-negative chemical species p-aminobenzoic acid was used to show that silver could be used to attract polar-negative chemicals or functional groups (carboxylate anion) and that gold could be used to attract polar-positive chemicals (amine) while a mixture of p-aminobenzoic acid and phenylacetylene was used to show that tetramethyl orthosilicate preferentially solvates polar chemicals and that tetraethyl orthosilicate preferentially solvates weakly polar chemicals This increased sample control was applied to cyanide and methyl phosphonic acid two hydrolysis products of chemical warfare agents Exceptional results were obtained for methyl phosphonic acid allowing measurement of 1x10-2 gL for a 3-min scan and 75 mW of incident laser power with an estimated detection limit of 245x10-4 gL and an enhancement factor of 87x106 However this detection limit is 76 times less sensitive than required for the JSAWM (32x10-6gL for the G-agents) Finally we note that the measurements performed here employed an FT-Raman spectrometer This instrumentation was chosen over dispersive Raman instrumentation for the high wavelength accuracy afforded by the HeNe reference laser (Connes Advantage25) which would allow reliable spectral subtraction matching of observed spectra to stored library spectra and confident use of chemometric approaches Such data analysis is likely to be required to identify the chemical agents as well as distinguish hydrolysis products However this instrumentation which employs 1064 nm excitation and InGaAs detection sacrifices sensitivity We believe that the measurement sensitivity can be improved by at least two orders of magnitude by using shorter laser excitation wavelengths (eg 532 nm) This would provide more efficient Raman scattering (fourth power dependence on laser excitation wavelength) more efficient generation of plasmon modes and allow using more efficient detector material (Si vs InGaAs) These modifications are underway
5 ACKNOWLEDGEMENTS The authors would like to thank Dr R Yin and J Jensen of the US Army for supporting this work (Contract Number DAAD13-01-C-0019) They would also like to thank Advanced Fuel Research for making their laboratory facilities available
6 REFERENCES 1 Tu Anthony ldquoOverview of Sarin Terrorist Incidents in Japan in 1994 and 1995rdquo 6th CBW Protection Symposium
Stockholm Sweden 10-15 May 1998 2 Jensen JL NC Wong W R Loerop SPIE 4775-03 (2001) 3 JSAWM Requirements at wwwsbccomapgeaarmymilRDAecbcrtPRODSERJSAWMjsawmhtml 4 ldquoThe Chemical Weapons Convention Redefines Analytical Challengerdquo Analytical Chemistry News amp Features June 1
397A (1998) 5 Johnston RL Hoefler CM Fargo JC and Moberley B ldquoThe Defense Nuclear Agencyrsquos ChemicalBiochemical
Weapons Agreements Verification Technology Research Development Test and Evaluation Program and its Requirements for On-Site Analysisrdquo AT-ONSITE 5-8 (1994)
6 Black RM Clarke RJ Read RW and Reid MT ldquoApplication of gas chromatography-mass spectrometry and gas chromatography-tandem mass spectrometry to the analysis of chemical warfare samples found to contain residues of the nerve agent sarin sulphur mustard and their degradation productsrdquo J Chromatography 662 301-321 (1994)
7 Hoffland LD Piffath RJ Bouck JBrdquoSpectral signatures of chemical agents and simulantsrdquo Optical Engineering 24 982-984 (1985)
8 Braue EHJ Pannella MGrdquoCIRCLE CELL FT-IR Analysis of Chemical Warfare Agents in Aqueous Solutionsrdquo Applied Spectroscopy 44 1513-1520 (1990)
9 Tseng C-H Mann CK and Vickers TJ ldquoDetermination of Organics on Metal Surfaces by Raman Spectroscopyrdquo Applied Spectroscopy 47 1767-1771 (1993)
10 Staff Reporter ldquoUS Army Tests Lidar to Detect Biological Toxinsrdquo Photonic Spectra pg 50 December 1998
Proc SPIE Vol 4577
Proc SPIE Vol 4577
173
11 Christesen SD Raman cross sections of chemical agents and simulants Applied Spectroscopy 42 318-321 (1988) 12 Alak AM and Vo-Dinh T ldquoSERS of Organophosphorous Chemical Agentsrdquo Analytical Chemistry 59 2149-2153
(1987) 13 Norrod KL Sudnik LM Rousell D and Rowlen KL ldquoQuantitative Comparison of Five SERS Substrates
Sensitivity and Detection Limitrdquo Applied Spectroscopy 51 994-1001 (1997) 14 Lee Y and Farquharson S ldquoSERS Sample Vials Based on Sol-Gel Process for Trace Pesticide Analysisrdquo SPIE 4206
140-146 (2000) 15 Farquharson S and Lee Y ldquoTrace Drug Analysis by Surface-Enhanced Raman Spectroscopyrdquo SPIE 4200-16 (2000) 16 Lee Y Farquharson S and Rainey P M Surface-Enhanced Raman Sensor for Trace Chemical Detection in Water
SPIE 3857 76-84 (1999) 17 Lee Y Farquharson S Kwong H and Shahriari M ldquoSurface-Enhanced Raman Sensor for Surface-Enhanced Raman
Spectroscopyrdquo SPIE 3537 252-260 (1998) 18 Farquharson S Smith W Carangelo R C and Brouillette C ldquoIndustrial Raman Providing Easy Immediate Cost
Effective Chemical Analysis Anywhererdquo SPIE 3859 14-23 (1999) 19 Narayanan VA JM Bello JD Stokes and T Vo-Dinh Analusis 19 307-310 (1991) 20 Laserna JJ E L Torres and JD Winefordner Analytica Chemica Acta 469-480 (1987) 21 Torres EL and JD Winefordner Analytical Chemistry 59 1626-1632 (1987) 22 Suh JS DP DiLella M Moskovits J Phys Chem 87 1540-1544 (1983) 23 Benner RE R Dornhaus R Chang and BL Laube Correlations in the Raman spectra of cyanide complexes adsorbed
at silver electrodes with voltammograms Surface Science 101 341 (1980) 24 Chase D B and JF Rabolt Fourier Transform Raman Spectroscopy Acad Press Ch1 p 131 (1994) 25 Connes J Rev Opt Theor Instrum 40 45 (1961)
APPLIED SPECTROSCOPY 351
focusing the 488 nm laser beam 2 mm deep into thebulk of the crystals to avoid contributions from potentialdepletion layers As shown in Fig 1b the dependence ofthe Raman shift on the degree of deuteration is almostperfectly linear and ts very well with D 5 22684cmR1 24526 where D is the degree of deuteration (in )and R is the spectral mean of the PO4 vibration in cm21A linear correlation coef cient of 0998 indicates an ex-cellent linear dependence of the Raman peak shift withdegree of deuteration This result shows that the shift ofthe PO 4 peak is simply caused by the linear increase inatomic mass due to isotope substitution which decreasesthe length of hydrogen-like bonds
This excellent linear dependence allows us to map thepro le of the DH exchange layer at the surface of DKDPcrystals by acquiring Raman spectra and determining theposition of the PO 4 peak for various depths This methodis preferable over other methods such as determining thestrength of the OD vibration directly (eg at 715 cm21)because the position of the most intense peak in the Ra-man spectrum can be measured more precisely than theintensity of some of the weakest peaks in the spectrumThis is demonstrated in Fig 2 where depth-dependentRaman spectra (Fig 2a) and the resulting exchange layerpro les for two DKDP crystals are shown (Fig 2b) Thespectra in Fig 2a were obtained from a depth scan of aDKDP crystal with 75 degree of deuteration in thebulk grown at 45 8C The spectra start out as DKDP with30 deuteration close to the surface and approach thebulk DKDP spectrum within a few micrometers of depthThe fact that the relative degree of deuteration does notextend to 0 D is due to the limited depth resolution ofthe Raman microprobe which averages over 4 mm indepth Figure 2b depicts the resulting DH exchange layerpro les for this and a second crystal grown at 63 8Crespectively Both crystals had the same exposure to am-bient conditions and their main difference is the temper-ature at which they were grown The different exchangelayer pro les indicate that crystals grown at differenttemperatures have differen t proton conductiv ities 13
which leads to a difference in their rate of deuteriumdepletion The parameters controlling this behavior arecurrently the objective of a detailed study the results ofwhich will be reported elsewhere
CONCLUSION
In conclusion we have shown that the shift of the to-tally symmetric PO 4 stretch mode in the Raman spectrumof DKDP crystals scales linearly with degree of deuter-ation This allows us to correlate Raman peak positionsto deuteration levels in these crystals We have presenteda new technique to determine DH diffusion pro les inDKDP frequency conversion crystals based on micro-Ra-man spectroscopy This technique is fast inexpensiveand works under various environmental conditionswhich will allow us to better understand and control deu-terium depletion in DKDP crystals
ACKNOWLEDGMENTS
We would like to thank M Runkel for rst discovering DKDP crack-ing R Floyd for providing DKDP crystals and L Chase and A Burn-ham for their support and helpful discussions This work was performedunder the auspices of the US Department of Energy by the University
of California Lawrence Livermore National Laboratory under contract W-7405-Eng-48
1 J J De Yoreo A K Burnham and P K Whitman Int Mat Rev47 113 (2002)
2 C E Barker R A Sacks B M Van Wonterghern J A Caird JR Murray J H Campbell K Kyle R B Ehrlich and N DNielsen Proc SPIE-Int Soc Opt Eng 2633 501 (1995)
3 T Suratwala paper to be published4 Cleveland Crystals Inc httpwwwclevelandcrystalscom
KDPshtmltable5 E A Popova I T Savatinova and I A Velichko Sov Phys Solid
State 12 1543 (1971)6 I P Kaminow R C C Leite and S P S Porto J Phys Chem
Solids 26 2085 (1965)7 J A Subramony B J Marquardt J W Macklin and B Kahr
Chem Mat 11 1312 (1999)8 H Tanaka M Tokunaga and I Tatsuzaki Solid State Commun
49 153 (1984)9 R J Nelmes G M Meyer and J E Tibballs J Phys C 15 59
(1982)10 M A Yakshin D W Kim Y S Kim Y Y Broslavets O E
Sidoryuk and S Goldstein Laser Physics 7 941 (1997)11 I Takenaga Y Tominaga S Endo and M Kobayashi Solid State
Commun 84 931 (1992)12 C Krenn personal communication13 M Sharon and A K Kalia J Solid State Chem 21 171 (1977)
Rapid Dipicolinic Acid Extractionfrom Bacillus Spores Detectedby Surface-EnhancedRaman Spectroscopy
STUART FARQUHARSON ALAND GIFT PAUL MAKSYMIUK andFRANK E INSCOREReal-Time Analyzers Inc East Hartford Connecticut06108
Index Headings Dipicolinic acid Bacillus spores Anthrax Surface-enhanced Raman spectroscopy
INTRODUCTION
The anxiety caused by the distribution of anthrax en-dospores through the US postal system in October 2001was exacerbated by the long time required for positiveidenti cation of the Bacillus anthracis spores and the un-known extent of their distribution Since that time manymethods capable of rapid eld analysis have been inves-tigated to augment or replace the laboratory method ofgrowing microorganisms in culture media which takesdays to perform12 Prominent among these approachesare polymerase chain reactions (PCR)3 immunoassaysand detection of calcium dipicolinate as a biochemicalsignature PCR employs primers to separate organism-
Received 10 October 2003 accepted 14 November 2003 Author to whom correspondence should be sent
352 Volume 58 Number 3 2004
speci c nucleic acid sequences (eg capsular protein en-coding gene for Bacillus anthracis)4 and polymerases toamplify the segment until it is detectable Recently am-pli cation times have been substantially reduced andcomplete analysis can now be performed in an hour orless Immunoassay methods are also being developed thatuse competitive binding of the bioagent (as an antigen)and its labeled conjugate for a limited number of anti-bodies Although analyses can be performed in under 30minutes a well-de ned anthrax antigen has not yet beenidenti ed5ndash7 and consequently the false-positive rate isunacceptably high8
A number of other methods are being developed witha focus on the detection of calcium dipicolinate (CaDPA)and its derivatives as a B anthracis signature This is sobecause only spore-forming bacteria contain CaDPA andthe most common potentially interfering spores such aspollen and mold spores do not Relatively fast methodshave been developed to chemically extract CaDPA andthen detect it directly by uorescence9 or indirectly byluminescence1011 In the latter case hot dodecylamine(DDA) has been used to extract dipicolinic acid (DPA)and terbium has been utilized to form a highly lumines-cent DPA complex11 Although measurements have beenperformed in as little as ve minutes it was found thatas many as three concentration-dependent complexes canform each with different lifetimes This coupled withthe fact that the Tb31 cation produces the same lumines-cence spectrum makes determinations of low spore con-centrations problematic
It has been long known that Raman spectra of Bacillispores are dominated by bands associated with CaDPA12
and that these spectra may provide a suitable anthrax sig-nature at the genus level13 Since that time considerableimprovements in Raman instrumentation have led to lab-oratory measurements of single Bacilli spores14 and to eld measurements of spores captured from a mail-sort-ing system15 However the single spore measurementsrequired complex instrumentation that is not ruggedwhile the eld measurements required milligrams of sam-ple Furthermore the Raman spectra of both measure-ments contained uorescence contributions that would in-crease uncertainty in quanti cation
In related research we demonstrated that nanogramquantities of DPA could be detected by uorescence-freesurface-enhanced Raman spectroscopy (SERS)16 We alsodemonstrated that microliter volumes of chemicals canbe detected by SERS using metal-doped sol-gel-packedglass capillaries17 Towards the goal of developing a rap-id eld SERS-based anthrax spore detector we havecombined our previous research and we now report thatDPA can be extracted from a 10 mg B cereus spore sam-ple using DDA in 1 minute and can be detected by SERSin an additional 1 minute
EXPERIMENTAL
Dipicolinic acid (26-pyridinedicarboxylic acid DPA)and dodecylamine (DDA) were used as received fromSigma-Aldrich (Milwaukee WI) Lyophilized B cereusspores prepared according to the literature13 were sup-plied by the University of Rhode Island and used as re-ceived Multiple particles approximately 01 mm3 each
were separated and weighed at 5 to 15 mg representing05 to 15 million spores The sample masses were con-sistent with a previous determination of spore density at0081 gmL that indicated a high degree of entrained air
All chemicals used to prepare the silver-doped sol-gelcoated capillaries were also obtained and used as receivedfrom Sigma-Aldrich According to previously publishedprocedures17 two precursor solutions were preparedmixed and then drawn into 1-mm-diameter glass capil-laries The silver amine precursor consisted of a 51 vvratio of 1 N AgNO 3 to 28 NH3OH while the alkoxideprecursor consisted of a 21 vv ratio of methanol to te-tramethyl orthosilicate The alkoxide precursors weremixed with silver amine precursor in an 81 vv ratioApproximately 015 mL was drawn into the capillarycoating a 15-mm length After sol-gel formation the in-corporated silver ions were reduced with dilute sodiumborohydride which was followed by a water wash to re-move residual reducing agent
A 100 mL drop of a 50 mM DDA solution in ethanolpre-heated to 78 8C was added to each of the B cereusparticles to digest the spore coat After 1 minute the re-sultant solution was drawn into a SER-active capillarythat was immediately xed horizontally to an XY posi-tioning stage (Conix Research Spring eld OR) just in-side the focal point of an f 07 aspheric lens The lensfocused the beam into the sample and collected the scat-tered radiation back along the same axis A dichroic lter(Omega Optical Brattleborough VT) was used to re ectthe excitation laser to the lens and pass the Raman scat-tered radiation collected by the lens An f 2 achromatwas used to collimate the laser beam exiting a 200-mm-core-diameter source ber optic while a second f 2 ach-romat was used to focus the scattered radiation into a 365mm ber optic (Spectran Avon CT) A short-pass lterwas placed in the excitation beam path to block the sil-icon Raman scattering generated in the source ber fromre ecting off sampling optics and reaching the detectorA long-pass lter was placed in the collection beam pathto block the sample Rayleigh scattering from reachingthe detector A 785 nm diode laser (Process InstrumentsInc model 785-600 Salt Lake City UT) was used todeliver 100 to 150 mW of power to the sample A Fouriertransform Raman spectrometer (Real-Time Analyzersmodel IRA-785 East Hartford CT) and a silicon photo-avalanche detector (Perkin Elmer model C30902S Stam-ford CT) were used to acquire the SER spectra
RESULTS AND DISCUSSION
As an initial experiment the SER spectrum of 1 gLof DPA in water was measured using the newly devel-oped silver-doped sol-gel-coated capillaries (Fig 1A) Atthis concentration a high signal-to-noise ratio (SN) isobtained in 1 min In fact a reasonable spectrum is ob-tained in the same time frame for 1 mgL (Fig 1B) TheSER spectra are reasonably similar to the normal Raman(NR) spectrum obtained for a saturated solution of DPAin 1 N KOH (Fig 1C) and the following band shifts areobserved (NR to SER) 647 to 657 cm21 817 to 815cm21 998 to 1008 cm21 1384 to 1382 cm21 1434 to1428 cm21 and 1569 to 1567 cm21 Many of these bandshave been previously assigned1213 such as 998 cm21 to
APPLIED SPECTROSCOPY 353
FIG 1 SERS of DPA in water using silver-doped sol-gel-coated glasscapillary for (A) 1 gL and (B) 1 mgL (C ) NR of saturated DPA in 1N KOH in a glass capillary Spectral conditions (A) and (B) 150 mWof 785 nm 1-min acquisition time (C) 450 mW of 785 nm 5-minacquisition time both 8 cm 21 resolution
FIG 2 SERS of DPA extracted from 10 mg B cereus particle using100 mL of 50 mM hot DDA acquired in (A) 1 minute and (B) 2 seconds(C ) Attempted SERS of 50 mM hot DDA in ethanol using silver-dopedsol-gel-coated glass capillary acquired in 1 min Spectral conditions150 mW of 785 nm 8 cm21 resolution
the symmetric ring stretch 1384 cm21 to the OndashCndashOsymmetric stretch 1428 cm21 to the symmetric ring CndashH bend and 1569 cm21 to the asymmetric OndashCndashOstretch
The rst B cereus samples consisted of 2 mg of sporesin 2 mL of 5 mM hot DDA The samples were main-tained at 78 8C for 40 min and while hot approximately10 mL was drawn into a SER-active capillary Since spec-tra of DPA were obtained for these initial samples small-er spore masses higher DDA concentrations and shorterheating periods were examined In due course it wasfound that 10 mg of spores could be digested by 100 mLof 50 mM hot DDA in one minute and detected (Fig2A) In fact the signal was suf ciently intense that it canbe observed in as little as two seconds (Fig 2B) Theamount of DPA that was extracted was estimated to bebetween 5 and 10 mgL by comparing the signal intensityof the 1008 cm21 band to that measured for DPA in waterThis is consistent with previous research that found thatthe majority of the DPA is extracted from spores usingDDA11 and that B cereus spores contain approximately10 DPA by weight18 The SN of 127 for the 1008 cm21
band in the 1-minute SER spectrum suggests a limit ofdetection of approximately 250 ng of B cereus sporesbased on a SN of 3 Finally it should be noted that DDAdid not produce a detectable SER spectrum as shown inFig 2C
CONCLUSION
We have demonstrated that by combining rapid extrac-tion of dipicolinic acid from Bacillus cereus spores withchemical identi cation by surface-enhanced Raman spec-troscopy as little as 10 mg of spores can be detected Infact the entire measurement from the time of adding hotdodecylamine to the spores to the time when the dipi-colinic acid SER spectrum is acquired and analyzedcould be performed in less than two minutes The abilityof this method to distinguish between spore-forming bac-teria such as Bacillus anthracis and non-DPA containingpowders could help prevent costly shutdowns associated
with the appearance of suspicious material or intentionalmailing of common substances as an anthrax hoax Thismethod could also prove useful in detecting the locationof anthrax endospores in mail distribution facilities if an-other veri ed attack should occur
Research continues to fully characterize the surface-enhanced Raman spectroscopy signal intensities as afunction of sample concentration and to explore otherextractants that do not require the use of elevated tem-perature
ACKNOWLEDGMENTS
The authors are grateful for the support of the National Science Foun-dation (DMI-0296116 and DMI-0215819) and the US Army(DAAD13-02-C-0015 Joint Service Agent Water Monitor program)The authors are indebted to Chetan Shende for preparing the sol-gelcapillaries The authors also thank James Gillespie Nicholas Fell andAugustus Fountain for providing important background informationMark Farquharson for laboratory support and Professor Jay Sperry ofthe University of Rhode Island for supplying B cereus spores
1 V A Pasechnik C C Shone and P Hambleton Bioseparations 3267 (1993)
2 P J Jackson M E Hugh-Jones D M Adair G Green K K HillC R Kuske L M Grinberg F A Abramova and P Keim ProcNatl Acad Sci USA 95 1224 (1998)
3 B R Glick and J J Pasternak Molecular Biology Principles andApplications of Recombinant DNA (ASM Press Washington DC1994)
4 C A Bell J R Uhl T L Had eld J C David R F Meyer TF Smith and F R Cockerill III J Clin Microbiol 40 2897(2002)
5 D L Gatto-Menking H Yu J G Bruno M T Goode M Millerand A W Zulich Biosens Bioelectron 10 501 (1995)
6 J J Quinlan and P M Foegeding J Rapid Methods AutomationMicrobiol 6 1 (1998)
7 A A Hindle and E A H Hall Analyst (Cambridge UK) 1241599 (1999)
8 M S Ascher US Department of Health amp Human Services(httpwwwhhsgovophppresentationsAscherdoc)
9 R Nudelman B V Bronk and S Efrima Appl Spectrosc 54 445(2000)
10 D L Rosen C Sharpless and L B McBrown Anal Chem 691082 (1997)
354 Volume 58 Number 3 2004
11 P M Pellegrino N F Fell Jr and J B Gillespie Anal ChimActa 455 167 (2002)
12 W H Woodruff T G Spiro and C Gilvarg Biochem BiophysRes Commun 58 197 (1974)
13 E Ghiamati R S Manoharan W H Nelson and J F SperryAppl Spectrosc 46 357 (1992)
14 A P Esposito C E Talley T Huser C W Hollars C M Schal-dach and S M Lane Appl Spectrosc 57 868 (2003)
15 S Farquharson L Grigely V Khitrov W W Smith J F Sperryand G Fenerty J Raman Spectrosc paper accep ted (2003)
16 S Farquharson W W Smith S Elliott and J F Sperry SPIE-IntSoc Opt Eng 3855 110 (1999)
17 S Farquharson and P Maksymiuk Appl Spectrosc 57 479(2003)
18 F W Janssen A J Lund and L E Anderson Science (Washing-ton DC) 127 26 (1958)
SPIE -2003-5269 117
pH dependence of methyl phosphonic acid dipicolinic acid and cyanide by surface-enhanced Raman spectroscopy
Stuart Farquharson Alan Gift Paul Maksymiuk Frank Inscore and Wayne Smith
Real-Time Analyzers 87 Church Street East Hartford CT 06108
ABSTRACT US and Coalition forces fighting terrorism in Afghanistan and Iraq must consider a wide range of attack scenarios in addition to car bombings Among these is the intentional poisoning of water supplies to obstruct military operations To counter such attacks the military is developing portable analyzers that can identify and quantify potential chemical agents in water supplies at microgram per liter concentrations within 10 minutes To aid this effort we have been investigating the value of a surface-enhanced Raman spectroscopy based portable analyzer In particular we have been developing silver-doped sol-gels to generate SER spectra of chemical agents and their hydrolysis products Here we present SER spectra of methyl phosphonic acid and cyanide as a function of pH an important factor affecting quantitation measurements which to our knowledge has not been examined In addition dipicolinic acid a chemical signature associated with anthrax-causing spores is also presented Keywords Chemical warfare agents agent detection agent hydrolysis SERS Raman spectroscopy homeland security
1 INTRODUCTION In the past decade the Unites States and its allies have been challenged by a different kind of warfare exemplified by the terrorist attacks of September 11 2001 Suicide bombings and the use of chemical agents are the norm and military personnel must consider a wide range of attack scenarios Among these is the intentional poisoning of water supplies to obstruct military operations in Afghanistan and Iraq Currently colorimetric paper is used to detect agents on-site while gas chromatography combined with mass spectrometry (GCMS) is used in mobile support laboratories However both methods have severe drawbacks The paper changes color in response to contact with many chemicals besides CWAs causing a high incidence of false positives1 while GCMS although very chemically specific requires hours to perform and constant re-calibration234 Military operations would be greatly aided by a portable analyzer that can identify and quantify potential chemical agents at concentrations that impact safety This includes the analysis of drinking water supplies distribution and storage systems To meet this goal the Department of Defense has been investigating numerous approaches under the auspices of the Joint Service Agent Water Monitor (JSAWM) program5 The JSAWM must be capable of identifying the chemical agents with no false-positives and quantifying the agents at microgram per liter concentrations within 10 minutes678 Recently we and others have been investigating the ability of surface-enhanced Raman spectroscopy (SERS) to measure chemical agents9-12 bioagents13-17 and their hydrolysis products in water SERS employs the interaction of surface plasmon modes of metal particles with target analytes to increase scattering efficiency by as much as 1 million times18 In 1987 the potential of this approach to measure CWAs was demonstrated by measuring a series of pesticides19 Several of these organophosphonates have chemical structures similar to CWAs in particular P=O functional groups In our studies we have been employing silver-doped sol-gels to promote the SER effect The porous silica network of the sol-gel matrix offers a unique environment for immobilizing and stabilizing SER-active metal particles20-23 The sol-gels have been coated on the internal walls of standard glass vials to allow rapid SER analysis of small sample quantities without preparation We have measured over 100 chemicals with enhancements of 104 to 106 demonstrated reversible measurements in a flowing system reproducible measurements from vial-to-vial and measurements in multiple solvents including water20-23 Previously we used these vials to perform preliminary measurements of cyanide (CN) methylphosphonic acid (MPA) and dipicolinic acid (DPA) MPA is a hydrolysis product of the nerve agents (eg sarin Reaction 1) and may be a valuable indicator of nerve agent usage particularly since the alkyl methylphosphonic acids are relatively more stable than their corresponding parent complexes24 DPA is
SPIE -2003-5269 118
a chemical signature of spore forming bacteria such as Bacillus anthracis And in light of the inability to rapidly detect the anthrax spores distributed through the US mail in October 2001 a number of methods are being developed to extract and analyze this signature Reaction 1 Stepwise hydrolysis of Sarin to form hydrofluoric acid (HF) isopropyl methylphosphonic acid (IMPA) then methyl phosphonic acid (MPA) and 2-propanol In our previous SERS investigations MPA and DPA were measured at 50 and 100 mgL respectively In both cases limits of detection (LOD) were estimated at 100 microgL providing encouragement in that SERS may satisfy the needs of the JSAWM Since it has been shown that pH can substantially influence the intensity of SER bands25 which would clearly influence quantitative analysis we undertook the present study to determine the severity of these effects for cyanide methyl phosphonic acid and dipicolinic acid Furthermore we previously observed a band at 1050 cm-1 for MPA6 possibly due to an anion formed at basic pH Here we investigate the source of this spectral anomaly
2 EXPERIMENTAL All chemicals including potassium cyanide methyl phosphonic acid dipicolinic acid and those used to prepare the silver-doped and gold-doped sol-gel coated vials were obtained and used as received from Sigma-Aldrich All samples were prepared in HPLC grade water (Fischer Scientific Fair Lawn NJ) for SERS measurements The pH of these samples was adjusted using dilute nitric acid or potassium hydroxide and verified using a pH electrode (Corning Inc Corning NY) that had been calibrated with pH 400 700 and 1000 buffered standards from Fischer Scientific Once prepared the samples were transferred into the silver-doped sol-gel vials (Real-Time Analyzers Inc East Hartford CT) The vials were coated in a manner similar to that previously reported by adding ammonium hydroxide to a solution of silver nitrate tetramethyl orthosilicate (TMOS) and methanol20 Gold-vials were coated by adding nitric acid to a solution of gold tetrachloride TMOS and methanol The two precursor solutions were prepared mixed and transferred to 2-mL glass vials dried and heated After sol-gel formation the incorporated metal ions were reduced with dilute sodium borohydride (1mgmL) which was followed by a water wash to remove residual reducing agent After the resultant analyte solution was introduced the SER-active vial was immediately fixed horizontally to an XY positioning stage (Conix Research Springfield OR) just inside the focal point of an f07 aspheric lens The lens focused the beam into the sample and collected the scattered radiation back along the same axis A dichroic filter (Omega Optical Brattleborough VT) was used to reflect the excitation laser to the lens and pass the Raman scattered radiation collected by the lens An f2 achromat was used to collimate the laser beam exiting a 200 microm core diameter source fiber optic while a second f2 achromat was used to focus the scattered radiation into a 365 microm fiber optic (Spectran Avon CT) A short pass filter was placed in the excitation beam path to block the silicon Raman scattering generated in the source fiber from reflecting off sampling optics and reaching the detector A long pass filter was placed in the collection beam path to block the sample Rayleigh scattering from reaching the detector A 785 nm diode laser (Process Instruments Inc model 785-600 Salt Lake City UT) was used to deliver 100 to 150 mW of power to the sample A Fourier transform Raman spectrometer (Real-Time Analyzers model IRA-785 East Hartford CT) and a silicon photo-avalanche detector (Perkin Elmer model C30902S Stamford CT) were used to acquire the SER spectra
3 RESULTS AND DISCUSSION In a previous study of MPA6 aimed at developing a concentration calibration curve and determining limits of detection (LOD) we observed an anomaly at 1050 cm-1 Since it was found that the band intensity changed as a function of concentration the band must be associated with a sample parameter Two possible parameters photon flux and pH are examined here The first parameter was investigated by irradiating a 1mgmL MPA sample in a SER-active vial with laser powers of 200 mW and above and monitoring spectral changes It was immediately found that the 1050 cm-1 band
2O+ H HF + +OH
OH3H C
OP
OF
CH
CH3
3
3H C
OCP
OOH
CH
CH3
3
3H C
O
CPHO
CH
CH
3
3
C
Sarin IMPA MPA 2-propanol
SPIE -2003-5269 119
appeared and grew as a function of time and that the higher the laser power the faster the growth Figure 1 shows the growth of the 1050 cm-1 band over the course of 30 minutes when using 150 mW of 785 nm excitation while Figure 2A shows that the growth can be fit with a first order exponential equation namely I1050 = 03+05e-013t Furthermore the 760 cm-1 band could be fit with a first order decay equation with an identical exponential rate constant ie I760 = 08-
08e-013t The rates represent classical first order kinetics and their correspondence allows one to conclude that MPA is being transformed one-for-one into a photo-generated product At this time the photoproduct has not been positively identified but phosphonic (phosphorous) acid and phosphonate are likely candidates since the symmetric P(OH)3 stretch occurs at ~1050cm-1 Our studies show that a reduction in laser power to 100 mW at the sample for MPA essentially eliminates this degradation process This laser power or lower was used for further measurements
Methyl phosphonic acid is a diprotic acid that stepwise dissociates into two anions MPA- and MPA= according to the following reactions26
MPA MPA- + H+ pKa1 = 212 Reaction 2
MPA- MPA= + H+ pKa2 = 729 Reaction 3 The relative concentrations of MPA MPA- and MPA= can be determined at any pH by expressing [MPA] and [MPA=] in terms of [MPA-] using Reactions 2 and 3 and summing all three to equal the total starting concentration here 2 mgmL (0021M MW = 9602) viz
[MPA] + [MPA-] + [MPA=] = 0021M Equation 1 substituting from Reactions 1 and 2
([H+][MPA-])K1a + [MPA-] + (K2a[MPA-])[H+] = 0021M Equation 2 rearranging [MPA-] = 0021M(1+[H+]K1a + K2a[H+]) Equation 3 The relative concentrations of MPA MPA- and MPA= as a function of pH are shown in Figure 3 It is worth noting that near neutral pH both MPA- and MPA= will be present To confirm that the SER signal followed this pH dependence a starting solution consisting of 20 mg of MPA in 10 mL HPLC grade water was prepared and brought to pH of 20 using dilute nitric acid From this solution 2 mL were added to a SER-active vial and the SER spectrum recorded At this pH a peak at 760 cm-1 was barely discernable The 2 mL solution was returned to the starting solution and the pH was re-measured to correct for any changes that the silver-doped sol-gel vials might cause In most cases the change was less than 02 pH units and the pH is reported as the before and after average Next the pH of the
Figure 2 A) Exponential growth of 1050 cm-1 band and B) exponential decay of 760 cm-1 band for spectral series in Figure 1
Figure 1 Growth of 1050 cm-1 band as a function of time due to exposure to 150 mW of 785 nm Spectra are 5 sec each collected every 100-sec from 0 to 30-min
B
570 770 970 1170 1370Raman Shift (cm-1)
Arbit
rary
Unit
s
MPA Photodegradation
600 800 1000 1200 1400 Wavenumber (cm-1)
A
0 10 20 30 0 10 20 30 time (min) time (min)
Ram
an In
tens
ity (r
elat
ive)
30
min
0
SPIE -2003-5269 120
starting solution was adjusted to 325 using dilute KOH Again 2 mL were added to a vial and the SER spectrum recorded At this pH a reasonably strong 760 cm-1 band was observed This process was repeated as spectra were recorded at pHs of 70 74 75 79 85 and 100 A total of 1 mL of KOH was added diluting the total concentration by 10 Next the pH of the starting solution was made acidic by adding dilute nitric acid dropwise This time spectra were recorded at pHs of 72 69 64 and 37 Figure 4 shows the SER spectra for representative pHs (spectra were left out to simplify the figure) while Figure 3 shows the 760 cm-1 peak intensities as a function of pH (The band intensities were adjusted to compensate for dilution effects caused by the addition of HNO3 and KOH then normalized to 0021 M for the most intense band observed at pH 37) It is clear from Figure 3 that the 760 cm-1 band follows the MPA- concentration as a function of pH and must be assigned to this anion No bands were observed that corresponded to MPA or MPA= The lack of an MPA SER spectrum may be due to the absence of an attraction between the neutral analyte and the electropositive silver surface The same reasoning suggests that a strong SER spectrum should be observed for MPA= but it is not and a satisfactory explanation has not been found
0000
0005
0010
0015
0020
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14pH
Con
cent
ratio
n [M
]
pK1 = 212 pK2 = 729
MPA- MPA=
MPA
Figure 4 SER spectra of 002M MPA as a function of pH Conditions 100 mW of 785 nm 36 scans (1 min) 8 cm-1 recorded 2 min after sample introduction pH 19 and 20 not apparent on this scale pH 69 and 74 near identical to 70 and 75 and not shown for clarity
00
02
04
06
08
10
12
14
16
18
0 5 10 15 20 25 30Measurement Number
Ram
an In
tens
ity (7
60 c
m-1
)
Figure 3 Concentration dependence of MPA MPA- and MPA= as a function of pH for a 002M sample Intensity of 760 cm-1
band from Figure 6 as a function of pH ( for increasing basic adjustment for increasing acidic adjustment error was measured at ~10 for pH 64)
Figure 5 SER spectra of 002M MPA at pH 64 measured around a vial at three heights (9 points per height) Conditions as in Figure 4 but 10-sec scans
Since these measurements involved the removal and replacement of the SER-active vial in the sample holder to remove and add sample variation in the intensity as a function of vial position was minimized by illuminating the exact same height along the vial wall But this does not account for variability of the SERS response of the sol-gel coating around the vial To analyze this effect a vial containing MPA at pH 64 was rotated at ~ 40o
intervals at the original height and 18rdquo above and below this value Figure 5 shows the intensity of the 760 cm-1
band for the 27 positions It was found that the average value was 137plusmn014 an RSD of 10 overall and 5 for each height An error bar is included in Figure 4 for the pH 64 measurement
SPIE -2003-5269 121
With the above analyses in mind a preliminary investigation of the SER spectral response for MPA (as MPA-) as a function of concentration was performed A single vial was used for these measurements beginning with 1 mgL followed by measurements of 10 100 and 1000 mgL In all cases the pH was ~7 and 3 positions around the vial were measured per concentration Since the 760 cm-1 band was not observed for concentrations of 1 or 10 mgL using 100 mW of 785 nm the laser power at the sample was raised to 200 mW beginning with the 10 mgL concentration Photo-degradation was largely avoided (and not observed) by exposing the sample for only 33 seconds per spectral acquisition Representative spectra for 10 100 and 1000 mgL are shown in Figure 6 while a plot of the 760 cm-1 band intensity as a function of concentration is shown in Figure 7 These values were also used to estimate limits of detection based on the signal-to-noise ratio (SN) of the 760 cm-1 band using 100 mW of 785 nm laser excitation and a 10-min acquisition time As summarized in the Figure 7 inset the lower the measured concentration the lower the predicted LOD Even if as estimated 210 microgL could be measured using the silver-doped sol-gel vials an improvement of a factor of 70 is still required to meet JSAWM goals of measuring 3 microgL in 10-minutes Similar to MPA DPA is a diprotic acid (pKa1 = 216 and pKa2 = 692) and variations in pH will effect the relative concentrations of DPA DPA- and DPA= and possibly the SER spectra and band intensities This could prove significant if an acid or base is used to denature anthrax spores with the goal of extracting and analyzing DPA The MPA pH study described above was mimicked for DPA except that the starting solution consisted of 20 mg of DPA in 20 mL HPLC grade water (60x10-3M MW = 1671) The initial solution had a pH of 245 which was made basic by dilute KOH to pHs of 355 433 487 559 1069 and 1166 SER spectra were recorded at each pH using 100 mW of 785 nm and a 44-sec acquisition time Next one drop of concentrated nitric acid was used to remake the solution acidic at a pH of 200 Again sequential pH measurements were performed at 383 510 735 and 822 The solution pH was made acidic a third time but to pH 219 171 then 135 Throughout this process no more than 20 drops of acid or base were added and therefore the concentration was diluted by no more than 10 Most of the spectral bands showed a minor decrease in intensity as a function of increasing pH values However the bands at 525 and 795 cm-1 showed the most dramatic changes which occurred at acid pH Figure 8 shows the SER spectra of DPA for the spectral region and pH range of interest The identity of the DPA species was determined by plotting the normalized peak intensities with the lowest value set to 0 and the highest to 0006 M as a function of pH and overlaying these values on a plot of the relative concentrations for DPA DPA- and DPA= as previously done for MPA (Figure 9) As can be seen the 525 cm-1 band clearly corresponds to DPA The correspondence of the 795 cm-1 band to this species is less clear as the band retains intensity until pH of 55 This can be attributed to contribution to the overlapping band at 810 cm-1 which does not change as a function of pH The fact that most bands are observed at all pHs suggest that the primary interaction with silver is through the ring nitrogen This is supported by the fact that the most intense band occurs at 1008 cm-1 attributed to a symmetric ring breathing mode and that this interaction has been characterized for pyridine in numerous papers27
0
1
2
3
4
5
6
7
0 200 400 600 800 1000 1200
MPA Concentration (mgL)
760
Ban
d In
tens
ity
Figure 6 SER spectra of MPA in water at A) 1000 B) 100 and C) 10 mgL Conditions pH of 7 silver-doped sol-gel coated vial 200 mW of 785 nm 33-sec 8 cm-1 resolution
Figure 7 Plot of SER intensity of 760 cm-1 band of MPA as a function of concentration using 200 mW of 785 nm Inset table includes average intensity LOD standard deviation and percent deviation for each concentration but for 100 mW and 10 min
conc (mgL) LOD ave stddev dev10 021 002 1021100 072 010 14301000 312 040 1280
A
B C
SPIE -2003-5269 122
The ability to detect dipicolinic acid (DPA) by SERS is demonstrated in Figure 10 Even at 1 mgL the primary bands are visible The intensity of the dominant ring breathing mode observed at 1008 cm-1 for DPA in water (pH = 25 to 55) is plotted as a function of concentration in Figure 11 These values were also used to estimate limits of detection based on the SN of the 1008 cm-1 band using 100 mW of 785 nm laser excitation and a 10-min acquisition time Again the lower the measured concentration the lower the predicted LOD (see Figure 11 inset) and detection of 160 microgL is possible Gastrointestinal anthrax requires significant more spores than inhalation anthrax28 and a limit of detection might be placed at 1 million spores in 1 liter of water or 10 microgL Since each spore contains ~10 CaDPA by weight29 a goal for DPA might be 1 microgL indicating that the present measurements must be improved by nearly two orders of magnitude Due to the increased hazards of handling HCN gas KCN salt was used for these experiments Nevertheless all sample preparations were performed in a chemical hood KCN completely dissolves in water but its conjugate acid HCN is formed and has a Ka of 615x10-1030 viz
HCN CN- + H+ pKa = 921 Reaction 4
Figure 8 SER spectra of 1 mgmL DPA as a function of pH Conditions 100 mW 785 nm 100 scans (44-sec)
Figure 9 Concentration dependence of DPA DPA- and DPA= as a function of pH for a 0006M sample Intensity of 525 () and 795 (diams) cm-1 bands from Fig 8 as a function of pH
pH
135 171 219 383
0
0001
0002
0003
0004
0005
0006
0007
0 2 4 6 8 10 12 14pH
Con
cent
ratio
n [M
]
DPADPA-DPA=795525
DPA DPA=DPA-
pK2 = 692pK1 = 216
795 525
0
05
1
15
2
25
0 200 400 600 800 1000 1200
DPA Concentration (mgL)
1008
Ban
d In
tens
ity
conc (mgL) lod-10min-100mw1 017
10 016100 103
1000 355
Figure 10 SER spectra of DPA in water at A) 1000 B) 100 C) 10 and D) 1 mgL Conditions pH of 25-55 silver-doped sol-gel coated vial 175 mW of 785 nm 1-min 8 cm-1 D) has been multiplied by x10 to make bands visible
A
B C
D
Figure 11 Plot of SER intensity of 1008 cm-1 band of DPA as a function of concentration using 175 mW of 785 nm Inset table includes LOD in mgL for each concentration but for 100 mW and 10 min
SPIE -2003-5269 123
Consequently the cyanide concentration must be determined for each initial KCN concentration Specifically the samples prepared with concentrations of 01 1 10 100 and 1000 mgL of KCN produced CN- concentrations of 63x10-3 033 69 89 and 964 mgL at pHs of 816 90 967 102 and 107 respectively The pH dependence for the HCN and CN- concentrations are shown in Figure 12 Thus as the amount of KCN added to the solution decreases so does the pH of the solution (becomes less basic) and according to Reaction 4 the relative amount of CN- to HCN also decreases For example in the preparation of a 01 mgL solution of KCN the pH is shifted from 7 for pure water to only 816 and only 63 of the starting material becomes CN- or 63x10-3 mgL In comparison for a solution of 1000 mgL the pH is shifted from 7 to 107 and 96 of the starting material becomes CN- This is significant since the cyanide ion is better able to adsorb onto the silver particles and become SERS active SER spectra of 10 100 and 1000 mgL of KCN samples are shown in Figure 13 and a corresponding plot of the 2100 cm-1 band intensity as a function of concentration is shown in Figure 14
000102030405060708091011
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14pH
Con
cent
ratio
n [m
gm
L]
pKa = 921
CN -HCN
The SER spectra of cyanide are dominated by the single CN stretch at 2100 cm-1 which occurs in normal Raman spectra of solutions at 2080 cm-1 However a low frequency mode occurs at 135 cm-1 due to a Ag-CN stretch (not shown) indicative of a strong surface interaction It is also observed that as the concentration decreases the CN stretch shifts to 2140 cm-1 This shift has been attribute to the formation of a tetrahedral Ag(CN)3
2- surface structure31 as well
Figure 14 Concentration dependence of KCN SERS measured under conditions in Fig 11 Concentrations are 1 01 and 001 mgml Intensities are measured for the CN stretch at 2100 cm-1 Inset table includes LOD in mgL for each concentration in Figs 13 and 15 but for 100 mW and 10 min
Figure 13 SER spectra of KCN in water at A) 1000 B) 100 and C) 10 mgL Conditions pHs of 107 102 and 97 silver-doped sol-gel coated vial 100 mW of 785 nm 1-min 8 cm-1 C) has been multiplied by x10 to make band visible
Figure 12 Concentration dependence of HCN and CN- as a function of pH for a 1 mgmL sample Calculated intensity of 2100 () cm-1 band for a 1 mgml sample at pHs of 816 90 967 102 and 107
A
B
C
0
20
40
60
80
100
120
140
0 200 400 600 800 1000 1200CN Concentration (mgL)
2100
Ban
d In
tens
ity
Figure 15 SER spectra of KCN in water at A) 10 B) 1 and C) 01 mgL Conditions pHs of 97 12 and 12 gold-doped sol-gel coated vial 100 mW of 785 nm 1-min 8 cm-1 C) has been multiplied by x10 to make band visible
A
B
C
conc (mgL) lod-10min-100mw Condition01 001 Au-pH 121 007 Au-pH 12
69 003 Au-pH 9769 016 Ag-pH 9789 022 Ag-pH 102946 113 Ag-pH 107
SPIE -2003-5269 124
as to CN adsorbed to two different surface sites32 Alternatively the 2140 cm-1 band could be attributed to HCN since this species dominates at lower concentrations However it is unlikely that this species would be attracted to the electropositive silver surface Further both peaks should be present at pHs between 85 and 105 but this is not observed It has also been suggested that at concentrations near and above monolayer coverage the CN- species is forced to adsorb end-on due to crowding and at lower concentrations the molecule can reorient to lie flat33 This suggests that the 2100 and 2140 cm-1 bands correspond to the end-on and flat orientations respectively As Figures 13 and 14 show the intensity of the CN stretch for the 89 mgL sample is nearly as intense as the 964 mgmL band This suggests that the Raman signal for the flat orientation is more enhanced However more extensive measurements are required to verify this point Since resent research has suggested that cyanide may be more effectively detected on gold measurements of KCN solutions were also performed using gold-doped sol-gel vials Preliminary measurements are shown in Figure 15 for samples prepared from 01 1 and 10 mgL KCN Since the pHs are 816 90 967 the resultant CN- concentrations are 63x10-3 033 and 69 mgL Initially only the highest concentration was observed and the signal intensity was significantly better than the equivalent concentration measured using silver In an effort to shift Reaction 4 to the left transforming HCN to CN- (Le Chatelierrsquos principle) KOH was added to the lower concentration samples producing solutions with pH 12 This effectively forces all of the cyanide in solution to be CN- or 01 and 10 mgL respectively More importantly the CN stretch is now observed in the SER spectra The band appears at 2125 cm-1 as has been previously reported for gold12 As calculated for MPA and DPA LODs can be estimated from this data For the three concentrations of cyanide on silver the LODs are 016 to 11 mgL for 100 mW of 785 nm laser excitation and a 10-min acquisition time For gold pH adjusted the LODs are10 to 70 microgL an improvement of more than 10 times silver Nevertheless either substrate is sufficient to meet the JSAWM goals of measuring 3 mgL in 10-minutes as the requirements form cyanide are much less stringent than the nerve agents
4 CONCLUSIONS Here we examined the surface-enhanced Raman spectral response for methyl phosphonic acid dipicolinic acid and cyanide as a function of pH It was determined that the most prominent peak at 760 cm-1 reaches a maximum intensity between pH 3 and 7 and corresponds to the MPA- species Neither the MPA nor MPA= species appear to generate a SER spectrum and consequently no spectra were observed below pH 2 or above pH 8 In this study we also found that higher laser powers could cause photodegradation of MPA signified by the exponential growth of a band at 1050 cm-1 which is tentatively assigned to phosphorous acid Unlike MPA DPA was observed at all pHs This is attributed to the dominant interaction of the pyridine functional group with silver Minor spectral changes were observed at acid pHs and were assigned to neutral DPA Like MPA SER spectra of cyanide were pH dependent No spectra were observed for the HCN species while CN- was best observed at pHs more basic than 8 Preliminary concentration studies for the three analytes allowed estimating limits of detection for MPA DPA and CN using 100 mW of 785 nm and a 10-min acquisition time of 210 165 and 70 microgL respectively Although the latter value suggests that cyanide can be measured at sufficiently low concentrations to meet JSAWM goal improvements by 100 to 200 times are required for MPA and DPA It is clear from this study that pH of the sample is important and must be taken into account when developing concentration calibration curves as well as fieldable analyzers Future work will include tailoring the sol-gel with specific functional groups to dictate sample pH and thereby optimize sensitivity
ACKNOWLEDGMENTS The authors are grateful for the support of the US Army (DAAD13-02-C-0015 Joint Service Agent Water Monitor program) The authors would also like to thank Dr Steve Christensen of the US Army SBCCOM for helpful discussions and Mr Chetan Shende of Real-Time Analyzers for assistance in development of the gold-doped sol-gels
REFERENCES 1 Erickson B Analytical Chemistry News amp Features June 1 397A (1998)
SPIE -2003-5269 125
2 Johnston RL Hoefler CM Fargo JC and Moberley B AT-ONSITE 5-8 (1994) 3 Black RM RJ Clarke RW Read and MT Reid J Chromatography 662 301-321 (1994) 4 USA Army Center for Health Promotion and Preventive Medicine (USACHPPM) Technical Guideline 230A 5 Jensen JL NC Wong W R Loerop SPIE 4775-03 (2001) 6 Hoenig SL Handbook of Chemical Warfare and Terrorism Greenwood Press Wesport CT (2002) 7 Munro NB SS Talmage GD Griffin LC Waters AP Watson JF King and V Hauschild Env Health
Persp 107 933-974 (1999) 8 Holstege CP Kirk M Sidell FR Crit Care Clin13 923-42 (1997) 9 Farquharson S P Maksymiuk K Ong and S Christesen SPIE 4577 166-173 (2001) 10 Lee Y and S Farquharson SPIE 4378 21-26 (2001) 11 Spencer KM J Sylvia S Clauson and J Janni SPIE 4577 158-165 (2001) 12 Tessier P S Christesen K Ong E Clemente A Lenhoff E Kaler and O Velev Applied Spectroscopy 56
1524-1530 (2002) 13 Farquharson S WW Smith S Elliott and JF Sperry SPIE 3533 207-214 (1998) 14 Farquharson S WW Smith S Elliott and JF Sperry SPIE 3855110-116 (1999) 15 Farquharson S WW Smith YH Lee S Elliott and JF Sperry SPIE 4575 62-72 (2002) 16 Guzelian AA J Sylvia J Janni S Clauson and KM Spencer SPIE 4577 182-192 (2001) 17 Shende C F Inscore A Gift P Maksymiuk and S Farquharson in press 18 Weaver MJ S Farquharson and MA Tadayyoni J Chem Phys 82 4867-4874 (1985) 19 Alak AM and T Vo-Dinh Analytical Chemistry 59 2149-2153 (1987) 20 Lee Y and S Farquharson SPIE 4206 140-146 (2000) 21 Farquharson S and Y Lee SPIE 4200-16 (2000) 22 Lee Y S Farquharson and P M Rainey SPIE 3857 76-84 (1999) 23 Lee Y S Farquharson H Kwong and M Shahriari SPIE 3537 252-260 (1998) 24 Wang J M Pumera G Collins and A Mulchandani Analytical Chemistry 74 6121-6125 (2002) 25 Dou X YM Jung Z-Q Cao and Y Ozaki Applied Spectroscopy 53 1440-1447 (1999) 26 Data supplied by S Christesen and K Ewing 27 Kerker M and B Thompson Eds SPIE MS 10 (1990) 28 Inglesby TV DA Henderson JG Bartlett JAMA 287 2236 (2002) 29 FW Janssen AJ Lund and LE Anderson Science 127 26 (1958) 30 Lide DR Ed Handbook of Chemistry and Physics CRC Press 77th Ed (1996-7) 31 Billmann J G Kovacs and A Otto Surf Sci 92 153 (1980) 32 Murray CA and S Bodoff Phys Rev B 32 671 (1985) 33 Kellogg D and J Pemberton J Phys Chem 91 1120 (1987)
SPIE-2003-5269 16
Chemical agent detection by surface-enhanced Raman spectroscopy
Stuart Farquharson Alan Gift Paul Maksymiuk Frank Inscore and Wayne Smith
Real-Time Analyzers 87 Church Street East Hartford CT 06108
Kevin Morrisey and Steven D Christesen US Army SBCCOM Aberdeen Proving Ground MD 21010
ABSTRACT
In the past decade the Unites States and its allies have been challenged by a different kind of warfare exemplified by the terrorist attacks of September 11 2001 Although suicide bombings are the most often used form of terror military personnel must consider a wide range of attack scenarios Among these is the intentional poisoning of water supplies to obstruct military operations in Afghanistan and Iraq To counter such attacks the military is developing portable analyzers that can identify and quantify potential chemical agents in water supplies at microgram per liter concentrations within 10 minutes To aid this effort we have been investigating the value of a surface-enhanced Raman spectroscopy based portable analyzer In particular we have been developing silver-doped sol-gels to generate SER spectra of chemical agents and their hydrolysis products Here we present SER spectra of several chemical agents measured in a generic tap water Repeat measurements were performed to establish statistical error associated with SERS obtained using the sol-gel coated vials Keywords Chemical agents chemical agent detection SERS Raman spectroscopy
1 INTRODUCTION
In the autumn of 2001 terrorism within US borders became a sobering reality While extensive efforts are being implemented to secure the homeland US and Coalition forces in Afghanistan and Iraq are constantly faced with terrorist attacks In addition to car-bombings the military has identified several non-traditional attack scenarios including poisoning of water supplies by chemical warfare agents (CWAs) To counter this threat the Department of Defense has been investigating numerous approaches to detect such attacks under the auspices of the Joint Service Agent Water Monitor (JSAWM) program1 The JSAWM must be capable of identifying the chemical agents with no false-positives and quantifying the agents at microgram per liter concentrations within 10 minutes2 This includes the analysis of drinking water supplies distribution and storage systems Currently colorimetric paper is used to detect agents on-site while gas chromatography combined with mass spectrometry (GCMS) is used in mobile support laboratories However both methods have severe drawbacks The paper changes color in response to contact with many chemicals besides CWAs causing a high incidence of false positives3 while GCMS although very chemically specific requires up to an hour to perform and regular re-calibration456 Vibrational spectroscopy has also been investigated as a potential method of rapidly detecting CWAs7-
11 as early as 1985 in the case of infrared spectroscopy8 and 1988 in the case of Raman spectroscopy11 Again however these techniques also have limitations when it comes to measuring trace poisons in water Infrared spectra would be dominated by the very strong absorption of water which would obscure absorptions by most other chemicals present Whereas Raman spectroscopy is simply not a very sensitive technique and detection limits are typically grams per liter Surface-enhanced Raman spectroscopy (SERS) offers several advantages over conventional vibrational methods and that may provide the necessary sensitivity required for detecting trace quantities of chemical agents in water SERS employs the interaction of surface plasmon modes of metal particles with target analytes to increase scattering efficiency by as much as 1 million times12 In 1987 the potential of this approach to measure CWAs was demonstrated by measuring a series of pesticides13 Several of these organophosphonates have chemical structures similar to CWAs in particular P=O functional groups In the past few years we and others have further explored the ability of SERS to detect CWAs14-17 and even bioagents 18-21 We have been employing silver-doped sol-gels to promote the SER effect
SPIE-2003-5269 17
in these studies The porous silica network of the sol-gel matrix offers a unique environment for immobilizing and stabilizing SER-active metal particles22-25 The silver-doped sol-gels have been coated on the internal walls of standard glass vials to allow rapid SER analysis of small sample quantities without preparation We have measured over 100 chemicals with enhancements of 104 to 106 demonstrated reversible measurements in a flowing system reproducible measurements from vial-to-vial and measurements in multiple solvents including water21-25 Previously we used these vials to perform preliminary measurements of cyanide (CN) and methylphosphonic acid (MPA) Most of the nerve agents form MPA during hydrolysis while Tabun forms CN a chemical agent in its own right In another paper including in these proceedings we examined the limits of detection (LOD) for MPA by measuring a series of concentrations down to 50 mgL and estimated a limit of detection of 100 microgL26 These measurements provide encouragement in that SERS may satisfy the needs of the JSAWM To further establish the viability of SERS in particular silver-doped sol-gels here we present analysis of cyanide mustard and VX in tap water The measurements performed at the US Armyrsquos Edgewood Chemical Biological Center Aberdeen MD also included numerous repeat measurements to establish reproducibility
2 EXPERIMENTAL 2a General All chemicals including potassium cyanide 2-chloroethylethyl sulfide and those used to prepare the silver-doped and gold-doped sol-gel coated vials were obtained and used as received from Sigma-Aldrich All samples were prepared in a chemical hood using HPLC grade water unless otherwise noted (Fischer Scientific Fair Lawn NJ) for SERS measurements Once prepared the samples were transferred into the silver-doped sol-gel vials (Real-Time Analyzers Inc East Hartford CT) The vials were coated in a manner similar to that previously reported by adding ammonium hydroxide to a solution of silver nitrate tetramethyl orthosilicate (TMOS) and methanol22 The two precursor solutions were prepared mixed and transferred to 2-mL glass vials dried and heated After sol-gel formation the incorporated metal ions were reduced with dilute sodium borohydride (1mgmL) which is followed by a water wash to remove residual reducing agent After the resultant analyte solution was introduced the SER-active vial was fixed horizontally to an XY positioning stage (Conix Research Springfield OR) just inside the focal point of an f07 aspheric lens The lens focused the beam into the sample and collected the scattered radiation back along the same axis A dichroic filter (Omega Optical Brattleborough VT) was used to reflect the excitation laser to the lens and pass the Raman scattered radiation collected by the lens An f2 achromat was used to collimate the laser beam exiting a 200 microm core diameter source fiber optic while a second f2 achromat was used to focus the scattered radiation into a 365 microm fiber optic (Spectran Avon CT) A short pass filter was placed in the excitation beam path to block the silicon Raman scattering generated in the source fiber from reflecting off sampling optics and reaching the detector A long pass filter was placed in the collection beam path to block the sample Rayleigh scattering from reaching the detector A 785 nm diode laser (Process Instruments Inc model 785-600 Salt Lake City UT) was used to deliver 100 to 150 mW of power to the sample A Fourier transform Raman spectrometer (Real-Time Analyzers model IRA-785 East Hartford CT) and a silicon photo-avalanche detector (Perkin Elmer model C30902S Stamford CT) were used to acquire the SER spectra
2b Edgewood Chemical Biological Center The surface-enhanced Raman spectral measurements at the US Armyrsquos Chemical Biological center presented here were all performed on September 12 2003 To expedite measurements a plate was machined to hold up to 12 SER-active sample vials (Figure 1) The plate fit a standard XY plate reader that could be programmed Pure KCN bis-(2-chloroethyl)sulfide (distilled mustard gas HD) and ethyl S-2-diisopropylamino ethyl methylphosphonothioate (VX) were obtained on-site and used to prepare 1 gL tap water solutions in a chemical hood with appropriate safety equipment Simulated tap water was prepared by adding 10 chemicals most often found in tap water at appropriate concentrations to distilled water (Table 1) SER measurements were also performed in a chemical hood For added safety the FT-Raman instrument was placed outside the laboratory and 30 foot fiber optic and electrical cables were used to allow remote SERS measurements and plate manipulation For each experiment 1gL samples were prepared and added to 9 individual vials which were then loaded on the plate In some cases a tenth vial was included as a blank
Table 1 Generic Tap Water Composition
Compound mgL NaHCO3 100 CaSO4 27 MgSO4bull7H2O 67 NaNO3 1 Fulvic Acid 1 K2HPO4 07 KH2PO4 03 (NH4)2HSO4 001 NaCl 001 FeSO4 0001 pH 76-78
SPIE-2003-5269 18
A software program was written that allowed selecting the sequence that the vials were measured the number of positions along the length of the vials to measure (1 to 5) and the number of scans to co-add During sample analysis the program displayed the vial being analyzed the point being analyzed and the spectrum as it was being acquired Once all the data was collected a second software program was written to rapidly analyze the data The spectra collected for all the vials on a plate could be loaded at one time and then the spectra for each point could be displayed simultaneously or separately The user could then select the Raman peak to analyze in terms of peak height or area This was accomplished by selecting points on either side of the peak to define a baseline of zero The peak height or area could then be computed for all of the spectra loaded and then exported to a spreadsheet for statistical analysis
Figure 1 A) Vial Holder 6 slots to hold 2 vials each end-to-end B) Measurement Configuration Program user selects vials to measure sequence number of points per vial (1 to 5) and number of scans per point C) Spectral Acquisition Program shows spectrum being collected which vial and position D) Spectra Analysis Program user selects spectra to analyze by plate vial and point (s) as well as two wavenumbers defining the peak and the baseline to subtract The image is of 5 repeat measurements of 10 mgL KCN in generic tap water 16 sec each 100 mW of 785 nm
3 RESULTS AND DISCUSSION Raman and surface-enhanced Raman spectra were obtained for potassium cyanide bis-(2-chloroethyl)sulfide and ethyl S-2-diisopropylamino ethyl methylphosphonothioate representing three classes of chemical agents cyanides mustards and nerve agents respectively Spectra were also obtained for 2-chloroethyl ethyl sulfide (CEES) a structural analogue to HD which was included in the study to aid in assigning spectral bands KCN salt was used for cyanide experiments to avoid the increased hazards of handling HCN gas KCN completely dissolves in water forming its conjugate acid HCN according to its Ka of 615x10-1027 and at a concentration of 1 mgmL results in a pH 107 solution This is significant in that only CN- appears to interact sufficiently with silver to produce a SER spectrum and no spectral signal is observed below pH 726 Figure 2 shows the SER and normal Raman spectra for KCN The SERS spectrum of 1mgml KCN in water shows a single intense somewhat broad feature at
A B
C D
SPIE-2003-5269 19
2100 cm-1 assigned to the single CequivN stretch The band is much sharper in the normal Raman spectra of the solid KCN salt at 2074 cm-1 This band does broaden and shift to 2080 cm-1 in solution (not shown) However the observed SERS frequency is attributed to interaction with silver and low frequency mode at 135 cm-1 attributed to a Ag-CN stretch (not shown) supports this conclusion
cm-1 band to a C-S stretch but the authors concede that it is in fact more likely a C-Cl stretch28 It appears that the most intense Raman bands at 648 692 and 747 cm-1 shift to 620 660 and 720 cm-1 in the SER spectra and are tentatively assigned as above The width of these bands suggests that they overlap underlying spectral features Additional bands in the Raman spectra occur at 972 1034 1049 1263 1286 1423 1442 2865 2935 and 2960 cm-1 Corresponding bands occur in the SER spectra at 964 1015 1054 1286 1410 1447 2865 and 2935 cm-1 Most of these bands are associated with alkane modes specifically the bands at approximately 1040 cm-1 to a C-C stretch 1290 cm-1 to a CH2 in-phase twist 1440 cm-1 to a CH2 wag 2865 cm-1 to a symmetric CH2 stretch and 2965 cm-1 to an asymmetric CH2 stretch The Raman and SER spectra of sulfur mustard were measured at the Edgewood center (Figure 4) Both spectra are largely similar to CEES The C-Cl and C-S bands in the Raman spectrum of HD now occur at 640 655 700 739 and 760 cm-1 and are more resolved possibly due to the increased molecular symmetry Theoretical calculations indicate that the first three bands are due to C-Cl stretching modes and the latter two to C-S stretching modes28 Only the C-Cl bands maintain significant intensity in the SER spectra occurring at 624 and 643 cm-1 which is attributed to the
Prior to measurements of HD CEES was examined by Raman and SER spectroscopy (Figure 3) CEES also known as half-mustard is essentially identical to HD except one of the chlorine end atoms is replaced by a hydrogen atom Again although not as toxic as HD CEES is a blister agent and dilute aqueous samples were prepared using appropriate safety equipment Both the Raman and SER spectra of CEES are similar and dominated by bands between 600 and 800 cm-1 These are associated with C-Cl and C-S stretching modes which are tentatively assigned to 648 and 747 cm-1 in the Raman spectra respectively The shoulder at 630 cm-1 the overlapped band at 660 cm-1 and the strong band at 692 cm-1 could also be due to these modes or their asymmetric counterparts It is worth noting that theoretical calculations assign the 692
Figure 2 A) SER and B) NR spectra of KCN Conditions A) 1 mgml in tap water 100 mW of 785 nm at sample 1-min acquisition time B) solid 300 mW of 785 nm 5-min All spectra are 8 cm-1 resolution
A
B
Figure 3 A) SER and B) NR spectra of CEES Conditions A) 1 vv (10 mgml) in MeOH 100 mW of 785 nm 1-min acquisition time B) neat 300 mW of 785 nm 5-min
A
B
Cl-CH2-CH2-S-CH2-CH3
A
B
Figure 4 A) SER and B) NR of HD Conditions A) 1mgml in tap water B) pure both 100 mW of 785 nm 1-min
Cl-CH2-CH2-S-CH2-CH2-Cl
SPIE-2003-5269 20
expected strong interaction between chlorine and silver and adds support to the assignment of this band to a C-Cl stretch Weaker overlapping bands occur at 670 692 and 724 cm-1 the latter possibly due to C-S stretching modes Again the alkane modes are apparent in the normal Raman spectra of HD but only a broad feature at 1300 to 1450 cm-1 suggests CH2 contributions in the SER spectrum Although the observed bands in the VX spectrum have not been assigned (Figure 4) a computer generated Raman spectrum29 predicts many of the same features with surprising accuracy and are used here Two intense bands at 460 and 530 cm-1 closely match predicted bands at 463 and 546 cm-1 assigned to a CH3-P=O bend and a PO2CS wag Three highly overlapped bands occur at 694 745 and 771 cm-1 matching predicted bands at 713 730 and 760 cm-1 The first
Table 2 Measured SER peak heights for the CN stretch at 2100 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
Figure 5 A) SERS and B) NR spectra of VX Conditions A) 1 vv (10 mgml) in MeOH B) pure sample both 100 mW of 785 nm 1-min acquisition time
A
B
two have been assigned to a C-S stretch and CH2bend respectively while the latter has been attributed to either a P-C stretch or an O-C-C stretch Although the 745 cm-1 band may alternatively be assigned to a C-S stretch based on the previous measurements of CEES and HD The relatively intense bands at 890 1106 1218 1445 and 1465 cm-1 also match predicted bands at 880 1108 1216 1440 and 1464 cm-1 that are assigned to a C-C stretch CH3 rock N-C3 stretch various C-H3 bends and C-H bends respectively Both the computer generated and the measured spectra contain numerous other less intense bands One is worth mentioning A unique band appears at 370 cm-1 that is predicted at 368 cm-1 and corresponds to an O-P=O bend The surface-enhanced Raman spectrum of VX is also rich with spectral features It has the unique low frequency band at 370 cm-1 as well as a second band at 380 cm-1 that is assigned tothe S-P-O bend predicted in the normal Raman spectrum at 388 cm-1 Based on the measured and predicted normal Raman spectra the following SERS assignments are given 460 cm-1 to the CH3-P=O bend 544 cm-1 to the PO2CS wag 738 cm-1 to a C-S stretch (based on arguments above) 890 cm-1
to a C-C stretch 1101 cm-1 to a CH3 rock and 1456
cm-1 to a C-H bend The ability of SERS to measure chemical agents in water containing real-world chemical interferents was tested by using the generic tap water described in Table 1 The ability to reproduce measurements was accomplished by preparing three separate water stock solutions which were used to prepare three sample solutions each of 1 mgmL KCN in the generic tap water The samples defined as vials 1-9 were then measured at 4 points per vial in a semi-random fashion such that errors associated with stock solution preparation and errors associated with instrument drift could be identified No trends were apparent that signified such systematic errors Each spectrum collected consisted of 20 averaged scans taking 16 seconds at 8 cm-1 resolution The laser power at the sample was measured periodically during the day and it ranged from 102 to 105 mW spacccc
spot Vial 1 Vial 2 Vial 41 13754 13004 128192 13519 12692 12909
1 3 13541 12721 126394 13462 12648 12651
ave 13569 12766 12755Vial 6 Vial 3 Vial 5
1 11586 14068 10757 2 11236 14402 115122 3 11376 14546 11559
4 10894 11763 11214ave 11273 13695 11261
Vial 7 Vial8 Vial 91 11198 15804 11023
2 11243 15716 88753 3 11562 15216 11218
4 11655 15084 11077ave 11415 15455 10548
CN AVG STDEV ERR1 mgmL pk ht 12526 1560 1245
stock solution
SPIE-2003-5269 21
The CN measurements consisting of 36 data points produced an average height of 125 for the 2100 cm-1 peak with a standard deviation of 156 or 125 (Table 2) The HD and VX measurements were performed precisely the same way (Tables 3 and 4) For HD the 624 cm-1 peak was used for analysis and it had an average height of 53 with a standard deviation of 068 or 129 while for VX the 544 cm-1 peak was used for analysis and it had an average height of 1051 with a standard deviation of 308 or 293 The greater error in the VX measurements can be somewhat attributed to Vial 4 which produced lower SER signal intensities But removing this vial from the data set changes the standard deviation to 233 only a modest improvement
spot Vial 1 Vial 2 Vial 41 525 609 4682 45 675 484
1 3 527 644 5094 559 807 669
ave 51525 68375 5325Vial 6 Vial 3 Vial 5
1 521 574 575 2 536 53 4492 3 509 508 451
4 565 594 379ave 53275 5515 4635
Vial 7 Vial 8 Vial 91 457 589 413
2 583 497 4273 3 544 505 4
4 528 5 509ave 528 52275 43725
HD AVG STDEV ERR1 mgmL pk ht 530 068 1291
stock solution
spot Vial 1 Vial 2 Vial 41 1464 1034 4672 1485 99 61
1 3 1491 989 5684 1041 777 553
ave 137025 9475 5495Vial 6 Vial 3 Vial 5
1 1058 942 1293 2 697 121 965
2 3 727 1206 11124 689 1254 746
ave 79275 1153 1029Vial 7 Vial 8 Vial 9
1 1127 783 163 2 1358 812 1614
3 3 1371 908 15124 1097 875 1397
ave 123825 8445 153825
VX AVG STDEV ERR1 mgmL pk ht 1051 308 2925
stock solution
4 CONCLUSIONS In this paper we examined the ability of surface-enhanced Raman spectroscopy to reproducibly measure CN HD and VX in tap water without chemical interference Both normal and surface-enhanced Raman spectra were examined to select unique bands suitable to identify and quantify these chemical agents For SER measurements the 2100 cm-1 C-N stretch was used for CN the 624 cm-1 C-Cl stretch was used for HD and the 544 cm-1 PO2CS wag was used for VX It was determined that 1 mgmL samples of each of these chemicals measured 36 times in glass vials coated with a silver-doped sol-gel reproduced measurements with standard deviations of 125 129 and 293 It was further found that the 10 chemicals added to simulate generic tap water did not interfere with or alter the SER spectra It should be noted that the concentrations used in this study were considerably greater than those required by the JSAWM program Current work involves improving SER sensitivity and designing sampling systems with better reproducibility This includes the development of fractal silver and gold structures within the sol-gel matrix and the development of chemically selective sol-gels
ACKNOWLEDGMENTS The authors are grateful for the support of the US Army (DAAD13-02-C-0015 Joint Service Agent Water Monitor program) The authors would also like to thank Janet Jensen Ronald Crosier and Kristina Gonser for helpful discussions
Table 4 Measured SER peak heights for the VX band at 544 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
Table 3 Measured SER peak heights for the HD band at 624 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
SPIE-2003-5269 22
REFERENCES 1 Jensen JL NC Wong W R Loerop SPIE 4775-03 (2001) 2 JSAWM Requirements at wwwsbccomapgeaarmymilRDAecbcrtPRODSERJSAWMjsawmhtml 3 Erickson B ldquoThe Chemical Weapons Convention Redefines Analytical Challengerdquo Anal Chem 70 397A (1998) 4 Johnston RL Hoefler CM Fargo JC and Moberley B ldquoThe Defense Nuclear Agencyrsquos
ChemicalBiochemical Weapons Agreements Verification Technology Research Development Test and Evaluation Program and its Requirements for On-Site Analysisrdquo AT-ONSITE 5-8 (1994)
5 Black RM Clarke RJ Read RW and Reid MT ldquoApplication of GC-MS and GC-tandem MS to the analysis of chemical warfare samples found to contain residues of the nerve agent sarin sulphur mustard and their degradation productsrdquo J Chrom 662 301-321 (1994)
6 USA Army Center for Health Promotion and Preventive Medicine (USACHPPM) Technical Guideline 230A 7 Hoffland LD Piffath RJ Bouck JB rdquoSpectral signatures of chemical agents and simulantsrdquo Optical
Engineering 24 982-984 (1985) 8 Braue EHJ Pannella MGrdquoCIRCLE CELL FT-IR Analysis of Chemical Warfare Agents in Aqueous Solutionsrdquo
App Spectrosc 44 1513-1520 (1990) 9 Tseng C-H Mann CK and Vickers TJ ldquoDetermination of Organics on Metal Surfaces by Raman
Spectroscopyrdquo App Spectrosc 47 1767-1771 (1993) 10 Staff Reporter ldquoUS Army Tests Lidar to Detect Biological Toxinsrdquo Photonic Spectra pg 50 December 1998 11 Christesen SD Raman cross sections of chemical agents and simulants App Spectrosc 42 318-321 (1988) 12 Weaver MJ Farquharson S Tadayyoni MA ldquoSurface-enhancement factors for Raman scattering at silver electrodesrdquo J Chem Phys 82 4867-4874 (1985) 13 Alak AM and Vo-Dinh T ldquoSERS of Organophosphorous Chemical Agentsrdquo Anal Chem 59 2149-2153 (1987) 14 Lee Y Farquharson S ldquoRapid chemical agent identification by SERSrdquo SPIE 4378 21-26 (2001) 15 Farquharson S Maksymiuk P Ong K Christesen S ldquoChemical agent identification by surface-enhanced Raman spectroscopyrdquo SPIE 4577 166-173 (2001) 16 Spencer KM Sylvia J Clauson S and Janni J ldquoSurface Enhanced Raman as a Water Monitor for Warfare
Agents in Waterrdquo SPIE 4577 158-165 (2001) 17 Tessier P Christesen S Ong K Clemente E Lenhoff A Kaler E Velev O ldquoOn-line spectroscopic characterization of sodium cyanide with nanostructured gold surface-enhanced Raman spectroscopy substratesrdquo
App Spectrosc 56 1524-1530 (2002) 18 Farquharson S W W Smith S Elliott and J F Sperry Rapid biological agent identification by surface- enhanced Raman spectroscopy SPIE 3855110-116 (1999) 19 Farquharson S WW Smith YH Lee S Elliott and J F Sperry Detection of biological signatures A
comparison of electrolytic and metal-doped sol-gel surface-enhanced Raman media SPIE 4575 62-72 (2002) 20 Guzelian AA Sylvia J Janni J Clauson S and Spencer KM ldquoSERS of whole cell bacteria and trace levels of biological moleculesrdquo SPIE 4577 182-192 (2001) 21 Shende C Inscore F Gift A Maksymiuk P Farquharson S ldquoRapid dipicolinic acid extraction from Bacillus spores detected by surface-enhanced Raman spectroscopyrdquo App Spectrosc 58 accepted 22 Lee Y and Farquharson S ldquoSERS Sample Vials Based on Sol-Gel Process for Trace Pesticide Analysisrdquo SPIE
4206 140-146 (2000) 23 Farquharson S and Lee Y ldquoTrace Drug Analysis by SERSrdquo SPIE 4200-16 (2000) 24 Lee Y Farquharson S and Rainey P M Surface-Enhanced Raman Sensor for Trace Chemical Detection in
Water SPIE 3857 76-84 (1999) 25 Lee Y Farquharson S Kwong H and Shahriari M ldquoSol-Gel Chemical Sensor for Surface-Enhanced Raman
Spectroscopyrdquo SPIE 3537 252-260 (1998) 26 Farquharson S A Gift P Maksymiuk F Inscore and W Smith ldquopH dependence of methyl phosphonic acid
dipicolinic acid and cyanide by surface-enhanced Raman spectroscopyrdquo SPIE 5269-19 (2003) 27 Lide DR Ed Handbook of Chemistry and Physics CRC Press 77th Ed (1996-7) 28 Sosa C RJ Bartlett K KuBulat and WB Person ldquoA theoretical study of harmonic vibrational frequencies and
infrared intensities of XCH2CH2SCH2CH2X and XCH2CH2SH (= H Cl)rdquo J Phys Chem 93 577-588 (1993) 29 Hameka HF and JO Jenson ldquoComputer-generated predictions of the structure and the IR and Raman spectra of
VXrdquo ERDEC-TR-065 May 1993
- Chemagents Appendicespdf
-
- SPIE2001-4575-Bioagent-SERSapdf
-
- Detection of bioagent signatures A comparison of electrolytic and metal-doped sol-gel surface-enhanced Raman media
- Stuart Farquharson Wayne Smith and Yuan Lee
-
- Real-Time Analyzers 87 Church Street East Hartford CT 06108
-
- ABSTRACT
- 4 CONCLUSIONS
- 5 ACKNOWLEDGEMENTS
- 6 REFERENCES
-
SPIE-4378-2001
26
14 Tseng C-H Mann CK and Vickers TJ ldquoDetermination of Organics on Metal Surfaces by Raman Spectroscopyrdquo
Applied Spectroscopy 47 1767-1771 (1993) 15 Staff Reporter ldquoUS Army Tests Lidar to Detect Biological Toxinsrdquo Photonic Spectra pg 50 December 1998 16 Christesen SD Raman cross sections of chemical agents and simulants Applied Spectroscopy 42 318-321 (1988) 17 Alak AM and Vo-Dinh T ldquoSERS of Organophosphorous Chemical Agentsrdquo Analytical Chemistry 59 2149-2153
(1987) 18 Norrod KL Sudnik LM Rousell D and Rowlen KL ldquoQuantitative Comparison of Five SERS Substrates
Sensitivity and Detection Limitrdquo Applied Spectroscopy 51 994-1001 (1997) 19 Lee Y and Farquharson S ldquoSERS Sample Vials Based on Sol-Gel Process for Trace Pesticide Analysisrdquo SPIE 4206
140-146 (2000) 20 Farquharson S and Lee Y ldquoTrace Drug Analysis by Surface-Enhanced Raman Spectroscopyrdquo SPIE 4200-16 (2000) 21 Lee Y Farquharson S and Rainey P M Surface-Enhanced Raman Sensor for Trace Chemical Detection in Water
SPIE 3857 76-84 (1999) 22 Lee Y Farquharson S Kwong H and Shahriari M ldquoSurface-Enhanced Raman Sensor for Surface-Enhanced Raman
Spectroscopyrdquo SPIE 3537 252-260 (1998) 23 see Material Safety Data Sheets for details 24 Farquharson S Smith W Carangelo R C and Brouillette C ldquoIndustrial Raman Providing Easy Immediate Cost
Effective Chemical Analysis Anywhererdquo SPIE 3859 14-23 (1999) 25 Bennett S Bane J Benford P and Pratt R ldquoEnvironmental Hazards of Chemical Agent Simulantsrdquo Aberdeen
Proving Ground Maryland Chemical Research and Development Center CRDC-TR-84055 (1984) 26 Mabey W and Mill T Critical Review of Hydrolysis of Organic Compounds in Water under Environmental Conditions
Journal of Physics and Chemistry Reference Data 7(2) 383-414 (1978) 27 Christesen S MacIver B Procell L Sorrick D Carabba M and Bello J ldquo Noninstrusive Analysis of Chemical Agent
Identification Sets Using a Portable Fiber-Optic Raman Spectrometerrdquo Applied Spectroscopy 53 850-855 (1999) 28 Meylan WM and Howard PH J Pharm Sci 84 83-92 (1995) 29 Jenkins A Uy O and Murray G ldquoPolymer-Based Lanthanide Luminescent Sensor for Detection of Hydrolysis Product
of the Nerve Agent Soman in Waterrdquo Analytical Chemistry 71 373-378 (1999) 30 Nassar A Lucas S and Hoffland L ldquoDetermination of Chemical Warfare Agent Degradation Products at Low-Part-
per-Billion Levels in Aqueous Samples and Sub-Part-per-Million Levels in Soils Using Capillary Electrophoresisrdquo Analytical Chemistry 71 1285-1292 (1999)
SPIE 2001-4575
62
Detection of bioagent signatures A comparison of electrolytic and metal-doped sol-gel surface-enhanced Raman media
Stuart Farquharson Wayne Smith and Yuan Lee
Real-Time Analyzers 87 Church Street East Hartford CT 06108
Susan Elliott and Jay F Sperry University of Rhode Island 45 Lower College Rd Kingston RI 02881
ABSTRACT Since September 11 2001 the threat of terrorist attacks and biological warfare within US borders has become a sobering reality In an effort to aid military personnel and the public at large we have been investigating the utility of surface-enhanced Raman spectroscopy (SERS) to provide rapid identification of chemical agents directly and biological agents through their chemical signatures This approach is based on the ability of Raman spectroscopy to identify molecular structure through the abundant vibration information provided in spectra and the ability of SERS to detect extremely low concentrations (eg part-per-billion) through the enhancement of Raman scattering by six orders of magnitude or more Towards the goal of developing a portable analyzer we have been studying the ability of two SER media to obtain continuous (ie reversible) and quantitative (ie reproducible) measurements Here we compare measurements of nucleic acid-bases adenosine monophosphate and ribonucleic acid extracted from Escherichia coli Bacillus subtilis and Staphylococcus aureus obtained by electrolytic SERS and metal-doped sol-gel SERS The capabilities of these SER media are summarized in terms of rapid detection of B anthracis and dipicolinic acid Keywords bioagent detection SERS RNA analysis bacterial analysis Raman spectroscopy
1 INTRODUCTION The recent distribution of anthrax through the US postal system and the subsequent infection and death of several postal and national media employees amplifies the need for methods to rapidly detect and identify this and other chemical and biological warfare agents (BWA) The primary methods currently used immunoassays for screening and nucleic acid (NA) sequencing for positive identification of BWAs (bacteria protozoa and viruses) have serious limitations123 Immunoassay methods employ competitive binding of the bioagent (as an antigen) and its labeled (eg fluorescence) conjugate for a limited number of antibodies Although this analysis method is fast and semi-quantitative other chemicals may compete for the antibodies interfere with the enzymatic reaction or interfere with the measurement (eg it fluoresces) resulting in a high number of false positive responses1 Furthermore the antibodies denature due to moisture and heat limiting shelf life and require sterile often refrigerated storage Positive identification of a BWA can be accomplished by sequencing deoxyribonucleic acid or ribonucleic acid (DNA and RNA)23 This requires enumeration of a nucleic acid sequence through polymerase chain reactions (PCR) or multiplication of the microorganism through culture growth to provide sufficient quantities of DNA or RNA for analysis Unfortunately PCR and culture growth require from several hours to several days23 Consequently a wide variety of technologies have been investigated for rapid identification of BWAs The Department of Defense is actively monitoring 200 such technologies4 This includes traditional methods such as gas chromatographic separation coupled with ion mobility spectrometry detection5 to exotic methods based on nature such as monitoring toxin induced color changes in fish scales6 Although all of these techniques have achieved varying degrees of success none are yet capable of detecting and identifying BWAs in 10 minutes or less Towards this goal we have been investigating the ability of SERS to detect sub-nanogram quantities of DNA or RNA (eliminating enumeration) determine relative NA base concentrations and identify BWA taxonomy
To whom correspondences should be addresses e-mailfarqureal-time-analyzerscom wwwreal-time-analyzerscom
SPIE 2001-4575
63
Raman spectroscopy has a rich history of investigating biochemical and biological processes7 Some of the earliest laser-Raman studies demonstrated that the five NA bases adenine (A) cytosine (C) guanine (G) thymine (T in DNA) and uracil (U in RNA) yielded distinct spectra with several bands suitable for identification and quantification8 Furthermore these studies included exceptional spectra of both DNA and RNA for which the NA bases as well as several phosphate bands were easily identified9 However since the Raman effect is very inefficient (very low conversion of incident radiation to inelastically scattered Raman radiation) these samples had to be highly concentrated Fortunately two phenomena exist that can increase the generation of Raman photons by six orders of magnitude or more known as the resonance Raman and surface-enhanced Raman effects1011 Resonance Raman scattering occurs when the laser excitation wavelength is in resonance with an electronic transition of a molecule (within the absorption envelope)10 Excitation at ultraviolet wavelengths has been used to obtain resonance Raman spectra of amino acids and nucleic acids in whole bacteria1213 For example excitation at 242 nm has been used to maximize the nucleic acid spectral band intensities and minimize the amino acids band intensities A peak at 1530 cm-1 was found to be proportional to the amount of the NA base cytosine while a peak at 1485 cm-1 was proportional to the combined amount of the NA bases adenine and guanine This quantitative behavior has been used to define an A+TG+C base-pair ratio and provide a level of bacterial identification as taxonomic markers13 In recent years SERS has also been used to analyze bacterial cell components14 including amino acids15 lipids16 nucleic acids151718 and the adenine derivatives192021 SERS has proven to be one of the most sensitive methods for trace chemical analysis through the detection of single molecules2223 including DNA (dye labeled 17-mer)24 Since its discovery in 197425 the mechanism responsible for the large increase in scattering efficiency has been the subject of considerable research2627 Briefly incident laser photons couple to free conducting electrons within a metal which confined by the particle surface collectively cause the electron cloud to resonate2628 These surface plasmons are known as the physical component of the SER effect These surface plasmons can transfer energy to the molecular vibrational modes of molecules through interactions with the molecular electron orbitals2629 This interaction is known as the chemical component of the SER effect This perturbation of the molecular polarizability generates surface-enhanced Raman photons26 A number of methods have been developed to produce surfaces or solutions containing one of these metals with optimum roughness or diameter to promote SERS30 These methods include preparation of activated electrodes in electrolytic cells 1131 activated silver and gold colloid reagents32 and metal coated substrates333435 Selecting a SER-active medium for chemical and biological agent detection requires consideration of the method of deployment and hence the method of sampling Chemical aerosols or airborne bacteria will require a collection device to concentrate and continuously present the sample to the SERS medium Poisoned water supplies will also require a flow through device for continuous monitoring or a grab-sample device for periodic analysis And contaminated surfaces will require a grab-sample extractive device A SERS-based device used for continuous monitoring (air or water) must be reversible and reproducible if quantitative measurements are desired while a SERS-based device used for periodic sampling (water or surfaces) must be reproducible Both reversible and reproducible measurements have been performed using electrolytic SERS (E-SERS)36 But this requires a three-electrode sample cell and an electrolyte of known concentration to perform the necessary oxidation-reduction cycles (ORCs) to re-activate the electrode surface with new uncontaminated sites from one measurement to the next Colloids are severely limited in that continuous measurements would require a continuous supply of colloids For periodic measurements vials of colloids one per measurement could be used However aggregate size and consequently SER intensity change with sample conditions (especially pH) and quantitative reproducible measurements are unlikely Substrates appear to have the greatest potential and designs range from silver evaporated on titania particles34 to periodic gold pyramids evaporated between polystyrene beads35 Most substrates require concentrating the sample on the surface through drying to obtain the largest signal enhancements in effect making the measurements irreproducible and irreversible However successful measurements using flow systems have been obtained with glass posts but manufacturing costs appear prohibitive In an effort to overcome these limitations we have developed metal-doped sol-gels to provide SERS measurements that are reproducible reversible and quantitative and yet not restricted to specific environments such as electrolytes solvents or evaporated surfaces3738 The porous silica network of the sol-gel offers a unique environment for stabilizing SER active metal particles and the high surface area increases the interaction between the analyte and metal particles The sol-gel can be coated on the end of fiber optics or on the internal walls of a glass flow tube for continuous measurements or standard glass sample vials for periodic measurements Previously we measured 100 mgL methylphosphonic acid (the primary hydrolysis product of nerve agents) in water with an estimated detection limit of 05 mgL (100 parts-per-billion) We have also
SPIE 2001-4575
64
demonstrated reversible and reproducible measurements of p-aminobenzoic acid (PABA) in a flow through system Here we investigate the ability of the sol-gel SERS (SG-SERS) to measure the NA bases adenosine monophosphate and RNA extracted from E coli B subtilis and S aureus The measurements are compared to those obtained by E-SERS
2 EXPERIMENTAL The inorganic chemicals and solvents used to prepare samples were spectroscopic grade and purchased from Aldrich (Milwaukee WI) Fisher (Pittsburgh PA) or Pfaltz amp Bauer (Waterbury CT) The nucleic acid bases and dipicolinic acid were purchased from Sigma (St Louis MO) Normal Raman samples were measured to establish enhancement factors In each case 1cm3 of sample was placed into a 1x1 cm glass cuvette weighed and measured Unpacked densities were typically 6-7 gcm3 For all SER measurements including RNA samples were prepared as ~01mgmL (see Figure captions for exact concentrations) in 01M KCl and buffered to a pH of 92 with Na2B4O7bullH2O Adenine pH dependence measurements used pH buffer solutions at 4 (potassium acid phthalate) 69 (potassium phosphate monobasicsodium phosphate dibasic) 92 (Na2B4O7bullH2O) and 104 (tris-hydroxymethyl amino methane) Escherichia coli Bacillus subtilis and Staphylococcus aureus cultures (250ml per 1000mL Erlenmeyer flask) were grown overnight in a Trypticase soy broth (TSB) medium containing 1 glucose in a shaking water bath at 37 oC The bacteria were harvested by centrifugation for 10 minutes at 8000 rpm in a GSA rotor at 5degC then washed once in 085 saline The gram-positive bacteria were concentrated to 20 ml and passed through a French pressure cell twice at 15000 psi to break open the cells RNA was extracted according to Protocol 44139 to ensure pristine samples for initial measurements Since this method takes approximately 4 hours a streamlined method was developed For vegetative bacteria the specimen was boiled for 30 sec in 1 ml of distilled water to lyse the cells and release the RNA For bacterial spores the specimen was first incubated in 1 ml of saline solution containing 02 mg lysozyme and phosphate-buffered to pH of 624 for 1 hr at 37 oC This solution was then boiled for 2-3 minutes in 4X loading buffer to release the RNA For both specimens RNA STAT-60TM was added to the supernatant which was centrifuged at 12000 g for 5 minutes to precipitate the ~15 water-soluble proteins This procedure allowed extracting RNA for SER analysis in ~ 10 minutes Electrophoresis shows high purity while the existence of chemicals that could interfere with the SER measurements is still under investigation The electrolytic sample cell has been described previously36 Briefly a three electrode design is incorporated into a Plexiglas structure containing a well for the reference electrode (a saturated calomel electrode Cole Parmer Vernon Hills IL) and a 5mL sample well containing the silver working electrode and platinum wire counter electrode (05 mm wire Alfa Ward Hill MA) A channel connecting the two wells contained a 2 mm diameter semi-porous membrane (10-20 micron pore Ace Glass) The silver electrode was made from a 3 mm length of 2 mm diameter silver wire (Alfa) soldered to a copper wire lead encased in a 4 mm diameter Pyrex tube A cap containing the silver electrode platinum wire and nitrogen purge and vent lines fixed the silver electrode surface 1 mm from a 1 mm thick glass plate attached to the bottom of the sample well The potentiostat used to control the three electrodes was built in-house and has been described in detail elsewhere36 A multifuntional analog digital and timing inputoutput interface card (DAQCard-1200 National Instruments) is used to both drive the electrolytic cell as well as read the current generated in the cell A LabVIEW software program is used set the oxidation potential reduction potential SER measurement potential hold times and sweep rates The amount of charge passed was plotted as a cyclic voltammogram For all spectra presented five oxidation-reduction cycles (ORCs) stepping from -03 VSCE to 03 VSCE and back to -3 VSCE at 50 mVsec were used The SG-SER measurements were accomplished by simply placing the identical samples prepared above into Simple SERS Sample VialsTM (RTA) These 2-mL glass vials are internally coated with ~ 01 micron thick silver-doped sol-gel A prototype Fourier transform Raman spectrometer (RTA) was used for these measurements40 The system consisted of a NdYAG laser (Brimrose or Spectra Physics) for excitation at 1064 nm an interferometer built by On-Line Technologies (OLT East Hartford CT) for frequency separation an uncooled InGaAs detector for signal detection (RTA) and an Intel 400 MHz Pentium II based laptop computer (Dell Round Rock TX) for interferometric control data acquisition (OLT) and analysis (LabVIEW by National Instruments Austin TX) Additional components included a Notch filter (Kaiser Ann Arbor MI) and interferometer entrance and exit optics (Edmund Scientific Barrington NJ) Fiber optics were used to deliver the excitation beam to the sample and the scattered radiation to the interferometer (2 meter lengths of 200 and 365 micron core diameter respectively Spectran Avon CT) A second Notch filter (Kaiser) was used as a beam splitter to direct the excitation beam along the same axis as the collected radiation A microscope object (20x magnification 04 numeric aperture Newport Irvine CA) was used to focus the beam into the sample and to collect the scattered radiation back along the same axis In this co-axial backscattering arrangement the excitation beam passed through the glass plate onto the silver
SPIE 2001-4575
65
electrode surface for E-SERS through the vial glass wall and into the silver-doped sol-gel film for SG-SERS or through the glass wall of the cuvette and into the solid sample for normal Raman spectroscopy All E-SERS and normal Raman spectra were obtained with 750 mW of laser power at the sample while all SG-SERS spectra were obtained with 75 mW of laser power at the system Incident powers above 200 mW in some cases degraded the sol-gel
3 RESULTS AND DISCUSSION The generation of surface-enhanced Raman scattering at electrode surfaces has been extensively researched and the optimum sample conditions are well developed2729 Several researches incorporated electrodes into flowing systems and demonstrated that continuous monitoring of chemicals is possible18 These successes suggested investigated the capability of E-SERS to measure the NA bases and RNA The E-SERS measurements also provided a benchmark to compare and evaluate SG-SERS measurements The molecular structure of adenine (as well as the other base pairs) which includes an aromatic nitrogen-containing heterocycle is ideally suited to interact with the surface plasmons and contribute substantially to the chemical component of the SER effect1119 Even with excitation at 1064 nm a 3-minute scan of 18x10-5M adenine yields high signal-to-noise (SN) E-SER spectra and all of the bands are revealed with clarity (Figure 1 Table 1) Spectra of similar quality were obtained by SG-SERS and the principal spectral bands are easily observed The identical 18x10-5M adenine sample was measured in the same 3-minute time frame but with 110th the laser power The lower power appears to reduce the SN The amount of adenine responsible for the SER spectra as well as enhancement factors for the two SER media can be determined The molecules producing the E-SERS spectrum are those on the electrode surface within the illumination area of the laser (The solution concentration only determines the number of molecules available to adsorb to the electrode surface) For the current experiments the laser illuminates an area of 28x10-7m2 or 56 x10-7m2 if we assume the ORCs increase the surface area by a factor of two Furthermore if we assume monolayer coverage on the electrode and each 3x5 angstrom molecule (lying flat) occupies 15x10-19m2 then there are ca 4x1012 molecules contributing to the Raman scattering This is ca twice the number of molecules measured at electrode surfaces using either differential capacitance-potential curve measurements or rapid linear sweep voltammetry (eg 3x1018 moleculesm2 for pyridine and pyrazine)29 Thus the adenine spectrum in Figure 1 is due to 87x10-10g (6x10-12 moles) A detection limit defined as a SN of 3 can also be calculated The SN for a 3-minute scan is 844 for the 735 cm-1 band suggesting a mass detection limit of 3x10-12g (2x10-14 moles) This is consistent with previous estimates for adenine by others of 25 x10-14 moles1530 However sub-monolayer concentrations must be measured to verify this The root-mean-squared (RMS) noise is measured between 4400-4600 cm-1 Since noise is distributed evenly throughout the spectrum when transformed this region does not have any
Wavenumbers (∆cm-1) Figure 1 A) Normal Raman spectrum of pure adenine powder B) E-SERS and C) SG-SERS of 18x10-5M adenine at pH 92 All spectra 8 cm-1 resolution 200 scans (3 min) and 1064 nm excitation A) and B) 750 mW C) 75 mW B) measurement potential of 11VSCE
A
B
C
A
B
C SG-SERS
pH 10
D
725
735
735
pH 4
Wavenumbers (∆cm-1) Figure 2 A) and C) E-SERS and B) and D) SG-SERS of adenine at A) and B) pH 104 and C) and D) pH 40 Note consistent appearance of bands at 1270 and 1375 cm-1 as the pH is changed to 10 for both SER media E-SERS used 750 mW SG-SERS used 75 mW of 1064 nm excitation
SPIE 2001-4575
66
contributions from signals or baseline offsets The measurement error is given as SplusmnRMS and for adenine this equals 234 The number of molecules contributing to the SG-SERS are those on the silver particles that are embedded in the sol-gel The total silver surface area can be determined from the average particle size (40 nm diameter) concentration (073 by weight based on molar conc and measured sol-gel density) and the scattering volume (a cylinder defined by the laser area 28x10-7m2 and sol-gel thickness10-4m) The 61x109 silver particles in this volume have a collective area of 31x10-5m2 However it may reasonably be assumed that at least half of the silver surface is in contact with the sol-gel matrix and unavailable for analyte interaction Then approximately 10x1014 molecules or 22x10-8g of adenine contribute to the SG-SER spectrum The slightly lower SN of 207 suggests a mass detection limit of 32x10-10g Determination of the enhancement factors for the two SER media requires estimating the number of adenine molecules contributing to the normal Raman spectrum Here a cylindrical scattering volume is assumed again based on the laser area (28x10-7m2) and the penetration depth (1x10-3 m)41 The density of the sample was measured at 064 gcm3 indicating that 18x10-4g (13x10-6 moles) of adenine produced the normal Raman signal The enhancement factor EF is defined by the following equation
EF = (ISERSINR)bull(MNRMSERS) bull(PNRPSERS) bull(TNRTSERS)12
where I is the spectral band intensity (here 735 cm-1) M is the sample mass P is the incident laser power and T is the measurement time (or number of scans) For the E-SERS measurement the enhancement factor is 22x105 (01780184) bull (18x10-487x10-10)) while the SG-SERS enhancement factor is 10x105 (0160184) bull(18x10-422x10-8) bull(75075) bull(315)12) The lower enhancement for the SG-SERS may be real or the available surface of the silver embedded in the sol-gel may have been overestimated In addition to enhancing the Raman scattering efficiency to an extent similar to E-SERS the SG-SER medium also yields an identical shift of the adenine ring-breathing mode from 725 cm-1 in the normal Raman to 735 cm-1 Furthermore in the course of optimizing the E-SERS sample conditions it was found that pH influenced the adenine interaction with the silver surface (Figure 2) In particular the relative band intensities of the pyrimidine ring skeletal vibrations at 1270 and 1375 cm-1 and the imidazol ring skeletal vibration at1335 cm-1 change At pH 4 adenine is protonated presumably the imidazol ring since the band at 1335 cm-1 increases in intensity while the pyrimidine bands are virtually absent Conversely at pH 10 the imidazol band decreases in intensity while the pyrimidine bands appear It is worth noting that the ring-breathing mode at 735 cm-1 changes little between pH 4 and 10 suggesting that the skeletal changes are more a function of molecule-plasmon interactions than reorientation of the molecule on the surface Measurements of the identical pH series of adenine samples by SG-SERS yielded virtually identical spectral changes This suggests that the sol-gel does not influence the measurement This is critical to reproducing measurements and performing quantitative analysis Next the remaining NA bases were measured by both E-SERS and SG-SERS and compared Previously we examined the optimum pH and electrode potentials for E-SERS measurements to determine if a common pH could be used that yielded good sensitivity for all the bases and if variations in potential could be used to provide an added degree of selectivity between the bases Primarily it was found that high quality spectra were obtained between pH 7 and 95 and that cytosine and uracil were best enhanced at potentials positive of the potential-of-zero charge (pzc ca 065VSCE for Ag) guanine and thymine near the pzc and adenine negative of the pzc In all cases the ring-breathing modes were the most intense and in general could be used to identify the NA bases (Figure 3 Table 1) Specifically adenine has an intense band at 735 cm-1 cytosine at 797 cm-1 guanine at 653 cm-1 thymine at 784 cm-1 and uracil at 800 cm-1 The adenine cytosine guanine and thymine bands are sufficiently separated that their contributions to DNA should be determinable Although adenine and guanine contributions to RNA should also be determinable cytosine and uracil are highly overlapped and unfortunately share the same potential dependence Alternate unique bands at 1183 cm-1 for cytosine and 1275 cm-1 for uracil might be suitable for calculating contributions The SG-SER spectra of the remaining NA bases faithfully reproduced the E-SER spectra In particular the primary identifying bands occur at virtually the same wavenumbers (see Table 1) However the spectra for both cytosine and thymine contain an intense band at ca 1040 cm-1 Initially this was attributed to the pH buffer but samples prepared without either the buffer or the 01M KCl electrolyte yielded identical spectra containing this band In fact the E-SER and SG-SER spectra of thymine are virtually identical except for this band Also the SG-SERS of guanine contains an intense band at 1551 cm-1 that is not observed in the E-SER spectrum This band may be due to a moderately intense band at 1553cm-1 in the normal Raman spectrum that is SG-SER active It was also found that the SG-SERS of cytosine was considerably better than the E-SERS while uracil showed the opposite relationship It is also worth noting that all of the SG-SERS were obtained with 110th the laser power Most importantly the primary ring-breathing modes in the SG-SER spectra are sufficiently intense and unique to be used in determining contributions to DNA and RNA as outlined above
SPIE 2001-4575
67
Table 1 Comparison of E-SER and SG-SER Spectral Band Positions for the NA Bases and Adenosine Monophosphate
Adenine Cytosine Guanine Thymine Uracil AMP E-SERS SG-SERS E-SERS SG-SERS E-SERS SG-SERS E-SERS SG-SERS E-SERS SG-SERS E-SERS SG-SERS1647 1638 1634 1656 1655 1630 1587 1585 1510 1580 1551 1539 1456 1456 1465 1460 1462 1480 1453 1459 1394 1398 1425 1431 1435 1399 1404 1392 1374 1375 1373 1383 1370 1335 1332 1311 1307 1333 1331 1353 1348 1331 1329 1265 1273 1280 1292 1278 1276 1275 1279 1271 1183 1195 1222 1232 1221 1219 1204 1205 1180 1144 1097 1033 1029 1038 1040 1035 1051 1037 1041 1035 963 963 957 1001 1000 961 944 884 819 817 859 866 735 737 797 799 784 782 800 800 72738 742 630 630 653 664 667 684 603 602 590 611 466 561
Bands unique to SG-SERS The next chemical to be analyzed by both E-SERS and SG-SERS was adenosine monophosphate (AMP) The E-SER spectrum yields bands due to the adenine chemical functionality at 727 961 1233 1279 1331 1381 and 1486 cm-1 In addition phosphate bands are observed at 860 1097 1453 1587 and 1705 cm-1 (Figure 5) Other researchers have noted that the ribose component does not appear to contribute to the spectrum19 The AMP spectrum also changes as a function of potential As the electrode is swept more positive (here from -09 to -03VSCE) the phosphate bands at 860 1097 1453 and 1587 cm-1 increase in intensity compared to the adenine bands while a band at 1705 cm-1 appears The adenine bands at 1233 1381 and 1486 cm-1 virtually disappear These potential dependent spectral changes are consistent with earlier studies that show that phosphate is attracted to silver at potentials positive of the pzc but repelled at potentials negative of the pzc19
Wavenumbers (∆cm-1) Figure 3 E-SERS of A) 21x10-3M cytosine at -03VSCE 1000 scans B) ~10x10-5M guanine at -06VSCE 500 scans C) 23x10-3M thymine at -06VSCE 500 scans and D) 12x10-3M uracil at -093VSCE 500 scans All spectra at pH 92 750 mW 1064 nm at 8 cm-1
Wavenumbers (∆cm-1) Figure 4 SG-SERS of A) 21x10-3M cytosine 200 scans B) ~10x10-5M guanine 200 scans C) 23x10-3M thymine 200 scans and D) 12x10-3M uracil 500 scans All spectra at pH 92 75 mW 1064 nm at 8 cm-1
A
B
C
A
B
C
D D
SPIE 2001-4575
68
The SG-SER spectrum of AMP is considerably different The adenine bands virtually disappear except for the two primary bands which shift to742 and 1329 cm-1 While the phosphate band at 1459 cm-1 has gained considerable intensity In addition two new intense bands appear at 684 and 1539 cm-1 as well as a moderately intense band at 1180 cm-1 The SG-SER spectrum has greater similarity to the E-SER spectrum at -03VSCE and suggests that the silver particles embedded in the sol-gel behave as if at a potential positive of the pzc RNA samples extracted from E coli B subtilis and S aureus were next examined by both E-SERS and SG-SERS E-SER spectra of these samples yielded quality spectra in 10 minutes in which all of the major features can be identified (Figures 6 and 7) This includes guanine at 650 cm-1 adenine at 791 cm-1 cytosine and uracil combining at 790 cm-1 and phosphate at 1100 1335 (in combination with adenine and guanine) 1465 and 1570 cm-1 Surprisingly adenine which demonstrated the greatest surface-enhanced Raman effect does not dominate the ring-breathing mode portion of the spectrum The intensities of the other base-pairs bands are of the same order of magnitude This suggests that when the base-pairs are linked together as in RNA they are enhanced in concert In fact the relative intensities are very similar to a normal Raman spectrum of E coli RNA which shows the combined cytosine and uracil band at ca twice the intensity of the adenine band and ca four times the intensity of the guanine band Unfortunately this means that the independent enhancement factors for the NA bases can not be used to estimate relative concentrations For example the relative 791 and 734 cm-1 bands for B subtilis would indicate that the cytosine andor uracil concentration was at least 20 times the adenine concentration whereas each of the four RNA bases are known to contribute 15-35 Nevertheless it is worth noting that the three RNA samples yield different relative band intensities that were reproduced in numerous measurements Although the relative concentrations of the NA bases for these samples have not been determined these differences can be quantified If it is assumed that the 650 cm-1 band represents 25 guanine the 791 cm-1 band represents 25 adenine and the 790 cm-1 50 cytosine plus uracil in the E coli RNA spectrum then the relative concentrations can be estimated for the other RNA samples To aid this calculation the three spectra were normalized to the phosphate band at 1100 cm-1 which has been shown to correlate to the total phosphate concentration and can be used as an internal standard In addition a simple baseline correction was applied (Figure 7) This yields 15 adenine 30 guanine and 55 cytosine plus uracil for B subtilis RNA and 18 adenine 25 guanine and 57 cytosine plus uracil for S aureus RNA The average SN of these measurements was 26 with an average error of 8 of the value (SplusmnN) It is also worth noting that the three RNA spectra show a marked shift in a band near 825 cm-1 This band is assigned to the symmetric stretch of the O-P-O ester linkage9 The band appears at 815 cm-1 for S aureus shifting to 820 cm-1 for B subtilis and 830 cm-1 for E coli Others have used the normal Raman intensity of the band at 815 cm-1 as a direct indication of the amount of A-class helix present while the intensity of the band at 830 cm-1 has been used as a direct indication of the amount of B-class helix present However the latter is more associated with DNA than RNA
Wavenumbers (∆cm-1) Figure 5 E-SER spectra of 020 mgmL adenosine monophosphate at A) -03 and B) -09VSCE and C) SG-SER spectra Conditions sample in 01M KCl buffered to pH 92 A) and B) 750 mW C) 75 mW of 1064 64 scans (1-min) at 8 cm-1
Wavenumbers (∆cm-1) Figure 6 E-SERS of 01 mgmL RNA from E coli 02 mgmL RNA from B subtilis and 02 mgmL RNA from S aureus Conditions 01M KCl pH 92 -03VSCE 750 mW of 1064 nm 640 scans (10 min) at 8 cm-1
AMP RNA
E coli A
B
C
B subtilis
S aureus
SPIE 2001-4575
69
SG-SER spectra of reasonable quality were also obtained for E coli and B subtilis especially the latter (Figure 8) However the spectra differ substantially from the E-SERS of the same samples Both SG-SER spectra are dominated by adenine at 735 cm-1 and a band at 1030 cm-1 Although unassigned the latter does appear in the RNA E-SER spectra Bands at 1105 and 1565 cm-1 are likely due to phosphate while bands at 1320 and 1470 cm-1 are less confidently assigned to phosphate They are significantly less intense and somewhat shifted from their SG-SERS counterparts (1335 and 1455 cm-1) A band at 670 cm-1 may be due to guanine which was observed at 664 cm-1 for SG-SERS of the pure sample However the SG-SER spectrum of AMP also had an intense 667 cm-1 band A number of other bands occur at 890 1070 1165 1245 1290 1420 1505 cm-1 and remain unassigned The SG-SER spectra are somewhat disappointing in that only adenine and guanine contributions can be positively identified This limits the ability to determine relative NA base concentrations and distinguish bacterial RNA However several of the unassigned bands may be due to the bases (eg 1030 and 1420 cm-1 due to cytosine) Further experiments will be required to clarify this point
Ecoli
B subtilis
S aureas
P G
A C+U
OPO
A B
Wavenumbers (∆cm-1) Wavenumbers (∆cm-1) Figure 7 SER spectra of RNA from A) B subtilis with contributions indicated and B) E coli B subtilis and S aureus with baseline correction and peak positions used to calculate contributions indicated G = guanine A = adenine C+U = cytosine plus uracil P = phosphate (backbone) OPO = phosphate ester linkage (A- vs B-class helix)
Wavenumbers (∆cm-1) Wavenumbers (∆cm-1) Figure 9 E-SER (-03VSCE) and SG-SER spectra of RNA from A) E coli and B) B subtilis Sample conditions as in Figure 6 E-SER spectra at 750 mW SG-SERS at 75 mW
A B E-SERS
SG-SERS
E-SERS
SG-SERS
SPIE 2001-4575
70
A final comparison was made between the two SER methods by measuring dipicolinic acid (DPA) This chemical may be invaluable as a test for spore forming bacteria specifically B anthracis 50 to 90 of B anthracis sporilates During spore formation dipicolinic acid is synthesized and once completed 10-15 of the dry spore weight is composed of the Ca2+ complex located in the spore core42 Heating in water can be used to initiate germination at which point the exosporium breaks and releases the Ca dipicolinate which becomes dipicolinic acid in water The structure of this chemical strongly
4 CONCLUSIONS Towards the goal of developing a method and technology to rapidly detect and identify bioagents we have been investigating surface-enhanced Raman spectroscopy as a tool to measure relative concentrations of nucleic acid bases in RNA and determine bioagent taxonomy Initially we investigated E-SERS since this method has been extensively researched and the optimum sample conditions are well developed However this method requires a three-electrode sample cell and electrolyte solution Incorporation of an electrolytic cell into sample systems used to analyze for BWA as aerosols in water or on surfaces can be designed using flow injection analysis technologies but cross-contamination and plugging of sample lines seems inevitable For this reason we also investigated metal-doped sol-gels as a SER-active medium Previous studies have shown this material to be active in all solvents particularly water capable of continuous measurements in flowing systems and reproducible (quantitative) between coated sample vials Here we compared SG-SER spectra to traditional E-SER spectra of the nucleic acid base pairs adenosine monophosphate and RNA High quality spectra of adenine cytosine guanine thymine and uracil were obtained by both E-SERS and SG-SERS Both methods yielded very similar spectra for the NA bases including a pH dependent study of adenine Enhancement factors and detection limits for adenine were determined as 2x105 and 16x10-11g and 1x105 and 12x10-10g for E-SERS and SG-SERS respectively (normalized to 75 mW and 10-min acquisition time) Fifty percent of the silver particle surface area in the sol-gel matrix was assumed covered by adenine which may have been overestimated yielding a lower EF and higher detection limit It should also be realized that each E-SER spectrum required several attempts to optimize the measurement conditions (pH electrode potential etc) While each SG-SER spectrum involved no sample preparation and often represents the first and only attempt to make the measurement Quality spectra of RNA extracted from Escherichia coli Bacillus subtilis and Staphylococcus aureus were obtained by E-SERS that were easily interpreted Bands due to adenine guanine cytosine plus uracil and phosphate were identified The SER band intensity of the NA bases in the RNA samples were of the same order of magnitude suggesting that their interaction with the silver surface is concerted as is their Raman enhancement Interestingly the relative SER band
Wavenumbers (∆cm-1) Figure 10 A) Raman spectrum of solid dipicolinic acid B) Ax20 C) electrolytic SERS of 6x10-3 M dipicolinic acid in 01 M KCl at a potential of +07VSCE and pH of 4 and D) sol-gel SERS of 6x10-3 M dipicolinic acid Conditions for A and C as in Figure 1 C) 100 mW of 1064 nm 50 scans 8 cm-1
A B (Ax20)
C
D
suggested that it would be SER active However the E-SER spectrum was unstable and varied considerably as a function of potential A consistent spectrum was obtained at +06VSCE (Figure 10) This potential is not recommended for measurement because the surface is actively dissolving in solution The SG-SER spectrum was considerably more stable of higher quality and easily reproduced Bands at 660 825 1010 1390 1430 1570 1590 and 3075 cm-1 were observed Enhancement factors were determined for the two media using the symmetric ring stretching mode at 995 and 1010 cm-1 for the normal Raman and SER spectra respectively E-SERS yielded an EF of 5x103 while SG-SERS yielded an EF of 2x105 for DPA The SN of the latter suggests a detection limit of 20x10-10g (based on adenine coverage 75 mW and 10-min) The differences in SER activity for these two media may be attributed to the combined electrolytic potential of the solution chemical and metal15 Again the E-SERS suggests that the SG-SERS is at a potential positive of the pzc While the instability in the E-SERS may also be associated with surface interactions of two carboxylic acid groups of dipicolinic acid during the ORCs
DPA
SPIE 2001-4575
71
intensities for RNA extracted from E coli are very similar to those measured by normal Raman spectroscopy Although the relative percent that each of the NA bases contributed to each RNA sample was not determined reproducible band intensities allowed noting the following trends The percent adenine decreases while the combined percent cytosine and guanine increase for both B subtilis and S aureus compared to E coli Quality spectra were also obtained for the RNA samples by SG-SERS but only a few bands were readily identified Calculations of NA base concentrations by SG-SERS will require further research In light of recent events we summarize the capabilities of these SERS media in terms of rapid detection of B anthracis and dipicolinic acid However these capabilities must be qualified First and foremost the level to which SERS can distinguish bacteria or viruses has not yet been determined Development of a database of both DNA and RNA base concentrations for BWAs and common bacteria to establish the level of taxonomic identification is ongoing Second rapid collection of aerosol water or surface samples is being addressed by others who report trapping particles on filters from 100 liters of air per minute Third although not presented here we have developed methods to extract RNA or DNA from cells and spores for SER analysis within 10 minutes Finally we assume a detection limit of 3600 spores per 100 liters of air is required although a 50 lethal dosage of anthrax has not been established With these qualifications a mass detection limit for RNA using SERS is estimated as follows A single measurement is performed in ca 20 minutes (140 liters collected in 14 min RNA extracted in 8 min spectral acquisition and analysis in 10 min) The average human breaths 7 liters per minute therefore the analyzer must at the very minimum detect 5000 spores in 140 liters of air One spore is approximately 2x10-18m3 (1x1x2 microm) and if a density of 075 gcm3 is assumed this corresponds to a mass of 15x10-12g Each spore contains 4-12 RNA or 12x10-13g RNA for 8 If we assume 23 of the RNA can be isolated for analysis during lysis then the proposed instrument must be able to detect 4x10-10g RNA from 5000 spores per 70 liters of air within 10 minutes As noted above the mass detection limits for adenine were estimated at 16x10-11g and 12x10-10g for E-SERS and SG-SERS respectively Although these detection limits suggests that RNA from 5000 spores is detectable with the current instrumentation it is highly likely that only a portion of an RNA segment (eg 120-nucleotide 5S rRNA) is in contact with the metal surface and will contribute to the SER effect The SN for the RNA spectra were 110th of the average SN for the four individual RNA bases suggesting a 10 contribution Furthermore effective taxonomy will likely require knowing the NA base concentrations to 1 of the value (eg 25plusmn025) Again the average measurement error for the bases is 12 These values suggest that the E-SERS is within a factor of 4 of the required detection limit whereas the SG-SERS detection limit must be improved by 25 times The same arguments can be applied to the detection of dipicolinic acid If we assume a spore releases 10 by weight DPA during germination then the proposed instrument must be able to detect 75x10-10g DPA from 5000 spores per 70 liters of air within 10 minutes The detection limit for SG-SERS was estimated at 20x10-10g and suggest that the vials are suitable to perform a rapid screen for anthrax A series of concentration dependent measurements are currently being performed to verify this assertion Finally we note that the measurements performed here employed an FT-Raman spectrometer This instrumentation was chosen over dispersive Raman instrumentation for the high wavelength accuracy afforded by the HeNe reference laser (Connes Advantage43) which would allow reliable spectral subtraction matching of observed spectra to stored library spectra and confident use of chemometric approaches Such data analysis is likely to be required to enhance BWA identification However this instrumentation sacrifices sensitivity due to a less efficient detector (InGaAs vs Si) less efficient Raman scattering and less efficient generation of plasmon modes Substantial improvements can be obtained using visible excitation and Si detection and these measurements are underway
5 ACKNOWLEDGEMENTS The authors are grateful to Drs D Cookmeyer and S Tove of the US Army Research Office (Contract Number DAAH04-96-C-0078) for their interest and support of this research The authors would also like top acknowledge Dr R Yin and J Jensen for supporting development of the metal-doped sol-gels (Contract Number DAAD13-01-C-0019) They also thank Dr Wilfred H Nelson for assistance in spectral interpretations
SPIE 2001-4575
72
6 REFERENCES 1 Roberts WL and Rainey PM Clin Chem 39 1872-1877 (1993) 2 Pasechnik VA CC Shone and P Hambleton Bioseperations 3 267-283 (1993) 3 Jackson PJ ME Hugh-Jones DM Adair G Green KK Hill CR Kuske LM Grinberg FA Abramova and P
Keim Proc Natl Acad Sci 95 1224-1229 (1998) 4 Jensen JL NC Wong W R Loerop SPIE 4775-03 (2001) 5 Snyder A Pet al SPIE 3853-15 (1999) 6 Danosky T R and McFadden P N in press (1997) 7 Woodruff WH Farquharson S Science 201 831-833 (1978) 8 Lord RC and Thomas GJJr Spectrochemica Acta 23A 2551-2591 (1967) 9 Thomas GJJr Biochim Biophys Acta 213 417-423 (1970) 10 Placzek G Handbuch der Radiologie 2 EMarx ed Akademische Verlagagescellschatt Liepzig 1934 UCRL
Trans No 526 (1959) 11 Jeanmaire DL and RP Van Duyne J Electroanalytical Chem 84 1-20 (1977) 12 Chada S Manoharan R Moenne-Loccoz P Nelson WH Peticolas WL and Sperry JF Applied Spectroscopy
47 38-43 (1993) 13 Manoharan R Ghiamati E Chada S Nelson WH and Sperry JF Applied Spectroscopy 47 2145-2150 (1993) 14 Todd EA Morris MD Applied Spectroscopy 48 545-548 (1994) 15 Wentrup-Byrne E Sarinas S and Fredericks PM Applied Spectroscopy 47 1192-1197 (1993) 16 Weldon MK VR Zhelyaskov and MD Morris Applied Spectroscopy 52 265-269 (1998) 17 Kneipp K and J Fleming J Mol Structure 145 173-179 (1986) 18 Pothier NJ and Force RK Applied Spectroscopy 46 147-151 (1992) 19 Ervin KM E Koglin JM Sequaris P Valenta and HW Nurnberg J Electroanal Chem 114 179-194 (1980) 20 Kim SK TH Joo SW Suh and MS Kim J Raman Spectrosc 17 381-386 (1986) 21 Pothier NJ and Force RK Analytical Chemistry 62 678-680 (1990) 22 Kneipp K Y Wang RR Dasari and MS Feld Applied Spectroscopy 49 780-784 (1995) 23 Nie S and Emory SR Science 275 1102 (1997) 24 Graham D WE Smith AMT Linacre CH Munro ND Watson and PC White Analytical Chemistry 69 4703-
4707 (1997) 25 Fleischmann M PJ Hendra and AJ McQuillan Chem Phys Lett 26 163-166 (1974) 26 Pettinger B J Chemical Phys 85 7442-7451 (1986) 27 Surface-Enhanced Raman Scattering Section Four Theory SPIE MS 10 M Kerker and B Thompson Eds (1990) 28 Wang D-S and Kerker M SPIE (M Kerker and B Thompson Eds) MS 10 417-429 (1990) 29 Weaver MJ Farquharson S Tadayyoni MA J Chem Phys 82 4867-4874 (1985) 30 Norrod KL Sudnik LM Rousell D and Rowlen KL Applied Spectroscopy 51 994-1001 (1997) 31 Farquharson S Weaver WJ Lay PA Magnuson RH and Taube H J Am Chem Soc 105 3350-3351 (1983) 32 Lee PC and Meisel D ldquoAdsorption and Surface-Enhanced Raman of Dyes on Silver and gold Solsrdquo J Phys Chem
86 3391-3395 (1982) 33 Li Y-S and Wang Y Applied Spectroscopy 46 142-146 (1992) 34 Bello JM DL Stokes and T Vo-Dinh Analytical Chemistry 61 1779-1783 (1989) 35 van Duyne RP J C Hulteen D A Treichel M T Smith M L Duval and T R Jensen J Phys Chem B1033854-
3863 (1999) 36 Farquharson S and W W Smith W H Nelson and J F Sperry SPIE 3533-27 207-214 (1998) 37 Lee Y H W Smith S Farquharson H C Kwon M R Shahriari and P M Rainey SPIE 3537 252-260 (1998) 38 Lee Y-H S Farquharson and P M Rainey SPIE 3857 76-84 (1999) 39 Current Protocols in Molecular Biology Wiley Interscience1003-1006 (1987) 40 Farquharson S Smith W Carangelo R C and Brouillette C SPIE 3859 14-23 (1999) 41 Chase D B and JF Rabolt Fourier Transform Raman Spectroscopy Acad Press Ch1 p 131 (1994) 42 Brock TD MT Madigan JM Martinko and J Parker Biology of Microorganisms 7th Ed Prentice Hall p 76-80
(1994) 43 Connes J Rev Opt Theor Instrum 40 45 (1961)
166
Chemical agent identification by surface-enhanced Raman spectroscopy
Stuart Farquharson and Paul Maksymiuk
Real-Time Analyzers 87 Church Street East Hartford CT 06108
Kate Ong and Steven D Christesen US Army SBCCOM Aberdeen Proving Ground MD 21010
ABSTRACT The recent distribution of anthrax through the US postal system and the subsequent infection and death of several postal and national media employees amplifies the need for methods to rapidly detect identify and quantify this and other chemical and biological warfare agents The US military has also identified water supplies as a likely method of warfare agent deployment and is funding the development of a Joint Service Agent Water Monitor (JSAWM) In an effort to aid military personnel and the public at large we are developing a portable analyzer capable of identifying and quantifying chemical agents rapidly either on-demand or continuously The approach is based on the ability of Raman spectroscopy to identify molecular structure through the abundant vibration information provided in spectra and the ability of surface-enhanced Raman spectroscopy (SERS) to detect extremely low concentrations (eg part-per-billion) through the enhancement of Raman scattering by six orders of magnitude or more A key element to the analyzer design is a new SER active medium that is capable of quantitative reversible measurements The medium consists of silver or gold nanoparticles incorporated into a sol-gel matrix The porous silica network offers a unique environment for stabilizing SER active metals and the high surface area increases the interaction between the analyte and metal particles Here we present the use of new sol-gels that also selectively enhance chemicals based on polarity and charge Base-line measurements of chemical agents and their hydrolysis products are presented and compared to the JSAWM goal of 30 micrograms per liter detection Keywords Chemical warfare agent hydrolysis product SERS Raman spectroscopy sol-gel nanoparticle
1 INTRODUCTION Since September 11 2001 the threat of terrorist attacks and biological warfare within US borders has become a sobering reality The simplicity in manufacturing ease of deployment and the relatively low cost of chemical warfare agents (CWAs) raises public concern that they may also be used by terrorists Indeed terrorists released sarin (GB) in the Tokyo subway in 19951 Countering terrorism and terrorist attacks requires recognizing likely deployment scenarios and having the required technology to rapidly detect the deployment event One method of deployment has been long identified by the US military distribution through water supplies To counter this threat the Department of Defense is funding or monitoring the capabilities of 200 technologies with the goal of developing a Joint Service Agent Water Monitor (JSAWM) that is field portable2 The JSAWM must be capable of identifying the chemical agents with no false-positives and quantifying the agents at microgram per liter concentrations within 10 minutes (Table 1)3 This includes the analysis of drinking water supplies distribution and storage systems as well as potable water supplies The earliest technologies developed for CWA detection were based on electrochemical ionization or colorimetric analysis (eg phosgene tape) Although these analyzers were easy to use they were not generally agent specific and suffered from false-positives4 More traditional laboratory methods have also been investigated and in particular combined gas chromatography and mass spectrometry (GCMS) has been very successful at eliminating false-positives56 However GCMS requires extraction repeated calibration and long analysis times (typically 20 to 60 minutes)6 making it labor intensive and less than desirable for field use More rapid analysis of agents in the solid liquid and gas phase has been To whom correspondence should be addressed emailfarqureal-time-analyzerscom
Vibrational Spectroscopy-based Sensor Systems Steven D Christesen Arthur J Sedlacek III EditorsProceedings of SPIE Vol 4557 (2002) copy 2002 SPIE middot0277-786X02$1500
167
Table 1 Chemical Agent Structures Hydrolysis Half-lives and JSAWM Thresholds Agent Short-Hand Chemical Structure Hydrolysis
Half-Life JSAWM
Thresholds
Sarin (GB) F-[O=P-CH3]-O-CH(CH3)2 213 hours 32 microgL Soman (GD) F-[O=P-CH3]-O-CH(CH3)-(C-(CH3)3) 23 hours 32 microgL Tabun (GA) (CH3)2-N-[O=P-CN]-O-C2H5 41 hours 32 microgL VX C2H5O-[O=P-CH3]-S-(CH2)2-N-(CH(CH2)2)2 821 hours 32 microgL EA2192 HO-[O=P-CH3]-S-(CH2)2-N-(CH(CH2)2)2 gt9 years 32 microgL Mustard (H) ClCH2CH2-S-CH2CH2Cl encapsulates 47 microgL Lewisite (L) ClCH=CH-As-Cl2 rapid 27 microgL HCN HCN rapid 20 mgL BZ C7NH12-O-[C=O]-COH(C6H5)2 23 microgL T-2 Toxin 87 microgL at pH 7 to 75 and 20 to 25 oC
demonstrated by vibrational spectroscopy7-10 Hoffland et al7 reported infrared absorbance spectra and absolute Raman cross sections for several chemical agents while Christesen measured Raman cross sections for sarin tabun mustard gas and VX (ethyl S-2-diisopropylamino ethyl methylphosphonothioate)11 Again however these techniques also have limitations Raman spectroscopy is simply not a very sensitive technique and detection limits are typically 01 (1000 ppm) While infrared spectroscopy would have limited value in analyzing poisoned water since the very strong infrared absorption of water would obscure most other chemicals present Nevertheless efforts to overcome these limitations have been demonstrated Braue and Pannella8 quantified the G-series nerve agents (tabun sarin and soman) in terms of infrared attenuated total reflectance using a circle-cell And Alak and Vo-Dinh demonstrated the possibility of surface-enhanced Raman spectroscopy (SERS) to identify CWAs by measuring several organophosphonates that simulate the nerve agents12 However quantitative measurements have not been demonstrated for the SER-active material used (silver coated on alumina particles) or other SER-active media13 Recently we developed silver-doped sol-gels to promote the SER effect14-17 The porous silica network of the sol-gel matrix offers a unique environment for stabilizing SER-active metal particles and the sol-gel provides a high surface area that effectively increases the number of molecules observed within the Raman scattering volume The silver-doped sol-gels have been coated on the internal walls of standard glass vials to allow rapid SER analysis of small sample quantities (lt 01 mL) without preparation We have used p-aminobenzoic acid (PABA) as a test chemical to demonstrate surface enhancements greater than 106 reversible measurements in a flowing system reproducible measurements from vial-to-vial and batch-to-batch and measurements in multiple solvents including water14-17 Recently we used these vials to measure Tabun (GB) and Sarin and several hydrolysis products pinacolyl methyl phosphonate (PMP from Soman) and methyl phosphonic acid (MPA from all G-agents Figure 1) Although a number of unique vibrational bands are observed (eg C-N stretch doublet and P-C stretch) the G-agents were only observed for 5 concentrations and all spectra required baseline corrections Figure 1 Surface enhanced Raman spectra of ~5 vv A) Tabun and B) Sarin C) 1 vv PMP and D) 10 ppm MPA using sol-gel sample vials 785 nm excitation 1-min scan and CCD detection Performed at Aberdeen Proving Ground
C A
Wavenumber (∆cm-1) Wavenumber (∆cm-1)
D B
790
545 C-N
2135 2190 P-C 770
760 1290
Proc SPIE Vol 4577
168
Nevertheless MPA was readily observed for a 10 ppm sample with an estimated detection limit of 04 ppm (based on a signal-to-noise ratio of 3 for the 760 cm-1 band intensity) This measurement provides encouragement in that SERS may satisfy the needs of a JSAWM Furthermore MPA is also a hydrolysis product of VX and V-gas and EA2192 (Figure 2) and may prove a valuable indicator of agent usage Figure 2 Hydrolysis of Sarin to form hydrofluoric acid (HF) methylphosphonic acid 1-methylethyl ester (MPAMME) methyl phosphonic acid (MPA) and 2-propanol With this initial albeit modest success we began analyzing chemicals with various sol-gel compositions that we have been developing Here we describe four sol-gel compositions that select for 1) polar-positive 2) polar-negative 3) weakly polar-positive and 4) weakly polar-negative chemical species The ability of these sol-gels to select and enhance Raman scattering is described for several test chemicals and MPA
2 EXPERIMENTAL The chemicals analyzed as well as all chemicals used to prepare the metal-doped sol-gels were obtained at their purest commercially available grade from Aldrich (Milwaukee WI) The sol-gel designed to select for polar-negative species was prepared from a silver amine complex tetramethyl orthosilicate (TMOS) and methanol After mixing 02 mL of the sol-gel solution was transferred into a glass vial (2 mL) dried and heated The incorporated silver ions were then reduced using dilute sodium borohydride The vials were washed and dried prior to the addition of a sample solution In a similar manner the sol-gel designed to select for polar-positive species was prepared from a gold salt TMOS and methanol The sol-gel designed to select for weakly polar-negative species was prepared from a silver amine complex tetraethyl orthosilicate (TEOS) and methanol And the last sol-gel designed to select for weakly polar-positive species was prepared from a gold salt TEOS and methanol All samples were prepared in a chemical hood and transferred into plain or SER-active vials for analysis Normal Raman spectral measurements employed 1-mL pure samples that were placed in a 1-cm3 cuvette and weighed This yielded a powder density that allowed accurate calculation of molecules in the optical collection field SERS measurements employed 1-mg sample per mL water concentrations unless otherwise stated Once prepared a 01 mL sample was placed into one of the four selective sample vials which in turn was placed into the sample compartment of a Raman spectrometer for analysis A prototype Fourier transform Raman spectrometer (RTA) was used for these measurements18 The system consisted of a NdYAG laser (Brimrose) for excitation at 1064 nm an interferometer built by On-Line Technologies (OLT East Hartford CT) for frequency separation an uncooled InGaAs detector for signal detection (RTA) and an Intel 400 MHz Pentium II based laptop computer (Dell Round Rock TX) for interferometric control data acquisition (OLT) and analysis (LabVIEW by National Instruments Austin TX) Additional components included a Notch filter (Kaiser Ann Arbor MI) and interferometer entrance and exit optics (Edmund Scientific Barrington NJ) Fiber optics were used to deliver the excitation beam to the sample and the scattered radiation to the interferometer (1 meter lengths of 200 and 365 micron core diameter respectively Spectran Avon CT) A second Notch filter (Kaiser) was used as a beam splitter to direct the excitation beam along the same axis as the collected radiation A microscope object (20x04 Newport Irvine CA) was used to focus the beam into the sample and to collect the scattered radiation back along the same axis In this co-axial backscattering arrangement the excitation beam was passed through the outside of a glass vial and focused onto the silver-doped sol-gel film (01-03 mm thickness) containing the sample
3 RESULTS AND DISCUSSION p-aminobenzoic acid (PABA) and phenyl acetylene (PA) and were used to refine the selectivity and SER-activity of the four different metal-doped sol-gels PABA is a popular chemical used to evaluate the performance of SER-active media Here the polar end groups can be used to test selectivity of the polar-negative and polar-positive sol-gels PA is essentially non-
2O+ H HF + +OH
OH3H C
OP
OF
CH
CH3
3
3H C
OCP
OOH
CH
CH3
3
3H C
O
CPHO
CH
CH
3
3
C
Sarin MPAMME MPA 2-propanol
Proc SPIE Vol 4577
169
polar but a high electron density in the cylindrical π cloud around the carbon-carbon triple bond allows testing the selectivity of the weakly polar-negative and weakly polar-positive sol-gels As Figure 3 illustrates PABA passes through the polar sol-gel and is enhanced by either the silver or gold particles At 1 mgml the concentration of neutral PABA is ca 20 times that of the ionized form (pKa = 48) For electropositive silver the PABA anion is expected to interact through the carboxylate group and the associated vibrational modes are expected to dominate the spectrum Conversely for electronegative gold either form of PABA is expected to interact through the amine group The clear differences in our spectra support this expectation Furthermore bands at 840 and 1405 cm-1 assigned to a COO- bend and stretch respectively are significantly more intense for silver than gold Additional bands at 1140 and 1195 cm-1 are assigned to CH bending modes while bands at 1450 1500 and 1605 cm-1 are assigned to ring vibrational modes A very similar SER spectrum for PABA on a silver-coated alumina substrate has previously been reported with similar assignments19 For the gold-doped sol-gel new bands appear at 690 1355 and 1585 cm-1 The first band is assigned to a ring-H bending mode the second band to a ring-N- stretching mode and the third band to a possible NH2 scissors mode or ring mode The second band is not observed in the normal Raman spectrum but infrared bands occur at this frequency for aromatic ring-secondary amine stretching modes The scissors mode occurs at this frequency in Raman spectra for several chemicals but is absent in the PABA Raman spectrum Alternatively this mode may be the1600 cm-1 ring mode that has been shifted by the gold interaction Again a very similar SER spectrum of PABA has been reported but surprisingly using silver (colloids)2021 not gold as the enhancement medium These researchers also assumed the primary interaction of PABA with silver was through the carboxylate anion and made assignments accordingly For example they assigned the 1359 cm-1 to a COO- stretch not to the amine group as we have They also favor the ring stretching mode assignment for the 1582 cm-1 band Finally it should be said that other researchers have argued that the most dominant band in the SER spectra at 1450 cm-1 a ring vibration mode suggests that PABA lies flat on the surface and the π-orbitals dominate the surface interaction22
Figure 3 SER spectra of A) PABA using polar-negative and B) polar-positive sol-gels and C) PA using weakly polar-negative and D) weakly polar-positive sol-gels PABA is 1 mgmL PA is 1 vv Spectral conditions 75 mw 1064 nm 100 scans (15 min) 8 cm-1 resolution Non-polar PA passes through the non-polar sol-gels and is also enhanced by both metals The spectra are easily understood For electropositive silver PA interacts through the cylindrical triple bond π electron cloud and a -CequivC- doublet occurs near 2000 cm-1 The interaction is reasonably strong since this band appears at 2112 cm-1 in the normal Raman spectrum For electronegative gold this interaction is unlikely and only very weak bands occur near 2000 cm-1 The remaining bands are at 1000 cm-1 1200 cm-1 doublet and 1595 cm-1 all appear in the normal Raman spectra at virtually the same frequencies and are assigned to the symmetric ring-breathing mode CH bending modes and the trigonal ring-breathing mode respectively The polarnon-polar selectivity of the polar-negative and weakly polar-negative sol-gels was tested by adding a 11 molar mixture of PABA and PA The selective enhancement is quite good (Figure 4) The spectrum obtained using the polar sol-gel represents 78 PABA and 22 PA while the spectrum obtained using the weakly polar sol-gel represents 9 PABA and 91 PA The band peak intensities at 2000 cm-1 for PA and 1450 cm-1 for PABA were used for these calculations and are expanded in Figure 4 for clarity
C A
Wavenumber (∆cm-1) Wavenumber (∆cm-1)
2NH COOH
D B
C CH
Proc SPIE Vol 4577
170
Figure 4 SERS of 11 MM of PABA and PA in A) polar-negative and C) weakly polar-negative sol-gels The lower traces compare the pure chemicals B) 1 mgml PABA in polar-negative sol-gel and D) 1 PA in weakly polar-negative sol-gel while the insets magnify the minority species for clarity (x5 in A and x10 in B) Spectral conditions as in Figure 3 Following this development of selective sol-gels that maintained SER activity we measured cyanide and MPA (Figure 5) Not surprisingly the best sensitivity for both hydrolysis products was obtained using the polar-negative sol-gel The interaction of the cyanide anion with the silver surface is sufficient to shift the CequivN stretch observed at 2080 cm-1 in the normal Raman spectrum to 2145 cm-1 in the surface-enhanced Raman spectrum Furthermore the band is substantially broadened This anion has been extensively studied by electrolytic SERS and this shift and broadening have been attributed to the formation of a tetrahedral Ag(CN)3
2- surface structure23 Figure 5 Surface-enhanced (upper traces) and normal Raman spectra (bottom traces) of A) CN- and B) MPA in silver-doped TMOS SERS conditions as in Figure 3 and 1 mgmL Note MPA yields two distinct spectra for neutral (top) and acidic pH (middle) The normal Raman spectra employed pure powders 500 scans and 900 mW of 1064 nm SER measurements of MPA with the polar-negative sol-gel yielded two unique spectral signatures that depended on solution pH (Figure 5) For more neutral solutions the P-C stretch of MPA at 762 cm-1 dominates and the CH2 stretch at 2922 cm-1 appears The SN is sufficiently high that the anti-Stokes Raman shift at -762 cm-1 is observed For deprotonated MPA an oxygen-surface mode appears at 325 cm-1 (as well as its anti-Stokes complement) suggesting a strong interaction This results in substantial enhancement of the P-O-C mode at 1051 cm-1 (upper trace) Others report that this mode dominates the infrared spectra of nerve agents measured in water8 Comparison of the two spectra suggests the following molecule-to-surface orientations The appearance of the oxygen-surface and P-O-C modes in the upper spectrum of Figure 5B indicates that the tetrahedral molecule interacts with the silver surface through the deprotonated oxygen and is oriented end-on The
A B
Wavenumber (∆cm-1)
C A
Wavenumber (∆cm-1) Wavenumber (∆cm-1)
D B
Wavenumber (∆cm-1)
Proc SPIE Vol 4577
171
dominance of the P-C and the CH2 stretches and the disappearance of the P-O-C mode in the upper spectrum suggest the molecule is oriented with the methyl-hydroxyl-hydroxyl face towards the surface However considerably more research must be performed to verify these points Both chemical forms were measured as a function of concentration to determine the onset of monolayer coverage and estimate expected detection limits (Figure 6) Below monolayer coverage the signal to concentration dependence should be linear and the SN of any spectral measurement in this range can be used to predict the detection limit In the spectra presented here the peak height was used as the signal while the noise as root-mean-squared (RMS) was measured between 4400-4600 cm-1 Since noise is distributed evenly throughout the spectrum when transformed this region was used since it does not have any contributions from signals or baseline offsets Figure 6 shows a series of spectra for MPA along with a plot of the 1050 cm-1 band intensity (the noise was identical for this series) as a function of sample concentration A clear discontinuity appears in the vicinity of 01 mgmL (19 ppm) indicating the onset of monolayer coverage A detection limit defined as a SN of 3 was calculated for the 01 and 005gmL samples at 24x10-4 and 25x10-4 gL respectively A more modest detection limit of 101x10-4 gL was obtained using the 760 cm-1 band in the second series of concentration measurements These detection limits correspond to 46 and 190 ppb for a 3-min scan and 75 mW of incident laser power
Figure 6 A) Concentration dependence of MPA SERS measured in silver-doped TMOS) B) Concentrations are 001 005 01 05 1 gL (188 94 188 94 188 ppm) I760 series (bull) and I1050 series (∆)
Enhancement factors for cyanide and methyl phosphonic acid can be determined by estimating the number of molecules contributing to the surface-enhanced and normal Raman spectra The enhancement factor EF can be defined by the following equation
EF = (ISERSINR)bull(MNRMSERS) bull(PNRPSERS) bull(TNRTSERS)12
where I is the spectral band intensity M is the sample mass P is the incident laser power and T is the measurement time (or number of scans) for the two measurements For the normal Raman spectra a cylindrical scattering volume is assumed based on the laser area (28x10-7m2 6x10-4m diameter spot) and the penetration depth (1x10-3 m)24 The density of KCN and MPA as powders were measured at 0572 and 0516 gcm3 indicating that 16x10-4 and 144x10-4 g produced the normal Raman signals in Figure 5 respectively The number of molecules contributing to the SER spectra are those on the silver particles embedded in the sol-gel The total silver surface area can be determined from the average particle size concentration and the scattering volume Previous scanning electron micrographs showed the average silver particle size to be 40 nm in diameter (335x10-23m3)17 The silver concentration is 012M based on the reactant molar concentrations and dilution factors And the scattering volume is 76x10-
11m3 again based on a cylindrical scattering volume defined by a laser area of 28x10-7m2 and a sol-gel thickness of 27x10-
4m This volume contains 123x10-6g of silver equivalent to 35x109 silver particles with a collective surface area of 18x10-
5m2 However it may reasonably be assumed that at least half of the silver surface is in contact with the sol-gel matrix and unavailable for analyte interaction If we assume monolayer coverage and that each CN molecule occupies 15x10-20m2 then approximately 62x1014 molecules or 27x10-8g of CN contribute to the SER spectrum (20x10-19m2 46x1013 molecules 74x10-9g for MPA) Accordingly the EF for cyanide equals 48x104 ((180599) bull(16x10-427x10-8) bull(90075) bull(500100)12) The EF for MPA is considerably higher at 87x106 ((60326) bull(144x10-474x10-9) bull(90075) bull(500200)12)
0
20
40
60
80
100
120
140
0 02 04 06 08 1 12[MPA] (mgmL)
I (76
0)
0
100
200
300
400
500
600
I (10
50)
Wavenumber (∆cm-1)
A B
Proc SPIE Vol 4577
172
4 CONCLUSIONS Here we present for the first time surface-enhanced Raman spectra of Tabun and Sarin obtained using silver-doped sol-gels However the inferior enhancement suggested employing SER-active sol-gels varying in composition to improve both detection limits and selectivity of the target analytes To this end we successfully demonstrated the capabilities of four sol-gels that select for 1) polar-positive 2) polar-negative 3) weakly polar-positive and 4) weakly polar-negative chemical species p-aminobenzoic acid was used to show that silver could be used to attract polar-negative chemicals or functional groups (carboxylate anion) and that gold could be used to attract polar-positive chemicals (amine) while a mixture of p-aminobenzoic acid and phenylacetylene was used to show that tetramethyl orthosilicate preferentially solvates polar chemicals and that tetraethyl orthosilicate preferentially solvates weakly polar chemicals This increased sample control was applied to cyanide and methyl phosphonic acid two hydrolysis products of chemical warfare agents Exceptional results were obtained for methyl phosphonic acid allowing measurement of 1x10-2 gL for a 3-min scan and 75 mW of incident laser power with an estimated detection limit of 245x10-4 gL and an enhancement factor of 87x106 However this detection limit is 76 times less sensitive than required for the JSAWM (32x10-6gL for the G-agents) Finally we note that the measurements performed here employed an FT-Raman spectrometer This instrumentation was chosen over dispersive Raman instrumentation for the high wavelength accuracy afforded by the HeNe reference laser (Connes Advantage25) which would allow reliable spectral subtraction matching of observed spectra to stored library spectra and confident use of chemometric approaches Such data analysis is likely to be required to identify the chemical agents as well as distinguish hydrolysis products However this instrumentation which employs 1064 nm excitation and InGaAs detection sacrifices sensitivity We believe that the measurement sensitivity can be improved by at least two orders of magnitude by using shorter laser excitation wavelengths (eg 532 nm) This would provide more efficient Raman scattering (fourth power dependence on laser excitation wavelength) more efficient generation of plasmon modes and allow using more efficient detector material (Si vs InGaAs) These modifications are underway
5 ACKNOWLEDGEMENTS The authors would like to thank Dr R Yin and J Jensen of the US Army for supporting this work (Contract Number DAAD13-01-C-0019) They would also like to thank Advanced Fuel Research for making their laboratory facilities available
6 REFERENCES 1 Tu Anthony ldquoOverview of Sarin Terrorist Incidents in Japan in 1994 and 1995rdquo 6th CBW Protection Symposium
Stockholm Sweden 10-15 May 1998 2 Jensen JL NC Wong W R Loerop SPIE 4775-03 (2001) 3 JSAWM Requirements at wwwsbccomapgeaarmymilRDAecbcrtPRODSERJSAWMjsawmhtml 4 ldquoThe Chemical Weapons Convention Redefines Analytical Challengerdquo Analytical Chemistry News amp Features June 1
397A (1998) 5 Johnston RL Hoefler CM Fargo JC and Moberley B ldquoThe Defense Nuclear Agencyrsquos ChemicalBiochemical
Weapons Agreements Verification Technology Research Development Test and Evaluation Program and its Requirements for On-Site Analysisrdquo AT-ONSITE 5-8 (1994)
6 Black RM Clarke RJ Read RW and Reid MT ldquoApplication of gas chromatography-mass spectrometry and gas chromatography-tandem mass spectrometry to the analysis of chemical warfare samples found to contain residues of the nerve agent sarin sulphur mustard and their degradation productsrdquo J Chromatography 662 301-321 (1994)
7 Hoffland LD Piffath RJ Bouck JBrdquoSpectral signatures of chemical agents and simulantsrdquo Optical Engineering 24 982-984 (1985)
8 Braue EHJ Pannella MGrdquoCIRCLE CELL FT-IR Analysis of Chemical Warfare Agents in Aqueous Solutionsrdquo Applied Spectroscopy 44 1513-1520 (1990)
9 Tseng C-H Mann CK and Vickers TJ ldquoDetermination of Organics on Metal Surfaces by Raman Spectroscopyrdquo Applied Spectroscopy 47 1767-1771 (1993)
10 Staff Reporter ldquoUS Army Tests Lidar to Detect Biological Toxinsrdquo Photonic Spectra pg 50 December 1998
Proc SPIE Vol 4577
Proc SPIE Vol 4577
173
11 Christesen SD Raman cross sections of chemical agents and simulants Applied Spectroscopy 42 318-321 (1988) 12 Alak AM and Vo-Dinh T ldquoSERS of Organophosphorous Chemical Agentsrdquo Analytical Chemistry 59 2149-2153
(1987) 13 Norrod KL Sudnik LM Rousell D and Rowlen KL ldquoQuantitative Comparison of Five SERS Substrates
Sensitivity and Detection Limitrdquo Applied Spectroscopy 51 994-1001 (1997) 14 Lee Y and Farquharson S ldquoSERS Sample Vials Based on Sol-Gel Process for Trace Pesticide Analysisrdquo SPIE 4206
140-146 (2000) 15 Farquharson S and Lee Y ldquoTrace Drug Analysis by Surface-Enhanced Raman Spectroscopyrdquo SPIE 4200-16 (2000) 16 Lee Y Farquharson S and Rainey P M Surface-Enhanced Raman Sensor for Trace Chemical Detection in Water
SPIE 3857 76-84 (1999) 17 Lee Y Farquharson S Kwong H and Shahriari M ldquoSurface-Enhanced Raman Sensor for Surface-Enhanced Raman
Spectroscopyrdquo SPIE 3537 252-260 (1998) 18 Farquharson S Smith W Carangelo R C and Brouillette C ldquoIndustrial Raman Providing Easy Immediate Cost
Effective Chemical Analysis Anywhererdquo SPIE 3859 14-23 (1999) 19 Narayanan VA JM Bello JD Stokes and T Vo-Dinh Analusis 19 307-310 (1991) 20 Laserna JJ E L Torres and JD Winefordner Analytica Chemica Acta 469-480 (1987) 21 Torres EL and JD Winefordner Analytical Chemistry 59 1626-1632 (1987) 22 Suh JS DP DiLella M Moskovits J Phys Chem 87 1540-1544 (1983) 23 Benner RE R Dornhaus R Chang and BL Laube Correlations in the Raman spectra of cyanide complexes adsorbed
at silver electrodes with voltammograms Surface Science 101 341 (1980) 24 Chase D B and JF Rabolt Fourier Transform Raman Spectroscopy Acad Press Ch1 p 131 (1994) 25 Connes J Rev Opt Theor Instrum 40 45 (1961)
APPLIED SPECTROSCOPY 351
focusing the 488 nm laser beam 2 mm deep into thebulk of the crystals to avoid contributions from potentialdepletion layers As shown in Fig 1b the dependence ofthe Raman shift on the degree of deuteration is almostperfectly linear and ts very well with D 5 22684cmR1 24526 where D is the degree of deuteration (in )and R is the spectral mean of the PO4 vibration in cm21A linear correlation coef cient of 0998 indicates an ex-cellent linear dependence of the Raman peak shift withdegree of deuteration This result shows that the shift ofthe PO 4 peak is simply caused by the linear increase inatomic mass due to isotope substitution which decreasesthe length of hydrogen-like bonds
This excellent linear dependence allows us to map thepro le of the DH exchange layer at the surface of DKDPcrystals by acquiring Raman spectra and determining theposition of the PO 4 peak for various depths This methodis preferable over other methods such as determining thestrength of the OD vibration directly (eg at 715 cm21)because the position of the most intense peak in the Ra-man spectrum can be measured more precisely than theintensity of some of the weakest peaks in the spectrumThis is demonstrated in Fig 2 where depth-dependentRaman spectra (Fig 2a) and the resulting exchange layerpro les for two DKDP crystals are shown (Fig 2b) Thespectra in Fig 2a were obtained from a depth scan of aDKDP crystal with 75 degree of deuteration in thebulk grown at 45 8C The spectra start out as DKDP with30 deuteration close to the surface and approach thebulk DKDP spectrum within a few micrometers of depthThe fact that the relative degree of deuteration does notextend to 0 D is due to the limited depth resolution ofthe Raman microprobe which averages over 4 mm indepth Figure 2b depicts the resulting DH exchange layerpro les for this and a second crystal grown at 63 8Crespectively Both crystals had the same exposure to am-bient conditions and their main difference is the temper-ature at which they were grown The different exchangelayer pro les indicate that crystals grown at differenttemperatures have differen t proton conductiv ities 13
which leads to a difference in their rate of deuteriumdepletion The parameters controlling this behavior arecurrently the objective of a detailed study the results ofwhich will be reported elsewhere
CONCLUSION
In conclusion we have shown that the shift of the to-tally symmetric PO 4 stretch mode in the Raman spectrumof DKDP crystals scales linearly with degree of deuter-ation This allows us to correlate Raman peak positionsto deuteration levels in these crystals We have presenteda new technique to determine DH diffusion pro les inDKDP frequency conversion crystals based on micro-Ra-man spectroscopy This technique is fast inexpensiveand works under various environmental conditionswhich will allow us to better understand and control deu-terium depletion in DKDP crystals
ACKNOWLEDGMENTS
We would like to thank M Runkel for rst discovering DKDP crack-ing R Floyd for providing DKDP crystals and L Chase and A Burn-ham for their support and helpful discussions This work was performedunder the auspices of the US Department of Energy by the University
of California Lawrence Livermore National Laboratory under contract W-7405-Eng-48
1 J J De Yoreo A K Burnham and P K Whitman Int Mat Rev47 113 (2002)
2 C E Barker R A Sacks B M Van Wonterghern J A Caird JR Murray J H Campbell K Kyle R B Ehrlich and N DNielsen Proc SPIE-Int Soc Opt Eng 2633 501 (1995)
3 T Suratwala paper to be published4 Cleveland Crystals Inc httpwwwclevelandcrystalscom
KDPshtmltable5 E A Popova I T Savatinova and I A Velichko Sov Phys Solid
State 12 1543 (1971)6 I P Kaminow R C C Leite and S P S Porto J Phys Chem
Solids 26 2085 (1965)7 J A Subramony B J Marquardt J W Macklin and B Kahr
Chem Mat 11 1312 (1999)8 H Tanaka M Tokunaga and I Tatsuzaki Solid State Commun
49 153 (1984)9 R J Nelmes G M Meyer and J E Tibballs J Phys C 15 59
(1982)10 M A Yakshin D W Kim Y S Kim Y Y Broslavets O E
Sidoryuk and S Goldstein Laser Physics 7 941 (1997)11 I Takenaga Y Tominaga S Endo and M Kobayashi Solid State
Commun 84 931 (1992)12 C Krenn personal communication13 M Sharon and A K Kalia J Solid State Chem 21 171 (1977)
Rapid Dipicolinic Acid Extractionfrom Bacillus Spores Detectedby Surface-EnhancedRaman Spectroscopy
STUART FARQUHARSON ALAND GIFT PAUL MAKSYMIUK andFRANK E INSCOREReal-Time Analyzers Inc East Hartford Connecticut06108
Index Headings Dipicolinic acid Bacillus spores Anthrax Surface-enhanced Raman spectroscopy
INTRODUCTION
The anxiety caused by the distribution of anthrax en-dospores through the US postal system in October 2001was exacerbated by the long time required for positiveidenti cation of the Bacillus anthracis spores and the un-known extent of their distribution Since that time manymethods capable of rapid eld analysis have been inves-tigated to augment or replace the laboratory method ofgrowing microorganisms in culture media which takesdays to perform12 Prominent among these approachesare polymerase chain reactions (PCR)3 immunoassaysand detection of calcium dipicolinate as a biochemicalsignature PCR employs primers to separate organism-
Received 10 October 2003 accepted 14 November 2003 Author to whom correspondence should be sent
352 Volume 58 Number 3 2004
speci c nucleic acid sequences (eg capsular protein en-coding gene for Bacillus anthracis)4 and polymerases toamplify the segment until it is detectable Recently am-pli cation times have been substantially reduced andcomplete analysis can now be performed in an hour orless Immunoassay methods are also being developed thatuse competitive binding of the bioagent (as an antigen)and its labeled conjugate for a limited number of anti-bodies Although analyses can be performed in under 30minutes a well-de ned anthrax antigen has not yet beenidenti ed5ndash7 and consequently the false-positive rate isunacceptably high8
A number of other methods are being developed witha focus on the detection of calcium dipicolinate (CaDPA)and its derivatives as a B anthracis signature This is sobecause only spore-forming bacteria contain CaDPA andthe most common potentially interfering spores such aspollen and mold spores do not Relatively fast methodshave been developed to chemically extract CaDPA andthen detect it directly by uorescence9 or indirectly byluminescence1011 In the latter case hot dodecylamine(DDA) has been used to extract dipicolinic acid (DPA)and terbium has been utilized to form a highly lumines-cent DPA complex11 Although measurements have beenperformed in as little as ve minutes it was found thatas many as three concentration-dependent complexes canform each with different lifetimes This coupled withthe fact that the Tb31 cation produces the same lumines-cence spectrum makes determinations of low spore con-centrations problematic
It has been long known that Raman spectra of Bacillispores are dominated by bands associated with CaDPA12
and that these spectra may provide a suitable anthrax sig-nature at the genus level13 Since that time considerableimprovements in Raman instrumentation have led to lab-oratory measurements of single Bacilli spores14 and to eld measurements of spores captured from a mail-sort-ing system15 However the single spore measurementsrequired complex instrumentation that is not ruggedwhile the eld measurements required milligrams of sam-ple Furthermore the Raman spectra of both measure-ments contained uorescence contributions that would in-crease uncertainty in quanti cation
In related research we demonstrated that nanogramquantities of DPA could be detected by uorescence-freesurface-enhanced Raman spectroscopy (SERS)16 We alsodemonstrated that microliter volumes of chemicals canbe detected by SERS using metal-doped sol-gel-packedglass capillaries17 Towards the goal of developing a rap-id eld SERS-based anthrax spore detector we havecombined our previous research and we now report thatDPA can be extracted from a 10 mg B cereus spore sam-ple using DDA in 1 minute and can be detected by SERSin an additional 1 minute
EXPERIMENTAL
Dipicolinic acid (26-pyridinedicarboxylic acid DPA)and dodecylamine (DDA) were used as received fromSigma-Aldrich (Milwaukee WI) Lyophilized B cereusspores prepared according to the literature13 were sup-plied by the University of Rhode Island and used as re-ceived Multiple particles approximately 01 mm3 each
were separated and weighed at 5 to 15 mg representing05 to 15 million spores The sample masses were con-sistent with a previous determination of spore density at0081 gmL that indicated a high degree of entrained air
All chemicals used to prepare the silver-doped sol-gelcoated capillaries were also obtained and used as receivedfrom Sigma-Aldrich According to previously publishedprocedures17 two precursor solutions were preparedmixed and then drawn into 1-mm-diameter glass capil-laries The silver amine precursor consisted of a 51 vvratio of 1 N AgNO 3 to 28 NH3OH while the alkoxideprecursor consisted of a 21 vv ratio of methanol to te-tramethyl orthosilicate The alkoxide precursors weremixed with silver amine precursor in an 81 vv ratioApproximately 015 mL was drawn into the capillarycoating a 15-mm length After sol-gel formation the in-corporated silver ions were reduced with dilute sodiumborohydride which was followed by a water wash to re-move residual reducing agent
A 100 mL drop of a 50 mM DDA solution in ethanolpre-heated to 78 8C was added to each of the B cereusparticles to digest the spore coat After 1 minute the re-sultant solution was drawn into a SER-active capillarythat was immediately xed horizontally to an XY posi-tioning stage (Conix Research Spring eld OR) just in-side the focal point of an f 07 aspheric lens The lensfocused the beam into the sample and collected the scat-tered radiation back along the same axis A dichroic lter(Omega Optical Brattleborough VT) was used to re ectthe excitation laser to the lens and pass the Raman scat-tered radiation collected by the lens An f 2 achromatwas used to collimate the laser beam exiting a 200-mm-core-diameter source ber optic while a second f 2 ach-romat was used to focus the scattered radiation into a 365mm ber optic (Spectran Avon CT) A short-pass lterwas placed in the excitation beam path to block the sil-icon Raman scattering generated in the source ber fromre ecting off sampling optics and reaching the detectorA long-pass lter was placed in the collection beam pathto block the sample Rayleigh scattering from reachingthe detector A 785 nm diode laser (Process InstrumentsInc model 785-600 Salt Lake City UT) was used todeliver 100 to 150 mW of power to the sample A Fouriertransform Raman spectrometer (Real-Time Analyzersmodel IRA-785 East Hartford CT) and a silicon photo-avalanche detector (Perkin Elmer model C30902S Stam-ford CT) were used to acquire the SER spectra
RESULTS AND DISCUSSION
As an initial experiment the SER spectrum of 1 gLof DPA in water was measured using the newly devel-oped silver-doped sol-gel-coated capillaries (Fig 1A) Atthis concentration a high signal-to-noise ratio (SN) isobtained in 1 min In fact a reasonable spectrum is ob-tained in the same time frame for 1 mgL (Fig 1B) TheSER spectra are reasonably similar to the normal Raman(NR) spectrum obtained for a saturated solution of DPAin 1 N KOH (Fig 1C) and the following band shifts areobserved (NR to SER) 647 to 657 cm21 817 to 815cm21 998 to 1008 cm21 1384 to 1382 cm21 1434 to1428 cm21 and 1569 to 1567 cm21 Many of these bandshave been previously assigned1213 such as 998 cm21 to
APPLIED SPECTROSCOPY 353
FIG 1 SERS of DPA in water using silver-doped sol-gel-coated glasscapillary for (A) 1 gL and (B) 1 mgL (C ) NR of saturated DPA in 1N KOH in a glass capillary Spectral conditions (A) and (B) 150 mWof 785 nm 1-min acquisition time (C) 450 mW of 785 nm 5-minacquisition time both 8 cm 21 resolution
FIG 2 SERS of DPA extracted from 10 mg B cereus particle using100 mL of 50 mM hot DDA acquired in (A) 1 minute and (B) 2 seconds(C ) Attempted SERS of 50 mM hot DDA in ethanol using silver-dopedsol-gel-coated glass capillary acquired in 1 min Spectral conditions150 mW of 785 nm 8 cm21 resolution
the symmetric ring stretch 1384 cm21 to the OndashCndashOsymmetric stretch 1428 cm21 to the symmetric ring CndashH bend and 1569 cm21 to the asymmetric OndashCndashOstretch
The rst B cereus samples consisted of 2 mg of sporesin 2 mL of 5 mM hot DDA The samples were main-tained at 78 8C for 40 min and while hot approximately10 mL was drawn into a SER-active capillary Since spec-tra of DPA were obtained for these initial samples small-er spore masses higher DDA concentrations and shorterheating periods were examined In due course it wasfound that 10 mg of spores could be digested by 100 mLof 50 mM hot DDA in one minute and detected (Fig2A) In fact the signal was suf ciently intense that it canbe observed in as little as two seconds (Fig 2B) Theamount of DPA that was extracted was estimated to bebetween 5 and 10 mgL by comparing the signal intensityof the 1008 cm21 band to that measured for DPA in waterThis is consistent with previous research that found thatthe majority of the DPA is extracted from spores usingDDA11 and that B cereus spores contain approximately10 DPA by weight18 The SN of 127 for the 1008 cm21
band in the 1-minute SER spectrum suggests a limit ofdetection of approximately 250 ng of B cereus sporesbased on a SN of 3 Finally it should be noted that DDAdid not produce a detectable SER spectrum as shown inFig 2C
CONCLUSION
We have demonstrated that by combining rapid extrac-tion of dipicolinic acid from Bacillus cereus spores withchemical identi cation by surface-enhanced Raman spec-troscopy as little as 10 mg of spores can be detected Infact the entire measurement from the time of adding hotdodecylamine to the spores to the time when the dipi-colinic acid SER spectrum is acquired and analyzedcould be performed in less than two minutes The abilityof this method to distinguish between spore-forming bac-teria such as Bacillus anthracis and non-DPA containingpowders could help prevent costly shutdowns associated
with the appearance of suspicious material or intentionalmailing of common substances as an anthrax hoax Thismethod could also prove useful in detecting the locationof anthrax endospores in mail distribution facilities if an-other veri ed attack should occur
Research continues to fully characterize the surface-enhanced Raman spectroscopy signal intensities as afunction of sample concentration and to explore otherextractants that do not require the use of elevated tem-perature
ACKNOWLEDGMENTS
The authors are grateful for the support of the National Science Foun-dation (DMI-0296116 and DMI-0215819) and the US Army(DAAD13-02-C-0015 Joint Service Agent Water Monitor program)The authors are indebted to Chetan Shende for preparing the sol-gelcapillaries The authors also thank James Gillespie Nicholas Fell andAugustus Fountain for providing important background informationMark Farquharson for laboratory support and Professor Jay Sperry ofthe University of Rhode Island for supplying B cereus spores
1 V A Pasechnik C C Shone and P Hambleton Bioseparations 3267 (1993)
2 P J Jackson M E Hugh-Jones D M Adair G Green K K HillC R Kuske L M Grinberg F A Abramova and P Keim ProcNatl Acad Sci USA 95 1224 (1998)
3 B R Glick and J J Pasternak Molecular Biology Principles andApplications of Recombinant DNA (ASM Press Washington DC1994)
4 C A Bell J R Uhl T L Had eld J C David R F Meyer TF Smith and F R Cockerill III J Clin Microbiol 40 2897(2002)
5 D L Gatto-Menking H Yu J G Bruno M T Goode M Millerand A W Zulich Biosens Bioelectron 10 501 (1995)
6 J J Quinlan and P M Foegeding J Rapid Methods AutomationMicrobiol 6 1 (1998)
7 A A Hindle and E A H Hall Analyst (Cambridge UK) 1241599 (1999)
8 M S Ascher US Department of Health amp Human Services(httpwwwhhsgovophppresentationsAscherdoc)
9 R Nudelman B V Bronk and S Efrima Appl Spectrosc 54 445(2000)
10 D L Rosen C Sharpless and L B McBrown Anal Chem 691082 (1997)
354 Volume 58 Number 3 2004
11 P M Pellegrino N F Fell Jr and J B Gillespie Anal ChimActa 455 167 (2002)
12 W H Woodruff T G Spiro and C Gilvarg Biochem BiophysRes Commun 58 197 (1974)
13 E Ghiamati R S Manoharan W H Nelson and J F SperryAppl Spectrosc 46 357 (1992)
14 A P Esposito C E Talley T Huser C W Hollars C M Schal-dach and S M Lane Appl Spectrosc 57 868 (2003)
15 S Farquharson L Grigely V Khitrov W W Smith J F Sperryand G Fenerty J Raman Spectrosc paper accep ted (2003)
16 S Farquharson W W Smith S Elliott and J F Sperry SPIE-IntSoc Opt Eng 3855 110 (1999)
17 S Farquharson and P Maksymiuk Appl Spectrosc 57 479(2003)
18 F W Janssen A J Lund and L E Anderson Science (Washing-ton DC) 127 26 (1958)
SPIE -2003-5269 117
pH dependence of methyl phosphonic acid dipicolinic acid and cyanide by surface-enhanced Raman spectroscopy
Stuart Farquharson Alan Gift Paul Maksymiuk Frank Inscore and Wayne Smith
Real-Time Analyzers 87 Church Street East Hartford CT 06108
ABSTRACT US and Coalition forces fighting terrorism in Afghanistan and Iraq must consider a wide range of attack scenarios in addition to car bombings Among these is the intentional poisoning of water supplies to obstruct military operations To counter such attacks the military is developing portable analyzers that can identify and quantify potential chemical agents in water supplies at microgram per liter concentrations within 10 minutes To aid this effort we have been investigating the value of a surface-enhanced Raman spectroscopy based portable analyzer In particular we have been developing silver-doped sol-gels to generate SER spectra of chemical agents and their hydrolysis products Here we present SER spectra of methyl phosphonic acid and cyanide as a function of pH an important factor affecting quantitation measurements which to our knowledge has not been examined In addition dipicolinic acid a chemical signature associated with anthrax-causing spores is also presented Keywords Chemical warfare agents agent detection agent hydrolysis SERS Raman spectroscopy homeland security
1 INTRODUCTION In the past decade the Unites States and its allies have been challenged by a different kind of warfare exemplified by the terrorist attacks of September 11 2001 Suicide bombings and the use of chemical agents are the norm and military personnel must consider a wide range of attack scenarios Among these is the intentional poisoning of water supplies to obstruct military operations in Afghanistan and Iraq Currently colorimetric paper is used to detect agents on-site while gas chromatography combined with mass spectrometry (GCMS) is used in mobile support laboratories However both methods have severe drawbacks The paper changes color in response to contact with many chemicals besides CWAs causing a high incidence of false positives1 while GCMS although very chemically specific requires hours to perform and constant re-calibration234 Military operations would be greatly aided by a portable analyzer that can identify and quantify potential chemical agents at concentrations that impact safety This includes the analysis of drinking water supplies distribution and storage systems To meet this goal the Department of Defense has been investigating numerous approaches under the auspices of the Joint Service Agent Water Monitor (JSAWM) program5 The JSAWM must be capable of identifying the chemical agents with no false-positives and quantifying the agents at microgram per liter concentrations within 10 minutes678 Recently we and others have been investigating the ability of surface-enhanced Raman spectroscopy (SERS) to measure chemical agents9-12 bioagents13-17 and their hydrolysis products in water SERS employs the interaction of surface plasmon modes of metal particles with target analytes to increase scattering efficiency by as much as 1 million times18 In 1987 the potential of this approach to measure CWAs was demonstrated by measuring a series of pesticides19 Several of these organophosphonates have chemical structures similar to CWAs in particular P=O functional groups In our studies we have been employing silver-doped sol-gels to promote the SER effect The porous silica network of the sol-gel matrix offers a unique environment for immobilizing and stabilizing SER-active metal particles20-23 The sol-gels have been coated on the internal walls of standard glass vials to allow rapid SER analysis of small sample quantities without preparation We have measured over 100 chemicals with enhancements of 104 to 106 demonstrated reversible measurements in a flowing system reproducible measurements from vial-to-vial and measurements in multiple solvents including water20-23 Previously we used these vials to perform preliminary measurements of cyanide (CN) methylphosphonic acid (MPA) and dipicolinic acid (DPA) MPA is a hydrolysis product of the nerve agents (eg sarin Reaction 1) and may be a valuable indicator of nerve agent usage particularly since the alkyl methylphosphonic acids are relatively more stable than their corresponding parent complexes24 DPA is
SPIE -2003-5269 118
a chemical signature of spore forming bacteria such as Bacillus anthracis And in light of the inability to rapidly detect the anthrax spores distributed through the US mail in October 2001 a number of methods are being developed to extract and analyze this signature Reaction 1 Stepwise hydrolysis of Sarin to form hydrofluoric acid (HF) isopropyl methylphosphonic acid (IMPA) then methyl phosphonic acid (MPA) and 2-propanol In our previous SERS investigations MPA and DPA were measured at 50 and 100 mgL respectively In both cases limits of detection (LOD) were estimated at 100 microgL providing encouragement in that SERS may satisfy the needs of the JSAWM Since it has been shown that pH can substantially influence the intensity of SER bands25 which would clearly influence quantitative analysis we undertook the present study to determine the severity of these effects for cyanide methyl phosphonic acid and dipicolinic acid Furthermore we previously observed a band at 1050 cm-1 for MPA6 possibly due to an anion formed at basic pH Here we investigate the source of this spectral anomaly
2 EXPERIMENTAL All chemicals including potassium cyanide methyl phosphonic acid dipicolinic acid and those used to prepare the silver-doped and gold-doped sol-gel coated vials were obtained and used as received from Sigma-Aldrich All samples were prepared in HPLC grade water (Fischer Scientific Fair Lawn NJ) for SERS measurements The pH of these samples was adjusted using dilute nitric acid or potassium hydroxide and verified using a pH electrode (Corning Inc Corning NY) that had been calibrated with pH 400 700 and 1000 buffered standards from Fischer Scientific Once prepared the samples were transferred into the silver-doped sol-gel vials (Real-Time Analyzers Inc East Hartford CT) The vials were coated in a manner similar to that previously reported by adding ammonium hydroxide to a solution of silver nitrate tetramethyl orthosilicate (TMOS) and methanol20 Gold-vials were coated by adding nitric acid to a solution of gold tetrachloride TMOS and methanol The two precursor solutions were prepared mixed and transferred to 2-mL glass vials dried and heated After sol-gel formation the incorporated metal ions were reduced with dilute sodium borohydride (1mgmL) which was followed by a water wash to remove residual reducing agent After the resultant analyte solution was introduced the SER-active vial was immediately fixed horizontally to an XY positioning stage (Conix Research Springfield OR) just inside the focal point of an f07 aspheric lens The lens focused the beam into the sample and collected the scattered radiation back along the same axis A dichroic filter (Omega Optical Brattleborough VT) was used to reflect the excitation laser to the lens and pass the Raman scattered radiation collected by the lens An f2 achromat was used to collimate the laser beam exiting a 200 microm core diameter source fiber optic while a second f2 achromat was used to focus the scattered radiation into a 365 microm fiber optic (Spectran Avon CT) A short pass filter was placed in the excitation beam path to block the silicon Raman scattering generated in the source fiber from reflecting off sampling optics and reaching the detector A long pass filter was placed in the collection beam path to block the sample Rayleigh scattering from reaching the detector A 785 nm diode laser (Process Instruments Inc model 785-600 Salt Lake City UT) was used to deliver 100 to 150 mW of power to the sample A Fourier transform Raman spectrometer (Real-Time Analyzers model IRA-785 East Hartford CT) and a silicon photo-avalanche detector (Perkin Elmer model C30902S Stamford CT) were used to acquire the SER spectra
3 RESULTS AND DISCUSSION In a previous study of MPA6 aimed at developing a concentration calibration curve and determining limits of detection (LOD) we observed an anomaly at 1050 cm-1 Since it was found that the band intensity changed as a function of concentration the band must be associated with a sample parameter Two possible parameters photon flux and pH are examined here The first parameter was investigated by irradiating a 1mgmL MPA sample in a SER-active vial with laser powers of 200 mW and above and monitoring spectral changes It was immediately found that the 1050 cm-1 band
2O+ H HF + +OH
OH3H C
OP
OF
CH
CH3
3
3H C
OCP
OOH
CH
CH3
3
3H C
O
CPHO
CH
CH
3
3
C
Sarin IMPA MPA 2-propanol
SPIE -2003-5269 119
appeared and grew as a function of time and that the higher the laser power the faster the growth Figure 1 shows the growth of the 1050 cm-1 band over the course of 30 minutes when using 150 mW of 785 nm excitation while Figure 2A shows that the growth can be fit with a first order exponential equation namely I1050 = 03+05e-013t Furthermore the 760 cm-1 band could be fit with a first order decay equation with an identical exponential rate constant ie I760 = 08-
08e-013t The rates represent classical first order kinetics and their correspondence allows one to conclude that MPA is being transformed one-for-one into a photo-generated product At this time the photoproduct has not been positively identified but phosphonic (phosphorous) acid and phosphonate are likely candidates since the symmetric P(OH)3 stretch occurs at ~1050cm-1 Our studies show that a reduction in laser power to 100 mW at the sample for MPA essentially eliminates this degradation process This laser power or lower was used for further measurements
Methyl phosphonic acid is a diprotic acid that stepwise dissociates into two anions MPA- and MPA= according to the following reactions26
MPA MPA- + H+ pKa1 = 212 Reaction 2
MPA- MPA= + H+ pKa2 = 729 Reaction 3 The relative concentrations of MPA MPA- and MPA= can be determined at any pH by expressing [MPA] and [MPA=] in terms of [MPA-] using Reactions 2 and 3 and summing all three to equal the total starting concentration here 2 mgmL (0021M MW = 9602) viz
[MPA] + [MPA-] + [MPA=] = 0021M Equation 1 substituting from Reactions 1 and 2
([H+][MPA-])K1a + [MPA-] + (K2a[MPA-])[H+] = 0021M Equation 2 rearranging [MPA-] = 0021M(1+[H+]K1a + K2a[H+]) Equation 3 The relative concentrations of MPA MPA- and MPA= as a function of pH are shown in Figure 3 It is worth noting that near neutral pH both MPA- and MPA= will be present To confirm that the SER signal followed this pH dependence a starting solution consisting of 20 mg of MPA in 10 mL HPLC grade water was prepared and brought to pH of 20 using dilute nitric acid From this solution 2 mL were added to a SER-active vial and the SER spectrum recorded At this pH a peak at 760 cm-1 was barely discernable The 2 mL solution was returned to the starting solution and the pH was re-measured to correct for any changes that the silver-doped sol-gel vials might cause In most cases the change was less than 02 pH units and the pH is reported as the before and after average Next the pH of the
Figure 2 A) Exponential growth of 1050 cm-1 band and B) exponential decay of 760 cm-1 band for spectral series in Figure 1
Figure 1 Growth of 1050 cm-1 band as a function of time due to exposure to 150 mW of 785 nm Spectra are 5 sec each collected every 100-sec from 0 to 30-min
B
570 770 970 1170 1370Raman Shift (cm-1)
Arbit
rary
Unit
s
MPA Photodegradation
600 800 1000 1200 1400 Wavenumber (cm-1)
A
0 10 20 30 0 10 20 30 time (min) time (min)
Ram
an In
tens
ity (r
elat
ive)
30
min
0
SPIE -2003-5269 120
starting solution was adjusted to 325 using dilute KOH Again 2 mL were added to a vial and the SER spectrum recorded At this pH a reasonably strong 760 cm-1 band was observed This process was repeated as spectra were recorded at pHs of 70 74 75 79 85 and 100 A total of 1 mL of KOH was added diluting the total concentration by 10 Next the pH of the starting solution was made acidic by adding dilute nitric acid dropwise This time spectra were recorded at pHs of 72 69 64 and 37 Figure 4 shows the SER spectra for representative pHs (spectra were left out to simplify the figure) while Figure 3 shows the 760 cm-1 peak intensities as a function of pH (The band intensities were adjusted to compensate for dilution effects caused by the addition of HNO3 and KOH then normalized to 0021 M for the most intense band observed at pH 37) It is clear from Figure 3 that the 760 cm-1 band follows the MPA- concentration as a function of pH and must be assigned to this anion No bands were observed that corresponded to MPA or MPA= The lack of an MPA SER spectrum may be due to the absence of an attraction between the neutral analyte and the electropositive silver surface The same reasoning suggests that a strong SER spectrum should be observed for MPA= but it is not and a satisfactory explanation has not been found
0000
0005
0010
0015
0020
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14pH
Con
cent
ratio
n [M
]
pK1 = 212 pK2 = 729
MPA- MPA=
MPA
Figure 4 SER spectra of 002M MPA as a function of pH Conditions 100 mW of 785 nm 36 scans (1 min) 8 cm-1 recorded 2 min after sample introduction pH 19 and 20 not apparent on this scale pH 69 and 74 near identical to 70 and 75 and not shown for clarity
00
02
04
06
08
10
12
14
16
18
0 5 10 15 20 25 30Measurement Number
Ram
an In
tens
ity (7
60 c
m-1
)
Figure 3 Concentration dependence of MPA MPA- and MPA= as a function of pH for a 002M sample Intensity of 760 cm-1
band from Figure 6 as a function of pH ( for increasing basic adjustment for increasing acidic adjustment error was measured at ~10 for pH 64)
Figure 5 SER spectra of 002M MPA at pH 64 measured around a vial at three heights (9 points per height) Conditions as in Figure 4 but 10-sec scans
Since these measurements involved the removal and replacement of the SER-active vial in the sample holder to remove and add sample variation in the intensity as a function of vial position was minimized by illuminating the exact same height along the vial wall But this does not account for variability of the SERS response of the sol-gel coating around the vial To analyze this effect a vial containing MPA at pH 64 was rotated at ~ 40o
intervals at the original height and 18rdquo above and below this value Figure 5 shows the intensity of the 760 cm-1
band for the 27 positions It was found that the average value was 137plusmn014 an RSD of 10 overall and 5 for each height An error bar is included in Figure 4 for the pH 64 measurement
SPIE -2003-5269 121
With the above analyses in mind a preliminary investigation of the SER spectral response for MPA (as MPA-) as a function of concentration was performed A single vial was used for these measurements beginning with 1 mgL followed by measurements of 10 100 and 1000 mgL In all cases the pH was ~7 and 3 positions around the vial were measured per concentration Since the 760 cm-1 band was not observed for concentrations of 1 or 10 mgL using 100 mW of 785 nm the laser power at the sample was raised to 200 mW beginning with the 10 mgL concentration Photo-degradation was largely avoided (and not observed) by exposing the sample for only 33 seconds per spectral acquisition Representative spectra for 10 100 and 1000 mgL are shown in Figure 6 while a plot of the 760 cm-1 band intensity as a function of concentration is shown in Figure 7 These values were also used to estimate limits of detection based on the signal-to-noise ratio (SN) of the 760 cm-1 band using 100 mW of 785 nm laser excitation and a 10-min acquisition time As summarized in the Figure 7 inset the lower the measured concentration the lower the predicted LOD Even if as estimated 210 microgL could be measured using the silver-doped sol-gel vials an improvement of a factor of 70 is still required to meet JSAWM goals of measuring 3 microgL in 10-minutes Similar to MPA DPA is a diprotic acid (pKa1 = 216 and pKa2 = 692) and variations in pH will effect the relative concentrations of DPA DPA- and DPA= and possibly the SER spectra and band intensities This could prove significant if an acid or base is used to denature anthrax spores with the goal of extracting and analyzing DPA The MPA pH study described above was mimicked for DPA except that the starting solution consisted of 20 mg of DPA in 20 mL HPLC grade water (60x10-3M MW = 1671) The initial solution had a pH of 245 which was made basic by dilute KOH to pHs of 355 433 487 559 1069 and 1166 SER spectra were recorded at each pH using 100 mW of 785 nm and a 44-sec acquisition time Next one drop of concentrated nitric acid was used to remake the solution acidic at a pH of 200 Again sequential pH measurements were performed at 383 510 735 and 822 The solution pH was made acidic a third time but to pH 219 171 then 135 Throughout this process no more than 20 drops of acid or base were added and therefore the concentration was diluted by no more than 10 Most of the spectral bands showed a minor decrease in intensity as a function of increasing pH values However the bands at 525 and 795 cm-1 showed the most dramatic changes which occurred at acid pH Figure 8 shows the SER spectra of DPA for the spectral region and pH range of interest The identity of the DPA species was determined by plotting the normalized peak intensities with the lowest value set to 0 and the highest to 0006 M as a function of pH and overlaying these values on a plot of the relative concentrations for DPA DPA- and DPA= as previously done for MPA (Figure 9) As can be seen the 525 cm-1 band clearly corresponds to DPA The correspondence of the 795 cm-1 band to this species is less clear as the band retains intensity until pH of 55 This can be attributed to contribution to the overlapping band at 810 cm-1 which does not change as a function of pH The fact that most bands are observed at all pHs suggest that the primary interaction with silver is through the ring nitrogen This is supported by the fact that the most intense band occurs at 1008 cm-1 attributed to a symmetric ring breathing mode and that this interaction has been characterized for pyridine in numerous papers27
0
1
2
3
4
5
6
7
0 200 400 600 800 1000 1200
MPA Concentration (mgL)
760
Ban
d In
tens
ity
Figure 6 SER spectra of MPA in water at A) 1000 B) 100 and C) 10 mgL Conditions pH of 7 silver-doped sol-gel coated vial 200 mW of 785 nm 33-sec 8 cm-1 resolution
Figure 7 Plot of SER intensity of 760 cm-1 band of MPA as a function of concentration using 200 mW of 785 nm Inset table includes average intensity LOD standard deviation and percent deviation for each concentration but for 100 mW and 10 min
conc (mgL) LOD ave stddev dev10 021 002 1021100 072 010 14301000 312 040 1280
A
B C
SPIE -2003-5269 122
The ability to detect dipicolinic acid (DPA) by SERS is demonstrated in Figure 10 Even at 1 mgL the primary bands are visible The intensity of the dominant ring breathing mode observed at 1008 cm-1 for DPA in water (pH = 25 to 55) is plotted as a function of concentration in Figure 11 These values were also used to estimate limits of detection based on the SN of the 1008 cm-1 band using 100 mW of 785 nm laser excitation and a 10-min acquisition time Again the lower the measured concentration the lower the predicted LOD (see Figure 11 inset) and detection of 160 microgL is possible Gastrointestinal anthrax requires significant more spores than inhalation anthrax28 and a limit of detection might be placed at 1 million spores in 1 liter of water or 10 microgL Since each spore contains ~10 CaDPA by weight29 a goal for DPA might be 1 microgL indicating that the present measurements must be improved by nearly two orders of magnitude Due to the increased hazards of handling HCN gas KCN salt was used for these experiments Nevertheless all sample preparations were performed in a chemical hood KCN completely dissolves in water but its conjugate acid HCN is formed and has a Ka of 615x10-1030 viz
HCN CN- + H+ pKa = 921 Reaction 4
Figure 8 SER spectra of 1 mgmL DPA as a function of pH Conditions 100 mW 785 nm 100 scans (44-sec)
Figure 9 Concentration dependence of DPA DPA- and DPA= as a function of pH for a 0006M sample Intensity of 525 () and 795 (diams) cm-1 bands from Fig 8 as a function of pH
pH
135 171 219 383
0
0001
0002
0003
0004
0005
0006
0007
0 2 4 6 8 10 12 14pH
Con
cent
ratio
n [M
]
DPADPA-DPA=795525
DPA DPA=DPA-
pK2 = 692pK1 = 216
795 525
0
05
1
15
2
25
0 200 400 600 800 1000 1200
DPA Concentration (mgL)
1008
Ban
d In
tens
ity
conc (mgL) lod-10min-100mw1 017
10 016100 103
1000 355
Figure 10 SER spectra of DPA in water at A) 1000 B) 100 C) 10 and D) 1 mgL Conditions pH of 25-55 silver-doped sol-gel coated vial 175 mW of 785 nm 1-min 8 cm-1 D) has been multiplied by x10 to make bands visible
A
B C
D
Figure 11 Plot of SER intensity of 1008 cm-1 band of DPA as a function of concentration using 175 mW of 785 nm Inset table includes LOD in mgL for each concentration but for 100 mW and 10 min
SPIE -2003-5269 123
Consequently the cyanide concentration must be determined for each initial KCN concentration Specifically the samples prepared with concentrations of 01 1 10 100 and 1000 mgL of KCN produced CN- concentrations of 63x10-3 033 69 89 and 964 mgL at pHs of 816 90 967 102 and 107 respectively The pH dependence for the HCN and CN- concentrations are shown in Figure 12 Thus as the amount of KCN added to the solution decreases so does the pH of the solution (becomes less basic) and according to Reaction 4 the relative amount of CN- to HCN also decreases For example in the preparation of a 01 mgL solution of KCN the pH is shifted from 7 for pure water to only 816 and only 63 of the starting material becomes CN- or 63x10-3 mgL In comparison for a solution of 1000 mgL the pH is shifted from 7 to 107 and 96 of the starting material becomes CN- This is significant since the cyanide ion is better able to adsorb onto the silver particles and become SERS active SER spectra of 10 100 and 1000 mgL of KCN samples are shown in Figure 13 and a corresponding plot of the 2100 cm-1 band intensity as a function of concentration is shown in Figure 14
000102030405060708091011
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14pH
Con
cent
ratio
n [m
gm
L]
pKa = 921
CN -HCN
The SER spectra of cyanide are dominated by the single CN stretch at 2100 cm-1 which occurs in normal Raman spectra of solutions at 2080 cm-1 However a low frequency mode occurs at 135 cm-1 due to a Ag-CN stretch (not shown) indicative of a strong surface interaction It is also observed that as the concentration decreases the CN stretch shifts to 2140 cm-1 This shift has been attribute to the formation of a tetrahedral Ag(CN)3
2- surface structure31 as well
Figure 14 Concentration dependence of KCN SERS measured under conditions in Fig 11 Concentrations are 1 01 and 001 mgml Intensities are measured for the CN stretch at 2100 cm-1 Inset table includes LOD in mgL for each concentration in Figs 13 and 15 but for 100 mW and 10 min
Figure 13 SER spectra of KCN in water at A) 1000 B) 100 and C) 10 mgL Conditions pHs of 107 102 and 97 silver-doped sol-gel coated vial 100 mW of 785 nm 1-min 8 cm-1 C) has been multiplied by x10 to make band visible
Figure 12 Concentration dependence of HCN and CN- as a function of pH for a 1 mgmL sample Calculated intensity of 2100 () cm-1 band for a 1 mgml sample at pHs of 816 90 967 102 and 107
A
B
C
0
20
40
60
80
100
120
140
0 200 400 600 800 1000 1200CN Concentration (mgL)
2100
Ban
d In
tens
ity
Figure 15 SER spectra of KCN in water at A) 10 B) 1 and C) 01 mgL Conditions pHs of 97 12 and 12 gold-doped sol-gel coated vial 100 mW of 785 nm 1-min 8 cm-1 C) has been multiplied by x10 to make band visible
A
B
C
conc (mgL) lod-10min-100mw Condition01 001 Au-pH 121 007 Au-pH 12
69 003 Au-pH 9769 016 Ag-pH 9789 022 Ag-pH 102946 113 Ag-pH 107
SPIE -2003-5269 124
as to CN adsorbed to two different surface sites32 Alternatively the 2140 cm-1 band could be attributed to HCN since this species dominates at lower concentrations However it is unlikely that this species would be attracted to the electropositive silver surface Further both peaks should be present at pHs between 85 and 105 but this is not observed It has also been suggested that at concentrations near and above monolayer coverage the CN- species is forced to adsorb end-on due to crowding and at lower concentrations the molecule can reorient to lie flat33 This suggests that the 2100 and 2140 cm-1 bands correspond to the end-on and flat orientations respectively As Figures 13 and 14 show the intensity of the CN stretch for the 89 mgL sample is nearly as intense as the 964 mgmL band This suggests that the Raman signal for the flat orientation is more enhanced However more extensive measurements are required to verify this point Since resent research has suggested that cyanide may be more effectively detected on gold measurements of KCN solutions were also performed using gold-doped sol-gel vials Preliminary measurements are shown in Figure 15 for samples prepared from 01 1 and 10 mgL KCN Since the pHs are 816 90 967 the resultant CN- concentrations are 63x10-3 033 and 69 mgL Initially only the highest concentration was observed and the signal intensity was significantly better than the equivalent concentration measured using silver In an effort to shift Reaction 4 to the left transforming HCN to CN- (Le Chatelierrsquos principle) KOH was added to the lower concentration samples producing solutions with pH 12 This effectively forces all of the cyanide in solution to be CN- or 01 and 10 mgL respectively More importantly the CN stretch is now observed in the SER spectra The band appears at 2125 cm-1 as has been previously reported for gold12 As calculated for MPA and DPA LODs can be estimated from this data For the three concentrations of cyanide on silver the LODs are 016 to 11 mgL for 100 mW of 785 nm laser excitation and a 10-min acquisition time For gold pH adjusted the LODs are10 to 70 microgL an improvement of more than 10 times silver Nevertheless either substrate is sufficient to meet the JSAWM goals of measuring 3 mgL in 10-minutes as the requirements form cyanide are much less stringent than the nerve agents
4 CONCLUSIONS Here we examined the surface-enhanced Raman spectral response for methyl phosphonic acid dipicolinic acid and cyanide as a function of pH It was determined that the most prominent peak at 760 cm-1 reaches a maximum intensity between pH 3 and 7 and corresponds to the MPA- species Neither the MPA nor MPA= species appear to generate a SER spectrum and consequently no spectra were observed below pH 2 or above pH 8 In this study we also found that higher laser powers could cause photodegradation of MPA signified by the exponential growth of a band at 1050 cm-1 which is tentatively assigned to phosphorous acid Unlike MPA DPA was observed at all pHs This is attributed to the dominant interaction of the pyridine functional group with silver Minor spectral changes were observed at acid pHs and were assigned to neutral DPA Like MPA SER spectra of cyanide were pH dependent No spectra were observed for the HCN species while CN- was best observed at pHs more basic than 8 Preliminary concentration studies for the three analytes allowed estimating limits of detection for MPA DPA and CN using 100 mW of 785 nm and a 10-min acquisition time of 210 165 and 70 microgL respectively Although the latter value suggests that cyanide can be measured at sufficiently low concentrations to meet JSAWM goal improvements by 100 to 200 times are required for MPA and DPA It is clear from this study that pH of the sample is important and must be taken into account when developing concentration calibration curves as well as fieldable analyzers Future work will include tailoring the sol-gel with specific functional groups to dictate sample pH and thereby optimize sensitivity
ACKNOWLEDGMENTS The authors are grateful for the support of the US Army (DAAD13-02-C-0015 Joint Service Agent Water Monitor program) The authors would also like to thank Dr Steve Christensen of the US Army SBCCOM for helpful discussions and Mr Chetan Shende of Real-Time Analyzers for assistance in development of the gold-doped sol-gels
REFERENCES 1 Erickson B Analytical Chemistry News amp Features June 1 397A (1998)
SPIE -2003-5269 125
2 Johnston RL Hoefler CM Fargo JC and Moberley B AT-ONSITE 5-8 (1994) 3 Black RM RJ Clarke RW Read and MT Reid J Chromatography 662 301-321 (1994) 4 USA Army Center for Health Promotion and Preventive Medicine (USACHPPM) Technical Guideline 230A 5 Jensen JL NC Wong W R Loerop SPIE 4775-03 (2001) 6 Hoenig SL Handbook of Chemical Warfare and Terrorism Greenwood Press Wesport CT (2002) 7 Munro NB SS Talmage GD Griffin LC Waters AP Watson JF King and V Hauschild Env Health
Persp 107 933-974 (1999) 8 Holstege CP Kirk M Sidell FR Crit Care Clin13 923-42 (1997) 9 Farquharson S P Maksymiuk K Ong and S Christesen SPIE 4577 166-173 (2001) 10 Lee Y and S Farquharson SPIE 4378 21-26 (2001) 11 Spencer KM J Sylvia S Clauson and J Janni SPIE 4577 158-165 (2001) 12 Tessier P S Christesen K Ong E Clemente A Lenhoff E Kaler and O Velev Applied Spectroscopy 56
1524-1530 (2002) 13 Farquharson S WW Smith S Elliott and JF Sperry SPIE 3533 207-214 (1998) 14 Farquharson S WW Smith S Elliott and JF Sperry SPIE 3855110-116 (1999) 15 Farquharson S WW Smith YH Lee S Elliott and JF Sperry SPIE 4575 62-72 (2002) 16 Guzelian AA J Sylvia J Janni S Clauson and KM Spencer SPIE 4577 182-192 (2001) 17 Shende C F Inscore A Gift P Maksymiuk and S Farquharson in press 18 Weaver MJ S Farquharson and MA Tadayyoni J Chem Phys 82 4867-4874 (1985) 19 Alak AM and T Vo-Dinh Analytical Chemistry 59 2149-2153 (1987) 20 Lee Y and S Farquharson SPIE 4206 140-146 (2000) 21 Farquharson S and Y Lee SPIE 4200-16 (2000) 22 Lee Y S Farquharson and P M Rainey SPIE 3857 76-84 (1999) 23 Lee Y S Farquharson H Kwong and M Shahriari SPIE 3537 252-260 (1998) 24 Wang J M Pumera G Collins and A Mulchandani Analytical Chemistry 74 6121-6125 (2002) 25 Dou X YM Jung Z-Q Cao and Y Ozaki Applied Spectroscopy 53 1440-1447 (1999) 26 Data supplied by S Christesen and K Ewing 27 Kerker M and B Thompson Eds SPIE MS 10 (1990) 28 Inglesby TV DA Henderson JG Bartlett JAMA 287 2236 (2002) 29 FW Janssen AJ Lund and LE Anderson Science 127 26 (1958) 30 Lide DR Ed Handbook of Chemistry and Physics CRC Press 77th Ed (1996-7) 31 Billmann J G Kovacs and A Otto Surf Sci 92 153 (1980) 32 Murray CA and S Bodoff Phys Rev B 32 671 (1985) 33 Kellogg D and J Pemberton J Phys Chem 91 1120 (1987)
SPIE-2003-5269 16
Chemical agent detection by surface-enhanced Raman spectroscopy
Stuart Farquharson Alan Gift Paul Maksymiuk Frank Inscore and Wayne Smith
Real-Time Analyzers 87 Church Street East Hartford CT 06108
Kevin Morrisey and Steven D Christesen US Army SBCCOM Aberdeen Proving Ground MD 21010
ABSTRACT
In the past decade the Unites States and its allies have been challenged by a different kind of warfare exemplified by the terrorist attacks of September 11 2001 Although suicide bombings are the most often used form of terror military personnel must consider a wide range of attack scenarios Among these is the intentional poisoning of water supplies to obstruct military operations in Afghanistan and Iraq To counter such attacks the military is developing portable analyzers that can identify and quantify potential chemical agents in water supplies at microgram per liter concentrations within 10 minutes To aid this effort we have been investigating the value of a surface-enhanced Raman spectroscopy based portable analyzer In particular we have been developing silver-doped sol-gels to generate SER spectra of chemical agents and their hydrolysis products Here we present SER spectra of several chemical agents measured in a generic tap water Repeat measurements were performed to establish statistical error associated with SERS obtained using the sol-gel coated vials Keywords Chemical agents chemical agent detection SERS Raman spectroscopy
1 INTRODUCTION
In the autumn of 2001 terrorism within US borders became a sobering reality While extensive efforts are being implemented to secure the homeland US and Coalition forces in Afghanistan and Iraq are constantly faced with terrorist attacks In addition to car-bombings the military has identified several non-traditional attack scenarios including poisoning of water supplies by chemical warfare agents (CWAs) To counter this threat the Department of Defense has been investigating numerous approaches to detect such attacks under the auspices of the Joint Service Agent Water Monitor (JSAWM) program1 The JSAWM must be capable of identifying the chemical agents with no false-positives and quantifying the agents at microgram per liter concentrations within 10 minutes2 This includes the analysis of drinking water supplies distribution and storage systems Currently colorimetric paper is used to detect agents on-site while gas chromatography combined with mass spectrometry (GCMS) is used in mobile support laboratories However both methods have severe drawbacks The paper changes color in response to contact with many chemicals besides CWAs causing a high incidence of false positives3 while GCMS although very chemically specific requires up to an hour to perform and regular re-calibration456 Vibrational spectroscopy has also been investigated as a potential method of rapidly detecting CWAs7-
11 as early as 1985 in the case of infrared spectroscopy8 and 1988 in the case of Raman spectroscopy11 Again however these techniques also have limitations when it comes to measuring trace poisons in water Infrared spectra would be dominated by the very strong absorption of water which would obscure absorptions by most other chemicals present Whereas Raman spectroscopy is simply not a very sensitive technique and detection limits are typically grams per liter Surface-enhanced Raman spectroscopy (SERS) offers several advantages over conventional vibrational methods and that may provide the necessary sensitivity required for detecting trace quantities of chemical agents in water SERS employs the interaction of surface plasmon modes of metal particles with target analytes to increase scattering efficiency by as much as 1 million times12 In 1987 the potential of this approach to measure CWAs was demonstrated by measuring a series of pesticides13 Several of these organophosphonates have chemical structures similar to CWAs in particular P=O functional groups In the past few years we and others have further explored the ability of SERS to detect CWAs14-17 and even bioagents 18-21 We have been employing silver-doped sol-gels to promote the SER effect
SPIE-2003-5269 17
in these studies The porous silica network of the sol-gel matrix offers a unique environment for immobilizing and stabilizing SER-active metal particles22-25 The silver-doped sol-gels have been coated on the internal walls of standard glass vials to allow rapid SER analysis of small sample quantities without preparation We have measured over 100 chemicals with enhancements of 104 to 106 demonstrated reversible measurements in a flowing system reproducible measurements from vial-to-vial and measurements in multiple solvents including water21-25 Previously we used these vials to perform preliminary measurements of cyanide (CN) and methylphosphonic acid (MPA) Most of the nerve agents form MPA during hydrolysis while Tabun forms CN a chemical agent in its own right In another paper including in these proceedings we examined the limits of detection (LOD) for MPA by measuring a series of concentrations down to 50 mgL and estimated a limit of detection of 100 microgL26 These measurements provide encouragement in that SERS may satisfy the needs of the JSAWM To further establish the viability of SERS in particular silver-doped sol-gels here we present analysis of cyanide mustard and VX in tap water The measurements performed at the US Armyrsquos Edgewood Chemical Biological Center Aberdeen MD also included numerous repeat measurements to establish reproducibility
2 EXPERIMENTAL 2a General All chemicals including potassium cyanide 2-chloroethylethyl sulfide and those used to prepare the silver-doped and gold-doped sol-gel coated vials were obtained and used as received from Sigma-Aldrich All samples were prepared in a chemical hood using HPLC grade water unless otherwise noted (Fischer Scientific Fair Lawn NJ) for SERS measurements Once prepared the samples were transferred into the silver-doped sol-gel vials (Real-Time Analyzers Inc East Hartford CT) The vials were coated in a manner similar to that previously reported by adding ammonium hydroxide to a solution of silver nitrate tetramethyl orthosilicate (TMOS) and methanol22 The two precursor solutions were prepared mixed and transferred to 2-mL glass vials dried and heated After sol-gel formation the incorporated metal ions were reduced with dilute sodium borohydride (1mgmL) which is followed by a water wash to remove residual reducing agent After the resultant analyte solution was introduced the SER-active vial was fixed horizontally to an XY positioning stage (Conix Research Springfield OR) just inside the focal point of an f07 aspheric lens The lens focused the beam into the sample and collected the scattered radiation back along the same axis A dichroic filter (Omega Optical Brattleborough VT) was used to reflect the excitation laser to the lens and pass the Raman scattered radiation collected by the lens An f2 achromat was used to collimate the laser beam exiting a 200 microm core diameter source fiber optic while a second f2 achromat was used to focus the scattered radiation into a 365 microm fiber optic (Spectran Avon CT) A short pass filter was placed in the excitation beam path to block the silicon Raman scattering generated in the source fiber from reflecting off sampling optics and reaching the detector A long pass filter was placed in the collection beam path to block the sample Rayleigh scattering from reaching the detector A 785 nm diode laser (Process Instruments Inc model 785-600 Salt Lake City UT) was used to deliver 100 to 150 mW of power to the sample A Fourier transform Raman spectrometer (Real-Time Analyzers model IRA-785 East Hartford CT) and a silicon photo-avalanche detector (Perkin Elmer model C30902S Stamford CT) were used to acquire the SER spectra
2b Edgewood Chemical Biological Center The surface-enhanced Raman spectral measurements at the US Armyrsquos Chemical Biological center presented here were all performed on September 12 2003 To expedite measurements a plate was machined to hold up to 12 SER-active sample vials (Figure 1) The plate fit a standard XY plate reader that could be programmed Pure KCN bis-(2-chloroethyl)sulfide (distilled mustard gas HD) and ethyl S-2-diisopropylamino ethyl methylphosphonothioate (VX) were obtained on-site and used to prepare 1 gL tap water solutions in a chemical hood with appropriate safety equipment Simulated tap water was prepared by adding 10 chemicals most often found in tap water at appropriate concentrations to distilled water (Table 1) SER measurements were also performed in a chemical hood For added safety the FT-Raman instrument was placed outside the laboratory and 30 foot fiber optic and electrical cables were used to allow remote SERS measurements and plate manipulation For each experiment 1gL samples were prepared and added to 9 individual vials which were then loaded on the plate In some cases a tenth vial was included as a blank
Table 1 Generic Tap Water Composition
Compound mgL NaHCO3 100 CaSO4 27 MgSO4bull7H2O 67 NaNO3 1 Fulvic Acid 1 K2HPO4 07 KH2PO4 03 (NH4)2HSO4 001 NaCl 001 FeSO4 0001 pH 76-78
SPIE-2003-5269 18
A software program was written that allowed selecting the sequence that the vials were measured the number of positions along the length of the vials to measure (1 to 5) and the number of scans to co-add During sample analysis the program displayed the vial being analyzed the point being analyzed and the spectrum as it was being acquired Once all the data was collected a second software program was written to rapidly analyze the data The spectra collected for all the vials on a plate could be loaded at one time and then the spectra for each point could be displayed simultaneously or separately The user could then select the Raman peak to analyze in terms of peak height or area This was accomplished by selecting points on either side of the peak to define a baseline of zero The peak height or area could then be computed for all of the spectra loaded and then exported to a spreadsheet for statistical analysis
Figure 1 A) Vial Holder 6 slots to hold 2 vials each end-to-end B) Measurement Configuration Program user selects vials to measure sequence number of points per vial (1 to 5) and number of scans per point C) Spectral Acquisition Program shows spectrum being collected which vial and position D) Spectra Analysis Program user selects spectra to analyze by plate vial and point (s) as well as two wavenumbers defining the peak and the baseline to subtract The image is of 5 repeat measurements of 10 mgL KCN in generic tap water 16 sec each 100 mW of 785 nm
3 RESULTS AND DISCUSSION Raman and surface-enhanced Raman spectra were obtained for potassium cyanide bis-(2-chloroethyl)sulfide and ethyl S-2-diisopropylamino ethyl methylphosphonothioate representing three classes of chemical agents cyanides mustards and nerve agents respectively Spectra were also obtained for 2-chloroethyl ethyl sulfide (CEES) a structural analogue to HD which was included in the study to aid in assigning spectral bands KCN salt was used for cyanide experiments to avoid the increased hazards of handling HCN gas KCN completely dissolves in water forming its conjugate acid HCN according to its Ka of 615x10-1027 and at a concentration of 1 mgmL results in a pH 107 solution This is significant in that only CN- appears to interact sufficiently with silver to produce a SER spectrum and no spectral signal is observed below pH 726 Figure 2 shows the SER and normal Raman spectra for KCN The SERS spectrum of 1mgml KCN in water shows a single intense somewhat broad feature at
A B
C D
SPIE-2003-5269 19
2100 cm-1 assigned to the single CequivN stretch The band is much sharper in the normal Raman spectra of the solid KCN salt at 2074 cm-1 This band does broaden and shift to 2080 cm-1 in solution (not shown) However the observed SERS frequency is attributed to interaction with silver and low frequency mode at 135 cm-1 attributed to a Ag-CN stretch (not shown) supports this conclusion
cm-1 band to a C-S stretch but the authors concede that it is in fact more likely a C-Cl stretch28 It appears that the most intense Raman bands at 648 692 and 747 cm-1 shift to 620 660 and 720 cm-1 in the SER spectra and are tentatively assigned as above The width of these bands suggests that they overlap underlying spectral features Additional bands in the Raman spectra occur at 972 1034 1049 1263 1286 1423 1442 2865 2935 and 2960 cm-1 Corresponding bands occur in the SER spectra at 964 1015 1054 1286 1410 1447 2865 and 2935 cm-1 Most of these bands are associated with alkane modes specifically the bands at approximately 1040 cm-1 to a C-C stretch 1290 cm-1 to a CH2 in-phase twist 1440 cm-1 to a CH2 wag 2865 cm-1 to a symmetric CH2 stretch and 2965 cm-1 to an asymmetric CH2 stretch The Raman and SER spectra of sulfur mustard were measured at the Edgewood center (Figure 4) Both spectra are largely similar to CEES The C-Cl and C-S bands in the Raman spectrum of HD now occur at 640 655 700 739 and 760 cm-1 and are more resolved possibly due to the increased molecular symmetry Theoretical calculations indicate that the first three bands are due to C-Cl stretching modes and the latter two to C-S stretching modes28 Only the C-Cl bands maintain significant intensity in the SER spectra occurring at 624 and 643 cm-1 which is attributed to the
Prior to measurements of HD CEES was examined by Raman and SER spectroscopy (Figure 3) CEES also known as half-mustard is essentially identical to HD except one of the chlorine end atoms is replaced by a hydrogen atom Again although not as toxic as HD CEES is a blister agent and dilute aqueous samples were prepared using appropriate safety equipment Both the Raman and SER spectra of CEES are similar and dominated by bands between 600 and 800 cm-1 These are associated with C-Cl and C-S stretching modes which are tentatively assigned to 648 and 747 cm-1 in the Raman spectra respectively The shoulder at 630 cm-1 the overlapped band at 660 cm-1 and the strong band at 692 cm-1 could also be due to these modes or their asymmetric counterparts It is worth noting that theoretical calculations assign the 692
Figure 2 A) SER and B) NR spectra of KCN Conditions A) 1 mgml in tap water 100 mW of 785 nm at sample 1-min acquisition time B) solid 300 mW of 785 nm 5-min All spectra are 8 cm-1 resolution
A
B
Figure 3 A) SER and B) NR spectra of CEES Conditions A) 1 vv (10 mgml) in MeOH 100 mW of 785 nm 1-min acquisition time B) neat 300 mW of 785 nm 5-min
A
B
Cl-CH2-CH2-S-CH2-CH3
A
B
Figure 4 A) SER and B) NR of HD Conditions A) 1mgml in tap water B) pure both 100 mW of 785 nm 1-min
Cl-CH2-CH2-S-CH2-CH2-Cl
SPIE-2003-5269 20
expected strong interaction between chlorine and silver and adds support to the assignment of this band to a C-Cl stretch Weaker overlapping bands occur at 670 692 and 724 cm-1 the latter possibly due to C-S stretching modes Again the alkane modes are apparent in the normal Raman spectra of HD but only a broad feature at 1300 to 1450 cm-1 suggests CH2 contributions in the SER spectrum Although the observed bands in the VX spectrum have not been assigned (Figure 4) a computer generated Raman spectrum29 predicts many of the same features with surprising accuracy and are used here Two intense bands at 460 and 530 cm-1 closely match predicted bands at 463 and 546 cm-1 assigned to a CH3-P=O bend and a PO2CS wag Three highly overlapped bands occur at 694 745 and 771 cm-1 matching predicted bands at 713 730 and 760 cm-1 The first
Table 2 Measured SER peak heights for the CN stretch at 2100 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
Figure 5 A) SERS and B) NR spectra of VX Conditions A) 1 vv (10 mgml) in MeOH B) pure sample both 100 mW of 785 nm 1-min acquisition time
A
B
two have been assigned to a C-S stretch and CH2bend respectively while the latter has been attributed to either a P-C stretch or an O-C-C stretch Although the 745 cm-1 band may alternatively be assigned to a C-S stretch based on the previous measurements of CEES and HD The relatively intense bands at 890 1106 1218 1445 and 1465 cm-1 also match predicted bands at 880 1108 1216 1440 and 1464 cm-1 that are assigned to a C-C stretch CH3 rock N-C3 stretch various C-H3 bends and C-H bends respectively Both the computer generated and the measured spectra contain numerous other less intense bands One is worth mentioning A unique band appears at 370 cm-1 that is predicted at 368 cm-1 and corresponds to an O-P=O bend The surface-enhanced Raman spectrum of VX is also rich with spectral features It has the unique low frequency band at 370 cm-1 as well as a second band at 380 cm-1 that is assigned tothe S-P-O bend predicted in the normal Raman spectrum at 388 cm-1 Based on the measured and predicted normal Raman spectra the following SERS assignments are given 460 cm-1 to the CH3-P=O bend 544 cm-1 to the PO2CS wag 738 cm-1 to a C-S stretch (based on arguments above) 890 cm-1
to a C-C stretch 1101 cm-1 to a CH3 rock and 1456
cm-1 to a C-H bend The ability of SERS to measure chemical agents in water containing real-world chemical interferents was tested by using the generic tap water described in Table 1 The ability to reproduce measurements was accomplished by preparing three separate water stock solutions which were used to prepare three sample solutions each of 1 mgmL KCN in the generic tap water The samples defined as vials 1-9 were then measured at 4 points per vial in a semi-random fashion such that errors associated with stock solution preparation and errors associated with instrument drift could be identified No trends were apparent that signified such systematic errors Each spectrum collected consisted of 20 averaged scans taking 16 seconds at 8 cm-1 resolution The laser power at the sample was measured periodically during the day and it ranged from 102 to 105 mW spacccc
spot Vial 1 Vial 2 Vial 41 13754 13004 128192 13519 12692 12909
1 3 13541 12721 126394 13462 12648 12651
ave 13569 12766 12755Vial 6 Vial 3 Vial 5
1 11586 14068 10757 2 11236 14402 115122 3 11376 14546 11559
4 10894 11763 11214ave 11273 13695 11261
Vial 7 Vial8 Vial 91 11198 15804 11023
2 11243 15716 88753 3 11562 15216 11218
4 11655 15084 11077ave 11415 15455 10548
CN AVG STDEV ERR1 mgmL pk ht 12526 1560 1245
stock solution
SPIE-2003-5269 21
The CN measurements consisting of 36 data points produced an average height of 125 for the 2100 cm-1 peak with a standard deviation of 156 or 125 (Table 2) The HD and VX measurements were performed precisely the same way (Tables 3 and 4) For HD the 624 cm-1 peak was used for analysis and it had an average height of 53 with a standard deviation of 068 or 129 while for VX the 544 cm-1 peak was used for analysis and it had an average height of 1051 with a standard deviation of 308 or 293 The greater error in the VX measurements can be somewhat attributed to Vial 4 which produced lower SER signal intensities But removing this vial from the data set changes the standard deviation to 233 only a modest improvement
spot Vial 1 Vial 2 Vial 41 525 609 4682 45 675 484
1 3 527 644 5094 559 807 669
ave 51525 68375 5325Vial 6 Vial 3 Vial 5
1 521 574 575 2 536 53 4492 3 509 508 451
4 565 594 379ave 53275 5515 4635
Vial 7 Vial 8 Vial 91 457 589 413
2 583 497 4273 3 544 505 4
4 528 5 509ave 528 52275 43725
HD AVG STDEV ERR1 mgmL pk ht 530 068 1291
stock solution
spot Vial 1 Vial 2 Vial 41 1464 1034 4672 1485 99 61
1 3 1491 989 5684 1041 777 553
ave 137025 9475 5495Vial 6 Vial 3 Vial 5
1 1058 942 1293 2 697 121 965
2 3 727 1206 11124 689 1254 746
ave 79275 1153 1029Vial 7 Vial 8 Vial 9
1 1127 783 163 2 1358 812 1614
3 3 1371 908 15124 1097 875 1397
ave 123825 8445 153825
VX AVG STDEV ERR1 mgmL pk ht 1051 308 2925
stock solution
4 CONCLUSIONS In this paper we examined the ability of surface-enhanced Raman spectroscopy to reproducibly measure CN HD and VX in tap water without chemical interference Both normal and surface-enhanced Raman spectra were examined to select unique bands suitable to identify and quantify these chemical agents For SER measurements the 2100 cm-1 C-N stretch was used for CN the 624 cm-1 C-Cl stretch was used for HD and the 544 cm-1 PO2CS wag was used for VX It was determined that 1 mgmL samples of each of these chemicals measured 36 times in glass vials coated with a silver-doped sol-gel reproduced measurements with standard deviations of 125 129 and 293 It was further found that the 10 chemicals added to simulate generic tap water did not interfere with or alter the SER spectra It should be noted that the concentrations used in this study were considerably greater than those required by the JSAWM program Current work involves improving SER sensitivity and designing sampling systems with better reproducibility This includes the development of fractal silver and gold structures within the sol-gel matrix and the development of chemically selective sol-gels
ACKNOWLEDGMENTS The authors are grateful for the support of the US Army (DAAD13-02-C-0015 Joint Service Agent Water Monitor program) The authors would also like to thank Janet Jensen Ronald Crosier and Kristina Gonser for helpful discussions
Table 4 Measured SER peak heights for the VX band at 544 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
Table 3 Measured SER peak heights for the HD band at 624 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
SPIE-2003-5269 22
REFERENCES 1 Jensen JL NC Wong W R Loerop SPIE 4775-03 (2001) 2 JSAWM Requirements at wwwsbccomapgeaarmymilRDAecbcrtPRODSERJSAWMjsawmhtml 3 Erickson B ldquoThe Chemical Weapons Convention Redefines Analytical Challengerdquo Anal Chem 70 397A (1998) 4 Johnston RL Hoefler CM Fargo JC and Moberley B ldquoThe Defense Nuclear Agencyrsquos
ChemicalBiochemical Weapons Agreements Verification Technology Research Development Test and Evaluation Program and its Requirements for On-Site Analysisrdquo AT-ONSITE 5-8 (1994)
5 Black RM Clarke RJ Read RW and Reid MT ldquoApplication of GC-MS and GC-tandem MS to the analysis of chemical warfare samples found to contain residues of the nerve agent sarin sulphur mustard and their degradation productsrdquo J Chrom 662 301-321 (1994)
6 USA Army Center for Health Promotion and Preventive Medicine (USACHPPM) Technical Guideline 230A 7 Hoffland LD Piffath RJ Bouck JB rdquoSpectral signatures of chemical agents and simulantsrdquo Optical
Engineering 24 982-984 (1985) 8 Braue EHJ Pannella MGrdquoCIRCLE CELL FT-IR Analysis of Chemical Warfare Agents in Aqueous Solutionsrdquo
App Spectrosc 44 1513-1520 (1990) 9 Tseng C-H Mann CK and Vickers TJ ldquoDetermination of Organics on Metal Surfaces by Raman
Spectroscopyrdquo App Spectrosc 47 1767-1771 (1993) 10 Staff Reporter ldquoUS Army Tests Lidar to Detect Biological Toxinsrdquo Photonic Spectra pg 50 December 1998 11 Christesen SD Raman cross sections of chemical agents and simulants App Spectrosc 42 318-321 (1988) 12 Weaver MJ Farquharson S Tadayyoni MA ldquoSurface-enhancement factors for Raman scattering at silver electrodesrdquo J Chem Phys 82 4867-4874 (1985) 13 Alak AM and Vo-Dinh T ldquoSERS of Organophosphorous Chemical Agentsrdquo Anal Chem 59 2149-2153 (1987) 14 Lee Y Farquharson S ldquoRapid chemical agent identification by SERSrdquo SPIE 4378 21-26 (2001) 15 Farquharson S Maksymiuk P Ong K Christesen S ldquoChemical agent identification by surface-enhanced Raman spectroscopyrdquo SPIE 4577 166-173 (2001) 16 Spencer KM Sylvia J Clauson S and Janni J ldquoSurface Enhanced Raman as a Water Monitor for Warfare
Agents in Waterrdquo SPIE 4577 158-165 (2001) 17 Tessier P Christesen S Ong K Clemente E Lenhoff A Kaler E Velev O ldquoOn-line spectroscopic characterization of sodium cyanide with nanostructured gold surface-enhanced Raman spectroscopy substratesrdquo
App Spectrosc 56 1524-1530 (2002) 18 Farquharson S W W Smith S Elliott and J F Sperry Rapid biological agent identification by surface- enhanced Raman spectroscopy SPIE 3855110-116 (1999) 19 Farquharson S WW Smith YH Lee S Elliott and J F Sperry Detection of biological signatures A
comparison of electrolytic and metal-doped sol-gel surface-enhanced Raman media SPIE 4575 62-72 (2002) 20 Guzelian AA Sylvia J Janni J Clauson S and Spencer KM ldquoSERS of whole cell bacteria and trace levels of biological moleculesrdquo SPIE 4577 182-192 (2001) 21 Shende C Inscore F Gift A Maksymiuk P Farquharson S ldquoRapid dipicolinic acid extraction from Bacillus spores detected by surface-enhanced Raman spectroscopyrdquo App Spectrosc 58 accepted 22 Lee Y and Farquharson S ldquoSERS Sample Vials Based on Sol-Gel Process for Trace Pesticide Analysisrdquo SPIE
4206 140-146 (2000) 23 Farquharson S and Lee Y ldquoTrace Drug Analysis by SERSrdquo SPIE 4200-16 (2000) 24 Lee Y Farquharson S and Rainey P M Surface-Enhanced Raman Sensor for Trace Chemical Detection in
Water SPIE 3857 76-84 (1999) 25 Lee Y Farquharson S Kwong H and Shahriari M ldquoSol-Gel Chemical Sensor for Surface-Enhanced Raman
Spectroscopyrdquo SPIE 3537 252-260 (1998) 26 Farquharson S A Gift P Maksymiuk F Inscore and W Smith ldquopH dependence of methyl phosphonic acid
dipicolinic acid and cyanide by surface-enhanced Raman spectroscopyrdquo SPIE 5269-19 (2003) 27 Lide DR Ed Handbook of Chemistry and Physics CRC Press 77th Ed (1996-7) 28 Sosa C RJ Bartlett K KuBulat and WB Person ldquoA theoretical study of harmonic vibrational frequencies and
infrared intensities of XCH2CH2SCH2CH2X and XCH2CH2SH (= H Cl)rdquo J Phys Chem 93 577-588 (1993) 29 Hameka HF and JO Jenson ldquoComputer-generated predictions of the structure and the IR and Raman spectra of
VXrdquo ERDEC-TR-065 May 1993
- Chemagents Appendicespdf
-
- SPIE2001-4575-Bioagent-SERSapdf
-
- Detection of bioagent signatures A comparison of electrolytic and metal-doped sol-gel surface-enhanced Raman media
- Stuart Farquharson Wayne Smith and Yuan Lee
-
- Real-Time Analyzers 87 Church Street East Hartford CT 06108
-
- ABSTRACT
- 4 CONCLUSIONS
- 5 ACKNOWLEDGEMENTS
- 6 REFERENCES
-
SPIE 2001-4575
62
Detection of bioagent signatures A comparison of electrolytic and metal-doped sol-gel surface-enhanced Raman media
Stuart Farquharson Wayne Smith and Yuan Lee
Real-Time Analyzers 87 Church Street East Hartford CT 06108
Susan Elliott and Jay F Sperry University of Rhode Island 45 Lower College Rd Kingston RI 02881
ABSTRACT Since September 11 2001 the threat of terrorist attacks and biological warfare within US borders has become a sobering reality In an effort to aid military personnel and the public at large we have been investigating the utility of surface-enhanced Raman spectroscopy (SERS) to provide rapid identification of chemical agents directly and biological agents through their chemical signatures This approach is based on the ability of Raman spectroscopy to identify molecular structure through the abundant vibration information provided in spectra and the ability of SERS to detect extremely low concentrations (eg part-per-billion) through the enhancement of Raman scattering by six orders of magnitude or more Towards the goal of developing a portable analyzer we have been studying the ability of two SER media to obtain continuous (ie reversible) and quantitative (ie reproducible) measurements Here we compare measurements of nucleic acid-bases adenosine monophosphate and ribonucleic acid extracted from Escherichia coli Bacillus subtilis and Staphylococcus aureus obtained by electrolytic SERS and metal-doped sol-gel SERS The capabilities of these SER media are summarized in terms of rapid detection of B anthracis and dipicolinic acid Keywords bioagent detection SERS RNA analysis bacterial analysis Raman spectroscopy
1 INTRODUCTION The recent distribution of anthrax through the US postal system and the subsequent infection and death of several postal and national media employees amplifies the need for methods to rapidly detect and identify this and other chemical and biological warfare agents (BWA) The primary methods currently used immunoassays for screening and nucleic acid (NA) sequencing for positive identification of BWAs (bacteria protozoa and viruses) have serious limitations123 Immunoassay methods employ competitive binding of the bioagent (as an antigen) and its labeled (eg fluorescence) conjugate for a limited number of antibodies Although this analysis method is fast and semi-quantitative other chemicals may compete for the antibodies interfere with the enzymatic reaction or interfere with the measurement (eg it fluoresces) resulting in a high number of false positive responses1 Furthermore the antibodies denature due to moisture and heat limiting shelf life and require sterile often refrigerated storage Positive identification of a BWA can be accomplished by sequencing deoxyribonucleic acid or ribonucleic acid (DNA and RNA)23 This requires enumeration of a nucleic acid sequence through polymerase chain reactions (PCR) or multiplication of the microorganism through culture growth to provide sufficient quantities of DNA or RNA for analysis Unfortunately PCR and culture growth require from several hours to several days23 Consequently a wide variety of technologies have been investigated for rapid identification of BWAs The Department of Defense is actively monitoring 200 such technologies4 This includes traditional methods such as gas chromatographic separation coupled with ion mobility spectrometry detection5 to exotic methods based on nature such as monitoring toxin induced color changes in fish scales6 Although all of these techniques have achieved varying degrees of success none are yet capable of detecting and identifying BWAs in 10 minutes or less Towards this goal we have been investigating the ability of SERS to detect sub-nanogram quantities of DNA or RNA (eliminating enumeration) determine relative NA base concentrations and identify BWA taxonomy
To whom correspondences should be addresses e-mailfarqureal-time-analyzerscom wwwreal-time-analyzerscom
SPIE 2001-4575
63
Raman spectroscopy has a rich history of investigating biochemical and biological processes7 Some of the earliest laser-Raman studies demonstrated that the five NA bases adenine (A) cytosine (C) guanine (G) thymine (T in DNA) and uracil (U in RNA) yielded distinct spectra with several bands suitable for identification and quantification8 Furthermore these studies included exceptional spectra of both DNA and RNA for which the NA bases as well as several phosphate bands were easily identified9 However since the Raman effect is very inefficient (very low conversion of incident radiation to inelastically scattered Raman radiation) these samples had to be highly concentrated Fortunately two phenomena exist that can increase the generation of Raman photons by six orders of magnitude or more known as the resonance Raman and surface-enhanced Raman effects1011 Resonance Raman scattering occurs when the laser excitation wavelength is in resonance with an electronic transition of a molecule (within the absorption envelope)10 Excitation at ultraviolet wavelengths has been used to obtain resonance Raman spectra of amino acids and nucleic acids in whole bacteria1213 For example excitation at 242 nm has been used to maximize the nucleic acid spectral band intensities and minimize the amino acids band intensities A peak at 1530 cm-1 was found to be proportional to the amount of the NA base cytosine while a peak at 1485 cm-1 was proportional to the combined amount of the NA bases adenine and guanine This quantitative behavior has been used to define an A+TG+C base-pair ratio and provide a level of bacterial identification as taxonomic markers13 In recent years SERS has also been used to analyze bacterial cell components14 including amino acids15 lipids16 nucleic acids151718 and the adenine derivatives192021 SERS has proven to be one of the most sensitive methods for trace chemical analysis through the detection of single molecules2223 including DNA (dye labeled 17-mer)24 Since its discovery in 197425 the mechanism responsible for the large increase in scattering efficiency has been the subject of considerable research2627 Briefly incident laser photons couple to free conducting electrons within a metal which confined by the particle surface collectively cause the electron cloud to resonate2628 These surface plasmons are known as the physical component of the SER effect These surface plasmons can transfer energy to the molecular vibrational modes of molecules through interactions with the molecular electron orbitals2629 This interaction is known as the chemical component of the SER effect This perturbation of the molecular polarizability generates surface-enhanced Raman photons26 A number of methods have been developed to produce surfaces or solutions containing one of these metals with optimum roughness or diameter to promote SERS30 These methods include preparation of activated electrodes in electrolytic cells 1131 activated silver and gold colloid reagents32 and metal coated substrates333435 Selecting a SER-active medium for chemical and biological agent detection requires consideration of the method of deployment and hence the method of sampling Chemical aerosols or airborne bacteria will require a collection device to concentrate and continuously present the sample to the SERS medium Poisoned water supplies will also require a flow through device for continuous monitoring or a grab-sample device for periodic analysis And contaminated surfaces will require a grab-sample extractive device A SERS-based device used for continuous monitoring (air or water) must be reversible and reproducible if quantitative measurements are desired while a SERS-based device used for periodic sampling (water or surfaces) must be reproducible Both reversible and reproducible measurements have been performed using electrolytic SERS (E-SERS)36 But this requires a three-electrode sample cell and an electrolyte of known concentration to perform the necessary oxidation-reduction cycles (ORCs) to re-activate the electrode surface with new uncontaminated sites from one measurement to the next Colloids are severely limited in that continuous measurements would require a continuous supply of colloids For periodic measurements vials of colloids one per measurement could be used However aggregate size and consequently SER intensity change with sample conditions (especially pH) and quantitative reproducible measurements are unlikely Substrates appear to have the greatest potential and designs range from silver evaporated on titania particles34 to periodic gold pyramids evaporated between polystyrene beads35 Most substrates require concentrating the sample on the surface through drying to obtain the largest signal enhancements in effect making the measurements irreproducible and irreversible However successful measurements using flow systems have been obtained with glass posts but manufacturing costs appear prohibitive In an effort to overcome these limitations we have developed metal-doped sol-gels to provide SERS measurements that are reproducible reversible and quantitative and yet not restricted to specific environments such as electrolytes solvents or evaporated surfaces3738 The porous silica network of the sol-gel offers a unique environment for stabilizing SER active metal particles and the high surface area increases the interaction between the analyte and metal particles The sol-gel can be coated on the end of fiber optics or on the internal walls of a glass flow tube for continuous measurements or standard glass sample vials for periodic measurements Previously we measured 100 mgL methylphosphonic acid (the primary hydrolysis product of nerve agents) in water with an estimated detection limit of 05 mgL (100 parts-per-billion) We have also
SPIE 2001-4575
64
demonstrated reversible and reproducible measurements of p-aminobenzoic acid (PABA) in a flow through system Here we investigate the ability of the sol-gel SERS (SG-SERS) to measure the NA bases adenosine monophosphate and RNA extracted from E coli B subtilis and S aureus The measurements are compared to those obtained by E-SERS
2 EXPERIMENTAL The inorganic chemicals and solvents used to prepare samples were spectroscopic grade and purchased from Aldrich (Milwaukee WI) Fisher (Pittsburgh PA) or Pfaltz amp Bauer (Waterbury CT) The nucleic acid bases and dipicolinic acid were purchased from Sigma (St Louis MO) Normal Raman samples were measured to establish enhancement factors In each case 1cm3 of sample was placed into a 1x1 cm glass cuvette weighed and measured Unpacked densities were typically 6-7 gcm3 For all SER measurements including RNA samples were prepared as ~01mgmL (see Figure captions for exact concentrations) in 01M KCl and buffered to a pH of 92 with Na2B4O7bullH2O Adenine pH dependence measurements used pH buffer solutions at 4 (potassium acid phthalate) 69 (potassium phosphate monobasicsodium phosphate dibasic) 92 (Na2B4O7bullH2O) and 104 (tris-hydroxymethyl amino methane) Escherichia coli Bacillus subtilis and Staphylococcus aureus cultures (250ml per 1000mL Erlenmeyer flask) were grown overnight in a Trypticase soy broth (TSB) medium containing 1 glucose in a shaking water bath at 37 oC The bacteria were harvested by centrifugation for 10 minutes at 8000 rpm in a GSA rotor at 5degC then washed once in 085 saline The gram-positive bacteria were concentrated to 20 ml and passed through a French pressure cell twice at 15000 psi to break open the cells RNA was extracted according to Protocol 44139 to ensure pristine samples for initial measurements Since this method takes approximately 4 hours a streamlined method was developed For vegetative bacteria the specimen was boiled for 30 sec in 1 ml of distilled water to lyse the cells and release the RNA For bacterial spores the specimen was first incubated in 1 ml of saline solution containing 02 mg lysozyme and phosphate-buffered to pH of 624 for 1 hr at 37 oC This solution was then boiled for 2-3 minutes in 4X loading buffer to release the RNA For both specimens RNA STAT-60TM was added to the supernatant which was centrifuged at 12000 g for 5 minutes to precipitate the ~15 water-soluble proteins This procedure allowed extracting RNA for SER analysis in ~ 10 minutes Electrophoresis shows high purity while the existence of chemicals that could interfere with the SER measurements is still under investigation The electrolytic sample cell has been described previously36 Briefly a three electrode design is incorporated into a Plexiglas structure containing a well for the reference electrode (a saturated calomel electrode Cole Parmer Vernon Hills IL) and a 5mL sample well containing the silver working electrode and platinum wire counter electrode (05 mm wire Alfa Ward Hill MA) A channel connecting the two wells contained a 2 mm diameter semi-porous membrane (10-20 micron pore Ace Glass) The silver electrode was made from a 3 mm length of 2 mm diameter silver wire (Alfa) soldered to a copper wire lead encased in a 4 mm diameter Pyrex tube A cap containing the silver electrode platinum wire and nitrogen purge and vent lines fixed the silver electrode surface 1 mm from a 1 mm thick glass plate attached to the bottom of the sample well The potentiostat used to control the three electrodes was built in-house and has been described in detail elsewhere36 A multifuntional analog digital and timing inputoutput interface card (DAQCard-1200 National Instruments) is used to both drive the electrolytic cell as well as read the current generated in the cell A LabVIEW software program is used set the oxidation potential reduction potential SER measurement potential hold times and sweep rates The amount of charge passed was plotted as a cyclic voltammogram For all spectra presented five oxidation-reduction cycles (ORCs) stepping from -03 VSCE to 03 VSCE and back to -3 VSCE at 50 mVsec were used The SG-SER measurements were accomplished by simply placing the identical samples prepared above into Simple SERS Sample VialsTM (RTA) These 2-mL glass vials are internally coated with ~ 01 micron thick silver-doped sol-gel A prototype Fourier transform Raman spectrometer (RTA) was used for these measurements40 The system consisted of a NdYAG laser (Brimrose or Spectra Physics) for excitation at 1064 nm an interferometer built by On-Line Technologies (OLT East Hartford CT) for frequency separation an uncooled InGaAs detector for signal detection (RTA) and an Intel 400 MHz Pentium II based laptop computer (Dell Round Rock TX) for interferometric control data acquisition (OLT) and analysis (LabVIEW by National Instruments Austin TX) Additional components included a Notch filter (Kaiser Ann Arbor MI) and interferometer entrance and exit optics (Edmund Scientific Barrington NJ) Fiber optics were used to deliver the excitation beam to the sample and the scattered radiation to the interferometer (2 meter lengths of 200 and 365 micron core diameter respectively Spectran Avon CT) A second Notch filter (Kaiser) was used as a beam splitter to direct the excitation beam along the same axis as the collected radiation A microscope object (20x magnification 04 numeric aperture Newport Irvine CA) was used to focus the beam into the sample and to collect the scattered radiation back along the same axis In this co-axial backscattering arrangement the excitation beam passed through the glass plate onto the silver
SPIE 2001-4575
65
electrode surface for E-SERS through the vial glass wall and into the silver-doped sol-gel film for SG-SERS or through the glass wall of the cuvette and into the solid sample for normal Raman spectroscopy All E-SERS and normal Raman spectra were obtained with 750 mW of laser power at the sample while all SG-SERS spectra were obtained with 75 mW of laser power at the system Incident powers above 200 mW in some cases degraded the sol-gel
3 RESULTS AND DISCUSSION The generation of surface-enhanced Raman scattering at electrode surfaces has been extensively researched and the optimum sample conditions are well developed2729 Several researches incorporated electrodes into flowing systems and demonstrated that continuous monitoring of chemicals is possible18 These successes suggested investigated the capability of E-SERS to measure the NA bases and RNA The E-SERS measurements also provided a benchmark to compare and evaluate SG-SERS measurements The molecular structure of adenine (as well as the other base pairs) which includes an aromatic nitrogen-containing heterocycle is ideally suited to interact with the surface plasmons and contribute substantially to the chemical component of the SER effect1119 Even with excitation at 1064 nm a 3-minute scan of 18x10-5M adenine yields high signal-to-noise (SN) E-SER spectra and all of the bands are revealed with clarity (Figure 1 Table 1) Spectra of similar quality were obtained by SG-SERS and the principal spectral bands are easily observed The identical 18x10-5M adenine sample was measured in the same 3-minute time frame but with 110th the laser power The lower power appears to reduce the SN The amount of adenine responsible for the SER spectra as well as enhancement factors for the two SER media can be determined The molecules producing the E-SERS spectrum are those on the electrode surface within the illumination area of the laser (The solution concentration only determines the number of molecules available to adsorb to the electrode surface) For the current experiments the laser illuminates an area of 28x10-7m2 or 56 x10-7m2 if we assume the ORCs increase the surface area by a factor of two Furthermore if we assume monolayer coverage on the electrode and each 3x5 angstrom molecule (lying flat) occupies 15x10-19m2 then there are ca 4x1012 molecules contributing to the Raman scattering This is ca twice the number of molecules measured at electrode surfaces using either differential capacitance-potential curve measurements or rapid linear sweep voltammetry (eg 3x1018 moleculesm2 for pyridine and pyrazine)29 Thus the adenine spectrum in Figure 1 is due to 87x10-10g (6x10-12 moles) A detection limit defined as a SN of 3 can also be calculated The SN for a 3-minute scan is 844 for the 735 cm-1 band suggesting a mass detection limit of 3x10-12g (2x10-14 moles) This is consistent with previous estimates for adenine by others of 25 x10-14 moles1530 However sub-monolayer concentrations must be measured to verify this The root-mean-squared (RMS) noise is measured between 4400-4600 cm-1 Since noise is distributed evenly throughout the spectrum when transformed this region does not have any
Wavenumbers (∆cm-1) Figure 1 A) Normal Raman spectrum of pure adenine powder B) E-SERS and C) SG-SERS of 18x10-5M adenine at pH 92 All spectra 8 cm-1 resolution 200 scans (3 min) and 1064 nm excitation A) and B) 750 mW C) 75 mW B) measurement potential of 11VSCE
A
B
C
A
B
C SG-SERS
pH 10
D
725
735
735
pH 4
Wavenumbers (∆cm-1) Figure 2 A) and C) E-SERS and B) and D) SG-SERS of adenine at A) and B) pH 104 and C) and D) pH 40 Note consistent appearance of bands at 1270 and 1375 cm-1 as the pH is changed to 10 for both SER media E-SERS used 750 mW SG-SERS used 75 mW of 1064 nm excitation
SPIE 2001-4575
66
contributions from signals or baseline offsets The measurement error is given as SplusmnRMS and for adenine this equals 234 The number of molecules contributing to the SG-SERS are those on the silver particles that are embedded in the sol-gel The total silver surface area can be determined from the average particle size (40 nm diameter) concentration (073 by weight based on molar conc and measured sol-gel density) and the scattering volume (a cylinder defined by the laser area 28x10-7m2 and sol-gel thickness10-4m) The 61x109 silver particles in this volume have a collective area of 31x10-5m2 However it may reasonably be assumed that at least half of the silver surface is in contact with the sol-gel matrix and unavailable for analyte interaction Then approximately 10x1014 molecules or 22x10-8g of adenine contribute to the SG-SER spectrum The slightly lower SN of 207 suggests a mass detection limit of 32x10-10g Determination of the enhancement factors for the two SER media requires estimating the number of adenine molecules contributing to the normal Raman spectrum Here a cylindrical scattering volume is assumed again based on the laser area (28x10-7m2) and the penetration depth (1x10-3 m)41 The density of the sample was measured at 064 gcm3 indicating that 18x10-4g (13x10-6 moles) of adenine produced the normal Raman signal The enhancement factor EF is defined by the following equation
EF = (ISERSINR)bull(MNRMSERS) bull(PNRPSERS) bull(TNRTSERS)12
where I is the spectral band intensity (here 735 cm-1) M is the sample mass P is the incident laser power and T is the measurement time (or number of scans) For the E-SERS measurement the enhancement factor is 22x105 (01780184) bull (18x10-487x10-10)) while the SG-SERS enhancement factor is 10x105 (0160184) bull(18x10-422x10-8) bull(75075) bull(315)12) The lower enhancement for the SG-SERS may be real or the available surface of the silver embedded in the sol-gel may have been overestimated In addition to enhancing the Raman scattering efficiency to an extent similar to E-SERS the SG-SER medium also yields an identical shift of the adenine ring-breathing mode from 725 cm-1 in the normal Raman to 735 cm-1 Furthermore in the course of optimizing the E-SERS sample conditions it was found that pH influenced the adenine interaction with the silver surface (Figure 2) In particular the relative band intensities of the pyrimidine ring skeletal vibrations at 1270 and 1375 cm-1 and the imidazol ring skeletal vibration at1335 cm-1 change At pH 4 adenine is protonated presumably the imidazol ring since the band at 1335 cm-1 increases in intensity while the pyrimidine bands are virtually absent Conversely at pH 10 the imidazol band decreases in intensity while the pyrimidine bands appear It is worth noting that the ring-breathing mode at 735 cm-1 changes little between pH 4 and 10 suggesting that the skeletal changes are more a function of molecule-plasmon interactions than reorientation of the molecule on the surface Measurements of the identical pH series of adenine samples by SG-SERS yielded virtually identical spectral changes This suggests that the sol-gel does not influence the measurement This is critical to reproducing measurements and performing quantitative analysis Next the remaining NA bases were measured by both E-SERS and SG-SERS and compared Previously we examined the optimum pH and electrode potentials for E-SERS measurements to determine if a common pH could be used that yielded good sensitivity for all the bases and if variations in potential could be used to provide an added degree of selectivity between the bases Primarily it was found that high quality spectra were obtained between pH 7 and 95 and that cytosine and uracil were best enhanced at potentials positive of the potential-of-zero charge (pzc ca 065VSCE for Ag) guanine and thymine near the pzc and adenine negative of the pzc In all cases the ring-breathing modes were the most intense and in general could be used to identify the NA bases (Figure 3 Table 1) Specifically adenine has an intense band at 735 cm-1 cytosine at 797 cm-1 guanine at 653 cm-1 thymine at 784 cm-1 and uracil at 800 cm-1 The adenine cytosine guanine and thymine bands are sufficiently separated that their contributions to DNA should be determinable Although adenine and guanine contributions to RNA should also be determinable cytosine and uracil are highly overlapped and unfortunately share the same potential dependence Alternate unique bands at 1183 cm-1 for cytosine and 1275 cm-1 for uracil might be suitable for calculating contributions The SG-SER spectra of the remaining NA bases faithfully reproduced the E-SER spectra In particular the primary identifying bands occur at virtually the same wavenumbers (see Table 1) However the spectra for both cytosine and thymine contain an intense band at ca 1040 cm-1 Initially this was attributed to the pH buffer but samples prepared without either the buffer or the 01M KCl electrolyte yielded identical spectra containing this band In fact the E-SER and SG-SER spectra of thymine are virtually identical except for this band Also the SG-SERS of guanine contains an intense band at 1551 cm-1 that is not observed in the E-SER spectrum This band may be due to a moderately intense band at 1553cm-1 in the normal Raman spectrum that is SG-SER active It was also found that the SG-SERS of cytosine was considerably better than the E-SERS while uracil showed the opposite relationship It is also worth noting that all of the SG-SERS were obtained with 110th the laser power Most importantly the primary ring-breathing modes in the SG-SER spectra are sufficiently intense and unique to be used in determining contributions to DNA and RNA as outlined above
SPIE 2001-4575
67
Table 1 Comparison of E-SER and SG-SER Spectral Band Positions for the NA Bases and Adenosine Monophosphate
Adenine Cytosine Guanine Thymine Uracil AMP E-SERS SG-SERS E-SERS SG-SERS E-SERS SG-SERS E-SERS SG-SERS E-SERS SG-SERS E-SERS SG-SERS1647 1638 1634 1656 1655 1630 1587 1585 1510 1580 1551 1539 1456 1456 1465 1460 1462 1480 1453 1459 1394 1398 1425 1431 1435 1399 1404 1392 1374 1375 1373 1383 1370 1335 1332 1311 1307 1333 1331 1353 1348 1331 1329 1265 1273 1280 1292 1278 1276 1275 1279 1271 1183 1195 1222 1232 1221 1219 1204 1205 1180 1144 1097 1033 1029 1038 1040 1035 1051 1037 1041 1035 963 963 957 1001 1000 961 944 884 819 817 859 866 735 737 797 799 784 782 800 800 72738 742 630 630 653 664 667 684 603 602 590 611 466 561
Bands unique to SG-SERS The next chemical to be analyzed by both E-SERS and SG-SERS was adenosine monophosphate (AMP) The E-SER spectrum yields bands due to the adenine chemical functionality at 727 961 1233 1279 1331 1381 and 1486 cm-1 In addition phosphate bands are observed at 860 1097 1453 1587 and 1705 cm-1 (Figure 5) Other researchers have noted that the ribose component does not appear to contribute to the spectrum19 The AMP spectrum also changes as a function of potential As the electrode is swept more positive (here from -09 to -03VSCE) the phosphate bands at 860 1097 1453 and 1587 cm-1 increase in intensity compared to the adenine bands while a band at 1705 cm-1 appears The adenine bands at 1233 1381 and 1486 cm-1 virtually disappear These potential dependent spectral changes are consistent with earlier studies that show that phosphate is attracted to silver at potentials positive of the pzc but repelled at potentials negative of the pzc19
Wavenumbers (∆cm-1) Figure 3 E-SERS of A) 21x10-3M cytosine at -03VSCE 1000 scans B) ~10x10-5M guanine at -06VSCE 500 scans C) 23x10-3M thymine at -06VSCE 500 scans and D) 12x10-3M uracil at -093VSCE 500 scans All spectra at pH 92 750 mW 1064 nm at 8 cm-1
Wavenumbers (∆cm-1) Figure 4 SG-SERS of A) 21x10-3M cytosine 200 scans B) ~10x10-5M guanine 200 scans C) 23x10-3M thymine 200 scans and D) 12x10-3M uracil 500 scans All spectra at pH 92 75 mW 1064 nm at 8 cm-1
A
B
C
A
B
C
D D
SPIE 2001-4575
68
The SG-SER spectrum of AMP is considerably different The adenine bands virtually disappear except for the two primary bands which shift to742 and 1329 cm-1 While the phosphate band at 1459 cm-1 has gained considerable intensity In addition two new intense bands appear at 684 and 1539 cm-1 as well as a moderately intense band at 1180 cm-1 The SG-SER spectrum has greater similarity to the E-SER spectrum at -03VSCE and suggests that the silver particles embedded in the sol-gel behave as if at a potential positive of the pzc RNA samples extracted from E coli B subtilis and S aureus were next examined by both E-SERS and SG-SERS E-SER spectra of these samples yielded quality spectra in 10 minutes in which all of the major features can be identified (Figures 6 and 7) This includes guanine at 650 cm-1 adenine at 791 cm-1 cytosine and uracil combining at 790 cm-1 and phosphate at 1100 1335 (in combination with adenine and guanine) 1465 and 1570 cm-1 Surprisingly adenine which demonstrated the greatest surface-enhanced Raman effect does not dominate the ring-breathing mode portion of the spectrum The intensities of the other base-pairs bands are of the same order of magnitude This suggests that when the base-pairs are linked together as in RNA they are enhanced in concert In fact the relative intensities are very similar to a normal Raman spectrum of E coli RNA which shows the combined cytosine and uracil band at ca twice the intensity of the adenine band and ca four times the intensity of the guanine band Unfortunately this means that the independent enhancement factors for the NA bases can not be used to estimate relative concentrations For example the relative 791 and 734 cm-1 bands for B subtilis would indicate that the cytosine andor uracil concentration was at least 20 times the adenine concentration whereas each of the four RNA bases are known to contribute 15-35 Nevertheless it is worth noting that the three RNA samples yield different relative band intensities that were reproduced in numerous measurements Although the relative concentrations of the NA bases for these samples have not been determined these differences can be quantified If it is assumed that the 650 cm-1 band represents 25 guanine the 791 cm-1 band represents 25 adenine and the 790 cm-1 50 cytosine plus uracil in the E coli RNA spectrum then the relative concentrations can be estimated for the other RNA samples To aid this calculation the three spectra were normalized to the phosphate band at 1100 cm-1 which has been shown to correlate to the total phosphate concentration and can be used as an internal standard In addition a simple baseline correction was applied (Figure 7) This yields 15 adenine 30 guanine and 55 cytosine plus uracil for B subtilis RNA and 18 adenine 25 guanine and 57 cytosine plus uracil for S aureus RNA The average SN of these measurements was 26 with an average error of 8 of the value (SplusmnN) It is also worth noting that the three RNA spectra show a marked shift in a band near 825 cm-1 This band is assigned to the symmetric stretch of the O-P-O ester linkage9 The band appears at 815 cm-1 for S aureus shifting to 820 cm-1 for B subtilis and 830 cm-1 for E coli Others have used the normal Raman intensity of the band at 815 cm-1 as a direct indication of the amount of A-class helix present while the intensity of the band at 830 cm-1 has been used as a direct indication of the amount of B-class helix present However the latter is more associated with DNA than RNA
Wavenumbers (∆cm-1) Figure 5 E-SER spectra of 020 mgmL adenosine monophosphate at A) -03 and B) -09VSCE and C) SG-SER spectra Conditions sample in 01M KCl buffered to pH 92 A) and B) 750 mW C) 75 mW of 1064 64 scans (1-min) at 8 cm-1
Wavenumbers (∆cm-1) Figure 6 E-SERS of 01 mgmL RNA from E coli 02 mgmL RNA from B subtilis and 02 mgmL RNA from S aureus Conditions 01M KCl pH 92 -03VSCE 750 mW of 1064 nm 640 scans (10 min) at 8 cm-1
AMP RNA
E coli A
B
C
B subtilis
S aureus
SPIE 2001-4575
69
SG-SER spectra of reasonable quality were also obtained for E coli and B subtilis especially the latter (Figure 8) However the spectra differ substantially from the E-SERS of the same samples Both SG-SER spectra are dominated by adenine at 735 cm-1 and a band at 1030 cm-1 Although unassigned the latter does appear in the RNA E-SER spectra Bands at 1105 and 1565 cm-1 are likely due to phosphate while bands at 1320 and 1470 cm-1 are less confidently assigned to phosphate They are significantly less intense and somewhat shifted from their SG-SERS counterparts (1335 and 1455 cm-1) A band at 670 cm-1 may be due to guanine which was observed at 664 cm-1 for SG-SERS of the pure sample However the SG-SER spectrum of AMP also had an intense 667 cm-1 band A number of other bands occur at 890 1070 1165 1245 1290 1420 1505 cm-1 and remain unassigned The SG-SER spectra are somewhat disappointing in that only adenine and guanine contributions can be positively identified This limits the ability to determine relative NA base concentrations and distinguish bacterial RNA However several of the unassigned bands may be due to the bases (eg 1030 and 1420 cm-1 due to cytosine) Further experiments will be required to clarify this point
Ecoli
B subtilis
S aureas
P G
A C+U
OPO
A B
Wavenumbers (∆cm-1) Wavenumbers (∆cm-1) Figure 7 SER spectra of RNA from A) B subtilis with contributions indicated and B) E coli B subtilis and S aureus with baseline correction and peak positions used to calculate contributions indicated G = guanine A = adenine C+U = cytosine plus uracil P = phosphate (backbone) OPO = phosphate ester linkage (A- vs B-class helix)
Wavenumbers (∆cm-1) Wavenumbers (∆cm-1) Figure 9 E-SER (-03VSCE) and SG-SER spectra of RNA from A) E coli and B) B subtilis Sample conditions as in Figure 6 E-SER spectra at 750 mW SG-SERS at 75 mW
A B E-SERS
SG-SERS
E-SERS
SG-SERS
SPIE 2001-4575
70
A final comparison was made between the two SER methods by measuring dipicolinic acid (DPA) This chemical may be invaluable as a test for spore forming bacteria specifically B anthracis 50 to 90 of B anthracis sporilates During spore formation dipicolinic acid is synthesized and once completed 10-15 of the dry spore weight is composed of the Ca2+ complex located in the spore core42 Heating in water can be used to initiate germination at which point the exosporium breaks and releases the Ca dipicolinate which becomes dipicolinic acid in water The structure of this chemical strongly
4 CONCLUSIONS Towards the goal of developing a method and technology to rapidly detect and identify bioagents we have been investigating surface-enhanced Raman spectroscopy as a tool to measure relative concentrations of nucleic acid bases in RNA and determine bioagent taxonomy Initially we investigated E-SERS since this method has been extensively researched and the optimum sample conditions are well developed However this method requires a three-electrode sample cell and electrolyte solution Incorporation of an electrolytic cell into sample systems used to analyze for BWA as aerosols in water or on surfaces can be designed using flow injection analysis technologies but cross-contamination and plugging of sample lines seems inevitable For this reason we also investigated metal-doped sol-gels as a SER-active medium Previous studies have shown this material to be active in all solvents particularly water capable of continuous measurements in flowing systems and reproducible (quantitative) between coated sample vials Here we compared SG-SER spectra to traditional E-SER spectra of the nucleic acid base pairs adenosine monophosphate and RNA High quality spectra of adenine cytosine guanine thymine and uracil were obtained by both E-SERS and SG-SERS Both methods yielded very similar spectra for the NA bases including a pH dependent study of adenine Enhancement factors and detection limits for adenine were determined as 2x105 and 16x10-11g and 1x105 and 12x10-10g for E-SERS and SG-SERS respectively (normalized to 75 mW and 10-min acquisition time) Fifty percent of the silver particle surface area in the sol-gel matrix was assumed covered by adenine which may have been overestimated yielding a lower EF and higher detection limit It should also be realized that each E-SER spectrum required several attempts to optimize the measurement conditions (pH electrode potential etc) While each SG-SER spectrum involved no sample preparation and often represents the first and only attempt to make the measurement Quality spectra of RNA extracted from Escherichia coli Bacillus subtilis and Staphylococcus aureus were obtained by E-SERS that were easily interpreted Bands due to adenine guanine cytosine plus uracil and phosphate were identified The SER band intensity of the NA bases in the RNA samples were of the same order of magnitude suggesting that their interaction with the silver surface is concerted as is their Raman enhancement Interestingly the relative SER band
Wavenumbers (∆cm-1) Figure 10 A) Raman spectrum of solid dipicolinic acid B) Ax20 C) electrolytic SERS of 6x10-3 M dipicolinic acid in 01 M KCl at a potential of +07VSCE and pH of 4 and D) sol-gel SERS of 6x10-3 M dipicolinic acid Conditions for A and C as in Figure 1 C) 100 mW of 1064 nm 50 scans 8 cm-1
A B (Ax20)
C
D
suggested that it would be SER active However the E-SER spectrum was unstable and varied considerably as a function of potential A consistent spectrum was obtained at +06VSCE (Figure 10) This potential is not recommended for measurement because the surface is actively dissolving in solution The SG-SER spectrum was considerably more stable of higher quality and easily reproduced Bands at 660 825 1010 1390 1430 1570 1590 and 3075 cm-1 were observed Enhancement factors were determined for the two media using the symmetric ring stretching mode at 995 and 1010 cm-1 for the normal Raman and SER spectra respectively E-SERS yielded an EF of 5x103 while SG-SERS yielded an EF of 2x105 for DPA The SN of the latter suggests a detection limit of 20x10-10g (based on adenine coverage 75 mW and 10-min) The differences in SER activity for these two media may be attributed to the combined electrolytic potential of the solution chemical and metal15 Again the E-SERS suggests that the SG-SERS is at a potential positive of the pzc While the instability in the E-SERS may also be associated with surface interactions of two carboxylic acid groups of dipicolinic acid during the ORCs
DPA
SPIE 2001-4575
71
intensities for RNA extracted from E coli are very similar to those measured by normal Raman spectroscopy Although the relative percent that each of the NA bases contributed to each RNA sample was not determined reproducible band intensities allowed noting the following trends The percent adenine decreases while the combined percent cytosine and guanine increase for both B subtilis and S aureus compared to E coli Quality spectra were also obtained for the RNA samples by SG-SERS but only a few bands were readily identified Calculations of NA base concentrations by SG-SERS will require further research In light of recent events we summarize the capabilities of these SERS media in terms of rapid detection of B anthracis and dipicolinic acid However these capabilities must be qualified First and foremost the level to which SERS can distinguish bacteria or viruses has not yet been determined Development of a database of both DNA and RNA base concentrations for BWAs and common bacteria to establish the level of taxonomic identification is ongoing Second rapid collection of aerosol water or surface samples is being addressed by others who report trapping particles on filters from 100 liters of air per minute Third although not presented here we have developed methods to extract RNA or DNA from cells and spores for SER analysis within 10 minutes Finally we assume a detection limit of 3600 spores per 100 liters of air is required although a 50 lethal dosage of anthrax has not been established With these qualifications a mass detection limit for RNA using SERS is estimated as follows A single measurement is performed in ca 20 minutes (140 liters collected in 14 min RNA extracted in 8 min spectral acquisition and analysis in 10 min) The average human breaths 7 liters per minute therefore the analyzer must at the very minimum detect 5000 spores in 140 liters of air One spore is approximately 2x10-18m3 (1x1x2 microm) and if a density of 075 gcm3 is assumed this corresponds to a mass of 15x10-12g Each spore contains 4-12 RNA or 12x10-13g RNA for 8 If we assume 23 of the RNA can be isolated for analysis during lysis then the proposed instrument must be able to detect 4x10-10g RNA from 5000 spores per 70 liters of air within 10 minutes As noted above the mass detection limits for adenine were estimated at 16x10-11g and 12x10-10g for E-SERS and SG-SERS respectively Although these detection limits suggests that RNA from 5000 spores is detectable with the current instrumentation it is highly likely that only a portion of an RNA segment (eg 120-nucleotide 5S rRNA) is in contact with the metal surface and will contribute to the SER effect The SN for the RNA spectra were 110th of the average SN for the four individual RNA bases suggesting a 10 contribution Furthermore effective taxonomy will likely require knowing the NA base concentrations to 1 of the value (eg 25plusmn025) Again the average measurement error for the bases is 12 These values suggest that the E-SERS is within a factor of 4 of the required detection limit whereas the SG-SERS detection limit must be improved by 25 times The same arguments can be applied to the detection of dipicolinic acid If we assume a spore releases 10 by weight DPA during germination then the proposed instrument must be able to detect 75x10-10g DPA from 5000 spores per 70 liters of air within 10 minutes The detection limit for SG-SERS was estimated at 20x10-10g and suggest that the vials are suitable to perform a rapid screen for anthrax A series of concentration dependent measurements are currently being performed to verify this assertion Finally we note that the measurements performed here employed an FT-Raman spectrometer This instrumentation was chosen over dispersive Raman instrumentation for the high wavelength accuracy afforded by the HeNe reference laser (Connes Advantage43) which would allow reliable spectral subtraction matching of observed spectra to stored library spectra and confident use of chemometric approaches Such data analysis is likely to be required to enhance BWA identification However this instrumentation sacrifices sensitivity due to a less efficient detector (InGaAs vs Si) less efficient Raman scattering and less efficient generation of plasmon modes Substantial improvements can be obtained using visible excitation and Si detection and these measurements are underway
5 ACKNOWLEDGEMENTS The authors are grateful to Drs D Cookmeyer and S Tove of the US Army Research Office (Contract Number DAAH04-96-C-0078) for their interest and support of this research The authors would also like top acknowledge Dr R Yin and J Jensen for supporting development of the metal-doped sol-gels (Contract Number DAAD13-01-C-0019) They also thank Dr Wilfred H Nelson for assistance in spectral interpretations
SPIE 2001-4575
72
6 REFERENCES 1 Roberts WL and Rainey PM Clin Chem 39 1872-1877 (1993) 2 Pasechnik VA CC Shone and P Hambleton Bioseperations 3 267-283 (1993) 3 Jackson PJ ME Hugh-Jones DM Adair G Green KK Hill CR Kuske LM Grinberg FA Abramova and P
Keim Proc Natl Acad Sci 95 1224-1229 (1998) 4 Jensen JL NC Wong W R Loerop SPIE 4775-03 (2001) 5 Snyder A Pet al SPIE 3853-15 (1999) 6 Danosky T R and McFadden P N in press (1997) 7 Woodruff WH Farquharson S Science 201 831-833 (1978) 8 Lord RC and Thomas GJJr Spectrochemica Acta 23A 2551-2591 (1967) 9 Thomas GJJr Biochim Biophys Acta 213 417-423 (1970) 10 Placzek G Handbuch der Radiologie 2 EMarx ed Akademische Verlagagescellschatt Liepzig 1934 UCRL
Trans No 526 (1959) 11 Jeanmaire DL and RP Van Duyne J Electroanalytical Chem 84 1-20 (1977) 12 Chada S Manoharan R Moenne-Loccoz P Nelson WH Peticolas WL and Sperry JF Applied Spectroscopy
47 38-43 (1993) 13 Manoharan R Ghiamati E Chada S Nelson WH and Sperry JF Applied Spectroscopy 47 2145-2150 (1993) 14 Todd EA Morris MD Applied Spectroscopy 48 545-548 (1994) 15 Wentrup-Byrne E Sarinas S and Fredericks PM Applied Spectroscopy 47 1192-1197 (1993) 16 Weldon MK VR Zhelyaskov and MD Morris Applied Spectroscopy 52 265-269 (1998) 17 Kneipp K and J Fleming J Mol Structure 145 173-179 (1986) 18 Pothier NJ and Force RK Applied Spectroscopy 46 147-151 (1992) 19 Ervin KM E Koglin JM Sequaris P Valenta and HW Nurnberg J Electroanal Chem 114 179-194 (1980) 20 Kim SK TH Joo SW Suh and MS Kim J Raman Spectrosc 17 381-386 (1986) 21 Pothier NJ and Force RK Analytical Chemistry 62 678-680 (1990) 22 Kneipp K Y Wang RR Dasari and MS Feld Applied Spectroscopy 49 780-784 (1995) 23 Nie S and Emory SR Science 275 1102 (1997) 24 Graham D WE Smith AMT Linacre CH Munro ND Watson and PC White Analytical Chemistry 69 4703-
4707 (1997) 25 Fleischmann M PJ Hendra and AJ McQuillan Chem Phys Lett 26 163-166 (1974) 26 Pettinger B J Chemical Phys 85 7442-7451 (1986) 27 Surface-Enhanced Raman Scattering Section Four Theory SPIE MS 10 M Kerker and B Thompson Eds (1990) 28 Wang D-S and Kerker M SPIE (M Kerker and B Thompson Eds) MS 10 417-429 (1990) 29 Weaver MJ Farquharson S Tadayyoni MA J Chem Phys 82 4867-4874 (1985) 30 Norrod KL Sudnik LM Rousell D and Rowlen KL Applied Spectroscopy 51 994-1001 (1997) 31 Farquharson S Weaver WJ Lay PA Magnuson RH and Taube H J Am Chem Soc 105 3350-3351 (1983) 32 Lee PC and Meisel D ldquoAdsorption and Surface-Enhanced Raman of Dyes on Silver and gold Solsrdquo J Phys Chem
86 3391-3395 (1982) 33 Li Y-S and Wang Y Applied Spectroscopy 46 142-146 (1992) 34 Bello JM DL Stokes and T Vo-Dinh Analytical Chemistry 61 1779-1783 (1989) 35 van Duyne RP J C Hulteen D A Treichel M T Smith M L Duval and T R Jensen J Phys Chem B1033854-
3863 (1999) 36 Farquharson S and W W Smith W H Nelson and J F Sperry SPIE 3533-27 207-214 (1998) 37 Lee Y H W Smith S Farquharson H C Kwon M R Shahriari and P M Rainey SPIE 3537 252-260 (1998) 38 Lee Y-H S Farquharson and P M Rainey SPIE 3857 76-84 (1999) 39 Current Protocols in Molecular Biology Wiley Interscience1003-1006 (1987) 40 Farquharson S Smith W Carangelo R C and Brouillette C SPIE 3859 14-23 (1999) 41 Chase D B and JF Rabolt Fourier Transform Raman Spectroscopy Acad Press Ch1 p 131 (1994) 42 Brock TD MT Madigan JM Martinko and J Parker Biology of Microorganisms 7th Ed Prentice Hall p 76-80
(1994) 43 Connes J Rev Opt Theor Instrum 40 45 (1961)
166
Chemical agent identification by surface-enhanced Raman spectroscopy
Stuart Farquharson and Paul Maksymiuk
Real-Time Analyzers 87 Church Street East Hartford CT 06108
Kate Ong and Steven D Christesen US Army SBCCOM Aberdeen Proving Ground MD 21010
ABSTRACT The recent distribution of anthrax through the US postal system and the subsequent infection and death of several postal and national media employees amplifies the need for methods to rapidly detect identify and quantify this and other chemical and biological warfare agents The US military has also identified water supplies as a likely method of warfare agent deployment and is funding the development of a Joint Service Agent Water Monitor (JSAWM) In an effort to aid military personnel and the public at large we are developing a portable analyzer capable of identifying and quantifying chemical agents rapidly either on-demand or continuously The approach is based on the ability of Raman spectroscopy to identify molecular structure through the abundant vibration information provided in spectra and the ability of surface-enhanced Raman spectroscopy (SERS) to detect extremely low concentrations (eg part-per-billion) through the enhancement of Raman scattering by six orders of magnitude or more A key element to the analyzer design is a new SER active medium that is capable of quantitative reversible measurements The medium consists of silver or gold nanoparticles incorporated into a sol-gel matrix The porous silica network offers a unique environment for stabilizing SER active metals and the high surface area increases the interaction between the analyte and metal particles Here we present the use of new sol-gels that also selectively enhance chemicals based on polarity and charge Base-line measurements of chemical agents and their hydrolysis products are presented and compared to the JSAWM goal of 30 micrograms per liter detection Keywords Chemical warfare agent hydrolysis product SERS Raman spectroscopy sol-gel nanoparticle
1 INTRODUCTION Since September 11 2001 the threat of terrorist attacks and biological warfare within US borders has become a sobering reality The simplicity in manufacturing ease of deployment and the relatively low cost of chemical warfare agents (CWAs) raises public concern that they may also be used by terrorists Indeed terrorists released sarin (GB) in the Tokyo subway in 19951 Countering terrorism and terrorist attacks requires recognizing likely deployment scenarios and having the required technology to rapidly detect the deployment event One method of deployment has been long identified by the US military distribution through water supplies To counter this threat the Department of Defense is funding or monitoring the capabilities of 200 technologies with the goal of developing a Joint Service Agent Water Monitor (JSAWM) that is field portable2 The JSAWM must be capable of identifying the chemical agents with no false-positives and quantifying the agents at microgram per liter concentrations within 10 minutes (Table 1)3 This includes the analysis of drinking water supplies distribution and storage systems as well as potable water supplies The earliest technologies developed for CWA detection were based on electrochemical ionization or colorimetric analysis (eg phosgene tape) Although these analyzers were easy to use they were not generally agent specific and suffered from false-positives4 More traditional laboratory methods have also been investigated and in particular combined gas chromatography and mass spectrometry (GCMS) has been very successful at eliminating false-positives56 However GCMS requires extraction repeated calibration and long analysis times (typically 20 to 60 minutes)6 making it labor intensive and less than desirable for field use More rapid analysis of agents in the solid liquid and gas phase has been To whom correspondence should be addressed emailfarqureal-time-analyzerscom
Vibrational Spectroscopy-based Sensor Systems Steven D Christesen Arthur J Sedlacek III EditorsProceedings of SPIE Vol 4557 (2002) copy 2002 SPIE middot0277-786X02$1500
167
Table 1 Chemical Agent Structures Hydrolysis Half-lives and JSAWM Thresholds Agent Short-Hand Chemical Structure Hydrolysis
Half-Life JSAWM
Thresholds
Sarin (GB) F-[O=P-CH3]-O-CH(CH3)2 213 hours 32 microgL Soman (GD) F-[O=P-CH3]-O-CH(CH3)-(C-(CH3)3) 23 hours 32 microgL Tabun (GA) (CH3)2-N-[O=P-CN]-O-C2H5 41 hours 32 microgL VX C2H5O-[O=P-CH3]-S-(CH2)2-N-(CH(CH2)2)2 821 hours 32 microgL EA2192 HO-[O=P-CH3]-S-(CH2)2-N-(CH(CH2)2)2 gt9 years 32 microgL Mustard (H) ClCH2CH2-S-CH2CH2Cl encapsulates 47 microgL Lewisite (L) ClCH=CH-As-Cl2 rapid 27 microgL HCN HCN rapid 20 mgL BZ C7NH12-O-[C=O]-COH(C6H5)2 23 microgL T-2 Toxin 87 microgL at pH 7 to 75 and 20 to 25 oC
demonstrated by vibrational spectroscopy7-10 Hoffland et al7 reported infrared absorbance spectra and absolute Raman cross sections for several chemical agents while Christesen measured Raman cross sections for sarin tabun mustard gas and VX (ethyl S-2-diisopropylamino ethyl methylphosphonothioate)11 Again however these techniques also have limitations Raman spectroscopy is simply not a very sensitive technique and detection limits are typically 01 (1000 ppm) While infrared spectroscopy would have limited value in analyzing poisoned water since the very strong infrared absorption of water would obscure most other chemicals present Nevertheless efforts to overcome these limitations have been demonstrated Braue and Pannella8 quantified the G-series nerve agents (tabun sarin and soman) in terms of infrared attenuated total reflectance using a circle-cell And Alak and Vo-Dinh demonstrated the possibility of surface-enhanced Raman spectroscopy (SERS) to identify CWAs by measuring several organophosphonates that simulate the nerve agents12 However quantitative measurements have not been demonstrated for the SER-active material used (silver coated on alumina particles) or other SER-active media13 Recently we developed silver-doped sol-gels to promote the SER effect14-17 The porous silica network of the sol-gel matrix offers a unique environment for stabilizing SER-active metal particles and the sol-gel provides a high surface area that effectively increases the number of molecules observed within the Raman scattering volume The silver-doped sol-gels have been coated on the internal walls of standard glass vials to allow rapid SER analysis of small sample quantities (lt 01 mL) without preparation We have used p-aminobenzoic acid (PABA) as a test chemical to demonstrate surface enhancements greater than 106 reversible measurements in a flowing system reproducible measurements from vial-to-vial and batch-to-batch and measurements in multiple solvents including water14-17 Recently we used these vials to measure Tabun (GB) and Sarin and several hydrolysis products pinacolyl methyl phosphonate (PMP from Soman) and methyl phosphonic acid (MPA from all G-agents Figure 1) Although a number of unique vibrational bands are observed (eg C-N stretch doublet and P-C stretch) the G-agents were only observed for 5 concentrations and all spectra required baseline corrections Figure 1 Surface enhanced Raman spectra of ~5 vv A) Tabun and B) Sarin C) 1 vv PMP and D) 10 ppm MPA using sol-gel sample vials 785 nm excitation 1-min scan and CCD detection Performed at Aberdeen Proving Ground
C A
Wavenumber (∆cm-1) Wavenumber (∆cm-1)
D B
790
545 C-N
2135 2190 P-C 770
760 1290
Proc SPIE Vol 4577
168
Nevertheless MPA was readily observed for a 10 ppm sample with an estimated detection limit of 04 ppm (based on a signal-to-noise ratio of 3 for the 760 cm-1 band intensity) This measurement provides encouragement in that SERS may satisfy the needs of a JSAWM Furthermore MPA is also a hydrolysis product of VX and V-gas and EA2192 (Figure 2) and may prove a valuable indicator of agent usage Figure 2 Hydrolysis of Sarin to form hydrofluoric acid (HF) methylphosphonic acid 1-methylethyl ester (MPAMME) methyl phosphonic acid (MPA) and 2-propanol With this initial albeit modest success we began analyzing chemicals with various sol-gel compositions that we have been developing Here we describe four sol-gel compositions that select for 1) polar-positive 2) polar-negative 3) weakly polar-positive and 4) weakly polar-negative chemical species The ability of these sol-gels to select and enhance Raman scattering is described for several test chemicals and MPA
2 EXPERIMENTAL The chemicals analyzed as well as all chemicals used to prepare the metal-doped sol-gels were obtained at their purest commercially available grade from Aldrich (Milwaukee WI) The sol-gel designed to select for polar-negative species was prepared from a silver amine complex tetramethyl orthosilicate (TMOS) and methanol After mixing 02 mL of the sol-gel solution was transferred into a glass vial (2 mL) dried and heated The incorporated silver ions were then reduced using dilute sodium borohydride The vials were washed and dried prior to the addition of a sample solution In a similar manner the sol-gel designed to select for polar-positive species was prepared from a gold salt TMOS and methanol The sol-gel designed to select for weakly polar-negative species was prepared from a silver amine complex tetraethyl orthosilicate (TEOS) and methanol And the last sol-gel designed to select for weakly polar-positive species was prepared from a gold salt TEOS and methanol All samples were prepared in a chemical hood and transferred into plain or SER-active vials for analysis Normal Raman spectral measurements employed 1-mL pure samples that were placed in a 1-cm3 cuvette and weighed This yielded a powder density that allowed accurate calculation of molecules in the optical collection field SERS measurements employed 1-mg sample per mL water concentrations unless otherwise stated Once prepared a 01 mL sample was placed into one of the four selective sample vials which in turn was placed into the sample compartment of a Raman spectrometer for analysis A prototype Fourier transform Raman spectrometer (RTA) was used for these measurements18 The system consisted of a NdYAG laser (Brimrose) for excitation at 1064 nm an interferometer built by On-Line Technologies (OLT East Hartford CT) for frequency separation an uncooled InGaAs detector for signal detection (RTA) and an Intel 400 MHz Pentium II based laptop computer (Dell Round Rock TX) for interferometric control data acquisition (OLT) and analysis (LabVIEW by National Instruments Austin TX) Additional components included a Notch filter (Kaiser Ann Arbor MI) and interferometer entrance and exit optics (Edmund Scientific Barrington NJ) Fiber optics were used to deliver the excitation beam to the sample and the scattered radiation to the interferometer (1 meter lengths of 200 and 365 micron core diameter respectively Spectran Avon CT) A second Notch filter (Kaiser) was used as a beam splitter to direct the excitation beam along the same axis as the collected radiation A microscope object (20x04 Newport Irvine CA) was used to focus the beam into the sample and to collect the scattered radiation back along the same axis In this co-axial backscattering arrangement the excitation beam was passed through the outside of a glass vial and focused onto the silver-doped sol-gel film (01-03 mm thickness) containing the sample
3 RESULTS AND DISCUSSION p-aminobenzoic acid (PABA) and phenyl acetylene (PA) and were used to refine the selectivity and SER-activity of the four different metal-doped sol-gels PABA is a popular chemical used to evaluate the performance of SER-active media Here the polar end groups can be used to test selectivity of the polar-negative and polar-positive sol-gels PA is essentially non-
2O+ H HF + +OH
OH3H C
OP
OF
CH
CH3
3
3H C
OCP
OOH
CH
CH3
3
3H C
O
CPHO
CH
CH
3
3
C
Sarin MPAMME MPA 2-propanol
Proc SPIE Vol 4577
169
polar but a high electron density in the cylindrical π cloud around the carbon-carbon triple bond allows testing the selectivity of the weakly polar-negative and weakly polar-positive sol-gels As Figure 3 illustrates PABA passes through the polar sol-gel and is enhanced by either the silver or gold particles At 1 mgml the concentration of neutral PABA is ca 20 times that of the ionized form (pKa = 48) For electropositive silver the PABA anion is expected to interact through the carboxylate group and the associated vibrational modes are expected to dominate the spectrum Conversely for electronegative gold either form of PABA is expected to interact through the amine group The clear differences in our spectra support this expectation Furthermore bands at 840 and 1405 cm-1 assigned to a COO- bend and stretch respectively are significantly more intense for silver than gold Additional bands at 1140 and 1195 cm-1 are assigned to CH bending modes while bands at 1450 1500 and 1605 cm-1 are assigned to ring vibrational modes A very similar SER spectrum for PABA on a silver-coated alumina substrate has previously been reported with similar assignments19 For the gold-doped sol-gel new bands appear at 690 1355 and 1585 cm-1 The first band is assigned to a ring-H bending mode the second band to a ring-N- stretching mode and the third band to a possible NH2 scissors mode or ring mode The second band is not observed in the normal Raman spectrum but infrared bands occur at this frequency for aromatic ring-secondary amine stretching modes The scissors mode occurs at this frequency in Raman spectra for several chemicals but is absent in the PABA Raman spectrum Alternatively this mode may be the1600 cm-1 ring mode that has been shifted by the gold interaction Again a very similar SER spectrum of PABA has been reported but surprisingly using silver (colloids)2021 not gold as the enhancement medium These researchers also assumed the primary interaction of PABA with silver was through the carboxylate anion and made assignments accordingly For example they assigned the 1359 cm-1 to a COO- stretch not to the amine group as we have They also favor the ring stretching mode assignment for the 1582 cm-1 band Finally it should be said that other researchers have argued that the most dominant band in the SER spectra at 1450 cm-1 a ring vibration mode suggests that PABA lies flat on the surface and the π-orbitals dominate the surface interaction22
Figure 3 SER spectra of A) PABA using polar-negative and B) polar-positive sol-gels and C) PA using weakly polar-negative and D) weakly polar-positive sol-gels PABA is 1 mgmL PA is 1 vv Spectral conditions 75 mw 1064 nm 100 scans (15 min) 8 cm-1 resolution Non-polar PA passes through the non-polar sol-gels and is also enhanced by both metals The spectra are easily understood For electropositive silver PA interacts through the cylindrical triple bond π electron cloud and a -CequivC- doublet occurs near 2000 cm-1 The interaction is reasonably strong since this band appears at 2112 cm-1 in the normal Raman spectrum For electronegative gold this interaction is unlikely and only very weak bands occur near 2000 cm-1 The remaining bands are at 1000 cm-1 1200 cm-1 doublet and 1595 cm-1 all appear in the normal Raman spectra at virtually the same frequencies and are assigned to the symmetric ring-breathing mode CH bending modes and the trigonal ring-breathing mode respectively The polarnon-polar selectivity of the polar-negative and weakly polar-negative sol-gels was tested by adding a 11 molar mixture of PABA and PA The selective enhancement is quite good (Figure 4) The spectrum obtained using the polar sol-gel represents 78 PABA and 22 PA while the spectrum obtained using the weakly polar sol-gel represents 9 PABA and 91 PA The band peak intensities at 2000 cm-1 for PA and 1450 cm-1 for PABA were used for these calculations and are expanded in Figure 4 for clarity
C A
Wavenumber (∆cm-1) Wavenumber (∆cm-1)
2NH COOH
D B
C CH
Proc SPIE Vol 4577
170
Figure 4 SERS of 11 MM of PABA and PA in A) polar-negative and C) weakly polar-negative sol-gels The lower traces compare the pure chemicals B) 1 mgml PABA in polar-negative sol-gel and D) 1 PA in weakly polar-negative sol-gel while the insets magnify the minority species for clarity (x5 in A and x10 in B) Spectral conditions as in Figure 3 Following this development of selective sol-gels that maintained SER activity we measured cyanide and MPA (Figure 5) Not surprisingly the best sensitivity for both hydrolysis products was obtained using the polar-negative sol-gel The interaction of the cyanide anion with the silver surface is sufficient to shift the CequivN stretch observed at 2080 cm-1 in the normal Raman spectrum to 2145 cm-1 in the surface-enhanced Raman spectrum Furthermore the band is substantially broadened This anion has been extensively studied by electrolytic SERS and this shift and broadening have been attributed to the formation of a tetrahedral Ag(CN)3
2- surface structure23 Figure 5 Surface-enhanced (upper traces) and normal Raman spectra (bottom traces) of A) CN- and B) MPA in silver-doped TMOS SERS conditions as in Figure 3 and 1 mgmL Note MPA yields two distinct spectra for neutral (top) and acidic pH (middle) The normal Raman spectra employed pure powders 500 scans and 900 mW of 1064 nm SER measurements of MPA with the polar-negative sol-gel yielded two unique spectral signatures that depended on solution pH (Figure 5) For more neutral solutions the P-C stretch of MPA at 762 cm-1 dominates and the CH2 stretch at 2922 cm-1 appears The SN is sufficiently high that the anti-Stokes Raman shift at -762 cm-1 is observed For deprotonated MPA an oxygen-surface mode appears at 325 cm-1 (as well as its anti-Stokes complement) suggesting a strong interaction This results in substantial enhancement of the P-O-C mode at 1051 cm-1 (upper trace) Others report that this mode dominates the infrared spectra of nerve agents measured in water8 Comparison of the two spectra suggests the following molecule-to-surface orientations The appearance of the oxygen-surface and P-O-C modes in the upper spectrum of Figure 5B indicates that the tetrahedral molecule interacts with the silver surface through the deprotonated oxygen and is oriented end-on The
A B
Wavenumber (∆cm-1)
C A
Wavenumber (∆cm-1) Wavenumber (∆cm-1)
D B
Wavenumber (∆cm-1)
Proc SPIE Vol 4577
171
dominance of the P-C and the CH2 stretches and the disappearance of the P-O-C mode in the upper spectrum suggest the molecule is oriented with the methyl-hydroxyl-hydroxyl face towards the surface However considerably more research must be performed to verify these points Both chemical forms were measured as a function of concentration to determine the onset of monolayer coverage and estimate expected detection limits (Figure 6) Below monolayer coverage the signal to concentration dependence should be linear and the SN of any spectral measurement in this range can be used to predict the detection limit In the spectra presented here the peak height was used as the signal while the noise as root-mean-squared (RMS) was measured between 4400-4600 cm-1 Since noise is distributed evenly throughout the spectrum when transformed this region was used since it does not have any contributions from signals or baseline offsets Figure 6 shows a series of spectra for MPA along with a plot of the 1050 cm-1 band intensity (the noise was identical for this series) as a function of sample concentration A clear discontinuity appears in the vicinity of 01 mgmL (19 ppm) indicating the onset of monolayer coverage A detection limit defined as a SN of 3 was calculated for the 01 and 005gmL samples at 24x10-4 and 25x10-4 gL respectively A more modest detection limit of 101x10-4 gL was obtained using the 760 cm-1 band in the second series of concentration measurements These detection limits correspond to 46 and 190 ppb for a 3-min scan and 75 mW of incident laser power
Figure 6 A) Concentration dependence of MPA SERS measured in silver-doped TMOS) B) Concentrations are 001 005 01 05 1 gL (188 94 188 94 188 ppm) I760 series (bull) and I1050 series (∆)
Enhancement factors for cyanide and methyl phosphonic acid can be determined by estimating the number of molecules contributing to the surface-enhanced and normal Raman spectra The enhancement factor EF can be defined by the following equation
EF = (ISERSINR)bull(MNRMSERS) bull(PNRPSERS) bull(TNRTSERS)12
where I is the spectral band intensity M is the sample mass P is the incident laser power and T is the measurement time (or number of scans) for the two measurements For the normal Raman spectra a cylindrical scattering volume is assumed based on the laser area (28x10-7m2 6x10-4m diameter spot) and the penetration depth (1x10-3 m)24 The density of KCN and MPA as powders were measured at 0572 and 0516 gcm3 indicating that 16x10-4 and 144x10-4 g produced the normal Raman signals in Figure 5 respectively The number of molecules contributing to the SER spectra are those on the silver particles embedded in the sol-gel The total silver surface area can be determined from the average particle size concentration and the scattering volume Previous scanning electron micrographs showed the average silver particle size to be 40 nm in diameter (335x10-23m3)17 The silver concentration is 012M based on the reactant molar concentrations and dilution factors And the scattering volume is 76x10-
11m3 again based on a cylindrical scattering volume defined by a laser area of 28x10-7m2 and a sol-gel thickness of 27x10-
4m This volume contains 123x10-6g of silver equivalent to 35x109 silver particles with a collective surface area of 18x10-
5m2 However it may reasonably be assumed that at least half of the silver surface is in contact with the sol-gel matrix and unavailable for analyte interaction If we assume monolayer coverage and that each CN molecule occupies 15x10-20m2 then approximately 62x1014 molecules or 27x10-8g of CN contribute to the SER spectrum (20x10-19m2 46x1013 molecules 74x10-9g for MPA) Accordingly the EF for cyanide equals 48x104 ((180599) bull(16x10-427x10-8) bull(90075) bull(500100)12) The EF for MPA is considerably higher at 87x106 ((60326) bull(144x10-474x10-9) bull(90075) bull(500200)12)
0
20
40
60
80
100
120
140
0 02 04 06 08 1 12[MPA] (mgmL)
I (76
0)
0
100
200
300
400
500
600
I (10
50)
Wavenumber (∆cm-1)
A B
Proc SPIE Vol 4577
172
4 CONCLUSIONS Here we present for the first time surface-enhanced Raman spectra of Tabun and Sarin obtained using silver-doped sol-gels However the inferior enhancement suggested employing SER-active sol-gels varying in composition to improve both detection limits and selectivity of the target analytes To this end we successfully demonstrated the capabilities of four sol-gels that select for 1) polar-positive 2) polar-negative 3) weakly polar-positive and 4) weakly polar-negative chemical species p-aminobenzoic acid was used to show that silver could be used to attract polar-negative chemicals or functional groups (carboxylate anion) and that gold could be used to attract polar-positive chemicals (amine) while a mixture of p-aminobenzoic acid and phenylacetylene was used to show that tetramethyl orthosilicate preferentially solvates polar chemicals and that tetraethyl orthosilicate preferentially solvates weakly polar chemicals This increased sample control was applied to cyanide and methyl phosphonic acid two hydrolysis products of chemical warfare agents Exceptional results were obtained for methyl phosphonic acid allowing measurement of 1x10-2 gL for a 3-min scan and 75 mW of incident laser power with an estimated detection limit of 245x10-4 gL and an enhancement factor of 87x106 However this detection limit is 76 times less sensitive than required for the JSAWM (32x10-6gL for the G-agents) Finally we note that the measurements performed here employed an FT-Raman spectrometer This instrumentation was chosen over dispersive Raman instrumentation for the high wavelength accuracy afforded by the HeNe reference laser (Connes Advantage25) which would allow reliable spectral subtraction matching of observed spectra to stored library spectra and confident use of chemometric approaches Such data analysis is likely to be required to identify the chemical agents as well as distinguish hydrolysis products However this instrumentation which employs 1064 nm excitation and InGaAs detection sacrifices sensitivity We believe that the measurement sensitivity can be improved by at least two orders of magnitude by using shorter laser excitation wavelengths (eg 532 nm) This would provide more efficient Raman scattering (fourth power dependence on laser excitation wavelength) more efficient generation of plasmon modes and allow using more efficient detector material (Si vs InGaAs) These modifications are underway
5 ACKNOWLEDGEMENTS The authors would like to thank Dr R Yin and J Jensen of the US Army for supporting this work (Contract Number DAAD13-01-C-0019) They would also like to thank Advanced Fuel Research for making their laboratory facilities available
6 REFERENCES 1 Tu Anthony ldquoOverview of Sarin Terrorist Incidents in Japan in 1994 and 1995rdquo 6th CBW Protection Symposium
Stockholm Sweden 10-15 May 1998 2 Jensen JL NC Wong W R Loerop SPIE 4775-03 (2001) 3 JSAWM Requirements at wwwsbccomapgeaarmymilRDAecbcrtPRODSERJSAWMjsawmhtml 4 ldquoThe Chemical Weapons Convention Redefines Analytical Challengerdquo Analytical Chemistry News amp Features June 1
397A (1998) 5 Johnston RL Hoefler CM Fargo JC and Moberley B ldquoThe Defense Nuclear Agencyrsquos ChemicalBiochemical
Weapons Agreements Verification Technology Research Development Test and Evaluation Program and its Requirements for On-Site Analysisrdquo AT-ONSITE 5-8 (1994)
6 Black RM Clarke RJ Read RW and Reid MT ldquoApplication of gas chromatography-mass spectrometry and gas chromatography-tandem mass spectrometry to the analysis of chemical warfare samples found to contain residues of the nerve agent sarin sulphur mustard and their degradation productsrdquo J Chromatography 662 301-321 (1994)
7 Hoffland LD Piffath RJ Bouck JBrdquoSpectral signatures of chemical agents and simulantsrdquo Optical Engineering 24 982-984 (1985)
8 Braue EHJ Pannella MGrdquoCIRCLE CELL FT-IR Analysis of Chemical Warfare Agents in Aqueous Solutionsrdquo Applied Spectroscopy 44 1513-1520 (1990)
9 Tseng C-H Mann CK and Vickers TJ ldquoDetermination of Organics on Metal Surfaces by Raman Spectroscopyrdquo Applied Spectroscopy 47 1767-1771 (1993)
10 Staff Reporter ldquoUS Army Tests Lidar to Detect Biological Toxinsrdquo Photonic Spectra pg 50 December 1998
Proc SPIE Vol 4577
Proc SPIE Vol 4577
173
11 Christesen SD Raman cross sections of chemical agents and simulants Applied Spectroscopy 42 318-321 (1988) 12 Alak AM and Vo-Dinh T ldquoSERS of Organophosphorous Chemical Agentsrdquo Analytical Chemistry 59 2149-2153
(1987) 13 Norrod KL Sudnik LM Rousell D and Rowlen KL ldquoQuantitative Comparison of Five SERS Substrates
Sensitivity and Detection Limitrdquo Applied Spectroscopy 51 994-1001 (1997) 14 Lee Y and Farquharson S ldquoSERS Sample Vials Based on Sol-Gel Process for Trace Pesticide Analysisrdquo SPIE 4206
140-146 (2000) 15 Farquharson S and Lee Y ldquoTrace Drug Analysis by Surface-Enhanced Raman Spectroscopyrdquo SPIE 4200-16 (2000) 16 Lee Y Farquharson S and Rainey P M Surface-Enhanced Raman Sensor for Trace Chemical Detection in Water
SPIE 3857 76-84 (1999) 17 Lee Y Farquharson S Kwong H and Shahriari M ldquoSurface-Enhanced Raman Sensor for Surface-Enhanced Raman
Spectroscopyrdquo SPIE 3537 252-260 (1998) 18 Farquharson S Smith W Carangelo R C and Brouillette C ldquoIndustrial Raman Providing Easy Immediate Cost
Effective Chemical Analysis Anywhererdquo SPIE 3859 14-23 (1999) 19 Narayanan VA JM Bello JD Stokes and T Vo-Dinh Analusis 19 307-310 (1991) 20 Laserna JJ E L Torres and JD Winefordner Analytica Chemica Acta 469-480 (1987) 21 Torres EL and JD Winefordner Analytical Chemistry 59 1626-1632 (1987) 22 Suh JS DP DiLella M Moskovits J Phys Chem 87 1540-1544 (1983) 23 Benner RE R Dornhaus R Chang and BL Laube Correlations in the Raman spectra of cyanide complexes adsorbed
at silver electrodes with voltammograms Surface Science 101 341 (1980) 24 Chase D B and JF Rabolt Fourier Transform Raman Spectroscopy Acad Press Ch1 p 131 (1994) 25 Connes J Rev Opt Theor Instrum 40 45 (1961)
APPLIED SPECTROSCOPY 351
focusing the 488 nm laser beam 2 mm deep into thebulk of the crystals to avoid contributions from potentialdepletion layers As shown in Fig 1b the dependence ofthe Raman shift on the degree of deuteration is almostperfectly linear and ts very well with D 5 22684cmR1 24526 where D is the degree of deuteration (in )and R is the spectral mean of the PO4 vibration in cm21A linear correlation coef cient of 0998 indicates an ex-cellent linear dependence of the Raman peak shift withdegree of deuteration This result shows that the shift ofthe PO 4 peak is simply caused by the linear increase inatomic mass due to isotope substitution which decreasesthe length of hydrogen-like bonds
This excellent linear dependence allows us to map thepro le of the DH exchange layer at the surface of DKDPcrystals by acquiring Raman spectra and determining theposition of the PO 4 peak for various depths This methodis preferable over other methods such as determining thestrength of the OD vibration directly (eg at 715 cm21)because the position of the most intense peak in the Ra-man spectrum can be measured more precisely than theintensity of some of the weakest peaks in the spectrumThis is demonstrated in Fig 2 where depth-dependentRaman spectra (Fig 2a) and the resulting exchange layerpro les for two DKDP crystals are shown (Fig 2b) Thespectra in Fig 2a were obtained from a depth scan of aDKDP crystal with 75 degree of deuteration in thebulk grown at 45 8C The spectra start out as DKDP with30 deuteration close to the surface and approach thebulk DKDP spectrum within a few micrometers of depthThe fact that the relative degree of deuteration does notextend to 0 D is due to the limited depth resolution ofthe Raman microprobe which averages over 4 mm indepth Figure 2b depicts the resulting DH exchange layerpro les for this and a second crystal grown at 63 8Crespectively Both crystals had the same exposure to am-bient conditions and their main difference is the temper-ature at which they were grown The different exchangelayer pro les indicate that crystals grown at differenttemperatures have differen t proton conductiv ities 13
which leads to a difference in their rate of deuteriumdepletion The parameters controlling this behavior arecurrently the objective of a detailed study the results ofwhich will be reported elsewhere
CONCLUSION
In conclusion we have shown that the shift of the to-tally symmetric PO 4 stretch mode in the Raman spectrumof DKDP crystals scales linearly with degree of deuter-ation This allows us to correlate Raman peak positionsto deuteration levels in these crystals We have presenteda new technique to determine DH diffusion pro les inDKDP frequency conversion crystals based on micro-Ra-man spectroscopy This technique is fast inexpensiveand works under various environmental conditionswhich will allow us to better understand and control deu-terium depletion in DKDP crystals
ACKNOWLEDGMENTS
We would like to thank M Runkel for rst discovering DKDP crack-ing R Floyd for providing DKDP crystals and L Chase and A Burn-ham for their support and helpful discussions This work was performedunder the auspices of the US Department of Energy by the University
of California Lawrence Livermore National Laboratory under contract W-7405-Eng-48
1 J J De Yoreo A K Burnham and P K Whitman Int Mat Rev47 113 (2002)
2 C E Barker R A Sacks B M Van Wonterghern J A Caird JR Murray J H Campbell K Kyle R B Ehrlich and N DNielsen Proc SPIE-Int Soc Opt Eng 2633 501 (1995)
3 T Suratwala paper to be published4 Cleveland Crystals Inc httpwwwclevelandcrystalscom
KDPshtmltable5 E A Popova I T Savatinova and I A Velichko Sov Phys Solid
State 12 1543 (1971)6 I P Kaminow R C C Leite and S P S Porto J Phys Chem
Solids 26 2085 (1965)7 J A Subramony B J Marquardt J W Macklin and B Kahr
Chem Mat 11 1312 (1999)8 H Tanaka M Tokunaga and I Tatsuzaki Solid State Commun
49 153 (1984)9 R J Nelmes G M Meyer and J E Tibballs J Phys C 15 59
(1982)10 M A Yakshin D W Kim Y S Kim Y Y Broslavets O E
Sidoryuk and S Goldstein Laser Physics 7 941 (1997)11 I Takenaga Y Tominaga S Endo and M Kobayashi Solid State
Commun 84 931 (1992)12 C Krenn personal communication13 M Sharon and A K Kalia J Solid State Chem 21 171 (1977)
Rapid Dipicolinic Acid Extractionfrom Bacillus Spores Detectedby Surface-EnhancedRaman Spectroscopy
STUART FARQUHARSON ALAND GIFT PAUL MAKSYMIUK andFRANK E INSCOREReal-Time Analyzers Inc East Hartford Connecticut06108
Index Headings Dipicolinic acid Bacillus spores Anthrax Surface-enhanced Raman spectroscopy
INTRODUCTION
The anxiety caused by the distribution of anthrax en-dospores through the US postal system in October 2001was exacerbated by the long time required for positiveidenti cation of the Bacillus anthracis spores and the un-known extent of their distribution Since that time manymethods capable of rapid eld analysis have been inves-tigated to augment or replace the laboratory method ofgrowing microorganisms in culture media which takesdays to perform12 Prominent among these approachesare polymerase chain reactions (PCR)3 immunoassaysand detection of calcium dipicolinate as a biochemicalsignature PCR employs primers to separate organism-
Received 10 October 2003 accepted 14 November 2003 Author to whom correspondence should be sent
352 Volume 58 Number 3 2004
speci c nucleic acid sequences (eg capsular protein en-coding gene for Bacillus anthracis)4 and polymerases toamplify the segment until it is detectable Recently am-pli cation times have been substantially reduced andcomplete analysis can now be performed in an hour orless Immunoassay methods are also being developed thatuse competitive binding of the bioagent (as an antigen)and its labeled conjugate for a limited number of anti-bodies Although analyses can be performed in under 30minutes a well-de ned anthrax antigen has not yet beenidenti ed5ndash7 and consequently the false-positive rate isunacceptably high8
A number of other methods are being developed witha focus on the detection of calcium dipicolinate (CaDPA)and its derivatives as a B anthracis signature This is sobecause only spore-forming bacteria contain CaDPA andthe most common potentially interfering spores such aspollen and mold spores do not Relatively fast methodshave been developed to chemically extract CaDPA andthen detect it directly by uorescence9 or indirectly byluminescence1011 In the latter case hot dodecylamine(DDA) has been used to extract dipicolinic acid (DPA)and terbium has been utilized to form a highly lumines-cent DPA complex11 Although measurements have beenperformed in as little as ve minutes it was found thatas many as three concentration-dependent complexes canform each with different lifetimes This coupled withthe fact that the Tb31 cation produces the same lumines-cence spectrum makes determinations of low spore con-centrations problematic
It has been long known that Raman spectra of Bacillispores are dominated by bands associated with CaDPA12
and that these spectra may provide a suitable anthrax sig-nature at the genus level13 Since that time considerableimprovements in Raman instrumentation have led to lab-oratory measurements of single Bacilli spores14 and to eld measurements of spores captured from a mail-sort-ing system15 However the single spore measurementsrequired complex instrumentation that is not ruggedwhile the eld measurements required milligrams of sam-ple Furthermore the Raman spectra of both measure-ments contained uorescence contributions that would in-crease uncertainty in quanti cation
In related research we demonstrated that nanogramquantities of DPA could be detected by uorescence-freesurface-enhanced Raman spectroscopy (SERS)16 We alsodemonstrated that microliter volumes of chemicals canbe detected by SERS using metal-doped sol-gel-packedglass capillaries17 Towards the goal of developing a rap-id eld SERS-based anthrax spore detector we havecombined our previous research and we now report thatDPA can be extracted from a 10 mg B cereus spore sam-ple using DDA in 1 minute and can be detected by SERSin an additional 1 minute
EXPERIMENTAL
Dipicolinic acid (26-pyridinedicarboxylic acid DPA)and dodecylamine (DDA) were used as received fromSigma-Aldrich (Milwaukee WI) Lyophilized B cereusspores prepared according to the literature13 were sup-plied by the University of Rhode Island and used as re-ceived Multiple particles approximately 01 mm3 each
were separated and weighed at 5 to 15 mg representing05 to 15 million spores The sample masses were con-sistent with a previous determination of spore density at0081 gmL that indicated a high degree of entrained air
All chemicals used to prepare the silver-doped sol-gelcoated capillaries were also obtained and used as receivedfrom Sigma-Aldrich According to previously publishedprocedures17 two precursor solutions were preparedmixed and then drawn into 1-mm-diameter glass capil-laries The silver amine precursor consisted of a 51 vvratio of 1 N AgNO 3 to 28 NH3OH while the alkoxideprecursor consisted of a 21 vv ratio of methanol to te-tramethyl orthosilicate The alkoxide precursors weremixed with silver amine precursor in an 81 vv ratioApproximately 015 mL was drawn into the capillarycoating a 15-mm length After sol-gel formation the in-corporated silver ions were reduced with dilute sodiumborohydride which was followed by a water wash to re-move residual reducing agent
A 100 mL drop of a 50 mM DDA solution in ethanolpre-heated to 78 8C was added to each of the B cereusparticles to digest the spore coat After 1 minute the re-sultant solution was drawn into a SER-active capillarythat was immediately xed horizontally to an XY posi-tioning stage (Conix Research Spring eld OR) just in-side the focal point of an f 07 aspheric lens The lensfocused the beam into the sample and collected the scat-tered radiation back along the same axis A dichroic lter(Omega Optical Brattleborough VT) was used to re ectthe excitation laser to the lens and pass the Raman scat-tered radiation collected by the lens An f 2 achromatwas used to collimate the laser beam exiting a 200-mm-core-diameter source ber optic while a second f 2 ach-romat was used to focus the scattered radiation into a 365mm ber optic (Spectran Avon CT) A short-pass lterwas placed in the excitation beam path to block the sil-icon Raman scattering generated in the source ber fromre ecting off sampling optics and reaching the detectorA long-pass lter was placed in the collection beam pathto block the sample Rayleigh scattering from reachingthe detector A 785 nm diode laser (Process InstrumentsInc model 785-600 Salt Lake City UT) was used todeliver 100 to 150 mW of power to the sample A Fouriertransform Raman spectrometer (Real-Time Analyzersmodel IRA-785 East Hartford CT) and a silicon photo-avalanche detector (Perkin Elmer model C30902S Stam-ford CT) were used to acquire the SER spectra
RESULTS AND DISCUSSION
As an initial experiment the SER spectrum of 1 gLof DPA in water was measured using the newly devel-oped silver-doped sol-gel-coated capillaries (Fig 1A) Atthis concentration a high signal-to-noise ratio (SN) isobtained in 1 min In fact a reasonable spectrum is ob-tained in the same time frame for 1 mgL (Fig 1B) TheSER spectra are reasonably similar to the normal Raman(NR) spectrum obtained for a saturated solution of DPAin 1 N KOH (Fig 1C) and the following band shifts areobserved (NR to SER) 647 to 657 cm21 817 to 815cm21 998 to 1008 cm21 1384 to 1382 cm21 1434 to1428 cm21 and 1569 to 1567 cm21 Many of these bandshave been previously assigned1213 such as 998 cm21 to
APPLIED SPECTROSCOPY 353
FIG 1 SERS of DPA in water using silver-doped sol-gel-coated glasscapillary for (A) 1 gL and (B) 1 mgL (C ) NR of saturated DPA in 1N KOH in a glass capillary Spectral conditions (A) and (B) 150 mWof 785 nm 1-min acquisition time (C) 450 mW of 785 nm 5-minacquisition time both 8 cm 21 resolution
FIG 2 SERS of DPA extracted from 10 mg B cereus particle using100 mL of 50 mM hot DDA acquired in (A) 1 minute and (B) 2 seconds(C ) Attempted SERS of 50 mM hot DDA in ethanol using silver-dopedsol-gel-coated glass capillary acquired in 1 min Spectral conditions150 mW of 785 nm 8 cm21 resolution
the symmetric ring stretch 1384 cm21 to the OndashCndashOsymmetric stretch 1428 cm21 to the symmetric ring CndashH bend and 1569 cm21 to the asymmetric OndashCndashOstretch
The rst B cereus samples consisted of 2 mg of sporesin 2 mL of 5 mM hot DDA The samples were main-tained at 78 8C for 40 min and while hot approximately10 mL was drawn into a SER-active capillary Since spec-tra of DPA were obtained for these initial samples small-er spore masses higher DDA concentrations and shorterheating periods were examined In due course it wasfound that 10 mg of spores could be digested by 100 mLof 50 mM hot DDA in one minute and detected (Fig2A) In fact the signal was suf ciently intense that it canbe observed in as little as two seconds (Fig 2B) Theamount of DPA that was extracted was estimated to bebetween 5 and 10 mgL by comparing the signal intensityof the 1008 cm21 band to that measured for DPA in waterThis is consistent with previous research that found thatthe majority of the DPA is extracted from spores usingDDA11 and that B cereus spores contain approximately10 DPA by weight18 The SN of 127 for the 1008 cm21
band in the 1-minute SER spectrum suggests a limit ofdetection of approximately 250 ng of B cereus sporesbased on a SN of 3 Finally it should be noted that DDAdid not produce a detectable SER spectrum as shown inFig 2C
CONCLUSION
We have demonstrated that by combining rapid extrac-tion of dipicolinic acid from Bacillus cereus spores withchemical identi cation by surface-enhanced Raman spec-troscopy as little as 10 mg of spores can be detected Infact the entire measurement from the time of adding hotdodecylamine to the spores to the time when the dipi-colinic acid SER spectrum is acquired and analyzedcould be performed in less than two minutes The abilityof this method to distinguish between spore-forming bac-teria such as Bacillus anthracis and non-DPA containingpowders could help prevent costly shutdowns associated
with the appearance of suspicious material or intentionalmailing of common substances as an anthrax hoax Thismethod could also prove useful in detecting the locationof anthrax endospores in mail distribution facilities if an-other veri ed attack should occur
Research continues to fully characterize the surface-enhanced Raman spectroscopy signal intensities as afunction of sample concentration and to explore otherextractants that do not require the use of elevated tem-perature
ACKNOWLEDGMENTS
The authors are grateful for the support of the National Science Foun-dation (DMI-0296116 and DMI-0215819) and the US Army(DAAD13-02-C-0015 Joint Service Agent Water Monitor program)The authors are indebted to Chetan Shende for preparing the sol-gelcapillaries The authors also thank James Gillespie Nicholas Fell andAugustus Fountain for providing important background informationMark Farquharson for laboratory support and Professor Jay Sperry ofthe University of Rhode Island for supplying B cereus spores
1 V A Pasechnik C C Shone and P Hambleton Bioseparations 3267 (1993)
2 P J Jackson M E Hugh-Jones D M Adair G Green K K HillC R Kuske L M Grinberg F A Abramova and P Keim ProcNatl Acad Sci USA 95 1224 (1998)
3 B R Glick and J J Pasternak Molecular Biology Principles andApplications of Recombinant DNA (ASM Press Washington DC1994)
4 C A Bell J R Uhl T L Had eld J C David R F Meyer TF Smith and F R Cockerill III J Clin Microbiol 40 2897(2002)
5 D L Gatto-Menking H Yu J G Bruno M T Goode M Millerand A W Zulich Biosens Bioelectron 10 501 (1995)
6 J J Quinlan and P M Foegeding J Rapid Methods AutomationMicrobiol 6 1 (1998)
7 A A Hindle and E A H Hall Analyst (Cambridge UK) 1241599 (1999)
8 M S Ascher US Department of Health amp Human Services(httpwwwhhsgovophppresentationsAscherdoc)
9 R Nudelman B V Bronk and S Efrima Appl Spectrosc 54 445(2000)
10 D L Rosen C Sharpless and L B McBrown Anal Chem 691082 (1997)
354 Volume 58 Number 3 2004
11 P M Pellegrino N F Fell Jr and J B Gillespie Anal ChimActa 455 167 (2002)
12 W H Woodruff T G Spiro and C Gilvarg Biochem BiophysRes Commun 58 197 (1974)
13 E Ghiamati R S Manoharan W H Nelson and J F SperryAppl Spectrosc 46 357 (1992)
14 A P Esposito C E Talley T Huser C W Hollars C M Schal-dach and S M Lane Appl Spectrosc 57 868 (2003)
15 S Farquharson L Grigely V Khitrov W W Smith J F Sperryand G Fenerty J Raman Spectrosc paper accep ted (2003)
16 S Farquharson W W Smith S Elliott and J F Sperry SPIE-IntSoc Opt Eng 3855 110 (1999)
17 S Farquharson and P Maksymiuk Appl Spectrosc 57 479(2003)
18 F W Janssen A J Lund and L E Anderson Science (Washing-ton DC) 127 26 (1958)
SPIE -2003-5269 117
pH dependence of methyl phosphonic acid dipicolinic acid and cyanide by surface-enhanced Raman spectroscopy
Stuart Farquharson Alan Gift Paul Maksymiuk Frank Inscore and Wayne Smith
Real-Time Analyzers 87 Church Street East Hartford CT 06108
ABSTRACT US and Coalition forces fighting terrorism in Afghanistan and Iraq must consider a wide range of attack scenarios in addition to car bombings Among these is the intentional poisoning of water supplies to obstruct military operations To counter such attacks the military is developing portable analyzers that can identify and quantify potential chemical agents in water supplies at microgram per liter concentrations within 10 minutes To aid this effort we have been investigating the value of a surface-enhanced Raman spectroscopy based portable analyzer In particular we have been developing silver-doped sol-gels to generate SER spectra of chemical agents and their hydrolysis products Here we present SER spectra of methyl phosphonic acid and cyanide as a function of pH an important factor affecting quantitation measurements which to our knowledge has not been examined In addition dipicolinic acid a chemical signature associated with anthrax-causing spores is also presented Keywords Chemical warfare agents agent detection agent hydrolysis SERS Raman spectroscopy homeland security
1 INTRODUCTION In the past decade the Unites States and its allies have been challenged by a different kind of warfare exemplified by the terrorist attacks of September 11 2001 Suicide bombings and the use of chemical agents are the norm and military personnel must consider a wide range of attack scenarios Among these is the intentional poisoning of water supplies to obstruct military operations in Afghanistan and Iraq Currently colorimetric paper is used to detect agents on-site while gas chromatography combined with mass spectrometry (GCMS) is used in mobile support laboratories However both methods have severe drawbacks The paper changes color in response to contact with many chemicals besides CWAs causing a high incidence of false positives1 while GCMS although very chemically specific requires hours to perform and constant re-calibration234 Military operations would be greatly aided by a portable analyzer that can identify and quantify potential chemical agents at concentrations that impact safety This includes the analysis of drinking water supplies distribution and storage systems To meet this goal the Department of Defense has been investigating numerous approaches under the auspices of the Joint Service Agent Water Monitor (JSAWM) program5 The JSAWM must be capable of identifying the chemical agents with no false-positives and quantifying the agents at microgram per liter concentrations within 10 minutes678 Recently we and others have been investigating the ability of surface-enhanced Raman spectroscopy (SERS) to measure chemical agents9-12 bioagents13-17 and their hydrolysis products in water SERS employs the interaction of surface plasmon modes of metal particles with target analytes to increase scattering efficiency by as much as 1 million times18 In 1987 the potential of this approach to measure CWAs was demonstrated by measuring a series of pesticides19 Several of these organophosphonates have chemical structures similar to CWAs in particular P=O functional groups In our studies we have been employing silver-doped sol-gels to promote the SER effect The porous silica network of the sol-gel matrix offers a unique environment for immobilizing and stabilizing SER-active metal particles20-23 The sol-gels have been coated on the internal walls of standard glass vials to allow rapid SER analysis of small sample quantities without preparation We have measured over 100 chemicals with enhancements of 104 to 106 demonstrated reversible measurements in a flowing system reproducible measurements from vial-to-vial and measurements in multiple solvents including water20-23 Previously we used these vials to perform preliminary measurements of cyanide (CN) methylphosphonic acid (MPA) and dipicolinic acid (DPA) MPA is a hydrolysis product of the nerve agents (eg sarin Reaction 1) and may be a valuable indicator of nerve agent usage particularly since the alkyl methylphosphonic acids are relatively more stable than their corresponding parent complexes24 DPA is
SPIE -2003-5269 118
a chemical signature of spore forming bacteria such as Bacillus anthracis And in light of the inability to rapidly detect the anthrax spores distributed through the US mail in October 2001 a number of methods are being developed to extract and analyze this signature Reaction 1 Stepwise hydrolysis of Sarin to form hydrofluoric acid (HF) isopropyl methylphosphonic acid (IMPA) then methyl phosphonic acid (MPA) and 2-propanol In our previous SERS investigations MPA and DPA were measured at 50 and 100 mgL respectively In both cases limits of detection (LOD) were estimated at 100 microgL providing encouragement in that SERS may satisfy the needs of the JSAWM Since it has been shown that pH can substantially influence the intensity of SER bands25 which would clearly influence quantitative analysis we undertook the present study to determine the severity of these effects for cyanide methyl phosphonic acid and dipicolinic acid Furthermore we previously observed a band at 1050 cm-1 for MPA6 possibly due to an anion formed at basic pH Here we investigate the source of this spectral anomaly
2 EXPERIMENTAL All chemicals including potassium cyanide methyl phosphonic acid dipicolinic acid and those used to prepare the silver-doped and gold-doped sol-gel coated vials were obtained and used as received from Sigma-Aldrich All samples were prepared in HPLC grade water (Fischer Scientific Fair Lawn NJ) for SERS measurements The pH of these samples was adjusted using dilute nitric acid or potassium hydroxide and verified using a pH electrode (Corning Inc Corning NY) that had been calibrated with pH 400 700 and 1000 buffered standards from Fischer Scientific Once prepared the samples were transferred into the silver-doped sol-gel vials (Real-Time Analyzers Inc East Hartford CT) The vials were coated in a manner similar to that previously reported by adding ammonium hydroxide to a solution of silver nitrate tetramethyl orthosilicate (TMOS) and methanol20 Gold-vials were coated by adding nitric acid to a solution of gold tetrachloride TMOS and methanol The two precursor solutions were prepared mixed and transferred to 2-mL glass vials dried and heated After sol-gel formation the incorporated metal ions were reduced with dilute sodium borohydride (1mgmL) which was followed by a water wash to remove residual reducing agent After the resultant analyte solution was introduced the SER-active vial was immediately fixed horizontally to an XY positioning stage (Conix Research Springfield OR) just inside the focal point of an f07 aspheric lens The lens focused the beam into the sample and collected the scattered radiation back along the same axis A dichroic filter (Omega Optical Brattleborough VT) was used to reflect the excitation laser to the lens and pass the Raman scattered radiation collected by the lens An f2 achromat was used to collimate the laser beam exiting a 200 microm core diameter source fiber optic while a second f2 achromat was used to focus the scattered radiation into a 365 microm fiber optic (Spectran Avon CT) A short pass filter was placed in the excitation beam path to block the silicon Raman scattering generated in the source fiber from reflecting off sampling optics and reaching the detector A long pass filter was placed in the collection beam path to block the sample Rayleigh scattering from reaching the detector A 785 nm diode laser (Process Instruments Inc model 785-600 Salt Lake City UT) was used to deliver 100 to 150 mW of power to the sample A Fourier transform Raman spectrometer (Real-Time Analyzers model IRA-785 East Hartford CT) and a silicon photo-avalanche detector (Perkin Elmer model C30902S Stamford CT) were used to acquire the SER spectra
3 RESULTS AND DISCUSSION In a previous study of MPA6 aimed at developing a concentration calibration curve and determining limits of detection (LOD) we observed an anomaly at 1050 cm-1 Since it was found that the band intensity changed as a function of concentration the band must be associated with a sample parameter Two possible parameters photon flux and pH are examined here The first parameter was investigated by irradiating a 1mgmL MPA sample in a SER-active vial with laser powers of 200 mW and above and monitoring spectral changes It was immediately found that the 1050 cm-1 band
2O+ H HF + +OH
OH3H C
OP
OF
CH
CH3
3
3H C
OCP
OOH
CH
CH3
3
3H C
O
CPHO
CH
CH
3
3
C
Sarin IMPA MPA 2-propanol
SPIE -2003-5269 119
appeared and grew as a function of time and that the higher the laser power the faster the growth Figure 1 shows the growth of the 1050 cm-1 band over the course of 30 minutes when using 150 mW of 785 nm excitation while Figure 2A shows that the growth can be fit with a first order exponential equation namely I1050 = 03+05e-013t Furthermore the 760 cm-1 band could be fit with a first order decay equation with an identical exponential rate constant ie I760 = 08-
08e-013t The rates represent classical first order kinetics and their correspondence allows one to conclude that MPA is being transformed one-for-one into a photo-generated product At this time the photoproduct has not been positively identified but phosphonic (phosphorous) acid and phosphonate are likely candidates since the symmetric P(OH)3 stretch occurs at ~1050cm-1 Our studies show that a reduction in laser power to 100 mW at the sample for MPA essentially eliminates this degradation process This laser power or lower was used for further measurements
Methyl phosphonic acid is a diprotic acid that stepwise dissociates into two anions MPA- and MPA= according to the following reactions26
MPA MPA- + H+ pKa1 = 212 Reaction 2
MPA- MPA= + H+ pKa2 = 729 Reaction 3 The relative concentrations of MPA MPA- and MPA= can be determined at any pH by expressing [MPA] and [MPA=] in terms of [MPA-] using Reactions 2 and 3 and summing all three to equal the total starting concentration here 2 mgmL (0021M MW = 9602) viz
[MPA] + [MPA-] + [MPA=] = 0021M Equation 1 substituting from Reactions 1 and 2
([H+][MPA-])K1a + [MPA-] + (K2a[MPA-])[H+] = 0021M Equation 2 rearranging [MPA-] = 0021M(1+[H+]K1a + K2a[H+]) Equation 3 The relative concentrations of MPA MPA- and MPA= as a function of pH are shown in Figure 3 It is worth noting that near neutral pH both MPA- and MPA= will be present To confirm that the SER signal followed this pH dependence a starting solution consisting of 20 mg of MPA in 10 mL HPLC grade water was prepared and brought to pH of 20 using dilute nitric acid From this solution 2 mL were added to a SER-active vial and the SER spectrum recorded At this pH a peak at 760 cm-1 was barely discernable The 2 mL solution was returned to the starting solution and the pH was re-measured to correct for any changes that the silver-doped sol-gel vials might cause In most cases the change was less than 02 pH units and the pH is reported as the before and after average Next the pH of the
Figure 2 A) Exponential growth of 1050 cm-1 band and B) exponential decay of 760 cm-1 band for spectral series in Figure 1
Figure 1 Growth of 1050 cm-1 band as a function of time due to exposure to 150 mW of 785 nm Spectra are 5 sec each collected every 100-sec from 0 to 30-min
B
570 770 970 1170 1370Raman Shift (cm-1)
Arbit
rary
Unit
s
MPA Photodegradation
600 800 1000 1200 1400 Wavenumber (cm-1)
A
0 10 20 30 0 10 20 30 time (min) time (min)
Ram
an In
tens
ity (r
elat
ive)
30
min
0
SPIE -2003-5269 120
starting solution was adjusted to 325 using dilute KOH Again 2 mL were added to a vial and the SER spectrum recorded At this pH a reasonably strong 760 cm-1 band was observed This process was repeated as spectra were recorded at pHs of 70 74 75 79 85 and 100 A total of 1 mL of KOH was added diluting the total concentration by 10 Next the pH of the starting solution was made acidic by adding dilute nitric acid dropwise This time spectra were recorded at pHs of 72 69 64 and 37 Figure 4 shows the SER spectra for representative pHs (spectra were left out to simplify the figure) while Figure 3 shows the 760 cm-1 peak intensities as a function of pH (The band intensities were adjusted to compensate for dilution effects caused by the addition of HNO3 and KOH then normalized to 0021 M for the most intense band observed at pH 37) It is clear from Figure 3 that the 760 cm-1 band follows the MPA- concentration as a function of pH and must be assigned to this anion No bands were observed that corresponded to MPA or MPA= The lack of an MPA SER spectrum may be due to the absence of an attraction between the neutral analyte and the electropositive silver surface The same reasoning suggests that a strong SER spectrum should be observed for MPA= but it is not and a satisfactory explanation has not been found
0000
0005
0010
0015
0020
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14pH
Con
cent
ratio
n [M
]
pK1 = 212 pK2 = 729
MPA- MPA=
MPA
Figure 4 SER spectra of 002M MPA as a function of pH Conditions 100 mW of 785 nm 36 scans (1 min) 8 cm-1 recorded 2 min after sample introduction pH 19 and 20 not apparent on this scale pH 69 and 74 near identical to 70 and 75 and not shown for clarity
00
02
04
06
08
10
12
14
16
18
0 5 10 15 20 25 30Measurement Number
Ram
an In
tens
ity (7
60 c
m-1
)
Figure 3 Concentration dependence of MPA MPA- and MPA= as a function of pH for a 002M sample Intensity of 760 cm-1
band from Figure 6 as a function of pH ( for increasing basic adjustment for increasing acidic adjustment error was measured at ~10 for pH 64)
Figure 5 SER spectra of 002M MPA at pH 64 measured around a vial at three heights (9 points per height) Conditions as in Figure 4 but 10-sec scans
Since these measurements involved the removal and replacement of the SER-active vial in the sample holder to remove and add sample variation in the intensity as a function of vial position was minimized by illuminating the exact same height along the vial wall But this does not account for variability of the SERS response of the sol-gel coating around the vial To analyze this effect a vial containing MPA at pH 64 was rotated at ~ 40o
intervals at the original height and 18rdquo above and below this value Figure 5 shows the intensity of the 760 cm-1
band for the 27 positions It was found that the average value was 137plusmn014 an RSD of 10 overall and 5 for each height An error bar is included in Figure 4 for the pH 64 measurement
SPIE -2003-5269 121
With the above analyses in mind a preliminary investigation of the SER spectral response for MPA (as MPA-) as a function of concentration was performed A single vial was used for these measurements beginning with 1 mgL followed by measurements of 10 100 and 1000 mgL In all cases the pH was ~7 and 3 positions around the vial were measured per concentration Since the 760 cm-1 band was not observed for concentrations of 1 or 10 mgL using 100 mW of 785 nm the laser power at the sample was raised to 200 mW beginning with the 10 mgL concentration Photo-degradation was largely avoided (and not observed) by exposing the sample for only 33 seconds per spectral acquisition Representative spectra for 10 100 and 1000 mgL are shown in Figure 6 while a plot of the 760 cm-1 band intensity as a function of concentration is shown in Figure 7 These values were also used to estimate limits of detection based on the signal-to-noise ratio (SN) of the 760 cm-1 band using 100 mW of 785 nm laser excitation and a 10-min acquisition time As summarized in the Figure 7 inset the lower the measured concentration the lower the predicted LOD Even if as estimated 210 microgL could be measured using the silver-doped sol-gel vials an improvement of a factor of 70 is still required to meet JSAWM goals of measuring 3 microgL in 10-minutes Similar to MPA DPA is a diprotic acid (pKa1 = 216 and pKa2 = 692) and variations in pH will effect the relative concentrations of DPA DPA- and DPA= and possibly the SER spectra and band intensities This could prove significant if an acid or base is used to denature anthrax spores with the goal of extracting and analyzing DPA The MPA pH study described above was mimicked for DPA except that the starting solution consisted of 20 mg of DPA in 20 mL HPLC grade water (60x10-3M MW = 1671) The initial solution had a pH of 245 which was made basic by dilute KOH to pHs of 355 433 487 559 1069 and 1166 SER spectra were recorded at each pH using 100 mW of 785 nm and a 44-sec acquisition time Next one drop of concentrated nitric acid was used to remake the solution acidic at a pH of 200 Again sequential pH measurements were performed at 383 510 735 and 822 The solution pH was made acidic a third time but to pH 219 171 then 135 Throughout this process no more than 20 drops of acid or base were added and therefore the concentration was diluted by no more than 10 Most of the spectral bands showed a minor decrease in intensity as a function of increasing pH values However the bands at 525 and 795 cm-1 showed the most dramatic changes which occurred at acid pH Figure 8 shows the SER spectra of DPA for the spectral region and pH range of interest The identity of the DPA species was determined by plotting the normalized peak intensities with the lowest value set to 0 and the highest to 0006 M as a function of pH and overlaying these values on a plot of the relative concentrations for DPA DPA- and DPA= as previously done for MPA (Figure 9) As can be seen the 525 cm-1 band clearly corresponds to DPA The correspondence of the 795 cm-1 band to this species is less clear as the band retains intensity until pH of 55 This can be attributed to contribution to the overlapping band at 810 cm-1 which does not change as a function of pH The fact that most bands are observed at all pHs suggest that the primary interaction with silver is through the ring nitrogen This is supported by the fact that the most intense band occurs at 1008 cm-1 attributed to a symmetric ring breathing mode and that this interaction has been characterized for pyridine in numerous papers27
0
1
2
3
4
5
6
7
0 200 400 600 800 1000 1200
MPA Concentration (mgL)
760
Ban
d In
tens
ity
Figure 6 SER spectra of MPA in water at A) 1000 B) 100 and C) 10 mgL Conditions pH of 7 silver-doped sol-gel coated vial 200 mW of 785 nm 33-sec 8 cm-1 resolution
Figure 7 Plot of SER intensity of 760 cm-1 band of MPA as a function of concentration using 200 mW of 785 nm Inset table includes average intensity LOD standard deviation and percent deviation for each concentration but for 100 mW and 10 min
conc (mgL) LOD ave stddev dev10 021 002 1021100 072 010 14301000 312 040 1280
A
B C
SPIE -2003-5269 122
The ability to detect dipicolinic acid (DPA) by SERS is demonstrated in Figure 10 Even at 1 mgL the primary bands are visible The intensity of the dominant ring breathing mode observed at 1008 cm-1 for DPA in water (pH = 25 to 55) is plotted as a function of concentration in Figure 11 These values were also used to estimate limits of detection based on the SN of the 1008 cm-1 band using 100 mW of 785 nm laser excitation and a 10-min acquisition time Again the lower the measured concentration the lower the predicted LOD (see Figure 11 inset) and detection of 160 microgL is possible Gastrointestinal anthrax requires significant more spores than inhalation anthrax28 and a limit of detection might be placed at 1 million spores in 1 liter of water or 10 microgL Since each spore contains ~10 CaDPA by weight29 a goal for DPA might be 1 microgL indicating that the present measurements must be improved by nearly two orders of magnitude Due to the increased hazards of handling HCN gas KCN salt was used for these experiments Nevertheless all sample preparations were performed in a chemical hood KCN completely dissolves in water but its conjugate acid HCN is formed and has a Ka of 615x10-1030 viz
HCN CN- + H+ pKa = 921 Reaction 4
Figure 8 SER spectra of 1 mgmL DPA as a function of pH Conditions 100 mW 785 nm 100 scans (44-sec)
Figure 9 Concentration dependence of DPA DPA- and DPA= as a function of pH for a 0006M sample Intensity of 525 () and 795 (diams) cm-1 bands from Fig 8 as a function of pH
pH
135 171 219 383
0
0001
0002
0003
0004
0005
0006
0007
0 2 4 6 8 10 12 14pH
Con
cent
ratio
n [M
]
DPADPA-DPA=795525
DPA DPA=DPA-
pK2 = 692pK1 = 216
795 525
0
05
1
15
2
25
0 200 400 600 800 1000 1200
DPA Concentration (mgL)
1008
Ban
d In
tens
ity
conc (mgL) lod-10min-100mw1 017
10 016100 103
1000 355
Figure 10 SER spectra of DPA in water at A) 1000 B) 100 C) 10 and D) 1 mgL Conditions pH of 25-55 silver-doped sol-gel coated vial 175 mW of 785 nm 1-min 8 cm-1 D) has been multiplied by x10 to make bands visible
A
B C
D
Figure 11 Plot of SER intensity of 1008 cm-1 band of DPA as a function of concentration using 175 mW of 785 nm Inset table includes LOD in mgL for each concentration but for 100 mW and 10 min
SPIE -2003-5269 123
Consequently the cyanide concentration must be determined for each initial KCN concentration Specifically the samples prepared with concentrations of 01 1 10 100 and 1000 mgL of KCN produced CN- concentrations of 63x10-3 033 69 89 and 964 mgL at pHs of 816 90 967 102 and 107 respectively The pH dependence for the HCN and CN- concentrations are shown in Figure 12 Thus as the amount of KCN added to the solution decreases so does the pH of the solution (becomes less basic) and according to Reaction 4 the relative amount of CN- to HCN also decreases For example in the preparation of a 01 mgL solution of KCN the pH is shifted from 7 for pure water to only 816 and only 63 of the starting material becomes CN- or 63x10-3 mgL In comparison for a solution of 1000 mgL the pH is shifted from 7 to 107 and 96 of the starting material becomes CN- This is significant since the cyanide ion is better able to adsorb onto the silver particles and become SERS active SER spectra of 10 100 and 1000 mgL of KCN samples are shown in Figure 13 and a corresponding plot of the 2100 cm-1 band intensity as a function of concentration is shown in Figure 14
000102030405060708091011
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14pH
Con
cent
ratio
n [m
gm
L]
pKa = 921
CN -HCN
The SER spectra of cyanide are dominated by the single CN stretch at 2100 cm-1 which occurs in normal Raman spectra of solutions at 2080 cm-1 However a low frequency mode occurs at 135 cm-1 due to a Ag-CN stretch (not shown) indicative of a strong surface interaction It is also observed that as the concentration decreases the CN stretch shifts to 2140 cm-1 This shift has been attribute to the formation of a tetrahedral Ag(CN)3
2- surface structure31 as well
Figure 14 Concentration dependence of KCN SERS measured under conditions in Fig 11 Concentrations are 1 01 and 001 mgml Intensities are measured for the CN stretch at 2100 cm-1 Inset table includes LOD in mgL for each concentration in Figs 13 and 15 but for 100 mW and 10 min
Figure 13 SER spectra of KCN in water at A) 1000 B) 100 and C) 10 mgL Conditions pHs of 107 102 and 97 silver-doped sol-gel coated vial 100 mW of 785 nm 1-min 8 cm-1 C) has been multiplied by x10 to make band visible
Figure 12 Concentration dependence of HCN and CN- as a function of pH for a 1 mgmL sample Calculated intensity of 2100 () cm-1 band for a 1 mgml sample at pHs of 816 90 967 102 and 107
A
B
C
0
20
40
60
80
100
120
140
0 200 400 600 800 1000 1200CN Concentration (mgL)
2100
Ban
d In
tens
ity
Figure 15 SER spectra of KCN in water at A) 10 B) 1 and C) 01 mgL Conditions pHs of 97 12 and 12 gold-doped sol-gel coated vial 100 mW of 785 nm 1-min 8 cm-1 C) has been multiplied by x10 to make band visible
A
B
C
conc (mgL) lod-10min-100mw Condition01 001 Au-pH 121 007 Au-pH 12
69 003 Au-pH 9769 016 Ag-pH 9789 022 Ag-pH 102946 113 Ag-pH 107
SPIE -2003-5269 124
as to CN adsorbed to two different surface sites32 Alternatively the 2140 cm-1 band could be attributed to HCN since this species dominates at lower concentrations However it is unlikely that this species would be attracted to the electropositive silver surface Further both peaks should be present at pHs between 85 and 105 but this is not observed It has also been suggested that at concentrations near and above monolayer coverage the CN- species is forced to adsorb end-on due to crowding and at lower concentrations the molecule can reorient to lie flat33 This suggests that the 2100 and 2140 cm-1 bands correspond to the end-on and flat orientations respectively As Figures 13 and 14 show the intensity of the CN stretch for the 89 mgL sample is nearly as intense as the 964 mgmL band This suggests that the Raman signal for the flat orientation is more enhanced However more extensive measurements are required to verify this point Since resent research has suggested that cyanide may be more effectively detected on gold measurements of KCN solutions were also performed using gold-doped sol-gel vials Preliminary measurements are shown in Figure 15 for samples prepared from 01 1 and 10 mgL KCN Since the pHs are 816 90 967 the resultant CN- concentrations are 63x10-3 033 and 69 mgL Initially only the highest concentration was observed and the signal intensity was significantly better than the equivalent concentration measured using silver In an effort to shift Reaction 4 to the left transforming HCN to CN- (Le Chatelierrsquos principle) KOH was added to the lower concentration samples producing solutions with pH 12 This effectively forces all of the cyanide in solution to be CN- or 01 and 10 mgL respectively More importantly the CN stretch is now observed in the SER spectra The band appears at 2125 cm-1 as has been previously reported for gold12 As calculated for MPA and DPA LODs can be estimated from this data For the three concentrations of cyanide on silver the LODs are 016 to 11 mgL for 100 mW of 785 nm laser excitation and a 10-min acquisition time For gold pH adjusted the LODs are10 to 70 microgL an improvement of more than 10 times silver Nevertheless either substrate is sufficient to meet the JSAWM goals of measuring 3 mgL in 10-minutes as the requirements form cyanide are much less stringent than the nerve agents
4 CONCLUSIONS Here we examined the surface-enhanced Raman spectral response for methyl phosphonic acid dipicolinic acid and cyanide as a function of pH It was determined that the most prominent peak at 760 cm-1 reaches a maximum intensity between pH 3 and 7 and corresponds to the MPA- species Neither the MPA nor MPA= species appear to generate a SER spectrum and consequently no spectra were observed below pH 2 or above pH 8 In this study we also found that higher laser powers could cause photodegradation of MPA signified by the exponential growth of a band at 1050 cm-1 which is tentatively assigned to phosphorous acid Unlike MPA DPA was observed at all pHs This is attributed to the dominant interaction of the pyridine functional group with silver Minor spectral changes were observed at acid pHs and were assigned to neutral DPA Like MPA SER spectra of cyanide were pH dependent No spectra were observed for the HCN species while CN- was best observed at pHs more basic than 8 Preliminary concentration studies for the three analytes allowed estimating limits of detection for MPA DPA and CN using 100 mW of 785 nm and a 10-min acquisition time of 210 165 and 70 microgL respectively Although the latter value suggests that cyanide can be measured at sufficiently low concentrations to meet JSAWM goal improvements by 100 to 200 times are required for MPA and DPA It is clear from this study that pH of the sample is important and must be taken into account when developing concentration calibration curves as well as fieldable analyzers Future work will include tailoring the sol-gel with specific functional groups to dictate sample pH and thereby optimize sensitivity
ACKNOWLEDGMENTS The authors are grateful for the support of the US Army (DAAD13-02-C-0015 Joint Service Agent Water Monitor program) The authors would also like to thank Dr Steve Christensen of the US Army SBCCOM for helpful discussions and Mr Chetan Shende of Real-Time Analyzers for assistance in development of the gold-doped sol-gels
REFERENCES 1 Erickson B Analytical Chemistry News amp Features June 1 397A (1998)
SPIE -2003-5269 125
2 Johnston RL Hoefler CM Fargo JC and Moberley B AT-ONSITE 5-8 (1994) 3 Black RM RJ Clarke RW Read and MT Reid J Chromatography 662 301-321 (1994) 4 USA Army Center for Health Promotion and Preventive Medicine (USACHPPM) Technical Guideline 230A 5 Jensen JL NC Wong W R Loerop SPIE 4775-03 (2001) 6 Hoenig SL Handbook of Chemical Warfare and Terrorism Greenwood Press Wesport CT (2002) 7 Munro NB SS Talmage GD Griffin LC Waters AP Watson JF King and V Hauschild Env Health
Persp 107 933-974 (1999) 8 Holstege CP Kirk M Sidell FR Crit Care Clin13 923-42 (1997) 9 Farquharson S P Maksymiuk K Ong and S Christesen SPIE 4577 166-173 (2001) 10 Lee Y and S Farquharson SPIE 4378 21-26 (2001) 11 Spencer KM J Sylvia S Clauson and J Janni SPIE 4577 158-165 (2001) 12 Tessier P S Christesen K Ong E Clemente A Lenhoff E Kaler and O Velev Applied Spectroscopy 56
1524-1530 (2002) 13 Farquharson S WW Smith S Elliott and JF Sperry SPIE 3533 207-214 (1998) 14 Farquharson S WW Smith S Elliott and JF Sperry SPIE 3855110-116 (1999) 15 Farquharson S WW Smith YH Lee S Elliott and JF Sperry SPIE 4575 62-72 (2002) 16 Guzelian AA J Sylvia J Janni S Clauson and KM Spencer SPIE 4577 182-192 (2001) 17 Shende C F Inscore A Gift P Maksymiuk and S Farquharson in press 18 Weaver MJ S Farquharson and MA Tadayyoni J Chem Phys 82 4867-4874 (1985) 19 Alak AM and T Vo-Dinh Analytical Chemistry 59 2149-2153 (1987) 20 Lee Y and S Farquharson SPIE 4206 140-146 (2000) 21 Farquharson S and Y Lee SPIE 4200-16 (2000) 22 Lee Y S Farquharson and P M Rainey SPIE 3857 76-84 (1999) 23 Lee Y S Farquharson H Kwong and M Shahriari SPIE 3537 252-260 (1998) 24 Wang J M Pumera G Collins and A Mulchandani Analytical Chemistry 74 6121-6125 (2002) 25 Dou X YM Jung Z-Q Cao and Y Ozaki Applied Spectroscopy 53 1440-1447 (1999) 26 Data supplied by S Christesen and K Ewing 27 Kerker M and B Thompson Eds SPIE MS 10 (1990) 28 Inglesby TV DA Henderson JG Bartlett JAMA 287 2236 (2002) 29 FW Janssen AJ Lund and LE Anderson Science 127 26 (1958) 30 Lide DR Ed Handbook of Chemistry and Physics CRC Press 77th Ed (1996-7) 31 Billmann J G Kovacs and A Otto Surf Sci 92 153 (1980) 32 Murray CA and S Bodoff Phys Rev B 32 671 (1985) 33 Kellogg D and J Pemberton J Phys Chem 91 1120 (1987)
SPIE-2003-5269 16
Chemical agent detection by surface-enhanced Raman spectroscopy
Stuart Farquharson Alan Gift Paul Maksymiuk Frank Inscore and Wayne Smith
Real-Time Analyzers 87 Church Street East Hartford CT 06108
Kevin Morrisey and Steven D Christesen US Army SBCCOM Aberdeen Proving Ground MD 21010
ABSTRACT
In the past decade the Unites States and its allies have been challenged by a different kind of warfare exemplified by the terrorist attacks of September 11 2001 Although suicide bombings are the most often used form of terror military personnel must consider a wide range of attack scenarios Among these is the intentional poisoning of water supplies to obstruct military operations in Afghanistan and Iraq To counter such attacks the military is developing portable analyzers that can identify and quantify potential chemical agents in water supplies at microgram per liter concentrations within 10 minutes To aid this effort we have been investigating the value of a surface-enhanced Raman spectroscopy based portable analyzer In particular we have been developing silver-doped sol-gels to generate SER spectra of chemical agents and their hydrolysis products Here we present SER spectra of several chemical agents measured in a generic tap water Repeat measurements were performed to establish statistical error associated with SERS obtained using the sol-gel coated vials Keywords Chemical agents chemical agent detection SERS Raman spectroscopy
1 INTRODUCTION
In the autumn of 2001 terrorism within US borders became a sobering reality While extensive efforts are being implemented to secure the homeland US and Coalition forces in Afghanistan and Iraq are constantly faced with terrorist attacks In addition to car-bombings the military has identified several non-traditional attack scenarios including poisoning of water supplies by chemical warfare agents (CWAs) To counter this threat the Department of Defense has been investigating numerous approaches to detect such attacks under the auspices of the Joint Service Agent Water Monitor (JSAWM) program1 The JSAWM must be capable of identifying the chemical agents with no false-positives and quantifying the agents at microgram per liter concentrations within 10 minutes2 This includes the analysis of drinking water supplies distribution and storage systems Currently colorimetric paper is used to detect agents on-site while gas chromatography combined with mass spectrometry (GCMS) is used in mobile support laboratories However both methods have severe drawbacks The paper changes color in response to contact with many chemicals besides CWAs causing a high incidence of false positives3 while GCMS although very chemically specific requires up to an hour to perform and regular re-calibration456 Vibrational spectroscopy has also been investigated as a potential method of rapidly detecting CWAs7-
11 as early as 1985 in the case of infrared spectroscopy8 and 1988 in the case of Raman spectroscopy11 Again however these techniques also have limitations when it comes to measuring trace poisons in water Infrared spectra would be dominated by the very strong absorption of water which would obscure absorptions by most other chemicals present Whereas Raman spectroscopy is simply not a very sensitive technique and detection limits are typically grams per liter Surface-enhanced Raman spectroscopy (SERS) offers several advantages over conventional vibrational methods and that may provide the necessary sensitivity required for detecting trace quantities of chemical agents in water SERS employs the interaction of surface plasmon modes of metal particles with target analytes to increase scattering efficiency by as much as 1 million times12 In 1987 the potential of this approach to measure CWAs was demonstrated by measuring a series of pesticides13 Several of these organophosphonates have chemical structures similar to CWAs in particular P=O functional groups In the past few years we and others have further explored the ability of SERS to detect CWAs14-17 and even bioagents 18-21 We have been employing silver-doped sol-gels to promote the SER effect
SPIE-2003-5269 17
in these studies The porous silica network of the sol-gel matrix offers a unique environment for immobilizing and stabilizing SER-active metal particles22-25 The silver-doped sol-gels have been coated on the internal walls of standard glass vials to allow rapid SER analysis of small sample quantities without preparation We have measured over 100 chemicals with enhancements of 104 to 106 demonstrated reversible measurements in a flowing system reproducible measurements from vial-to-vial and measurements in multiple solvents including water21-25 Previously we used these vials to perform preliminary measurements of cyanide (CN) and methylphosphonic acid (MPA) Most of the nerve agents form MPA during hydrolysis while Tabun forms CN a chemical agent in its own right In another paper including in these proceedings we examined the limits of detection (LOD) for MPA by measuring a series of concentrations down to 50 mgL and estimated a limit of detection of 100 microgL26 These measurements provide encouragement in that SERS may satisfy the needs of the JSAWM To further establish the viability of SERS in particular silver-doped sol-gels here we present analysis of cyanide mustard and VX in tap water The measurements performed at the US Armyrsquos Edgewood Chemical Biological Center Aberdeen MD also included numerous repeat measurements to establish reproducibility
2 EXPERIMENTAL 2a General All chemicals including potassium cyanide 2-chloroethylethyl sulfide and those used to prepare the silver-doped and gold-doped sol-gel coated vials were obtained and used as received from Sigma-Aldrich All samples were prepared in a chemical hood using HPLC grade water unless otherwise noted (Fischer Scientific Fair Lawn NJ) for SERS measurements Once prepared the samples were transferred into the silver-doped sol-gel vials (Real-Time Analyzers Inc East Hartford CT) The vials were coated in a manner similar to that previously reported by adding ammonium hydroxide to a solution of silver nitrate tetramethyl orthosilicate (TMOS) and methanol22 The two precursor solutions were prepared mixed and transferred to 2-mL glass vials dried and heated After sol-gel formation the incorporated metal ions were reduced with dilute sodium borohydride (1mgmL) which is followed by a water wash to remove residual reducing agent After the resultant analyte solution was introduced the SER-active vial was fixed horizontally to an XY positioning stage (Conix Research Springfield OR) just inside the focal point of an f07 aspheric lens The lens focused the beam into the sample and collected the scattered radiation back along the same axis A dichroic filter (Omega Optical Brattleborough VT) was used to reflect the excitation laser to the lens and pass the Raman scattered radiation collected by the lens An f2 achromat was used to collimate the laser beam exiting a 200 microm core diameter source fiber optic while a second f2 achromat was used to focus the scattered radiation into a 365 microm fiber optic (Spectran Avon CT) A short pass filter was placed in the excitation beam path to block the silicon Raman scattering generated in the source fiber from reflecting off sampling optics and reaching the detector A long pass filter was placed in the collection beam path to block the sample Rayleigh scattering from reaching the detector A 785 nm diode laser (Process Instruments Inc model 785-600 Salt Lake City UT) was used to deliver 100 to 150 mW of power to the sample A Fourier transform Raman spectrometer (Real-Time Analyzers model IRA-785 East Hartford CT) and a silicon photo-avalanche detector (Perkin Elmer model C30902S Stamford CT) were used to acquire the SER spectra
2b Edgewood Chemical Biological Center The surface-enhanced Raman spectral measurements at the US Armyrsquos Chemical Biological center presented here were all performed on September 12 2003 To expedite measurements a plate was machined to hold up to 12 SER-active sample vials (Figure 1) The plate fit a standard XY plate reader that could be programmed Pure KCN bis-(2-chloroethyl)sulfide (distilled mustard gas HD) and ethyl S-2-diisopropylamino ethyl methylphosphonothioate (VX) were obtained on-site and used to prepare 1 gL tap water solutions in a chemical hood with appropriate safety equipment Simulated tap water was prepared by adding 10 chemicals most often found in tap water at appropriate concentrations to distilled water (Table 1) SER measurements were also performed in a chemical hood For added safety the FT-Raman instrument was placed outside the laboratory and 30 foot fiber optic and electrical cables were used to allow remote SERS measurements and plate manipulation For each experiment 1gL samples were prepared and added to 9 individual vials which were then loaded on the plate In some cases a tenth vial was included as a blank
Table 1 Generic Tap Water Composition
Compound mgL NaHCO3 100 CaSO4 27 MgSO4bull7H2O 67 NaNO3 1 Fulvic Acid 1 K2HPO4 07 KH2PO4 03 (NH4)2HSO4 001 NaCl 001 FeSO4 0001 pH 76-78
SPIE-2003-5269 18
A software program was written that allowed selecting the sequence that the vials were measured the number of positions along the length of the vials to measure (1 to 5) and the number of scans to co-add During sample analysis the program displayed the vial being analyzed the point being analyzed and the spectrum as it was being acquired Once all the data was collected a second software program was written to rapidly analyze the data The spectra collected for all the vials on a plate could be loaded at one time and then the spectra for each point could be displayed simultaneously or separately The user could then select the Raman peak to analyze in terms of peak height or area This was accomplished by selecting points on either side of the peak to define a baseline of zero The peak height or area could then be computed for all of the spectra loaded and then exported to a spreadsheet for statistical analysis
Figure 1 A) Vial Holder 6 slots to hold 2 vials each end-to-end B) Measurement Configuration Program user selects vials to measure sequence number of points per vial (1 to 5) and number of scans per point C) Spectral Acquisition Program shows spectrum being collected which vial and position D) Spectra Analysis Program user selects spectra to analyze by plate vial and point (s) as well as two wavenumbers defining the peak and the baseline to subtract The image is of 5 repeat measurements of 10 mgL KCN in generic tap water 16 sec each 100 mW of 785 nm
3 RESULTS AND DISCUSSION Raman and surface-enhanced Raman spectra were obtained for potassium cyanide bis-(2-chloroethyl)sulfide and ethyl S-2-diisopropylamino ethyl methylphosphonothioate representing three classes of chemical agents cyanides mustards and nerve agents respectively Spectra were also obtained for 2-chloroethyl ethyl sulfide (CEES) a structural analogue to HD which was included in the study to aid in assigning spectral bands KCN salt was used for cyanide experiments to avoid the increased hazards of handling HCN gas KCN completely dissolves in water forming its conjugate acid HCN according to its Ka of 615x10-1027 and at a concentration of 1 mgmL results in a pH 107 solution This is significant in that only CN- appears to interact sufficiently with silver to produce a SER spectrum and no spectral signal is observed below pH 726 Figure 2 shows the SER and normal Raman spectra for KCN The SERS spectrum of 1mgml KCN in water shows a single intense somewhat broad feature at
A B
C D
SPIE-2003-5269 19
2100 cm-1 assigned to the single CequivN stretch The band is much sharper in the normal Raman spectra of the solid KCN salt at 2074 cm-1 This band does broaden and shift to 2080 cm-1 in solution (not shown) However the observed SERS frequency is attributed to interaction with silver and low frequency mode at 135 cm-1 attributed to a Ag-CN stretch (not shown) supports this conclusion
cm-1 band to a C-S stretch but the authors concede that it is in fact more likely a C-Cl stretch28 It appears that the most intense Raman bands at 648 692 and 747 cm-1 shift to 620 660 and 720 cm-1 in the SER spectra and are tentatively assigned as above The width of these bands suggests that they overlap underlying spectral features Additional bands in the Raman spectra occur at 972 1034 1049 1263 1286 1423 1442 2865 2935 and 2960 cm-1 Corresponding bands occur in the SER spectra at 964 1015 1054 1286 1410 1447 2865 and 2935 cm-1 Most of these bands are associated with alkane modes specifically the bands at approximately 1040 cm-1 to a C-C stretch 1290 cm-1 to a CH2 in-phase twist 1440 cm-1 to a CH2 wag 2865 cm-1 to a symmetric CH2 stretch and 2965 cm-1 to an asymmetric CH2 stretch The Raman and SER spectra of sulfur mustard were measured at the Edgewood center (Figure 4) Both spectra are largely similar to CEES The C-Cl and C-S bands in the Raman spectrum of HD now occur at 640 655 700 739 and 760 cm-1 and are more resolved possibly due to the increased molecular symmetry Theoretical calculations indicate that the first three bands are due to C-Cl stretching modes and the latter two to C-S stretching modes28 Only the C-Cl bands maintain significant intensity in the SER spectra occurring at 624 and 643 cm-1 which is attributed to the
Prior to measurements of HD CEES was examined by Raman and SER spectroscopy (Figure 3) CEES also known as half-mustard is essentially identical to HD except one of the chlorine end atoms is replaced by a hydrogen atom Again although not as toxic as HD CEES is a blister agent and dilute aqueous samples were prepared using appropriate safety equipment Both the Raman and SER spectra of CEES are similar and dominated by bands between 600 and 800 cm-1 These are associated with C-Cl and C-S stretching modes which are tentatively assigned to 648 and 747 cm-1 in the Raman spectra respectively The shoulder at 630 cm-1 the overlapped band at 660 cm-1 and the strong band at 692 cm-1 could also be due to these modes or their asymmetric counterparts It is worth noting that theoretical calculations assign the 692
Figure 2 A) SER and B) NR spectra of KCN Conditions A) 1 mgml in tap water 100 mW of 785 nm at sample 1-min acquisition time B) solid 300 mW of 785 nm 5-min All spectra are 8 cm-1 resolution
A
B
Figure 3 A) SER and B) NR spectra of CEES Conditions A) 1 vv (10 mgml) in MeOH 100 mW of 785 nm 1-min acquisition time B) neat 300 mW of 785 nm 5-min
A
B
Cl-CH2-CH2-S-CH2-CH3
A
B
Figure 4 A) SER and B) NR of HD Conditions A) 1mgml in tap water B) pure both 100 mW of 785 nm 1-min
Cl-CH2-CH2-S-CH2-CH2-Cl
SPIE-2003-5269 20
expected strong interaction between chlorine and silver and adds support to the assignment of this band to a C-Cl stretch Weaker overlapping bands occur at 670 692 and 724 cm-1 the latter possibly due to C-S stretching modes Again the alkane modes are apparent in the normal Raman spectra of HD but only a broad feature at 1300 to 1450 cm-1 suggests CH2 contributions in the SER spectrum Although the observed bands in the VX spectrum have not been assigned (Figure 4) a computer generated Raman spectrum29 predicts many of the same features with surprising accuracy and are used here Two intense bands at 460 and 530 cm-1 closely match predicted bands at 463 and 546 cm-1 assigned to a CH3-P=O bend and a PO2CS wag Three highly overlapped bands occur at 694 745 and 771 cm-1 matching predicted bands at 713 730 and 760 cm-1 The first
Table 2 Measured SER peak heights for the CN stretch at 2100 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
Figure 5 A) SERS and B) NR spectra of VX Conditions A) 1 vv (10 mgml) in MeOH B) pure sample both 100 mW of 785 nm 1-min acquisition time
A
B
two have been assigned to a C-S stretch and CH2bend respectively while the latter has been attributed to either a P-C stretch or an O-C-C stretch Although the 745 cm-1 band may alternatively be assigned to a C-S stretch based on the previous measurements of CEES and HD The relatively intense bands at 890 1106 1218 1445 and 1465 cm-1 also match predicted bands at 880 1108 1216 1440 and 1464 cm-1 that are assigned to a C-C stretch CH3 rock N-C3 stretch various C-H3 bends and C-H bends respectively Both the computer generated and the measured spectra contain numerous other less intense bands One is worth mentioning A unique band appears at 370 cm-1 that is predicted at 368 cm-1 and corresponds to an O-P=O bend The surface-enhanced Raman spectrum of VX is also rich with spectral features It has the unique low frequency band at 370 cm-1 as well as a second band at 380 cm-1 that is assigned tothe S-P-O bend predicted in the normal Raman spectrum at 388 cm-1 Based on the measured and predicted normal Raman spectra the following SERS assignments are given 460 cm-1 to the CH3-P=O bend 544 cm-1 to the PO2CS wag 738 cm-1 to a C-S stretch (based on arguments above) 890 cm-1
to a C-C stretch 1101 cm-1 to a CH3 rock and 1456
cm-1 to a C-H bend The ability of SERS to measure chemical agents in water containing real-world chemical interferents was tested by using the generic tap water described in Table 1 The ability to reproduce measurements was accomplished by preparing three separate water stock solutions which were used to prepare three sample solutions each of 1 mgmL KCN in the generic tap water The samples defined as vials 1-9 were then measured at 4 points per vial in a semi-random fashion such that errors associated with stock solution preparation and errors associated with instrument drift could be identified No trends were apparent that signified such systematic errors Each spectrum collected consisted of 20 averaged scans taking 16 seconds at 8 cm-1 resolution The laser power at the sample was measured periodically during the day and it ranged from 102 to 105 mW spacccc
spot Vial 1 Vial 2 Vial 41 13754 13004 128192 13519 12692 12909
1 3 13541 12721 126394 13462 12648 12651
ave 13569 12766 12755Vial 6 Vial 3 Vial 5
1 11586 14068 10757 2 11236 14402 115122 3 11376 14546 11559
4 10894 11763 11214ave 11273 13695 11261
Vial 7 Vial8 Vial 91 11198 15804 11023
2 11243 15716 88753 3 11562 15216 11218
4 11655 15084 11077ave 11415 15455 10548
CN AVG STDEV ERR1 mgmL pk ht 12526 1560 1245
stock solution
SPIE-2003-5269 21
The CN measurements consisting of 36 data points produced an average height of 125 for the 2100 cm-1 peak with a standard deviation of 156 or 125 (Table 2) The HD and VX measurements were performed precisely the same way (Tables 3 and 4) For HD the 624 cm-1 peak was used for analysis and it had an average height of 53 with a standard deviation of 068 or 129 while for VX the 544 cm-1 peak was used for analysis and it had an average height of 1051 with a standard deviation of 308 or 293 The greater error in the VX measurements can be somewhat attributed to Vial 4 which produced lower SER signal intensities But removing this vial from the data set changes the standard deviation to 233 only a modest improvement
spot Vial 1 Vial 2 Vial 41 525 609 4682 45 675 484
1 3 527 644 5094 559 807 669
ave 51525 68375 5325Vial 6 Vial 3 Vial 5
1 521 574 575 2 536 53 4492 3 509 508 451
4 565 594 379ave 53275 5515 4635
Vial 7 Vial 8 Vial 91 457 589 413
2 583 497 4273 3 544 505 4
4 528 5 509ave 528 52275 43725
HD AVG STDEV ERR1 mgmL pk ht 530 068 1291
stock solution
spot Vial 1 Vial 2 Vial 41 1464 1034 4672 1485 99 61
1 3 1491 989 5684 1041 777 553
ave 137025 9475 5495Vial 6 Vial 3 Vial 5
1 1058 942 1293 2 697 121 965
2 3 727 1206 11124 689 1254 746
ave 79275 1153 1029Vial 7 Vial 8 Vial 9
1 1127 783 163 2 1358 812 1614
3 3 1371 908 15124 1097 875 1397
ave 123825 8445 153825
VX AVG STDEV ERR1 mgmL pk ht 1051 308 2925
stock solution
4 CONCLUSIONS In this paper we examined the ability of surface-enhanced Raman spectroscopy to reproducibly measure CN HD and VX in tap water without chemical interference Both normal and surface-enhanced Raman spectra were examined to select unique bands suitable to identify and quantify these chemical agents For SER measurements the 2100 cm-1 C-N stretch was used for CN the 624 cm-1 C-Cl stretch was used for HD and the 544 cm-1 PO2CS wag was used for VX It was determined that 1 mgmL samples of each of these chemicals measured 36 times in glass vials coated with a silver-doped sol-gel reproduced measurements with standard deviations of 125 129 and 293 It was further found that the 10 chemicals added to simulate generic tap water did not interfere with or alter the SER spectra It should be noted that the concentrations used in this study were considerably greater than those required by the JSAWM program Current work involves improving SER sensitivity and designing sampling systems with better reproducibility This includes the development of fractal silver and gold structures within the sol-gel matrix and the development of chemically selective sol-gels
ACKNOWLEDGMENTS The authors are grateful for the support of the US Army (DAAD13-02-C-0015 Joint Service Agent Water Monitor program) The authors would also like to thank Janet Jensen Ronald Crosier and Kristina Gonser for helpful discussions
Table 4 Measured SER peak heights for the VX band at 544 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
Table 3 Measured SER peak heights for the HD band at 624 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
SPIE-2003-5269 22
REFERENCES 1 Jensen JL NC Wong W R Loerop SPIE 4775-03 (2001) 2 JSAWM Requirements at wwwsbccomapgeaarmymilRDAecbcrtPRODSERJSAWMjsawmhtml 3 Erickson B ldquoThe Chemical Weapons Convention Redefines Analytical Challengerdquo Anal Chem 70 397A (1998) 4 Johnston RL Hoefler CM Fargo JC and Moberley B ldquoThe Defense Nuclear Agencyrsquos
ChemicalBiochemical Weapons Agreements Verification Technology Research Development Test and Evaluation Program and its Requirements for On-Site Analysisrdquo AT-ONSITE 5-8 (1994)
5 Black RM Clarke RJ Read RW and Reid MT ldquoApplication of GC-MS and GC-tandem MS to the analysis of chemical warfare samples found to contain residues of the nerve agent sarin sulphur mustard and their degradation productsrdquo J Chrom 662 301-321 (1994)
6 USA Army Center for Health Promotion and Preventive Medicine (USACHPPM) Technical Guideline 230A 7 Hoffland LD Piffath RJ Bouck JB rdquoSpectral signatures of chemical agents and simulantsrdquo Optical
Engineering 24 982-984 (1985) 8 Braue EHJ Pannella MGrdquoCIRCLE CELL FT-IR Analysis of Chemical Warfare Agents in Aqueous Solutionsrdquo
App Spectrosc 44 1513-1520 (1990) 9 Tseng C-H Mann CK and Vickers TJ ldquoDetermination of Organics on Metal Surfaces by Raman
Spectroscopyrdquo App Spectrosc 47 1767-1771 (1993) 10 Staff Reporter ldquoUS Army Tests Lidar to Detect Biological Toxinsrdquo Photonic Spectra pg 50 December 1998 11 Christesen SD Raman cross sections of chemical agents and simulants App Spectrosc 42 318-321 (1988) 12 Weaver MJ Farquharson S Tadayyoni MA ldquoSurface-enhancement factors for Raman scattering at silver electrodesrdquo J Chem Phys 82 4867-4874 (1985) 13 Alak AM and Vo-Dinh T ldquoSERS of Organophosphorous Chemical Agentsrdquo Anal Chem 59 2149-2153 (1987) 14 Lee Y Farquharson S ldquoRapid chemical agent identification by SERSrdquo SPIE 4378 21-26 (2001) 15 Farquharson S Maksymiuk P Ong K Christesen S ldquoChemical agent identification by surface-enhanced Raman spectroscopyrdquo SPIE 4577 166-173 (2001) 16 Spencer KM Sylvia J Clauson S and Janni J ldquoSurface Enhanced Raman as a Water Monitor for Warfare
Agents in Waterrdquo SPIE 4577 158-165 (2001) 17 Tessier P Christesen S Ong K Clemente E Lenhoff A Kaler E Velev O ldquoOn-line spectroscopic characterization of sodium cyanide with nanostructured gold surface-enhanced Raman spectroscopy substratesrdquo
App Spectrosc 56 1524-1530 (2002) 18 Farquharson S W W Smith S Elliott and J F Sperry Rapid biological agent identification by surface- enhanced Raman spectroscopy SPIE 3855110-116 (1999) 19 Farquharson S WW Smith YH Lee S Elliott and J F Sperry Detection of biological signatures A
comparison of electrolytic and metal-doped sol-gel surface-enhanced Raman media SPIE 4575 62-72 (2002) 20 Guzelian AA Sylvia J Janni J Clauson S and Spencer KM ldquoSERS of whole cell bacteria and trace levels of biological moleculesrdquo SPIE 4577 182-192 (2001) 21 Shende C Inscore F Gift A Maksymiuk P Farquharson S ldquoRapid dipicolinic acid extraction from Bacillus spores detected by surface-enhanced Raman spectroscopyrdquo App Spectrosc 58 accepted 22 Lee Y and Farquharson S ldquoSERS Sample Vials Based on Sol-Gel Process for Trace Pesticide Analysisrdquo SPIE
4206 140-146 (2000) 23 Farquharson S and Lee Y ldquoTrace Drug Analysis by SERSrdquo SPIE 4200-16 (2000) 24 Lee Y Farquharson S and Rainey P M Surface-Enhanced Raman Sensor for Trace Chemical Detection in
Water SPIE 3857 76-84 (1999) 25 Lee Y Farquharson S Kwong H and Shahriari M ldquoSol-Gel Chemical Sensor for Surface-Enhanced Raman
Spectroscopyrdquo SPIE 3537 252-260 (1998) 26 Farquharson S A Gift P Maksymiuk F Inscore and W Smith ldquopH dependence of methyl phosphonic acid
dipicolinic acid and cyanide by surface-enhanced Raman spectroscopyrdquo SPIE 5269-19 (2003) 27 Lide DR Ed Handbook of Chemistry and Physics CRC Press 77th Ed (1996-7) 28 Sosa C RJ Bartlett K KuBulat and WB Person ldquoA theoretical study of harmonic vibrational frequencies and
infrared intensities of XCH2CH2SCH2CH2X and XCH2CH2SH (= H Cl)rdquo J Phys Chem 93 577-588 (1993) 29 Hameka HF and JO Jenson ldquoComputer-generated predictions of the structure and the IR and Raman spectra of
VXrdquo ERDEC-TR-065 May 1993
- Chemagents Appendicespdf
-
- SPIE2001-4575-Bioagent-SERSapdf
-
- Detection of bioagent signatures A comparison of electrolytic and metal-doped sol-gel surface-enhanced Raman media
- Stuart Farquharson Wayne Smith and Yuan Lee
-
- Real-Time Analyzers 87 Church Street East Hartford CT 06108
-
- ABSTRACT
- 4 CONCLUSIONS
- 5 ACKNOWLEDGEMENTS
- 6 REFERENCES
-
SPIE 2001-4575
63
Raman spectroscopy has a rich history of investigating biochemical and biological processes7 Some of the earliest laser-Raman studies demonstrated that the five NA bases adenine (A) cytosine (C) guanine (G) thymine (T in DNA) and uracil (U in RNA) yielded distinct spectra with several bands suitable for identification and quantification8 Furthermore these studies included exceptional spectra of both DNA and RNA for which the NA bases as well as several phosphate bands were easily identified9 However since the Raman effect is very inefficient (very low conversion of incident radiation to inelastically scattered Raman radiation) these samples had to be highly concentrated Fortunately two phenomena exist that can increase the generation of Raman photons by six orders of magnitude or more known as the resonance Raman and surface-enhanced Raman effects1011 Resonance Raman scattering occurs when the laser excitation wavelength is in resonance with an electronic transition of a molecule (within the absorption envelope)10 Excitation at ultraviolet wavelengths has been used to obtain resonance Raman spectra of amino acids and nucleic acids in whole bacteria1213 For example excitation at 242 nm has been used to maximize the nucleic acid spectral band intensities and minimize the amino acids band intensities A peak at 1530 cm-1 was found to be proportional to the amount of the NA base cytosine while a peak at 1485 cm-1 was proportional to the combined amount of the NA bases adenine and guanine This quantitative behavior has been used to define an A+TG+C base-pair ratio and provide a level of bacterial identification as taxonomic markers13 In recent years SERS has also been used to analyze bacterial cell components14 including amino acids15 lipids16 nucleic acids151718 and the adenine derivatives192021 SERS has proven to be one of the most sensitive methods for trace chemical analysis through the detection of single molecules2223 including DNA (dye labeled 17-mer)24 Since its discovery in 197425 the mechanism responsible for the large increase in scattering efficiency has been the subject of considerable research2627 Briefly incident laser photons couple to free conducting electrons within a metal which confined by the particle surface collectively cause the electron cloud to resonate2628 These surface plasmons are known as the physical component of the SER effect These surface plasmons can transfer energy to the molecular vibrational modes of molecules through interactions with the molecular electron orbitals2629 This interaction is known as the chemical component of the SER effect This perturbation of the molecular polarizability generates surface-enhanced Raman photons26 A number of methods have been developed to produce surfaces or solutions containing one of these metals with optimum roughness or diameter to promote SERS30 These methods include preparation of activated electrodes in electrolytic cells 1131 activated silver and gold colloid reagents32 and metal coated substrates333435 Selecting a SER-active medium for chemical and biological agent detection requires consideration of the method of deployment and hence the method of sampling Chemical aerosols or airborne bacteria will require a collection device to concentrate and continuously present the sample to the SERS medium Poisoned water supplies will also require a flow through device for continuous monitoring or a grab-sample device for periodic analysis And contaminated surfaces will require a grab-sample extractive device A SERS-based device used for continuous monitoring (air or water) must be reversible and reproducible if quantitative measurements are desired while a SERS-based device used for periodic sampling (water or surfaces) must be reproducible Both reversible and reproducible measurements have been performed using electrolytic SERS (E-SERS)36 But this requires a three-electrode sample cell and an electrolyte of known concentration to perform the necessary oxidation-reduction cycles (ORCs) to re-activate the electrode surface with new uncontaminated sites from one measurement to the next Colloids are severely limited in that continuous measurements would require a continuous supply of colloids For periodic measurements vials of colloids one per measurement could be used However aggregate size and consequently SER intensity change with sample conditions (especially pH) and quantitative reproducible measurements are unlikely Substrates appear to have the greatest potential and designs range from silver evaporated on titania particles34 to periodic gold pyramids evaporated between polystyrene beads35 Most substrates require concentrating the sample on the surface through drying to obtain the largest signal enhancements in effect making the measurements irreproducible and irreversible However successful measurements using flow systems have been obtained with glass posts but manufacturing costs appear prohibitive In an effort to overcome these limitations we have developed metal-doped sol-gels to provide SERS measurements that are reproducible reversible and quantitative and yet not restricted to specific environments such as electrolytes solvents or evaporated surfaces3738 The porous silica network of the sol-gel offers a unique environment for stabilizing SER active metal particles and the high surface area increases the interaction between the analyte and metal particles The sol-gel can be coated on the end of fiber optics or on the internal walls of a glass flow tube for continuous measurements or standard glass sample vials for periodic measurements Previously we measured 100 mgL methylphosphonic acid (the primary hydrolysis product of nerve agents) in water with an estimated detection limit of 05 mgL (100 parts-per-billion) We have also
SPIE 2001-4575
64
demonstrated reversible and reproducible measurements of p-aminobenzoic acid (PABA) in a flow through system Here we investigate the ability of the sol-gel SERS (SG-SERS) to measure the NA bases adenosine monophosphate and RNA extracted from E coli B subtilis and S aureus The measurements are compared to those obtained by E-SERS
2 EXPERIMENTAL The inorganic chemicals and solvents used to prepare samples were spectroscopic grade and purchased from Aldrich (Milwaukee WI) Fisher (Pittsburgh PA) or Pfaltz amp Bauer (Waterbury CT) The nucleic acid bases and dipicolinic acid were purchased from Sigma (St Louis MO) Normal Raman samples were measured to establish enhancement factors In each case 1cm3 of sample was placed into a 1x1 cm glass cuvette weighed and measured Unpacked densities were typically 6-7 gcm3 For all SER measurements including RNA samples were prepared as ~01mgmL (see Figure captions for exact concentrations) in 01M KCl and buffered to a pH of 92 with Na2B4O7bullH2O Adenine pH dependence measurements used pH buffer solutions at 4 (potassium acid phthalate) 69 (potassium phosphate monobasicsodium phosphate dibasic) 92 (Na2B4O7bullH2O) and 104 (tris-hydroxymethyl amino methane) Escherichia coli Bacillus subtilis and Staphylococcus aureus cultures (250ml per 1000mL Erlenmeyer flask) were grown overnight in a Trypticase soy broth (TSB) medium containing 1 glucose in a shaking water bath at 37 oC The bacteria were harvested by centrifugation for 10 minutes at 8000 rpm in a GSA rotor at 5degC then washed once in 085 saline The gram-positive bacteria were concentrated to 20 ml and passed through a French pressure cell twice at 15000 psi to break open the cells RNA was extracted according to Protocol 44139 to ensure pristine samples for initial measurements Since this method takes approximately 4 hours a streamlined method was developed For vegetative bacteria the specimen was boiled for 30 sec in 1 ml of distilled water to lyse the cells and release the RNA For bacterial spores the specimen was first incubated in 1 ml of saline solution containing 02 mg lysozyme and phosphate-buffered to pH of 624 for 1 hr at 37 oC This solution was then boiled for 2-3 minutes in 4X loading buffer to release the RNA For both specimens RNA STAT-60TM was added to the supernatant which was centrifuged at 12000 g for 5 minutes to precipitate the ~15 water-soluble proteins This procedure allowed extracting RNA for SER analysis in ~ 10 minutes Electrophoresis shows high purity while the existence of chemicals that could interfere with the SER measurements is still under investigation The electrolytic sample cell has been described previously36 Briefly a three electrode design is incorporated into a Plexiglas structure containing a well for the reference electrode (a saturated calomel electrode Cole Parmer Vernon Hills IL) and a 5mL sample well containing the silver working electrode and platinum wire counter electrode (05 mm wire Alfa Ward Hill MA) A channel connecting the two wells contained a 2 mm diameter semi-porous membrane (10-20 micron pore Ace Glass) The silver electrode was made from a 3 mm length of 2 mm diameter silver wire (Alfa) soldered to a copper wire lead encased in a 4 mm diameter Pyrex tube A cap containing the silver electrode platinum wire and nitrogen purge and vent lines fixed the silver electrode surface 1 mm from a 1 mm thick glass plate attached to the bottom of the sample well The potentiostat used to control the three electrodes was built in-house and has been described in detail elsewhere36 A multifuntional analog digital and timing inputoutput interface card (DAQCard-1200 National Instruments) is used to both drive the electrolytic cell as well as read the current generated in the cell A LabVIEW software program is used set the oxidation potential reduction potential SER measurement potential hold times and sweep rates The amount of charge passed was plotted as a cyclic voltammogram For all spectra presented five oxidation-reduction cycles (ORCs) stepping from -03 VSCE to 03 VSCE and back to -3 VSCE at 50 mVsec were used The SG-SER measurements were accomplished by simply placing the identical samples prepared above into Simple SERS Sample VialsTM (RTA) These 2-mL glass vials are internally coated with ~ 01 micron thick silver-doped sol-gel A prototype Fourier transform Raman spectrometer (RTA) was used for these measurements40 The system consisted of a NdYAG laser (Brimrose or Spectra Physics) for excitation at 1064 nm an interferometer built by On-Line Technologies (OLT East Hartford CT) for frequency separation an uncooled InGaAs detector for signal detection (RTA) and an Intel 400 MHz Pentium II based laptop computer (Dell Round Rock TX) for interferometric control data acquisition (OLT) and analysis (LabVIEW by National Instruments Austin TX) Additional components included a Notch filter (Kaiser Ann Arbor MI) and interferometer entrance and exit optics (Edmund Scientific Barrington NJ) Fiber optics were used to deliver the excitation beam to the sample and the scattered radiation to the interferometer (2 meter lengths of 200 and 365 micron core diameter respectively Spectran Avon CT) A second Notch filter (Kaiser) was used as a beam splitter to direct the excitation beam along the same axis as the collected radiation A microscope object (20x magnification 04 numeric aperture Newport Irvine CA) was used to focus the beam into the sample and to collect the scattered radiation back along the same axis In this co-axial backscattering arrangement the excitation beam passed through the glass plate onto the silver
SPIE 2001-4575
65
electrode surface for E-SERS through the vial glass wall and into the silver-doped sol-gel film for SG-SERS or through the glass wall of the cuvette and into the solid sample for normal Raman spectroscopy All E-SERS and normal Raman spectra were obtained with 750 mW of laser power at the sample while all SG-SERS spectra were obtained with 75 mW of laser power at the system Incident powers above 200 mW in some cases degraded the sol-gel
3 RESULTS AND DISCUSSION The generation of surface-enhanced Raman scattering at electrode surfaces has been extensively researched and the optimum sample conditions are well developed2729 Several researches incorporated electrodes into flowing systems and demonstrated that continuous monitoring of chemicals is possible18 These successes suggested investigated the capability of E-SERS to measure the NA bases and RNA The E-SERS measurements also provided a benchmark to compare and evaluate SG-SERS measurements The molecular structure of adenine (as well as the other base pairs) which includes an aromatic nitrogen-containing heterocycle is ideally suited to interact with the surface plasmons and contribute substantially to the chemical component of the SER effect1119 Even with excitation at 1064 nm a 3-minute scan of 18x10-5M adenine yields high signal-to-noise (SN) E-SER spectra and all of the bands are revealed with clarity (Figure 1 Table 1) Spectra of similar quality were obtained by SG-SERS and the principal spectral bands are easily observed The identical 18x10-5M adenine sample was measured in the same 3-minute time frame but with 110th the laser power The lower power appears to reduce the SN The amount of adenine responsible for the SER spectra as well as enhancement factors for the two SER media can be determined The molecules producing the E-SERS spectrum are those on the electrode surface within the illumination area of the laser (The solution concentration only determines the number of molecules available to adsorb to the electrode surface) For the current experiments the laser illuminates an area of 28x10-7m2 or 56 x10-7m2 if we assume the ORCs increase the surface area by a factor of two Furthermore if we assume monolayer coverage on the electrode and each 3x5 angstrom molecule (lying flat) occupies 15x10-19m2 then there are ca 4x1012 molecules contributing to the Raman scattering This is ca twice the number of molecules measured at electrode surfaces using either differential capacitance-potential curve measurements or rapid linear sweep voltammetry (eg 3x1018 moleculesm2 for pyridine and pyrazine)29 Thus the adenine spectrum in Figure 1 is due to 87x10-10g (6x10-12 moles) A detection limit defined as a SN of 3 can also be calculated The SN for a 3-minute scan is 844 for the 735 cm-1 band suggesting a mass detection limit of 3x10-12g (2x10-14 moles) This is consistent with previous estimates for adenine by others of 25 x10-14 moles1530 However sub-monolayer concentrations must be measured to verify this The root-mean-squared (RMS) noise is measured between 4400-4600 cm-1 Since noise is distributed evenly throughout the spectrum when transformed this region does not have any
Wavenumbers (∆cm-1) Figure 1 A) Normal Raman spectrum of pure adenine powder B) E-SERS and C) SG-SERS of 18x10-5M adenine at pH 92 All spectra 8 cm-1 resolution 200 scans (3 min) and 1064 nm excitation A) and B) 750 mW C) 75 mW B) measurement potential of 11VSCE
A
B
C
A
B
C SG-SERS
pH 10
D
725
735
735
pH 4
Wavenumbers (∆cm-1) Figure 2 A) and C) E-SERS and B) and D) SG-SERS of adenine at A) and B) pH 104 and C) and D) pH 40 Note consistent appearance of bands at 1270 and 1375 cm-1 as the pH is changed to 10 for both SER media E-SERS used 750 mW SG-SERS used 75 mW of 1064 nm excitation
SPIE 2001-4575
66
contributions from signals or baseline offsets The measurement error is given as SplusmnRMS and for adenine this equals 234 The number of molecules contributing to the SG-SERS are those on the silver particles that are embedded in the sol-gel The total silver surface area can be determined from the average particle size (40 nm diameter) concentration (073 by weight based on molar conc and measured sol-gel density) and the scattering volume (a cylinder defined by the laser area 28x10-7m2 and sol-gel thickness10-4m) The 61x109 silver particles in this volume have a collective area of 31x10-5m2 However it may reasonably be assumed that at least half of the silver surface is in contact with the sol-gel matrix and unavailable for analyte interaction Then approximately 10x1014 molecules or 22x10-8g of adenine contribute to the SG-SER spectrum The slightly lower SN of 207 suggests a mass detection limit of 32x10-10g Determination of the enhancement factors for the two SER media requires estimating the number of adenine molecules contributing to the normal Raman spectrum Here a cylindrical scattering volume is assumed again based on the laser area (28x10-7m2) and the penetration depth (1x10-3 m)41 The density of the sample was measured at 064 gcm3 indicating that 18x10-4g (13x10-6 moles) of adenine produced the normal Raman signal The enhancement factor EF is defined by the following equation
EF = (ISERSINR)bull(MNRMSERS) bull(PNRPSERS) bull(TNRTSERS)12
where I is the spectral band intensity (here 735 cm-1) M is the sample mass P is the incident laser power and T is the measurement time (or number of scans) For the E-SERS measurement the enhancement factor is 22x105 (01780184) bull (18x10-487x10-10)) while the SG-SERS enhancement factor is 10x105 (0160184) bull(18x10-422x10-8) bull(75075) bull(315)12) The lower enhancement for the SG-SERS may be real or the available surface of the silver embedded in the sol-gel may have been overestimated In addition to enhancing the Raman scattering efficiency to an extent similar to E-SERS the SG-SER medium also yields an identical shift of the adenine ring-breathing mode from 725 cm-1 in the normal Raman to 735 cm-1 Furthermore in the course of optimizing the E-SERS sample conditions it was found that pH influenced the adenine interaction with the silver surface (Figure 2) In particular the relative band intensities of the pyrimidine ring skeletal vibrations at 1270 and 1375 cm-1 and the imidazol ring skeletal vibration at1335 cm-1 change At pH 4 adenine is protonated presumably the imidazol ring since the band at 1335 cm-1 increases in intensity while the pyrimidine bands are virtually absent Conversely at pH 10 the imidazol band decreases in intensity while the pyrimidine bands appear It is worth noting that the ring-breathing mode at 735 cm-1 changes little between pH 4 and 10 suggesting that the skeletal changes are more a function of molecule-plasmon interactions than reorientation of the molecule on the surface Measurements of the identical pH series of adenine samples by SG-SERS yielded virtually identical spectral changes This suggests that the sol-gel does not influence the measurement This is critical to reproducing measurements and performing quantitative analysis Next the remaining NA bases were measured by both E-SERS and SG-SERS and compared Previously we examined the optimum pH and electrode potentials for E-SERS measurements to determine if a common pH could be used that yielded good sensitivity for all the bases and if variations in potential could be used to provide an added degree of selectivity between the bases Primarily it was found that high quality spectra were obtained between pH 7 and 95 and that cytosine and uracil were best enhanced at potentials positive of the potential-of-zero charge (pzc ca 065VSCE for Ag) guanine and thymine near the pzc and adenine negative of the pzc In all cases the ring-breathing modes were the most intense and in general could be used to identify the NA bases (Figure 3 Table 1) Specifically adenine has an intense band at 735 cm-1 cytosine at 797 cm-1 guanine at 653 cm-1 thymine at 784 cm-1 and uracil at 800 cm-1 The adenine cytosine guanine and thymine bands are sufficiently separated that their contributions to DNA should be determinable Although adenine and guanine contributions to RNA should also be determinable cytosine and uracil are highly overlapped and unfortunately share the same potential dependence Alternate unique bands at 1183 cm-1 for cytosine and 1275 cm-1 for uracil might be suitable for calculating contributions The SG-SER spectra of the remaining NA bases faithfully reproduced the E-SER spectra In particular the primary identifying bands occur at virtually the same wavenumbers (see Table 1) However the spectra for both cytosine and thymine contain an intense band at ca 1040 cm-1 Initially this was attributed to the pH buffer but samples prepared without either the buffer or the 01M KCl electrolyte yielded identical spectra containing this band In fact the E-SER and SG-SER spectra of thymine are virtually identical except for this band Also the SG-SERS of guanine contains an intense band at 1551 cm-1 that is not observed in the E-SER spectrum This band may be due to a moderately intense band at 1553cm-1 in the normal Raman spectrum that is SG-SER active It was also found that the SG-SERS of cytosine was considerably better than the E-SERS while uracil showed the opposite relationship It is also worth noting that all of the SG-SERS were obtained with 110th the laser power Most importantly the primary ring-breathing modes in the SG-SER spectra are sufficiently intense and unique to be used in determining contributions to DNA and RNA as outlined above
SPIE 2001-4575
67
Table 1 Comparison of E-SER and SG-SER Spectral Band Positions for the NA Bases and Adenosine Monophosphate
Adenine Cytosine Guanine Thymine Uracil AMP E-SERS SG-SERS E-SERS SG-SERS E-SERS SG-SERS E-SERS SG-SERS E-SERS SG-SERS E-SERS SG-SERS1647 1638 1634 1656 1655 1630 1587 1585 1510 1580 1551 1539 1456 1456 1465 1460 1462 1480 1453 1459 1394 1398 1425 1431 1435 1399 1404 1392 1374 1375 1373 1383 1370 1335 1332 1311 1307 1333 1331 1353 1348 1331 1329 1265 1273 1280 1292 1278 1276 1275 1279 1271 1183 1195 1222 1232 1221 1219 1204 1205 1180 1144 1097 1033 1029 1038 1040 1035 1051 1037 1041 1035 963 963 957 1001 1000 961 944 884 819 817 859 866 735 737 797 799 784 782 800 800 72738 742 630 630 653 664 667 684 603 602 590 611 466 561
Bands unique to SG-SERS The next chemical to be analyzed by both E-SERS and SG-SERS was adenosine monophosphate (AMP) The E-SER spectrum yields bands due to the adenine chemical functionality at 727 961 1233 1279 1331 1381 and 1486 cm-1 In addition phosphate bands are observed at 860 1097 1453 1587 and 1705 cm-1 (Figure 5) Other researchers have noted that the ribose component does not appear to contribute to the spectrum19 The AMP spectrum also changes as a function of potential As the electrode is swept more positive (here from -09 to -03VSCE) the phosphate bands at 860 1097 1453 and 1587 cm-1 increase in intensity compared to the adenine bands while a band at 1705 cm-1 appears The adenine bands at 1233 1381 and 1486 cm-1 virtually disappear These potential dependent spectral changes are consistent with earlier studies that show that phosphate is attracted to silver at potentials positive of the pzc but repelled at potentials negative of the pzc19
Wavenumbers (∆cm-1) Figure 3 E-SERS of A) 21x10-3M cytosine at -03VSCE 1000 scans B) ~10x10-5M guanine at -06VSCE 500 scans C) 23x10-3M thymine at -06VSCE 500 scans and D) 12x10-3M uracil at -093VSCE 500 scans All spectra at pH 92 750 mW 1064 nm at 8 cm-1
Wavenumbers (∆cm-1) Figure 4 SG-SERS of A) 21x10-3M cytosine 200 scans B) ~10x10-5M guanine 200 scans C) 23x10-3M thymine 200 scans and D) 12x10-3M uracil 500 scans All spectra at pH 92 75 mW 1064 nm at 8 cm-1
A
B
C
A
B
C
D D
SPIE 2001-4575
68
The SG-SER spectrum of AMP is considerably different The adenine bands virtually disappear except for the two primary bands which shift to742 and 1329 cm-1 While the phosphate band at 1459 cm-1 has gained considerable intensity In addition two new intense bands appear at 684 and 1539 cm-1 as well as a moderately intense band at 1180 cm-1 The SG-SER spectrum has greater similarity to the E-SER spectrum at -03VSCE and suggests that the silver particles embedded in the sol-gel behave as if at a potential positive of the pzc RNA samples extracted from E coli B subtilis and S aureus were next examined by both E-SERS and SG-SERS E-SER spectra of these samples yielded quality spectra in 10 minutes in which all of the major features can be identified (Figures 6 and 7) This includes guanine at 650 cm-1 adenine at 791 cm-1 cytosine and uracil combining at 790 cm-1 and phosphate at 1100 1335 (in combination with adenine and guanine) 1465 and 1570 cm-1 Surprisingly adenine which demonstrated the greatest surface-enhanced Raman effect does not dominate the ring-breathing mode portion of the spectrum The intensities of the other base-pairs bands are of the same order of magnitude This suggests that when the base-pairs are linked together as in RNA they are enhanced in concert In fact the relative intensities are very similar to a normal Raman spectrum of E coli RNA which shows the combined cytosine and uracil band at ca twice the intensity of the adenine band and ca four times the intensity of the guanine band Unfortunately this means that the independent enhancement factors for the NA bases can not be used to estimate relative concentrations For example the relative 791 and 734 cm-1 bands for B subtilis would indicate that the cytosine andor uracil concentration was at least 20 times the adenine concentration whereas each of the four RNA bases are known to contribute 15-35 Nevertheless it is worth noting that the three RNA samples yield different relative band intensities that were reproduced in numerous measurements Although the relative concentrations of the NA bases for these samples have not been determined these differences can be quantified If it is assumed that the 650 cm-1 band represents 25 guanine the 791 cm-1 band represents 25 adenine and the 790 cm-1 50 cytosine plus uracil in the E coli RNA spectrum then the relative concentrations can be estimated for the other RNA samples To aid this calculation the three spectra were normalized to the phosphate band at 1100 cm-1 which has been shown to correlate to the total phosphate concentration and can be used as an internal standard In addition a simple baseline correction was applied (Figure 7) This yields 15 adenine 30 guanine and 55 cytosine plus uracil for B subtilis RNA and 18 adenine 25 guanine and 57 cytosine plus uracil for S aureus RNA The average SN of these measurements was 26 with an average error of 8 of the value (SplusmnN) It is also worth noting that the three RNA spectra show a marked shift in a band near 825 cm-1 This band is assigned to the symmetric stretch of the O-P-O ester linkage9 The band appears at 815 cm-1 for S aureus shifting to 820 cm-1 for B subtilis and 830 cm-1 for E coli Others have used the normal Raman intensity of the band at 815 cm-1 as a direct indication of the amount of A-class helix present while the intensity of the band at 830 cm-1 has been used as a direct indication of the amount of B-class helix present However the latter is more associated with DNA than RNA
Wavenumbers (∆cm-1) Figure 5 E-SER spectra of 020 mgmL adenosine monophosphate at A) -03 and B) -09VSCE and C) SG-SER spectra Conditions sample in 01M KCl buffered to pH 92 A) and B) 750 mW C) 75 mW of 1064 64 scans (1-min) at 8 cm-1
Wavenumbers (∆cm-1) Figure 6 E-SERS of 01 mgmL RNA from E coli 02 mgmL RNA from B subtilis and 02 mgmL RNA from S aureus Conditions 01M KCl pH 92 -03VSCE 750 mW of 1064 nm 640 scans (10 min) at 8 cm-1
AMP RNA
E coli A
B
C
B subtilis
S aureus
SPIE 2001-4575
69
SG-SER spectra of reasonable quality were also obtained for E coli and B subtilis especially the latter (Figure 8) However the spectra differ substantially from the E-SERS of the same samples Both SG-SER spectra are dominated by adenine at 735 cm-1 and a band at 1030 cm-1 Although unassigned the latter does appear in the RNA E-SER spectra Bands at 1105 and 1565 cm-1 are likely due to phosphate while bands at 1320 and 1470 cm-1 are less confidently assigned to phosphate They are significantly less intense and somewhat shifted from their SG-SERS counterparts (1335 and 1455 cm-1) A band at 670 cm-1 may be due to guanine which was observed at 664 cm-1 for SG-SERS of the pure sample However the SG-SER spectrum of AMP also had an intense 667 cm-1 band A number of other bands occur at 890 1070 1165 1245 1290 1420 1505 cm-1 and remain unassigned The SG-SER spectra are somewhat disappointing in that only adenine and guanine contributions can be positively identified This limits the ability to determine relative NA base concentrations and distinguish bacterial RNA However several of the unassigned bands may be due to the bases (eg 1030 and 1420 cm-1 due to cytosine) Further experiments will be required to clarify this point
Ecoli
B subtilis
S aureas
P G
A C+U
OPO
A B
Wavenumbers (∆cm-1) Wavenumbers (∆cm-1) Figure 7 SER spectra of RNA from A) B subtilis with contributions indicated and B) E coli B subtilis and S aureus with baseline correction and peak positions used to calculate contributions indicated G = guanine A = adenine C+U = cytosine plus uracil P = phosphate (backbone) OPO = phosphate ester linkage (A- vs B-class helix)
Wavenumbers (∆cm-1) Wavenumbers (∆cm-1) Figure 9 E-SER (-03VSCE) and SG-SER spectra of RNA from A) E coli and B) B subtilis Sample conditions as in Figure 6 E-SER spectra at 750 mW SG-SERS at 75 mW
A B E-SERS
SG-SERS
E-SERS
SG-SERS
SPIE 2001-4575
70
A final comparison was made between the two SER methods by measuring dipicolinic acid (DPA) This chemical may be invaluable as a test for spore forming bacteria specifically B anthracis 50 to 90 of B anthracis sporilates During spore formation dipicolinic acid is synthesized and once completed 10-15 of the dry spore weight is composed of the Ca2+ complex located in the spore core42 Heating in water can be used to initiate germination at which point the exosporium breaks and releases the Ca dipicolinate which becomes dipicolinic acid in water The structure of this chemical strongly
4 CONCLUSIONS Towards the goal of developing a method and technology to rapidly detect and identify bioagents we have been investigating surface-enhanced Raman spectroscopy as a tool to measure relative concentrations of nucleic acid bases in RNA and determine bioagent taxonomy Initially we investigated E-SERS since this method has been extensively researched and the optimum sample conditions are well developed However this method requires a three-electrode sample cell and electrolyte solution Incorporation of an electrolytic cell into sample systems used to analyze for BWA as aerosols in water or on surfaces can be designed using flow injection analysis technologies but cross-contamination and plugging of sample lines seems inevitable For this reason we also investigated metal-doped sol-gels as a SER-active medium Previous studies have shown this material to be active in all solvents particularly water capable of continuous measurements in flowing systems and reproducible (quantitative) between coated sample vials Here we compared SG-SER spectra to traditional E-SER spectra of the nucleic acid base pairs adenosine monophosphate and RNA High quality spectra of adenine cytosine guanine thymine and uracil were obtained by both E-SERS and SG-SERS Both methods yielded very similar spectra for the NA bases including a pH dependent study of adenine Enhancement factors and detection limits for adenine were determined as 2x105 and 16x10-11g and 1x105 and 12x10-10g for E-SERS and SG-SERS respectively (normalized to 75 mW and 10-min acquisition time) Fifty percent of the silver particle surface area in the sol-gel matrix was assumed covered by adenine which may have been overestimated yielding a lower EF and higher detection limit It should also be realized that each E-SER spectrum required several attempts to optimize the measurement conditions (pH electrode potential etc) While each SG-SER spectrum involved no sample preparation and often represents the first and only attempt to make the measurement Quality spectra of RNA extracted from Escherichia coli Bacillus subtilis and Staphylococcus aureus were obtained by E-SERS that were easily interpreted Bands due to adenine guanine cytosine plus uracil and phosphate were identified The SER band intensity of the NA bases in the RNA samples were of the same order of magnitude suggesting that their interaction with the silver surface is concerted as is their Raman enhancement Interestingly the relative SER band
Wavenumbers (∆cm-1) Figure 10 A) Raman spectrum of solid dipicolinic acid B) Ax20 C) electrolytic SERS of 6x10-3 M dipicolinic acid in 01 M KCl at a potential of +07VSCE and pH of 4 and D) sol-gel SERS of 6x10-3 M dipicolinic acid Conditions for A and C as in Figure 1 C) 100 mW of 1064 nm 50 scans 8 cm-1
A B (Ax20)
C
D
suggested that it would be SER active However the E-SER spectrum was unstable and varied considerably as a function of potential A consistent spectrum was obtained at +06VSCE (Figure 10) This potential is not recommended for measurement because the surface is actively dissolving in solution The SG-SER spectrum was considerably more stable of higher quality and easily reproduced Bands at 660 825 1010 1390 1430 1570 1590 and 3075 cm-1 were observed Enhancement factors were determined for the two media using the symmetric ring stretching mode at 995 and 1010 cm-1 for the normal Raman and SER spectra respectively E-SERS yielded an EF of 5x103 while SG-SERS yielded an EF of 2x105 for DPA The SN of the latter suggests a detection limit of 20x10-10g (based on adenine coverage 75 mW and 10-min) The differences in SER activity for these two media may be attributed to the combined electrolytic potential of the solution chemical and metal15 Again the E-SERS suggests that the SG-SERS is at a potential positive of the pzc While the instability in the E-SERS may also be associated with surface interactions of two carboxylic acid groups of dipicolinic acid during the ORCs
DPA
SPIE 2001-4575
71
intensities for RNA extracted from E coli are very similar to those measured by normal Raman spectroscopy Although the relative percent that each of the NA bases contributed to each RNA sample was not determined reproducible band intensities allowed noting the following trends The percent adenine decreases while the combined percent cytosine and guanine increase for both B subtilis and S aureus compared to E coli Quality spectra were also obtained for the RNA samples by SG-SERS but only a few bands were readily identified Calculations of NA base concentrations by SG-SERS will require further research In light of recent events we summarize the capabilities of these SERS media in terms of rapid detection of B anthracis and dipicolinic acid However these capabilities must be qualified First and foremost the level to which SERS can distinguish bacteria or viruses has not yet been determined Development of a database of both DNA and RNA base concentrations for BWAs and common bacteria to establish the level of taxonomic identification is ongoing Second rapid collection of aerosol water or surface samples is being addressed by others who report trapping particles on filters from 100 liters of air per minute Third although not presented here we have developed methods to extract RNA or DNA from cells and spores for SER analysis within 10 minutes Finally we assume a detection limit of 3600 spores per 100 liters of air is required although a 50 lethal dosage of anthrax has not been established With these qualifications a mass detection limit for RNA using SERS is estimated as follows A single measurement is performed in ca 20 minutes (140 liters collected in 14 min RNA extracted in 8 min spectral acquisition and analysis in 10 min) The average human breaths 7 liters per minute therefore the analyzer must at the very minimum detect 5000 spores in 140 liters of air One spore is approximately 2x10-18m3 (1x1x2 microm) and if a density of 075 gcm3 is assumed this corresponds to a mass of 15x10-12g Each spore contains 4-12 RNA or 12x10-13g RNA for 8 If we assume 23 of the RNA can be isolated for analysis during lysis then the proposed instrument must be able to detect 4x10-10g RNA from 5000 spores per 70 liters of air within 10 minutes As noted above the mass detection limits for adenine were estimated at 16x10-11g and 12x10-10g for E-SERS and SG-SERS respectively Although these detection limits suggests that RNA from 5000 spores is detectable with the current instrumentation it is highly likely that only a portion of an RNA segment (eg 120-nucleotide 5S rRNA) is in contact with the metal surface and will contribute to the SER effect The SN for the RNA spectra were 110th of the average SN for the four individual RNA bases suggesting a 10 contribution Furthermore effective taxonomy will likely require knowing the NA base concentrations to 1 of the value (eg 25plusmn025) Again the average measurement error for the bases is 12 These values suggest that the E-SERS is within a factor of 4 of the required detection limit whereas the SG-SERS detection limit must be improved by 25 times The same arguments can be applied to the detection of dipicolinic acid If we assume a spore releases 10 by weight DPA during germination then the proposed instrument must be able to detect 75x10-10g DPA from 5000 spores per 70 liters of air within 10 minutes The detection limit for SG-SERS was estimated at 20x10-10g and suggest that the vials are suitable to perform a rapid screen for anthrax A series of concentration dependent measurements are currently being performed to verify this assertion Finally we note that the measurements performed here employed an FT-Raman spectrometer This instrumentation was chosen over dispersive Raman instrumentation for the high wavelength accuracy afforded by the HeNe reference laser (Connes Advantage43) which would allow reliable spectral subtraction matching of observed spectra to stored library spectra and confident use of chemometric approaches Such data analysis is likely to be required to enhance BWA identification However this instrumentation sacrifices sensitivity due to a less efficient detector (InGaAs vs Si) less efficient Raman scattering and less efficient generation of plasmon modes Substantial improvements can be obtained using visible excitation and Si detection and these measurements are underway
5 ACKNOWLEDGEMENTS The authors are grateful to Drs D Cookmeyer and S Tove of the US Army Research Office (Contract Number DAAH04-96-C-0078) for their interest and support of this research The authors would also like top acknowledge Dr R Yin and J Jensen for supporting development of the metal-doped sol-gels (Contract Number DAAD13-01-C-0019) They also thank Dr Wilfred H Nelson for assistance in spectral interpretations
SPIE 2001-4575
72
6 REFERENCES 1 Roberts WL and Rainey PM Clin Chem 39 1872-1877 (1993) 2 Pasechnik VA CC Shone and P Hambleton Bioseperations 3 267-283 (1993) 3 Jackson PJ ME Hugh-Jones DM Adair G Green KK Hill CR Kuske LM Grinberg FA Abramova and P
Keim Proc Natl Acad Sci 95 1224-1229 (1998) 4 Jensen JL NC Wong W R Loerop SPIE 4775-03 (2001) 5 Snyder A Pet al SPIE 3853-15 (1999) 6 Danosky T R and McFadden P N in press (1997) 7 Woodruff WH Farquharson S Science 201 831-833 (1978) 8 Lord RC and Thomas GJJr Spectrochemica Acta 23A 2551-2591 (1967) 9 Thomas GJJr Biochim Biophys Acta 213 417-423 (1970) 10 Placzek G Handbuch der Radiologie 2 EMarx ed Akademische Verlagagescellschatt Liepzig 1934 UCRL
Trans No 526 (1959) 11 Jeanmaire DL and RP Van Duyne J Electroanalytical Chem 84 1-20 (1977) 12 Chada S Manoharan R Moenne-Loccoz P Nelson WH Peticolas WL and Sperry JF Applied Spectroscopy
47 38-43 (1993) 13 Manoharan R Ghiamati E Chada S Nelson WH and Sperry JF Applied Spectroscopy 47 2145-2150 (1993) 14 Todd EA Morris MD Applied Spectroscopy 48 545-548 (1994) 15 Wentrup-Byrne E Sarinas S and Fredericks PM Applied Spectroscopy 47 1192-1197 (1993) 16 Weldon MK VR Zhelyaskov and MD Morris Applied Spectroscopy 52 265-269 (1998) 17 Kneipp K and J Fleming J Mol Structure 145 173-179 (1986) 18 Pothier NJ and Force RK Applied Spectroscopy 46 147-151 (1992) 19 Ervin KM E Koglin JM Sequaris P Valenta and HW Nurnberg J Electroanal Chem 114 179-194 (1980) 20 Kim SK TH Joo SW Suh and MS Kim J Raman Spectrosc 17 381-386 (1986) 21 Pothier NJ and Force RK Analytical Chemistry 62 678-680 (1990) 22 Kneipp K Y Wang RR Dasari and MS Feld Applied Spectroscopy 49 780-784 (1995) 23 Nie S and Emory SR Science 275 1102 (1997) 24 Graham D WE Smith AMT Linacre CH Munro ND Watson and PC White Analytical Chemistry 69 4703-
4707 (1997) 25 Fleischmann M PJ Hendra and AJ McQuillan Chem Phys Lett 26 163-166 (1974) 26 Pettinger B J Chemical Phys 85 7442-7451 (1986) 27 Surface-Enhanced Raman Scattering Section Four Theory SPIE MS 10 M Kerker and B Thompson Eds (1990) 28 Wang D-S and Kerker M SPIE (M Kerker and B Thompson Eds) MS 10 417-429 (1990) 29 Weaver MJ Farquharson S Tadayyoni MA J Chem Phys 82 4867-4874 (1985) 30 Norrod KL Sudnik LM Rousell D and Rowlen KL Applied Spectroscopy 51 994-1001 (1997) 31 Farquharson S Weaver WJ Lay PA Magnuson RH and Taube H J Am Chem Soc 105 3350-3351 (1983) 32 Lee PC and Meisel D ldquoAdsorption and Surface-Enhanced Raman of Dyes on Silver and gold Solsrdquo J Phys Chem
86 3391-3395 (1982) 33 Li Y-S and Wang Y Applied Spectroscopy 46 142-146 (1992) 34 Bello JM DL Stokes and T Vo-Dinh Analytical Chemistry 61 1779-1783 (1989) 35 van Duyne RP J C Hulteen D A Treichel M T Smith M L Duval and T R Jensen J Phys Chem B1033854-
3863 (1999) 36 Farquharson S and W W Smith W H Nelson and J F Sperry SPIE 3533-27 207-214 (1998) 37 Lee Y H W Smith S Farquharson H C Kwon M R Shahriari and P M Rainey SPIE 3537 252-260 (1998) 38 Lee Y-H S Farquharson and P M Rainey SPIE 3857 76-84 (1999) 39 Current Protocols in Molecular Biology Wiley Interscience1003-1006 (1987) 40 Farquharson S Smith W Carangelo R C and Brouillette C SPIE 3859 14-23 (1999) 41 Chase D B and JF Rabolt Fourier Transform Raman Spectroscopy Acad Press Ch1 p 131 (1994) 42 Brock TD MT Madigan JM Martinko and J Parker Biology of Microorganisms 7th Ed Prentice Hall p 76-80
(1994) 43 Connes J Rev Opt Theor Instrum 40 45 (1961)
166
Chemical agent identification by surface-enhanced Raman spectroscopy
Stuart Farquharson and Paul Maksymiuk
Real-Time Analyzers 87 Church Street East Hartford CT 06108
Kate Ong and Steven D Christesen US Army SBCCOM Aberdeen Proving Ground MD 21010
ABSTRACT The recent distribution of anthrax through the US postal system and the subsequent infection and death of several postal and national media employees amplifies the need for methods to rapidly detect identify and quantify this and other chemical and biological warfare agents The US military has also identified water supplies as a likely method of warfare agent deployment and is funding the development of a Joint Service Agent Water Monitor (JSAWM) In an effort to aid military personnel and the public at large we are developing a portable analyzer capable of identifying and quantifying chemical agents rapidly either on-demand or continuously The approach is based on the ability of Raman spectroscopy to identify molecular structure through the abundant vibration information provided in spectra and the ability of surface-enhanced Raman spectroscopy (SERS) to detect extremely low concentrations (eg part-per-billion) through the enhancement of Raman scattering by six orders of magnitude or more A key element to the analyzer design is a new SER active medium that is capable of quantitative reversible measurements The medium consists of silver or gold nanoparticles incorporated into a sol-gel matrix The porous silica network offers a unique environment for stabilizing SER active metals and the high surface area increases the interaction between the analyte and metal particles Here we present the use of new sol-gels that also selectively enhance chemicals based on polarity and charge Base-line measurements of chemical agents and their hydrolysis products are presented and compared to the JSAWM goal of 30 micrograms per liter detection Keywords Chemical warfare agent hydrolysis product SERS Raman spectroscopy sol-gel nanoparticle
1 INTRODUCTION Since September 11 2001 the threat of terrorist attacks and biological warfare within US borders has become a sobering reality The simplicity in manufacturing ease of deployment and the relatively low cost of chemical warfare agents (CWAs) raises public concern that they may also be used by terrorists Indeed terrorists released sarin (GB) in the Tokyo subway in 19951 Countering terrorism and terrorist attacks requires recognizing likely deployment scenarios and having the required technology to rapidly detect the deployment event One method of deployment has been long identified by the US military distribution through water supplies To counter this threat the Department of Defense is funding or monitoring the capabilities of 200 technologies with the goal of developing a Joint Service Agent Water Monitor (JSAWM) that is field portable2 The JSAWM must be capable of identifying the chemical agents with no false-positives and quantifying the agents at microgram per liter concentrations within 10 minutes (Table 1)3 This includes the analysis of drinking water supplies distribution and storage systems as well as potable water supplies The earliest technologies developed for CWA detection were based on electrochemical ionization or colorimetric analysis (eg phosgene tape) Although these analyzers were easy to use they were not generally agent specific and suffered from false-positives4 More traditional laboratory methods have also been investigated and in particular combined gas chromatography and mass spectrometry (GCMS) has been very successful at eliminating false-positives56 However GCMS requires extraction repeated calibration and long analysis times (typically 20 to 60 minutes)6 making it labor intensive and less than desirable for field use More rapid analysis of agents in the solid liquid and gas phase has been To whom correspondence should be addressed emailfarqureal-time-analyzerscom
Vibrational Spectroscopy-based Sensor Systems Steven D Christesen Arthur J Sedlacek III EditorsProceedings of SPIE Vol 4557 (2002) copy 2002 SPIE middot0277-786X02$1500
167
Table 1 Chemical Agent Structures Hydrolysis Half-lives and JSAWM Thresholds Agent Short-Hand Chemical Structure Hydrolysis
Half-Life JSAWM
Thresholds
Sarin (GB) F-[O=P-CH3]-O-CH(CH3)2 213 hours 32 microgL Soman (GD) F-[O=P-CH3]-O-CH(CH3)-(C-(CH3)3) 23 hours 32 microgL Tabun (GA) (CH3)2-N-[O=P-CN]-O-C2H5 41 hours 32 microgL VX C2H5O-[O=P-CH3]-S-(CH2)2-N-(CH(CH2)2)2 821 hours 32 microgL EA2192 HO-[O=P-CH3]-S-(CH2)2-N-(CH(CH2)2)2 gt9 years 32 microgL Mustard (H) ClCH2CH2-S-CH2CH2Cl encapsulates 47 microgL Lewisite (L) ClCH=CH-As-Cl2 rapid 27 microgL HCN HCN rapid 20 mgL BZ C7NH12-O-[C=O]-COH(C6H5)2 23 microgL T-2 Toxin 87 microgL at pH 7 to 75 and 20 to 25 oC
demonstrated by vibrational spectroscopy7-10 Hoffland et al7 reported infrared absorbance spectra and absolute Raman cross sections for several chemical agents while Christesen measured Raman cross sections for sarin tabun mustard gas and VX (ethyl S-2-diisopropylamino ethyl methylphosphonothioate)11 Again however these techniques also have limitations Raman spectroscopy is simply not a very sensitive technique and detection limits are typically 01 (1000 ppm) While infrared spectroscopy would have limited value in analyzing poisoned water since the very strong infrared absorption of water would obscure most other chemicals present Nevertheless efforts to overcome these limitations have been demonstrated Braue and Pannella8 quantified the G-series nerve agents (tabun sarin and soman) in terms of infrared attenuated total reflectance using a circle-cell And Alak and Vo-Dinh demonstrated the possibility of surface-enhanced Raman spectroscopy (SERS) to identify CWAs by measuring several organophosphonates that simulate the nerve agents12 However quantitative measurements have not been demonstrated for the SER-active material used (silver coated on alumina particles) or other SER-active media13 Recently we developed silver-doped sol-gels to promote the SER effect14-17 The porous silica network of the sol-gel matrix offers a unique environment for stabilizing SER-active metal particles and the sol-gel provides a high surface area that effectively increases the number of molecules observed within the Raman scattering volume The silver-doped sol-gels have been coated on the internal walls of standard glass vials to allow rapid SER analysis of small sample quantities (lt 01 mL) without preparation We have used p-aminobenzoic acid (PABA) as a test chemical to demonstrate surface enhancements greater than 106 reversible measurements in a flowing system reproducible measurements from vial-to-vial and batch-to-batch and measurements in multiple solvents including water14-17 Recently we used these vials to measure Tabun (GB) and Sarin and several hydrolysis products pinacolyl methyl phosphonate (PMP from Soman) and methyl phosphonic acid (MPA from all G-agents Figure 1) Although a number of unique vibrational bands are observed (eg C-N stretch doublet and P-C stretch) the G-agents were only observed for 5 concentrations and all spectra required baseline corrections Figure 1 Surface enhanced Raman spectra of ~5 vv A) Tabun and B) Sarin C) 1 vv PMP and D) 10 ppm MPA using sol-gel sample vials 785 nm excitation 1-min scan and CCD detection Performed at Aberdeen Proving Ground
C A
Wavenumber (∆cm-1) Wavenumber (∆cm-1)
D B
790
545 C-N
2135 2190 P-C 770
760 1290
Proc SPIE Vol 4577
168
Nevertheless MPA was readily observed for a 10 ppm sample with an estimated detection limit of 04 ppm (based on a signal-to-noise ratio of 3 for the 760 cm-1 band intensity) This measurement provides encouragement in that SERS may satisfy the needs of a JSAWM Furthermore MPA is also a hydrolysis product of VX and V-gas and EA2192 (Figure 2) and may prove a valuable indicator of agent usage Figure 2 Hydrolysis of Sarin to form hydrofluoric acid (HF) methylphosphonic acid 1-methylethyl ester (MPAMME) methyl phosphonic acid (MPA) and 2-propanol With this initial albeit modest success we began analyzing chemicals with various sol-gel compositions that we have been developing Here we describe four sol-gel compositions that select for 1) polar-positive 2) polar-negative 3) weakly polar-positive and 4) weakly polar-negative chemical species The ability of these sol-gels to select and enhance Raman scattering is described for several test chemicals and MPA
2 EXPERIMENTAL The chemicals analyzed as well as all chemicals used to prepare the metal-doped sol-gels were obtained at their purest commercially available grade from Aldrich (Milwaukee WI) The sol-gel designed to select for polar-negative species was prepared from a silver amine complex tetramethyl orthosilicate (TMOS) and methanol After mixing 02 mL of the sol-gel solution was transferred into a glass vial (2 mL) dried and heated The incorporated silver ions were then reduced using dilute sodium borohydride The vials were washed and dried prior to the addition of a sample solution In a similar manner the sol-gel designed to select for polar-positive species was prepared from a gold salt TMOS and methanol The sol-gel designed to select for weakly polar-negative species was prepared from a silver amine complex tetraethyl orthosilicate (TEOS) and methanol And the last sol-gel designed to select for weakly polar-positive species was prepared from a gold salt TEOS and methanol All samples were prepared in a chemical hood and transferred into plain or SER-active vials for analysis Normal Raman spectral measurements employed 1-mL pure samples that were placed in a 1-cm3 cuvette and weighed This yielded a powder density that allowed accurate calculation of molecules in the optical collection field SERS measurements employed 1-mg sample per mL water concentrations unless otherwise stated Once prepared a 01 mL sample was placed into one of the four selective sample vials which in turn was placed into the sample compartment of a Raman spectrometer for analysis A prototype Fourier transform Raman spectrometer (RTA) was used for these measurements18 The system consisted of a NdYAG laser (Brimrose) for excitation at 1064 nm an interferometer built by On-Line Technologies (OLT East Hartford CT) for frequency separation an uncooled InGaAs detector for signal detection (RTA) and an Intel 400 MHz Pentium II based laptop computer (Dell Round Rock TX) for interferometric control data acquisition (OLT) and analysis (LabVIEW by National Instruments Austin TX) Additional components included a Notch filter (Kaiser Ann Arbor MI) and interferometer entrance and exit optics (Edmund Scientific Barrington NJ) Fiber optics were used to deliver the excitation beam to the sample and the scattered radiation to the interferometer (1 meter lengths of 200 and 365 micron core diameter respectively Spectran Avon CT) A second Notch filter (Kaiser) was used as a beam splitter to direct the excitation beam along the same axis as the collected radiation A microscope object (20x04 Newport Irvine CA) was used to focus the beam into the sample and to collect the scattered radiation back along the same axis In this co-axial backscattering arrangement the excitation beam was passed through the outside of a glass vial and focused onto the silver-doped sol-gel film (01-03 mm thickness) containing the sample
3 RESULTS AND DISCUSSION p-aminobenzoic acid (PABA) and phenyl acetylene (PA) and were used to refine the selectivity and SER-activity of the four different metal-doped sol-gels PABA is a popular chemical used to evaluate the performance of SER-active media Here the polar end groups can be used to test selectivity of the polar-negative and polar-positive sol-gels PA is essentially non-
2O+ H HF + +OH
OH3H C
OP
OF
CH
CH3
3
3H C
OCP
OOH
CH
CH3
3
3H C
O
CPHO
CH
CH
3
3
C
Sarin MPAMME MPA 2-propanol
Proc SPIE Vol 4577
169
polar but a high electron density in the cylindrical π cloud around the carbon-carbon triple bond allows testing the selectivity of the weakly polar-negative and weakly polar-positive sol-gels As Figure 3 illustrates PABA passes through the polar sol-gel and is enhanced by either the silver or gold particles At 1 mgml the concentration of neutral PABA is ca 20 times that of the ionized form (pKa = 48) For electropositive silver the PABA anion is expected to interact through the carboxylate group and the associated vibrational modes are expected to dominate the spectrum Conversely for electronegative gold either form of PABA is expected to interact through the amine group The clear differences in our spectra support this expectation Furthermore bands at 840 and 1405 cm-1 assigned to a COO- bend and stretch respectively are significantly more intense for silver than gold Additional bands at 1140 and 1195 cm-1 are assigned to CH bending modes while bands at 1450 1500 and 1605 cm-1 are assigned to ring vibrational modes A very similar SER spectrum for PABA on a silver-coated alumina substrate has previously been reported with similar assignments19 For the gold-doped sol-gel new bands appear at 690 1355 and 1585 cm-1 The first band is assigned to a ring-H bending mode the second band to a ring-N- stretching mode and the third band to a possible NH2 scissors mode or ring mode The second band is not observed in the normal Raman spectrum but infrared bands occur at this frequency for aromatic ring-secondary amine stretching modes The scissors mode occurs at this frequency in Raman spectra for several chemicals but is absent in the PABA Raman spectrum Alternatively this mode may be the1600 cm-1 ring mode that has been shifted by the gold interaction Again a very similar SER spectrum of PABA has been reported but surprisingly using silver (colloids)2021 not gold as the enhancement medium These researchers also assumed the primary interaction of PABA with silver was through the carboxylate anion and made assignments accordingly For example they assigned the 1359 cm-1 to a COO- stretch not to the amine group as we have They also favor the ring stretching mode assignment for the 1582 cm-1 band Finally it should be said that other researchers have argued that the most dominant band in the SER spectra at 1450 cm-1 a ring vibration mode suggests that PABA lies flat on the surface and the π-orbitals dominate the surface interaction22
Figure 3 SER spectra of A) PABA using polar-negative and B) polar-positive sol-gels and C) PA using weakly polar-negative and D) weakly polar-positive sol-gels PABA is 1 mgmL PA is 1 vv Spectral conditions 75 mw 1064 nm 100 scans (15 min) 8 cm-1 resolution Non-polar PA passes through the non-polar sol-gels and is also enhanced by both metals The spectra are easily understood For electropositive silver PA interacts through the cylindrical triple bond π electron cloud and a -CequivC- doublet occurs near 2000 cm-1 The interaction is reasonably strong since this band appears at 2112 cm-1 in the normal Raman spectrum For electronegative gold this interaction is unlikely and only very weak bands occur near 2000 cm-1 The remaining bands are at 1000 cm-1 1200 cm-1 doublet and 1595 cm-1 all appear in the normal Raman spectra at virtually the same frequencies and are assigned to the symmetric ring-breathing mode CH bending modes and the trigonal ring-breathing mode respectively The polarnon-polar selectivity of the polar-negative and weakly polar-negative sol-gels was tested by adding a 11 molar mixture of PABA and PA The selective enhancement is quite good (Figure 4) The spectrum obtained using the polar sol-gel represents 78 PABA and 22 PA while the spectrum obtained using the weakly polar sol-gel represents 9 PABA and 91 PA The band peak intensities at 2000 cm-1 for PA and 1450 cm-1 for PABA were used for these calculations and are expanded in Figure 4 for clarity
C A
Wavenumber (∆cm-1) Wavenumber (∆cm-1)
2NH COOH
D B
C CH
Proc SPIE Vol 4577
170
Figure 4 SERS of 11 MM of PABA and PA in A) polar-negative and C) weakly polar-negative sol-gels The lower traces compare the pure chemicals B) 1 mgml PABA in polar-negative sol-gel and D) 1 PA in weakly polar-negative sol-gel while the insets magnify the minority species for clarity (x5 in A and x10 in B) Spectral conditions as in Figure 3 Following this development of selective sol-gels that maintained SER activity we measured cyanide and MPA (Figure 5) Not surprisingly the best sensitivity for both hydrolysis products was obtained using the polar-negative sol-gel The interaction of the cyanide anion with the silver surface is sufficient to shift the CequivN stretch observed at 2080 cm-1 in the normal Raman spectrum to 2145 cm-1 in the surface-enhanced Raman spectrum Furthermore the band is substantially broadened This anion has been extensively studied by electrolytic SERS and this shift and broadening have been attributed to the formation of a tetrahedral Ag(CN)3
2- surface structure23 Figure 5 Surface-enhanced (upper traces) and normal Raman spectra (bottom traces) of A) CN- and B) MPA in silver-doped TMOS SERS conditions as in Figure 3 and 1 mgmL Note MPA yields two distinct spectra for neutral (top) and acidic pH (middle) The normal Raman spectra employed pure powders 500 scans and 900 mW of 1064 nm SER measurements of MPA with the polar-negative sol-gel yielded two unique spectral signatures that depended on solution pH (Figure 5) For more neutral solutions the P-C stretch of MPA at 762 cm-1 dominates and the CH2 stretch at 2922 cm-1 appears The SN is sufficiently high that the anti-Stokes Raman shift at -762 cm-1 is observed For deprotonated MPA an oxygen-surface mode appears at 325 cm-1 (as well as its anti-Stokes complement) suggesting a strong interaction This results in substantial enhancement of the P-O-C mode at 1051 cm-1 (upper trace) Others report that this mode dominates the infrared spectra of nerve agents measured in water8 Comparison of the two spectra suggests the following molecule-to-surface orientations The appearance of the oxygen-surface and P-O-C modes in the upper spectrum of Figure 5B indicates that the tetrahedral molecule interacts with the silver surface through the deprotonated oxygen and is oriented end-on The
A B
Wavenumber (∆cm-1)
C A
Wavenumber (∆cm-1) Wavenumber (∆cm-1)
D B
Wavenumber (∆cm-1)
Proc SPIE Vol 4577
171
dominance of the P-C and the CH2 stretches and the disappearance of the P-O-C mode in the upper spectrum suggest the molecule is oriented with the methyl-hydroxyl-hydroxyl face towards the surface However considerably more research must be performed to verify these points Both chemical forms were measured as a function of concentration to determine the onset of monolayer coverage and estimate expected detection limits (Figure 6) Below monolayer coverage the signal to concentration dependence should be linear and the SN of any spectral measurement in this range can be used to predict the detection limit In the spectra presented here the peak height was used as the signal while the noise as root-mean-squared (RMS) was measured between 4400-4600 cm-1 Since noise is distributed evenly throughout the spectrum when transformed this region was used since it does not have any contributions from signals or baseline offsets Figure 6 shows a series of spectra for MPA along with a plot of the 1050 cm-1 band intensity (the noise was identical for this series) as a function of sample concentration A clear discontinuity appears in the vicinity of 01 mgmL (19 ppm) indicating the onset of monolayer coverage A detection limit defined as a SN of 3 was calculated for the 01 and 005gmL samples at 24x10-4 and 25x10-4 gL respectively A more modest detection limit of 101x10-4 gL was obtained using the 760 cm-1 band in the second series of concentration measurements These detection limits correspond to 46 and 190 ppb for a 3-min scan and 75 mW of incident laser power
Figure 6 A) Concentration dependence of MPA SERS measured in silver-doped TMOS) B) Concentrations are 001 005 01 05 1 gL (188 94 188 94 188 ppm) I760 series (bull) and I1050 series (∆)
Enhancement factors for cyanide and methyl phosphonic acid can be determined by estimating the number of molecules contributing to the surface-enhanced and normal Raman spectra The enhancement factor EF can be defined by the following equation
EF = (ISERSINR)bull(MNRMSERS) bull(PNRPSERS) bull(TNRTSERS)12
where I is the spectral band intensity M is the sample mass P is the incident laser power and T is the measurement time (or number of scans) for the two measurements For the normal Raman spectra a cylindrical scattering volume is assumed based on the laser area (28x10-7m2 6x10-4m diameter spot) and the penetration depth (1x10-3 m)24 The density of KCN and MPA as powders were measured at 0572 and 0516 gcm3 indicating that 16x10-4 and 144x10-4 g produced the normal Raman signals in Figure 5 respectively The number of molecules contributing to the SER spectra are those on the silver particles embedded in the sol-gel The total silver surface area can be determined from the average particle size concentration and the scattering volume Previous scanning electron micrographs showed the average silver particle size to be 40 nm in diameter (335x10-23m3)17 The silver concentration is 012M based on the reactant molar concentrations and dilution factors And the scattering volume is 76x10-
11m3 again based on a cylindrical scattering volume defined by a laser area of 28x10-7m2 and a sol-gel thickness of 27x10-
4m This volume contains 123x10-6g of silver equivalent to 35x109 silver particles with a collective surface area of 18x10-
5m2 However it may reasonably be assumed that at least half of the silver surface is in contact with the sol-gel matrix and unavailable for analyte interaction If we assume monolayer coverage and that each CN molecule occupies 15x10-20m2 then approximately 62x1014 molecules or 27x10-8g of CN contribute to the SER spectrum (20x10-19m2 46x1013 molecules 74x10-9g for MPA) Accordingly the EF for cyanide equals 48x104 ((180599) bull(16x10-427x10-8) bull(90075) bull(500100)12) The EF for MPA is considerably higher at 87x106 ((60326) bull(144x10-474x10-9) bull(90075) bull(500200)12)
0
20
40
60
80
100
120
140
0 02 04 06 08 1 12[MPA] (mgmL)
I (76
0)
0
100
200
300
400
500
600
I (10
50)
Wavenumber (∆cm-1)
A B
Proc SPIE Vol 4577
172
4 CONCLUSIONS Here we present for the first time surface-enhanced Raman spectra of Tabun and Sarin obtained using silver-doped sol-gels However the inferior enhancement suggested employing SER-active sol-gels varying in composition to improve both detection limits and selectivity of the target analytes To this end we successfully demonstrated the capabilities of four sol-gels that select for 1) polar-positive 2) polar-negative 3) weakly polar-positive and 4) weakly polar-negative chemical species p-aminobenzoic acid was used to show that silver could be used to attract polar-negative chemicals or functional groups (carboxylate anion) and that gold could be used to attract polar-positive chemicals (amine) while a mixture of p-aminobenzoic acid and phenylacetylene was used to show that tetramethyl orthosilicate preferentially solvates polar chemicals and that tetraethyl orthosilicate preferentially solvates weakly polar chemicals This increased sample control was applied to cyanide and methyl phosphonic acid two hydrolysis products of chemical warfare agents Exceptional results were obtained for methyl phosphonic acid allowing measurement of 1x10-2 gL for a 3-min scan and 75 mW of incident laser power with an estimated detection limit of 245x10-4 gL and an enhancement factor of 87x106 However this detection limit is 76 times less sensitive than required for the JSAWM (32x10-6gL for the G-agents) Finally we note that the measurements performed here employed an FT-Raman spectrometer This instrumentation was chosen over dispersive Raman instrumentation for the high wavelength accuracy afforded by the HeNe reference laser (Connes Advantage25) which would allow reliable spectral subtraction matching of observed spectra to stored library spectra and confident use of chemometric approaches Such data analysis is likely to be required to identify the chemical agents as well as distinguish hydrolysis products However this instrumentation which employs 1064 nm excitation and InGaAs detection sacrifices sensitivity We believe that the measurement sensitivity can be improved by at least two orders of magnitude by using shorter laser excitation wavelengths (eg 532 nm) This would provide more efficient Raman scattering (fourth power dependence on laser excitation wavelength) more efficient generation of plasmon modes and allow using more efficient detector material (Si vs InGaAs) These modifications are underway
5 ACKNOWLEDGEMENTS The authors would like to thank Dr R Yin and J Jensen of the US Army for supporting this work (Contract Number DAAD13-01-C-0019) They would also like to thank Advanced Fuel Research for making their laboratory facilities available
6 REFERENCES 1 Tu Anthony ldquoOverview of Sarin Terrorist Incidents in Japan in 1994 and 1995rdquo 6th CBW Protection Symposium
Stockholm Sweden 10-15 May 1998 2 Jensen JL NC Wong W R Loerop SPIE 4775-03 (2001) 3 JSAWM Requirements at wwwsbccomapgeaarmymilRDAecbcrtPRODSERJSAWMjsawmhtml 4 ldquoThe Chemical Weapons Convention Redefines Analytical Challengerdquo Analytical Chemistry News amp Features June 1
397A (1998) 5 Johnston RL Hoefler CM Fargo JC and Moberley B ldquoThe Defense Nuclear Agencyrsquos ChemicalBiochemical
Weapons Agreements Verification Technology Research Development Test and Evaluation Program and its Requirements for On-Site Analysisrdquo AT-ONSITE 5-8 (1994)
6 Black RM Clarke RJ Read RW and Reid MT ldquoApplication of gas chromatography-mass spectrometry and gas chromatography-tandem mass spectrometry to the analysis of chemical warfare samples found to contain residues of the nerve agent sarin sulphur mustard and their degradation productsrdquo J Chromatography 662 301-321 (1994)
7 Hoffland LD Piffath RJ Bouck JBrdquoSpectral signatures of chemical agents and simulantsrdquo Optical Engineering 24 982-984 (1985)
8 Braue EHJ Pannella MGrdquoCIRCLE CELL FT-IR Analysis of Chemical Warfare Agents in Aqueous Solutionsrdquo Applied Spectroscopy 44 1513-1520 (1990)
9 Tseng C-H Mann CK and Vickers TJ ldquoDetermination of Organics on Metal Surfaces by Raman Spectroscopyrdquo Applied Spectroscopy 47 1767-1771 (1993)
10 Staff Reporter ldquoUS Army Tests Lidar to Detect Biological Toxinsrdquo Photonic Spectra pg 50 December 1998
Proc SPIE Vol 4577
Proc SPIE Vol 4577
173
11 Christesen SD Raman cross sections of chemical agents and simulants Applied Spectroscopy 42 318-321 (1988) 12 Alak AM and Vo-Dinh T ldquoSERS of Organophosphorous Chemical Agentsrdquo Analytical Chemistry 59 2149-2153
(1987) 13 Norrod KL Sudnik LM Rousell D and Rowlen KL ldquoQuantitative Comparison of Five SERS Substrates
Sensitivity and Detection Limitrdquo Applied Spectroscopy 51 994-1001 (1997) 14 Lee Y and Farquharson S ldquoSERS Sample Vials Based on Sol-Gel Process for Trace Pesticide Analysisrdquo SPIE 4206
140-146 (2000) 15 Farquharson S and Lee Y ldquoTrace Drug Analysis by Surface-Enhanced Raman Spectroscopyrdquo SPIE 4200-16 (2000) 16 Lee Y Farquharson S and Rainey P M Surface-Enhanced Raman Sensor for Trace Chemical Detection in Water
SPIE 3857 76-84 (1999) 17 Lee Y Farquharson S Kwong H and Shahriari M ldquoSurface-Enhanced Raman Sensor for Surface-Enhanced Raman
Spectroscopyrdquo SPIE 3537 252-260 (1998) 18 Farquharson S Smith W Carangelo R C and Brouillette C ldquoIndustrial Raman Providing Easy Immediate Cost
Effective Chemical Analysis Anywhererdquo SPIE 3859 14-23 (1999) 19 Narayanan VA JM Bello JD Stokes and T Vo-Dinh Analusis 19 307-310 (1991) 20 Laserna JJ E L Torres and JD Winefordner Analytica Chemica Acta 469-480 (1987) 21 Torres EL and JD Winefordner Analytical Chemistry 59 1626-1632 (1987) 22 Suh JS DP DiLella M Moskovits J Phys Chem 87 1540-1544 (1983) 23 Benner RE R Dornhaus R Chang and BL Laube Correlations in the Raman spectra of cyanide complexes adsorbed
at silver electrodes with voltammograms Surface Science 101 341 (1980) 24 Chase D B and JF Rabolt Fourier Transform Raman Spectroscopy Acad Press Ch1 p 131 (1994) 25 Connes J Rev Opt Theor Instrum 40 45 (1961)
APPLIED SPECTROSCOPY 351
focusing the 488 nm laser beam 2 mm deep into thebulk of the crystals to avoid contributions from potentialdepletion layers As shown in Fig 1b the dependence ofthe Raman shift on the degree of deuteration is almostperfectly linear and ts very well with D 5 22684cmR1 24526 where D is the degree of deuteration (in )and R is the spectral mean of the PO4 vibration in cm21A linear correlation coef cient of 0998 indicates an ex-cellent linear dependence of the Raman peak shift withdegree of deuteration This result shows that the shift ofthe PO 4 peak is simply caused by the linear increase inatomic mass due to isotope substitution which decreasesthe length of hydrogen-like bonds
This excellent linear dependence allows us to map thepro le of the DH exchange layer at the surface of DKDPcrystals by acquiring Raman spectra and determining theposition of the PO 4 peak for various depths This methodis preferable over other methods such as determining thestrength of the OD vibration directly (eg at 715 cm21)because the position of the most intense peak in the Ra-man spectrum can be measured more precisely than theintensity of some of the weakest peaks in the spectrumThis is demonstrated in Fig 2 where depth-dependentRaman spectra (Fig 2a) and the resulting exchange layerpro les for two DKDP crystals are shown (Fig 2b) Thespectra in Fig 2a were obtained from a depth scan of aDKDP crystal with 75 degree of deuteration in thebulk grown at 45 8C The spectra start out as DKDP with30 deuteration close to the surface and approach thebulk DKDP spectrum within a few micrometers of depthThe fact that the relative degree of deuteration does notextend to 0 D is due to the limited depth resolution ofthe Raman microprobe which averages over 4 mm indepth Figure 2b depicts the resulting DH exchange layerpro les for this and a second crystal grown at 63 8Crespectively Both crystals had the same exposure to am-bient conditions and their main difference is the temper-ature at which they were grown The different exchangelayer pro les indicate that crystals grown at differenttemperatures have differen t proton conductiv ities 13
which leads to a difference in their rate of deuteriumdepletion The parameters controlling this behavior arecurrently the objective of a detailed study the results ofwhich will be reported elsewhere
CONCLUSION
In conclusion we have shown that the shift of the to-tally symmetric PO 4 stretch mode in the Raman spectrumof DKDP crystals scales linearly with degree of deuter-ation This allows us to correlate Raman peak positionsto deuteration levels in these crystals We have presenteda new technique to determine DH diffusion pro les inDKDP frequency conversion crystals based on micro-Ra-man spectroscopy This technique is fast inexpensiveand works under various environmental conditionswhich will allow us to better understand and control deu-terium depletion in DKDP crystals
ACKNOWLEDGMENTS
We would like to thank M Runkel for rst discovering DKDP crack-ing R Floyd for providing DKDP crystals and L Chase and A Burn-ham for their support and helpful discussions This work was performedunder the auspices of the US Department of Energy by the University
of California Lawrence Livermore National Laboratory under contract W-7405-Eng-48
1 J J De Yoreo A K Burnham and P K Whitman Int Mat Rev47 113 (2002)
2 C E Barker R A Sacks B M Van Wonterghern J A Caird JR Murray J H Campbell K Kyle R B Ehrlich and N DNielsen Proc SPIE-Int Soc Opt Eng 2633 501 (1995)
3 T Suratwala paper to be published4 Cleveland Crystals Inc httpwwwclevelandcrystalscom
KDPshtmltable5 E A Popova I T Savatinova and I A Velichko Sov Phys Solid
State 12 1543 (1971)6 I P Kaminow R C C Leite and S P S Porto J Phys Chem
Solids 26 2085 (1965)7 J A Subramony B J Marquardt J W Macklin and B Kahr
Chem Mat 11 1312 (1999)8 H Tanaka M Tokunaga and I Tatsuzaki Solid State Commun
49 153 (1984)9 R J Nelmes G M Meyer and J E Tibballs J Phys C 15 59
(1982)10 M A Yakshin D W Kim Y S Kim Y Y Broslavets O E
Sidoryuk and S Goldstein Laser Physics 7 941 (1997)11 I Takenaga Y Tominaga S Endo and M Kobayashi Solid State
Commun 84 931 (1992)12 C Krenn personal communication13 M Sharon and A K Kalia J Solid State Chem 21 171 (1977)
Rapid Dipicolinic Acid Extractionfrom Bacillus Spores Detectedby Surface-EnhancedRaman Spectroscopy
STUART FARQUHARSON ALAND GIFT PAUL MAKSYMIUK andFRANK E INSCOREReal-Time Analyzers Inc East Hartford Connecticut06108
Index Headings Dipicolinic acid Bacillus spores Anthrax Surface-enhanced Raman spectroscopy
INTRODUCTION
The anxiety caused by the distribution of anthrax en-dospores through the US postal system in October 2001was exacerbated by the long time required for positiveidenti cation of the Bacillus anthracis spores and the un-known extent of their distribution Since that time manymethods capable of rapid eld analysis have been inves-tigated to augment or replace the laboratory method ofgrowing microorganisms in culture media which takesdays to perform12 Prominent among these approachesare polymerase chain reactions (PCR)3 immunoassaysand detection of calcium dipicolinate as a biochemicalsignature PCR employs primers to separate organism-
Received 10 October 2003 accepted 14 November 2003 Author to whom correspondence should be sent
352 Volume 58 Number 3 2004
speci c nucleic acid sequences (eg capsular protein en-coding gene for Bacillus anthracis)4 and polymerases toamplify the segment until it is detectable Recently am-pli cation times have been substantially reduced andcomplete analysis can now be performed in an hour orless Immunoassay methods are also being developed thatuse competitive binding of the bioagent (as an antigen)and its labeled conjugate for a limited number of anti-bodies Although analyses can be performed in under 30minutes a well-de ned anthrax antigen has not yet beenidenti ed5ndash7 and consequently the false-positive rate isunacceptably high8
A number of other methods are being developed witha focus on the detection of calcium dipicolinate (CaDPA)and its derivatives as a B anthracis signature This is sobecause only spore-forming bacteria contain CaDPA andthe most common potentially interfering spores such aspollen and mold spores do not Relatively fast methodshave been developed to chemically extract CaDPA andthen detect it directly by uorescence9 or indirectly byluminescence1011 In the latter case hot dodecylamine(DDA) has been used to extract dipicolinic acid (DPA)and terbium has been utilized to form a highly lumines-cent DPA complex11 Although measurements have beenperformed in as little as ve minutes it was found thatas many as three concentration-dependent complexes canform each with different lifetimes This coupled withthe fact that the Tb31 cation produces the same lumines-cence spectrum makes determinations of low spore con-centrations problematic
It has been long known that Raman spectra of Bacillispores are dominated by bands associated with CaDPA12
and that these spectra may provide a suitable anthrax sig-nature at the genus level13 Since that time considerableimprovements in Raman instrumentation have led to lab-oratory measurements of single Bacilli spores14 and to eld measurements of spores captured from a mail-sort-ing system15 However the single spore measurementsrequired complex instrumentation that is not ruggedwhile the eld measurements required milligrams of sam-ple Furthermore the Raman spectra of both measure-ments contained uorescence contributions that would in-crease uncertainty in quanti cation
In related research we demonstrated that nanogramquantities of DPA could be detected by uorescence-freesurface-enhanced Raman spectroscopy (SERS)16 We alsodemonstrated that microliter volumes of chemicals canbe detected by SERS using metal-doped sol-gel-packedglass capillaries17 Towards the goal of developing a rap-id eld SERS-based anthrax spore detector we havecombined our previous research and we now report thatDPA can be extracted from a 10 mg B cereus spore sam-ple using DDA in 1 minute and can be detected by SERSin an additional 1 minute
EXPERIMENTAL
Dipicolinic acid (26-pyridinedicarboxylic acid DPA)and dodecylamine (DDA) were used as received fromSigma-Aldrich (Milwaukee WI) Lyophilized B cereusspores prepared according to the literature13 were sup-plied by the University of Rhode Island and used as re-ceived Multiple particles approximately 01 mm3 each
were separated and weighed at 5 to 15 mg representing05 to 15 million spores The sample masses were con-sistent with a previous determination of spore density at0081 gmL that indicated a high degree of entrained air
All chemicals used to prepare the silver-doped sol-gelcoated capillaries were also obtained and used as receivedfrom Sigma-Aldrich According to previously publishedprocedures17 two precursor solutions were preparedmixed and then drawn into 1-mm-diameter glass capil-laries The silver amine precursor consisted of a 51 vvratio of 1 N AgNO 3 to 28 NH3OH while the alkoxideprecursor consisted of a 21 vv ratio of methanol to te-tramethyl orthosilicate The alkoxide precursors weremixed with silver amine precursor in an 81 vv ratioApproximately 015 mL was drawn into the capillarycoating a 15-mm length After sol-gel formation the in-corporated silver ions were reduced with dilute sodiumborohydride which was followed by a water wash to re-move residual reducing agent
A 100 mL drop of a 50 mM DDA solution in ethanolpre-heated to 78 8C was added to each of the B cereusparticles to digest the spore coat After 1 minute the re-sultant solution was drawn into a SER-active capillarythat was immediately xed horizontally to an XY posi-tioning stage (Conix Research Spring eld OR) just in-side the focal point of an f 07 aspheric lens The lensfocused the beam into the sample and collected the scat-tered radiation back along the same axis A dichroic lter(Omega Optical Brattleborough VT) was used to re ectthe excitation laser to the lens and pass the Raman scat-tered radiation collected by the lens An f 2 achromatwas used to collimate the laser beam exiting a 200-mm-core-diameter source ber optic while a second f 2 ach-romat was used to focus the scattered radiation into a 365mm ber optic (Spectran Avon CT) A short-pass lterwas placed in the excitation beam path to block the sil-icon Raman scattering generated in the source ber fromre ecting off sampling optics and reaching the detectorA long-pass lter was placed in the collection beam pathto block the sample Rayleigh scattering from reachingthe detector A 785 nm diode laser (Process InstrumentsInc model 785-600 Salt Lake City UT) was used todeliver 100 to 150 mW of power to the sample A Fouriertransform Raman spectrometer (Real-Time Analyzersmodel IRA-785 East Hartford CT) and a silicon photo-avalanche detector (Perkin Elmer model C30902S Stam-ford CT) were used to acquire the SER spectra
RESULTS AND DISCUSSION
As an initial experiment the SER spectrum of 1 gLof DPA in water was measured using the newly devel-oped silver-doped sol-gel-coated capillaries (Fig 1A) Atthis concentration a high signal-to-noise ratio (SN) isobtained in 1 min In fact a reasonable spectrum is ob-tained in the same time frame for 1 mgL (Fig 1B) TheSER spectra are reasonably similar to the normal Raman(NR) spectrum obtained for a saturated solution of DPAin 1 N KOH (Fig 1C) and the following band shifts areobserved (NR to SER) 647 to 657 cm21 817 to 815cm21 998 to 1008 cm21 1384 to 1382 cm21 1434 to1428 cm21 and 1569 to 1567 cm21 Many of these bandshave been previously assigned1213 such as 998 cm21 to
APPLIED SPECTROSCOPY 353
FIG 1 SERS of DPA in water using silver-doped sol-gel-coated glasscapillary for (A) 1 gL and (B) 1 mgL (C ) NR of saturated DPA in 1N KOH in a glass capillary Spectral conditions (A) and (B) 150 mWof 785 nm 1-min acquisition time (C) 450 mW of 785 nm 5-minacquisition time both 8 cm 21 resolution
FIG 2 SERS of DPA extracted from 10 mg B cereus particle using100 mL of 50 mM hot DDA acquired in (A) 1 minute and (B) 2 seconds(C ) Attempted SERS of 50 mM hot DDA in ethanol using silver-dopedsol-gel-coated glass capillary acquired in 1 min Spectral conditions150 mW of 785 nm 8 cm21 resolution
the symmetric ring stretch 1384 cm21 to the OndashCndashOsymmetric stretch 1428 cm21 to the symmetric ring CndashH bend and 1569 cm21 to the asymmetric OndashCndashOstretch
The rst B cereus samples consisted of 2 mg of sporesin 2 mL of 5 mM hot DDA The samples were main-tained at 78 8C for 40 min and while hot approximately10 mL was drawn into a SER-active capillary Since spec-tra of DPA were obtained for these initial samples small-er spore masses higher DDA concentrations and shorterheating periods were examined In due course it wasfound that 10 mg of spores could be digested by 100 mLof 50 mM hot DDA in one minute and detected (Fig2A) In fact the signal was suf ciently intense that it canbe observed in as little as two seconds (Fig 2B) Theamount of DPA that was extracted was estimated to bebetween 5 and 10 mgL by comparing the signal intensityof the 1008 cm21 band to that measured for DPA in waterThis is consistent with previous research that found thatthe majority of the DPA is extracted from spores usingDDA11 and that B cereus spores contain approximately10 DPA by weight18 The SN of 127 for the 1008 cm21
band in the 1-minute SER spectrum suggests a limit ofdetection of approximately 250 ng of B cereus sporesbased on a SN of 3 Finally it should be noted that DDAdid not produce a detectable SER spectrum as shown inFig 2C
CONCLUSION
We have demonstrated that by combining rapid extrac-tion of dipicolinic acid from Bacillus cereus spores withchemical identi cation by surface-enhanced Raman spec-troscopy as little as 10 mg of spores can be detected Infact the entire measurement from the time of adding hotdodecylamine to the spores to the time when the dipi-colinic acid SER spectrum is acquired and analyzedcould be performed in less than two minutes The abilityof this method to distinguish between spore-forming bac-teria such as Bacillus anthracis and non-DPA containingpowders could help prevent costly shutdowns associated
with the appearance of suspicious material or intentionalmailing of common substances as an anthrax hoax Thismethod could also prove useful in detecting the locationof anthrax endospores in mail distribution facilities if an-other veri ed attack should occur
Research continues to fully characterize the surface-enhanced Raman spectroscopy signal intensities as afunction of sample concentration and to explore otherextractants that do not require the use of elevated tem-perature
ACKNOWLEDGMENTS
The authors are grateful for the support of the National Science Foun-dation (DMI-0296116 and DMI-0215819) and the US Army(DAAD13-02-C-0015 Joint Service Agent Water Monitor program)The authors are indebted to Chetan Shende for preparing the sol-gelcapillaries The authors also thank James Gillespie Nicholas Fell andAugustus Fountain for providing important background informationMark Farquharson for laboratory support and Professor Jay Sperry ofthe University of Rhode Island for supplying B cereus spores
1 V A Pasechnik C C Shone and P Hambleton Bioseparations 3267 (1993)
2 P J Jackson M E Hugh-Jones D M Adair G Green K K HillC R Kuske L M Grinberg F A Abramova and P Keim ProcNatl Acad Sci USA 95 1224 (1998)
3 B R Glick and J J Pasternak Molecular Biology Principles andApplications of Recombinant DNA (ASM Press Washington DC1994)
4 C A Bell J R Uhl T L Had eld J C David R F Meyer TF Smith and F R Cockerill III J Clin Microbiol 40 2897(2002)
5 D L Gatto-Menking H Yu J G Bruno M T Goode M Millerand A W Zulich Biosens Bioelectron 10 501 (1995)
6 J J Quinlan and P M Foegeding J Rapid Methods AutomationMicrobiol 6 1 (1998)
7 A A Hindle and E A H Hall Analyst (Cambridge UK) 1241599 (1999)
8 M S Ascher US Department of Health amp Human Services(httpwwwhhsgovophppresentationsAscherdoc)
9 R Nudelman B V Bronk and S Efrima Appl Spectrosc 54 445(2000)
10 D L Rosen C Sharpless and L B McBrown Anal Chem 691082 (1997)
354 Volume 58 Number 3 2004
11 P M Pellegrino N F Fell Jr and J B Gillespie Anal ChimActa 455 167 (2002)
12 W H Woodruff T G Spiro and C Gilvarg Biochem BiophysRes Commun 58 197 (1974)
13 E Ghiamati R S Manoharan W H Nelson and J F SperryAppl Spectrosc 46 357 (1992)
14 A P Esposito C E Talley T Huser C W Hollars C M Schal-dach and S M Lane Appl Spectrosc 57 868 (2003)
15 S Farquharson L Grigely V Khitrov W W Smith J F Sperryand G Fenerty J Raman Spectrosc paper accep ted (2003)
16 S Farquharson W W Smith S Elliott and J F Sperry SPIE-IntSoc Opt Eng 3855 110 (1999)
17 S Farquharson and P Maksymiuk Appl Spectrosc 57 479(2003)
18 F W Janssen A J Lund and L E Anderson Science (Washing-ton DC) 127 26 (1958)
SPIE -2003-5269 117
pH dependence of methyl phosphonic acid dipicolinic acid and cyanide by surface-enhanced Raman spectroscopy
Stuart Farquharson Alan Gift Paul Maksymiuk Frank Inscore and Wayne Smith
Real-Time Analyzers 87 Church Street East Hartford CT 06108
ABSTRACT US and Coalition forces fighting terrorism in Afghanistan and Iraq must consider a wide range of attack scenarios in addition to car bombings Among these is the intentional poisoning of water supplies to obstruct military operations To counter such attacks the military is developing portable analyzers that can identify and quantify potential chemical agents in water supplies at microgram per liter concentrations within 10 minutes To aid this effort we have been investigating the value of a surface-enhanced Raman spectroscopy based portable analyzer In particular we have been developing silver-doped sol-gels to generate SER spectra of chemical agents and their hydrolysis products Here we present SER spectra of methyl phosphonic acid and cyanide as a function of pH an important factor affecting quantitation measurements which to our knowledge has not been examined In addition dipicolinic acid a chemical signature associated with anthrax-causing spores is also presented Keywords Chemical warfare agents agent detection agent hydrolysis SERS Raman spectroscopy homeland security
1 INTRODUCTION In the past decade the Unites States and its allies have been challenged by a different kind of warfare exemplified by the terrorist attacks of September 11 2001 Suicide bombings and the use of chemical agents are the norm and military personnel must consider a wide range of attack scenarios Among these is the intentional poisoning of water supplies to obstruct military operations in Afghanistan and Iraq Currently colorimetric paper is used to detect agents on-site while gas chromatography combined with mass spectrometry (GCMS) is used in mobile support laboratories However both methods have severe drawbacks The paper changes color in response to contact with many chemicals besides CWAs causing a high incidence of false positives1 while GCMS although very chemically specific requires hours to perform and constant re-calibration234 Military operations would be greatly aided by a portable analyzer that can identify and quantify potential chemical agents at concentrations that impact safety This includes the analysis of drinking water supplies distribution and storage systems To meet this goal the Department of Defense has been investigating numerous approaches under the auspices of the Joint Service Agent Water Monitor (JSAWM) program5 The JSAWM must be capable of identifying the chemical agents with no false-positives and quantifying the agents at microgram per liter concentrations within 10 minutes678 Recently we and others have been investigating the ability of surface-enhanced Raman spectroscopy (SERS) to measure chemical agents9-12 bioagents13-17 and their hydrolysis products in water SERS employs the interaction of surface plasmon modes of metal particles with target analytes to increase scattering efficiency by as much as 1 million times18 In 1987 the potential of this approach to measure CWAs was demonstrated by measuring a series of pesticides19 Several of these organophosphonates have chemical structures similar to CWAs in particular P=O functional groups In our studies we have been employing silver-doped sol-gels to promote the SER effect The porous silica network of the sol-gel matrix offers a unique environment for immobilizing and stabilizing SER-active metal particles20-23 The sol-gels have been coated on the internal walls of standard glass vials to allow rapid SER analysis of small sample quantities without preparation We have measured over 100 chemicals with enhancements of 104 to 106 demonstrated reversible measurements in a flowing system reproducible measurements from vial-to-vial and measurements in multiple solvents including water20-23 Previously we used these vials to perform preliminary measurements of cyanide (CN) methylphosphonic acid (MPA) and dipicolinic acid (DPA) MPA is a hydrolysis product of the nerve agents (eg sarin Reaction 1) and may be a valuable indicator of nerve agent usage particularly since the alkyl methylphosphonic acids are relatively more stable than their corresponding parent complexes24 DPA is
SPIE -2003-5269 118
a chemical signature of spore forming bacteria such as Bacillus anthracis And in light of the inability to rapidly detect the anthrax spores distributed through the US mail in October 2001 a number of methods are being developed to extract and analyze this signature Reaction 1 Stepwise hydrolysis of Sarin to form hydrofluoric acid (HF) isopropyl methylphosphonic acid (IMPA) then methyl phosphonic acid (MPA) and 2-propanol In our previous SERS investigations MPA and DPA were measured at 50 and 100 mgL respectively In both cases limits of detection (LOD) were estimated at 100 microgL providing encouragement in that SERS may satisfy the needs of the JSAWM Since it has been shown that pH can substantially influence the intensity of SER bands25 which would clearly influence quantitative analysis we undertook the present study to determine the severity of these effects for cyanide methyl phosphonic acid and dipicolinic acid Furthermore we previously observed a band at 1050 cm-1 for MPA6 possibly due to an anion formed at basic pH Here we investigate the source of this spectral anomaly
2 EXPERIMENTAL All chemicals including potassium cyanide methyl phosphonic acid dipicolinic acid and those used to prepare the silver-doped and gold-doped sol-gel coated vials were obtained and used as received from Sigma-Aldrich All samples were prepared in HPLC grade water (Fischer Scientific Fair Lawn NJ) for SERS measurements The pH of these samples was adjusted using dilute nitric acid or potassium hydroxide and verified using a pH electrode (Corning Inc Corning NY) that had been calibrated with pH 400 700 and 1000 buffered standards from Fischer Scientific Once prepared the samples were transferred into the silver-doped sol-gel vials (Real-Time Analyzers Inc East Hartford CT) The vials were coated in a manner similar to that previously reported by adding ammonium hydroxide to a solution of silver nitrate tetramethyl orthosilicate (TMOS) and methanol20 Gold-vials were coated by adding nitric acid to a solution of gold tetrachloride TMOS and methanol The two precursor solutions were prepared mixed and transferred to 2-mL glass vials dried and heated After sol-gel formation the incorporated metal ions were reduced with dilute sodium borohydride (1mgmL) which was followed by a water wash to remove residual reducing agent After the resultant analyte solution was introduced the SER-active vial was immediately fixed horizontally to an XY positioning stage (Conix Research Springfield OR) just inside the focal point of an f07 aspheric lens The lens focused the beam into the sample and collected the scattered radiation back along the same axis A dichroic filter (Omega Optical Brattleborough VT) was used to reflect the excitation laser to the lens and pass the Raman scattered radiation collected by the lens An f2 achromat was used to collimate the laser beam exiting a 200 microm core diameter source fiber optic while a second f2 achromat was used to focus the scattered radiation into a 365 microm fiber optic (Spectran Avon CT) A short pass filter was placed in the excitation beam path to block the silicon Raman scattering generated in the source fiber from reflecting off sampling optics and reaching the detector A long pass filter was placed in the collection beam path to block the sample Rayleigh scattering from reaching the detector A 785 nm diode laser (Process Instruments Inc model 785-600 Salt Lake City UT) was used to deliver 100 to 150 mW of power to the sample A Fourier transform Raman spectrometer (Real-Time Analyzers model IRA-785 East Hartford CT) and a silicon photo-avalanche detector (Perkin Elmer model C30902S Stamford CT) were used to acquire the SER spectra
3 RESULTS AND DISCUSSION In a previous study of MPA6 aimed at developing a concentration calibration curve and determining limits of detection (LOD) we observed an anomaly at 1050 cm-1 Since it was found that the band intensity changed as a function of concentration the band must be associated with a sample parameter Two possible parameters photon flux and pH are examined here The first parameter was investigated by irradiating a 1mgmL MPA sample in a SER-active vial with laser powers of 200 mW and above and monitoring spectral changes It was immediately found that the 1050 cm-1 band
2O+ H HF + +OH
OH3H C
OP
OF
CH
CH3
3
3H C
OCP
OOH
CH
CH3
3
3H C
O
CPHO
CH
CH
3
3
C
Sarin IMPA MPA 2-propanol
SPIE -2003-5269 119
appeared and grew as a function of time and that the higher the laser power the faster the growth Figure 1 shows the growth of the 1050 cm-1 band over the course of 30 minutes when using 150 mW of 785 nm excitation while Figure 2A shows that the growth can be fit with a first order exponential equation namely I1050 = 03+05e-013t Furthermore the 760 cm-1 band could be fit with a first order decay equation with an identical exponential rate constant ie I760 = 08-
08e-013t The rates represent classical first order kinetics and their correspondence allows one to conclude that MPA is being transformed one-for-one into a photo-generated product At this time the photoproduct has not been positively identified but phosphonic (phosphorous) acid and phosphonate are likely candidates since the symmetric P(OH)3 stretch occurs at ~1050cm-1 Our studies show that a reduction in laser power to 100 mW at the sample for MPA essentially eliminates this degradation process This laser power or lower was used for further measurements
Methyl phosphonic acid is a diprotic acid that stepwise dissociates into two anions MPA- and MPA= according to the following reactions26
MPA MPA- + H+ pKa1 = 212 Reaction 2
MPA- MPA= + H+ pKa2 = 729 Reaction 3 The relative concentrations of MPA MPA- and MPA= can be determined at any pH by expressing [MPA] and [MPA=] in terms of [MPA-] using Reactions 2 and 3 and summing all three to equal the total starting concentration here 2 mgmL (0021M MW = 9602) viz
[MPA] + [MPA-] + [MPA=] = 0021M Equation 1 substituting from Reactions 1 and 2
([H+][MPA-])K1a + [MPA-] + (K2a[MPA-])[H+] = 0021M Equation 2 rearranging [MPA-] = 0021M(1+[H+]K1a + K2a[H+]) Equation 3 The relative concentrations of MPA MPA- and MPA= as a function of pH are shown in Figure 3 It is worth noting that near neutral pH both MPA- and MPA= will be present To confirm that the SER signal followed this pH dependence a starting solution consisting of 20 mg of MPA in 10 mL HPLC grade water was prepared and brought to pH of 20 using dilute nitric acid From this solution 2 mL were added to a SER-active vial and the SER spectrum recorded At this pH a peak at 760 cm-1 was barely discernable The 2 mL solution was returned to the starting solution and the pH was re-measured to correct for any changes that the silver-doped sol-gel vials might cause In most cases the change was less than 02 pH units and the pH is reported as the before and after average Next the pH of the
Figure 2 A) Exponential growth of 1050 cm-1 band and B) exponential decay of 760 cm-1 band for spectral series in Figure 1
Figure 1 Growth of 1050 cm-1 band as a function of time due to exposure to 150 mW of 785 nm Spectra are 5 sec each collected every 100-sec from 0 to 30-min
B
570 770 970 1170 1370Raman Shift (cm-1)
Arbit
rary
Unit
s
MPA Photodegradation
600 800 1000 1200 1400 Wavenumber (cm-1)
A
0 10 20 30 0 10 20 30 time (min) time (min)
Ram
an In
tens
ity (r
elat
ive)
30
min
0
SPIE -2003-5269 120
starting solution was adjusted to 325 using dilute KOH Again 2 mL were added to a vial and the SER spectrum recorded At this pH a reasonably strong 760 cm-1 band was observed This process was repeated as spectra were recorded at pHs of 70 74 75 79 85 and 100 A total of 1 mL of KOH was added diluting the total concentration by 10 Next the pH of the starting solution was made acidic by adding dilute nitric acid dropwise This time spectra were recorded at pHs of 72 69 64 and 37 Figure 4 shows the SER spectra for representative pHs (spectra were left out to simplify the figure) while Figure 3 shows the 760 cm-1 peak intensities as a function of pH (The band intensities were adjusted to compensate for dilution effects caused by the addition of HNO3 and KOH then normalized to 0021 M for the most intense band observed at pH 37) It is clear from Figure 3 that the 760 cm-1 band follows the MPA- concentration as a function of pH and must be assigned to this anion No bands were observed that corresponded to MPA or MPA= The lack of an MPA SER spectrum may be due to the absence of an attraction between the neutral analyte and the electropositive silver surface The same reasoning suggests that a strong SER spectrum should be observed for MPA= but it is not and a satisfactory explanation has not been found
0000
0005
0010
0015
0020
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14pH
Con
cent
ratio
n [M
]
pK1 = 212 pK2 = 729
MPA- MPA=
MPA
Figure 4 SER spectra of 002M MPA as a function of pH Conditions 100 mW of 785 nm 36 scans (1 min) 8 cm-1 recorded 2 min after sample introduction pH 19 and 20 not apparent on this scale pH 69 and 74 near identical to 70 and 75 and not shown for clarity
00
02
04
06
08
10
12
14
16
18
0 5 10 15 20 25 30Measurement Number
Ram
an In
tens
ity (7
60 c
m-1
)
Figure 3 Concentration dependence of MPA MPA- and MPA= as a function of pH for a 002M sample Intensity of 760 cm-1
band from Figure 6 as a function of pH ( for increasing basic adjustment for increasing acidic adjustment error was measured at ~10 for pH 64)
Figure 5 SER spectra of 002M MPA at pH 64 measured around a vial at three heights (9 points per height) Conditions as in Figure 4 but 10-sec scans
Since these measurements involved the removal and replacement of the SER-active vial in the sample holder to remove and add sample variation in the intensity as a function of vial position was minimized by illuminating the exact same height along the vial wall But this does not account for variability of the SERS response of the sol-gel coating around the vial To analyze this effect a vial containing MPA at pH 64 was rotated at ~ 40o
intervals at the original height and 18rdquo above and below this value Figure 5 shows the intensity of the 760 cm-1
band for the 27 positions It was found that the average value was 137plusmn014 an RSD of 10 overall and 5 for each height An error bar is included in Figure 4 for the pH 64 measurement
SPIE -2003-5269 121
With the above analyses in mind a preliminary investigation of the SER spectral response for MPA (as MPA-) as a function of concentration was performed A single vial was used for these measurements beginning with 1 mgL followed by measurements of 10 100 and 1000 mgL In all cases the pH was ~7 and 3 positions around the vial were measured per concentration Since the 760 cm-1 band was not observed for concentrations of 1 or 10 mgL using 100 mW of 785 nm the laser power at the sample was raised to 200 mW beginning with the 10 mgL concentration Photo-degradation was largely avoided (and not observed) by exposing the sample for only 33 seconds per spectral acquisition Representative spectra for 10 100 and 1000 mgL are shown in Figure 6 while a plot of the 760 cm-1 band intensity as a function of concentration is shown in Figure 7 These values were also used to estimate limits of detection based on the signal-to-noise ratio (SN) of the 760 cm-1 band using 100 mW of 785 nm laser excitation and a 10-min acquisition time As summarized in the Figure 7 inset the lower the measured concentration the lower the predicted LOD Even if as estimated 210 microgL could be measured using the silver-doped sol-gel vials an improvement of a factor of 70 is still required to meet JSAWM goals of measuring 3 microgL in 10-minutes Similar to MPA DPA is a diprotic acid (pKa1 = 216 and pKa2 = 692) and variations in pH will effect the relative concentrations of DPA DPA- and DPA= and possibly the SER spectra and band intensities This could prove significant if an acid or base is used to denature anthrax spores with the goal of extracting and analyzing DPA The MPA pH study described above was mimicked for DPA except that the starting solution consisted of 20 mg of DPA in 20 mL HPLC grade water (60x10-3M MW = 1671) The initial solution had a pH of 245 which was made basic by dilute KOH to pHs of 355 433 487 559 1069 and 1166 SER spectra were recorded at each pH using 100 mW of 785 nm and a 44-sec acquisition time Next one drop of concentrated nitric acid was used to remake the solution acidic at a pH of 200 Again sequential pH measurements were performed at 383 510 735 and 822 The solution pH was made acidic a third time but to pH 219 171 then 135 Throughout this process no more than 20 drops of acid or base were added and therefore the concentration was diluted by no more than 10 Most of the spectral bands showed a minor decrease in intensity as a function of increasing pH values However the bands at 525 and 795 cm-1 showed the most dramatic changes which occurred at acid pH Figure 8 shows the SER spectra of DPA for the spectral region and pH range of interest The identity of the DPA species was determined by plotting the normalized peak intensities with the lowest value set to 0 and the highest to 0006 M as a function of pH and overlaying these values on a plot of the relative concentrations for DPA DPA- and DPA= as previously done for MPA (Figure 9) As can be seen the 525 cm-1 band clearly corresponds to DPA The correspondence of the 795 cm-1 band to this species is less clear as the band retains intensity until pH of 55 This can be attributed to contribution to the overlapping band at 810 cm-1 which does not change as a function of pH The fact that most bands are observed at all pHs suggest that the primary interaction with silver is through the ring nitrogen This is supported by the fact that the most intense band occurs at 1008 cm-1 attributed to a symmetric ring breathing mode and that this interaction has been characterized for pyridine in numerous papers27
0
1
2
3
4
5
6
7
0 200 400 600 800 1000 1200
MPA Concentration (mgL)
760
Ban
d In
tens
ity
Figure 6 SER spectra of MPA in water at A) 1000 B) 100 and C) 10 mgL Conditions pH of 7 silver-doped sol-gel coated vial 200 mW of 785 nm 33-sec 8 cm-1 resolution
Figure 7 Plot of SER intensity of 760 cm-1 band of MPA as a function of concentration using 200 mW of 785 nm Inset table includes average intensity LOD standard deviation and percent deviation for each concentration but for 100 mW and 10 min
conc (mgL) LOD ave stddev dev10 021 002 1021100 072 010 14301000 312 040 1280
A
B C
SPIE -2003-5269 122
The ability to detect dipicolinic acid (DPA) by SERS is demonstrated in Figure 10 Even at 1 mgL the primary bands are visible The intensity of the dominant ring breathing mode observed at 1008 cm-1 for DPA in water (pH = 25 to 55) is plotted as a function of concentration in Figure 11 These values were also used to estimate limits of detection based on the SN of the 1008 cm-1 band using 100 mW of 785 nm laser excitation and a 10-min acquisition time Again the lower the measured concentration the lower the predicted LOD (see Figure 11 inset) and detection of 160 microgL is possible Gastrointestinal anthrax requires significant more spores than inhalation anthrax28 and a limit of detection might be placed at 1 million spores in 1 liter of water or 10 microgL Since each spore contains ~10 CaDPA by weight29 a goal for DPA might be 1 microgL indicating that the present measurements must be improved by nearly two orders of magnitude Due to the increased hazards of handling HCN gas KCN salt was used for these experiments Nevertheless all sample preparations were performed in a chemical hood KCN completely dissolves in water but its conjugate acid HCN is formed and has a Ka of 615x10-1030 viz
HCN CN- + H+ pKa = 921 Reaction 4
Figure 8 SER spectra of 1 mgmL DPA as a function of pH Conditions 100 mW 785 nm 100 scans (44-sec)
Figure 9 Concentration dependence of DPA DPA- and DPA= as a function of pH for a 0006M sample Intensity of 525 () and 795 (diams) cm-1 bands from Fig 8 as a function of pH
pH
135 171 219 383
0
0001
0002
0003
0004
0005
0006
0007
0 2 4 6 8 10 12 14pH
Con
cent
ratio
n [M
]
DPADPA-DPA=795525
DPA DPA=DPA-
pK2 = 692pK1 = 216
795 525
0
05
1
15
2
25
0 200 400 600 800 1000 1200
DPA Concentration (mgL)
1008
Ban
d In
tens
ity
conc (mgL) lod-10min-100mw1 017
10 016100 103
1000 355
Figure 10 SER spectra of DPA in water at A) 1000 B) 100 C) 10 and D) 1 mgL Conditions pH of 25-55 silver-doped sol-gel coated vial 175 mW of 785 nm 1-min 8 cm-1 D) has been multiplied by x10 to make bands visible
A
B C
D
Figure 11 Plot of SER intensity of 1008 cm-1 band of DPA as a function of concentration using 175 mW of 785 nm Inset table includes LOD in mgL for each concentration but for 100 mW and 10 min
SPIE -2003-5269 123
Consequently the cyanide concentration must be determined for each initial KCN concentration Specifically the samples prepared with concentrations of 01 1 10 100 and 1000 mgL of KCN produced CN- concentrations of 63x10-3 033 69 89 and 964 mgL at pHs of 816 90 967 102 and 107 respectively The pH dependence for the HCN and CN- concentrations are shown in Figure 12 Thus as the amount of KCN added to the solution decreases so does the pH of the solution (becomes less basic) and according to Reaction 4 the relative amount of CN- to HCN also decreases For example in the preparation of a 01 mgL solution of KCN the pH is shifted from 7 for pure water to only 816 and only 63 of the starting material becomes CN- or 63x10-3 mgL In comparison for a solution of 1000 mgL the pH is shifted from 7 to 107 and 96 of the starting material becomes CN- This is significant since the cyanide ion is better able to adsorb onto the silver particles and become SERS active SER spectra of 10 100 and 1000 mgL of KCN samples are shown in Figure 13 and a corresponding plot of the 2100 cm-1 band intensity as a function of concentration is shown in Figure 14
000102030405060708091011
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14pH
Con
cent
ratio
n [m
gm
L]
pKa = 921
CN -HCN
The SER spectra of cyanide are dominated by the single CN stretch at 2100 cm-1 which occurs in normal Raman spectra of solutions at 2080 cm-1 However a low frequency mode occurs at 135 cm-1 due to a Ag-CN stretch (not shown) indicative of a strong surface interaction It is also observed that as the concentration decreases the CN stretch shifts to 2140 cm-1 This shift has been attribute to the formation of a tetrahedral Ag(CN)3
2- surface structure31 as well
Figure 14 Concentration dependence of KCN SERS measured under conditions in Fig 11 Concentrations are 1 01 and 001 mgml Intensities are measured for the CN stretch at 2100 cm-1 Inset table includes LOD in mgL for each concentration in Figs 13 and 15 but for 100 mW and 10 min
Figure 13 SER spectra of KCN in water at A) 1000 B) 100 and C) 10 mgL Conditions pHs of 107 102 and 97 silver-doped sol-gel coated vial 100 mW of 785 nm 1-min 8 cm-1 C) has been multiplied by x10 to make band visible
Figure 12 Concentration dependence of HCN and CN- as a function of pH for a 1 mgmL sample Calculated intensity of 2100 () cm-1 band for a 1 mgml sample at pHs of 816 90 967 102 and 107
A
B
C
0
20
40
60
80
100
120
140
0 200 400 600 800 1000 1200CN Concentration (mgL)
2100
Ban
d In
tens
ity
Figure 15 SER spectra of KCN in water at A) 10 B) 1 and C) 01 mgL Conditions pHs of 97 12 and 12 gold-doped sol-gel coated vial 100 mW of 785 nm 1-min 8 cm-1 C) has been multiplied by x10 to make band visible
A
B
C
conc (mgL) lod-10min-100mw Condition01 001 Au-pH 121 007 Au-pH 12
69 003 Au-pH 9769 016 Ag-pH 9789 022 Ag-pH 102946 113 Ag-pH 107
SPIE -2003-5269 124
as to CN adsorbed to two different surface sites32 Alternatively the 2140 cm-1 band could be attributed to HCN since this species dominates at lower concentrations However it is unlikely that this species would be attracted to the electropositive silver surface Further both peaks should be present at pHs between 85 and 105 but this is not observed It has also been suggested that at concentrations near and above monolayer coverage the CN- species is forced to adsorb end-on due to crowding and at lower concentrations the molecule can reorient to lie flat33 This suggests that the 2100 and 2140 cm-1 bands correspond to the end-on and flat orientations respectively As Figures 13 and 14 show the intensity of the CN stretch for the 89 mgL sample is nearly as intense as the 964 mgmL band This suggests that the Raman signal for the flat orientation is more enhanced However more extensive measurements are required to verify this point Since resent research has suggested that cyanide may be more effectively detected on gold measurements of KCN solutions were also performed using gold-doped sol-gel vials Preliminary measurements are shown in Figure 15 for samples prepared from 01 1 and 10 mgL KCN Since the pHs are 816 90 967 the resultant CN- concentrations are 63x10-3 033 and 69 mgL Initially only the highest concentration was observed and the signal intensity was significantly better than the equivalent concentration measured using silver In an effort to shift Reaction 4 to the left transforming HCN to CN- (Le Chatelierrsquos principle) KOH was added to the lower concentration samples producing solutions with pH 12 This effectively forces all of the cyanide in solution to be CN- or 01 and 10 mgL respectively More importantly the CN stretch is now observed in the SER spectra The band appears at 2125 cm-1 as has been previously reported for gold12 As calculated for MPA and DPA LODs can be estimated from this data For the three concentrations of cyanide on silver the LODs are 016 to 11 mgL for 100 mW of 785 nm laser excitation and a 10-min acquisition time For gold pH adjusted the LODs are10 to 70 microgL an improvement of more than 10 times silver Nevertheless either substrate is sufficient to meet the JSAWM goals of measuring 3 mgL in 10-minutes as the requirements form cyanide are much less stringent than the nerve agents
4 CONCLUSIONS Here we examined the surface-enhanced Raman spectral response for methyl phosphonic acid dipicolinic acid and cyanide as a function of pH It was determined that the most prominent peak at 760 cm-1 reaches a maximum intensity between pH 3 and 7 and corresponds to the MPA- species Neither the MPA nor MPA= species appear to generate a SER spectrum and consequently no spectra were observed below pH 2 or above pH 8 In this study we also found that higher laser powers could cause photodegradation of MPA signified by the exponential growth of a band at 1050 cm-1 which is tentatively assigned to phosphorous acid Unlike MPA DPA was observed at all pHs This is attributed to the dominant interaction of the pyridine functional group with silver Minor spectral changes were observed at acid pHs and were assigned to neutral DPA Like MPA SER spectra of cyanide were pH dependent No spectra were observed for the HCN species while CN- was best observed at pHs more basic than 8 Preliminary concentration studies for the three analytes allowed estimating limits of detection for MPA DPA and CN using 100 mW of 785 nm and a 10-min acquisition time of 210 165 and 70 microgL respectively Although the latter value suggests that cyanide can be measured at sufficiently low concentrations to meet JSAWM goal improvements by 100 to 200 times are required for MPA and DPA It is clear from this study that pH of the sample is important and must be taken into account when developing concentration calibration curves as well as fieldable analyzers Future work will include tailoring the sol-gel with specific functional groups to dictate sample pH and thereby optimize sensitivity
ACKNOWLEDGMENTS The authors are grateful for the support of the US Army (DAAD13-02-C-0015 Joint Service Agent Water Monitor program) The authors would also like to thank Dr Steve Christensen of the US Army SBCCOM for helpful discussions and Mr Chetan Shende of Real-Time Analyzers for assistance in development of the gold-doped sol-gels
REFERENCES 1 Erickson B Analytical Chemistry News amp Features June 1 397A (1998)
SPIE -2003-5269 125
2 Johnston RL Hoefler CM Fargo JC and Moberley B AT-ONSITE 5-8 (1994) 3 Black RM RJ Clarke RW Read and MT Reid J Chromatography 662 301-321 (1994) 4 USA Army Center for Health Promotion and Preventive Medicine (USACHPPM) Technical Guideline 230A 5 Jensen JL NC Wong W R Loerop SPIE 4775-03 (2001) 6 Hoenig SL Handbook of Chemical Warfare and Terrorism Greenwood Press Wesport CT (2002) 7 Munro NB SS Talmage GD Griffin LC Waters AP Watson JF King and V Hauschild Env Health
Persp 107 933-974 (1999) 8 Holstege CP Kirk M Sidell FR Crit Care Clin13 923-42 (1997) 9 Farquharson S P Maksymiuk K Ong and S Christesen SPIE 4577 166-173 (2001) 10 Lee Y and S Farquharson SPIE 4378 21-26 (2001) 11 Spencer KM J Sylvia S Clauson and J Janni SPIE 4577 158-165 (2001) 12 Tessier P S Christesen K Ong E Clemente A Lenhoff E Kaler and O Velev Applied Spectroscopy 56
1524-1530 (2002) 13 Farquharson S WW Smith S Elliott and JF Sperry SPIE 3533 207-214 (1998) 14 Farquharson S WW Smith S Elliott and JF Sperry SPIE 3855110-116 (1999) 15 Farquharson S WW Smith YH Lee S Elliott and JF Sperry SPIE 4575 62-72 (2002) 16 Guzelian AA J Sylvia J Janni S Clauson and KM Spencer SPIE 4577 182-192 (2001) 17 Shende C F Inscore A Gift P Maksymiuk and S Farquharson in press 18 Weaver MJ S Farquharson and MA Tadayyoni J Chem Phys 82 4867-4874 (1985) 19 Alak AM and T Vo-Dinh Analytical Chemistry 59 2149-2153 (1987) 20 Lee Y and S Farquharson SPIE 4206 140-146 (2000) 21 Farquharson S and Y Lee SPIE 4200-16 (2000) 22 Lee Y S Farquharson and P M Rainey SPIE 3857 76-84 (1999) 23 Lee Y S Farquharson H Kwong and M Shahriari SPIE 3537 252-260 (1998) 24 Wang J M Pumera G Collins and A Mulchandani Analytical Chemistry 74 6121-6125 (2002) 25 Dou X YM Jung Z-Q Cao and Y Ozaki Applied Spectroscopy 53 1440-1447 (1999) 26 Data supplied by S Christesen and K Ewing 27 Kerker M and B Thompson Eds SPIE MS 10 (1990) 28 Inglesby TV DA Henderson JG Bartlett JAMA 287 2236 (2002) 29 FW Janssen AJ Lund and LE Anderson Science 127 26 (1958) 30 Lide DR Ed Handbook of Chemistry and Physics CRC Press 77th Ed (1996-7) 31 Billmann J G Kovacs and A Otto Surf Sci 92 153 (1980) 32 Murray CA and S Bodoff Phys Rev B 32 671 (1985) 33 Kellogg D and J Pemberton J Phys Chem 91 1120 (1987)
SPIE-2003-5269 16
Chemical agent detection by surface-enhanced Raman spectroscopy
Stuart Farquharson Alan Gift Paul Maksymiuk Frank Inscore and Wayne Smith
Real-Time Analyzers 87 Church Street East Hartford CT 06108
Kevin Morrisey and Steven D Christesen US Army SBCCOM Aberdeen Proving Ground MD 21010
ABSTRACT
In the past decade the Unites States and its allies have been challenged by a different kind of warfare exemplified by the terrorist attacks of September 11 2001 Although suicide bombings are the most often used form of terror military personnel must consider a wide range of attack scenarios Among these is the intentional poisoning of water supplies to obstruct military operations in Afghanistan and Iraq To counter such attacks the military is developing portable analyzers that can identify and quantify potential chemical agents in water supplies at microgram per liter concentrations within 10 minutes To aid this effort we have been investigating the value of a surface-enhanced Raman spectroscopy based portable analyzer In particular we have been developing silver-doped sol-gels to generate SER spectra of chemical agents and their hydrolysis products Here we present SER spectra of several chemical agents measured in a generic tap water Repeat measurements were performed to establish statistical error associated with SERS obtained using the sol-gel coated vials Keywords Chemical agents chemical agent detection SERS Raman spectroscopy
1 INTRODUCTION
In the autumn of 2001 terrorism within US borders became a sobering reality While extensive efforts are being implemented to secure the homeland US and Coalition forces in Afghanistan and Iraq are constantly faced with terrorist attacks In addition to car-bombings the military has identified several non-traditional attack scenarios including poisoning of water supplies by chemical warfare agents (CWAs) To counter this threat the Department of Defense has been investigating numerous approaches to detect such attacks under the auspices of the Joint Service Agent Water Monitor (JSAWM) program1 The JSAWM must be capable of identifying the chemical agents with no false-positives and quantifying the agents at microgram per liter concentrations within 10 minutes2 This includes the analysis of drinking water supplies distribution and storage systems Currently colorimetric paper is used to detect agents on-site while gas chromatography combined with mass spectrometry (GCMS) is used in mobile support laboratories However both methods have severe drawbacks The paper changes color in response to contact with many chemicals besides CWAs causing a high incidence of false positives3 while GCMS although very chemically specific requires up to an hour to perform and regular re-calibration456 Vibrational spectroscopy has also been investigated as a potential method of rapidly detecting CWAs7-
11 as early as 1985 in the case of infrared spectroscopy8 and 1988 in the case of Raman spectroscopy11 Again however these techniques also have limitations when it comes to measuring trace poisons in water Infrared spectra would be dominated by the very strong absorption of water which would obscure absorptions by most other chemicals present Whereas Raman spectroscopy is simply not a very sensitive technique and detection limits are typically grams per liter Surface-enhanced Raman spectroscopy (SERS) offers several advantages over conventional vibrational methods and that may provide the necessary sensitivity required for detecting trace quantities of chemical agents in water SERS employs the interaction of surface plasmon modes of metal particles with target analytes to increase scattering efficiency by as much as 1 million times12 In 1987 the potential of this approach to measure CWAs was demonstrated by measuring a series of pesticides13 Several of these organophosphonates have chemical structures similar to CWAs in particular P=O functional groups In the past few years we and others have further explored the ability of SERS to detect CWAs14-17 and even bioagents 18-21 We have been employing silver-doped sol-gels to promote the SER effect
SPIE-2003-5269 17
in these studies The porous silica network of the sol-gel matrix offers a unique environment for immobilizing and stabilizing SER-active metal particles22-25 The silver-doped sol-gels have been coated on the internal walls of standard glass vials to allow rapid SER analysis of small sample quantities without preparation We have measured over 100 chemicals with enhancements of 104 to 106 demonstrated reversible measurements in a flowing system reproducible measurements from vial-to-vial and measurements in multiple solvents including water21-25 Previously we used these vials to perform preliminary measurements of cyanide (CN) and methylphosphonic acid (MPA) Most of the nerve agents form MPA during hydrolysis while Tabun forms CN a chemical agent in its own right In another paper including in these proceedings we examined the limits of detection (LOD) for MPA by measuring a series of concentrations down to 50 mgL and estimated a limit of detection of 100 microgL26 These measurements provide encouragement in that SERS may satisfy the needs of the JSAWM To further establish the viability of SERS in particular silver-doped sol-gels here we present analysis of cyanide mustard and VX in tap water The measurements performed at the US Armyrsquos Edgewood Chemical Biological Center Aberdeen MD also included numerous repeat measurements to establish reproducibility
2 EXPERIMENTAL 2a General All chemicals including potassium cyanide 2-chloroethylethyl sulfide and those used to prepare the silver-doped and gold-doped sol-gel coated vials were obtained and used as received from Sigma-Aldrich All samples were prepared in a chemical hood using HPLC grade water unless otherwise noted (Fischer Scientific Fair Lawn NJ) for SERS measurements Once prepared the samples were transferred into the silver-doped sol-gel vials (Real-Time Analyzers Inc East Hartford CT) The vials were coated in a manner similar to that previously reported by adding ammonium hydroxide to a solution of silver nitrate tetramethyl orthosilicate (TMOS) and methanol22 The two precursor solutions were prepared mixed and transferred to 2-mL glass vials dried and heated After sol-gel formation the incorporated metal ions were reduced with dilute sodium borohydride (1mgmL) which is followed by a water wash to remove residual reducing agent After the resultant analyte solution was introduced the SER-active vial was fixed horizontally to an XY positioning stage (Conix Research Springfield OR) just inside the focal point of an f07 aspheric lens The lens focused the beam into the sample and collected the scattered radiation back along the same axis A dichroic filter (Omega Optical Brattleborough VT) was used to reflect the excitation laser to the lens and pass the Raman scattered radiation collected by the lens An f2 achromat was used to collimate the laser beam exiting a 200 microm core diameter source fiber optic while a second f2 achromat was used to focus the scattered radiation into a 365 microm fiber optic (Spectran Avon CT) A short pass filter was placed in the excitation beam path to block the silicon Raman scattering generated in the source fiber from reflecting off sampling optics and reaching the detector A long pass filter was placed in the collection beam path to block the sample Rayleigh scattering from reaching the detector A 785 nm diode laser (Process Instruments Inc model 785-600 Salt Lake City UT) was used to deliver 100 to 150 mW of power to the sample A Fourier transform Raman spectrometer (Real-Time Analyzers model IRA-785 East Hartford CT) and a silicon photo-avalanche detector (Perkin Elmer model C30902S Stamford CT) were used to acquire the SER spectra
2b Edgewood Chemical Biological Center The surface-enhanced Raman spectral measurements at the US Armyrsquos Chemical Biological center presented here were all performed on September 12 2003 To expedite measurements a plate was machined to hold up to 12 SER-active sample vials (Figure 1) The plate fit a standard XY plate reader that could be programmed Pure KCN bis-(2-chloroethyl)sulfide (distilled mustard gas HD) and ethyl S-2-diisopropylamino ethyl methylphosphonothioate (VX) were obtained on-site and used to prepare 1 gL tap water solutions in a chemical hood with appropriate safety equipment Simulated tap water was prepared by adding 10 chemicals most often found in tap water at appropriate concentrations to distilled water (Table 1) SER measurements were also performed in a chemical hood For added safety the FT-Raman instrument was placed outside the laboratory and 30 foot fiber optic and electrical cables were used to allow remote SERS measurements and plate manipulation For each experiment 1gL samples were prepared and added to 9 individual vials which were then loaded on the plate In some cases a tenth vial was included as a blank
Table 1 Generic Tap Water Composition
Compound mgL NaHCO3 100 CaSO4 27 MgSO4bull7H2O 67 NaNO3 1 Fulvic Acid 1 K2HPO4 07 KH2PO4 03 (NH4)2HSO4 001 NaCl 001 FeSO4 0001 pH 76-78
SPIE-2003-5269 18
A software program was written that allowed selecting the sequence that the vials were measured the number of positions along the length of the vials to measure (1 to 5) and the number of scans to co-add During sample analysis the program displayed the vial being analyzed the point being analyzed and the spectrum as it was being acquired Once all the data was collected a second software program was written to rapidly analyze the data The spectra collected for all the vials on a plate could be loaded at one time and then the spectra for each point could be displayed simultaneously or separately The user could then select the Raman peak to analyze in terms of peak height or area This was accomplished by selecting points on either side of the peak to define a baseline of zero The peak height or area could then be computed for all of the spectra loaded and then exported to a spreadsheet for statistical analysis
Figure 1 A) Vial Holder 6 slots to hold 2 vials each end-to-end B) Measurement Configuration Program user selects vials to measure sequence number of points per vial (1 to 5) and number of scans per point C) Spectral Acquisition Program shows spectrum being collected which vial and position D) Spectra Analysis Program user selects spectra to analyze by plate vial and point (s) as well as two wavenumbers defining the peak and the baseline to subtract The image is of 5 repeat measurements of 10 mgL KCN in generic tap water 16 sec each 100 mW of 785 nm
3 RESULTS AND DISCUSSION Raman and surface-enhanced Raman spectra were obtained for potassium cyanide bis-(2-chloroethyl)sulfide and ethyl S-2-diisopropylamino ethyl methylphosphonothioate representing three classes of chemical agents cyanides mustards and nerve agents respectively Spectra were also obtained for 2-chloroethyl ethyl sulfide (CEES) a structural analogue to HD which was included in the study to aid in assigning spectral bands KCN salt was used for cyanide experiments to avoid the increased hazards of handling HCN gas KCN completely dissolves in water forming its conjugate acid HCN according to its Ka of 615x10-1027 and at a concentration of 1 mgmL results in a pH 107 solution This is significant in that only CN- appears to interact sufficiently with silver to produce a SER spectrum and no spectral signal is observed below pH 726 Figure 2 shows the SER and normal Raman spectra for KCN The SERS spectrum of 1mgml KCN in water shows a single intense somewhat broad feature at
A B
C D
SPIE-2003-5269 19
2100 cm-1 assigned to the single CequivN stretch The band is much sharper in the normal Raman spectra of the solid KCN salt at 2074 cm-1 This band does broaden and shift to 2080 cm-1 in solution (not shown) However the observed SERS frequency is attributed to interaction with silver and low frequency mode at 135 cm-1 attributed to a Ag-CN stretch (not shown) supports this conclusion
cm-1 band to a C-S stretch but the authors concede that it is in fact more likely a C-Cl stretch28 It appears that the most intense Raman bands at 648 692 and 747 cm-1 shift to 620 660 and 720 cm-1 in the SER spectra and are tentatively assigned as above The width of these bands suggests that they overlap underlying spectral features Additional bands in the Raman spectra occur at 972 1034 1049 1263 1286 1423 1442 2865 2935 and 2960 cm-1 Corresponding bands occur in the SER spectra at 964 1015 1054 1286 1410 1447 2865 and 2935 cm-1 Most of these bands are associated with alkane modes specifically the bands at approximately 1040 cm-1 to a C-C stretch 1290 cm-1 to a CH2 in-phase twist 1440 cm-1 to a CH2 wag 2865 cm-1 to a symmetric CH2 stretch and 2965 cm-1 to an asymmetric CH2 stretch The Raman and SER spectra of sulfur mustard were measured at the Edgewood center (Figure 4) Both spectra are largely similar to CEES The C-Cl and C-S bands in the Raman spectrum of HD now occur at 640 655 700 739 and 760 cm-1 and are more resolved possibly due to the increased molecular symmetry Theoretical calculations indicate that the first three bands are due to C-Cl stretching modes and the latter two to C-S stretching modes28 Only the C-Cl bands maintain significant intensity in the SER spectra occurring at 624 and 643 cm-1 which is attributed to the
Prior to measurements of HD CEES was examined by Raman and SER spectroscopy (Figure 3) CEES also known as half-mustard is essentially identical to HD except one of the chlorine end atoms is replaced by a hydrogen atom Again although not as toxic as HD CEES is a blister agent and dilute aqueous samples were prepared using appropriate safety equipment Both the Raman and SER spectra of CEES are similar and dominated by bands between 600 and 800 cm-1 These are associated with C-Cl and C-S stretching modes which are tentatively assigned to 648 and 747 cm-1 in the Raman spectra respectively The shoulder at 630 cm-1 the overlapped band at 660 cm-1 and the strong band at 692 cm-1 could also be due to these modes or their asymmetric counterparts It is worth noting that theoretical calculations assign the 692
Figure 2 A) SER and B) NR spectra of KCN Conditions A) 1 mgml in tap water 100 mW of 785 nm at sample 1-min acquisition time B) solid 300 mW of 785 nm 5-min All spectra are 8 cm-1 resolution
A
B
Figure 3 A) SER and B) NR spectra of CEES Conditions A) 1 vv (10 mgml) in MeOH 100 mW of 785 nm 1-min acquisition time B) neat 300 mW of 785 nm 5-min
A
B
Cl-CH2-CH2-S-CH2-CH3
A
B
Figure 4 A) SER and B) NR of HD Conditions A) 1mgml in tap water B) pure both 100 mW of 785 nm 1-min
Cl-CH2-CH2-S-CH2-CH2-Cl
SPIE-2003-5269 20
expected strong interaction between chlorine and silver and adds support to the assignment of this band to a C-Cl stretch Weaker overlapping bands occur at 670 692 and 724 cm-1 the latter possibly due to C-S stretching modes Again the alkane modes are apparent in the normal Raman spectra of HD but only a broad feature at 1300 to 1450 cm-1 suggests CH2 contributions in the SER spectrum Although the observed bands in the VX spectrum have not been assigned (Figure 4) a computer generated Raman spectrum29 predicts many of the same features with surprising accuracy and are used here Two intense bands at 460 and 530 cm-1 closely match predicted bands at 463 and 546 cm-1 assigned to a CH3-P=O bend and a PO2CS wag Three highly overlapped bands occur at 694 745 and 771 cm-1 matching predicted bands at 713 730 and 760 cm-1 The first
Table 2 Measured SER peak heights for the CN stretch at 2100 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
Figure 5 A) SERS and B) NR spectra of VX Conditions A) 1 vv (10 mgml) in MeOH B) pure sample both 100 mW of 785 nm 1-min acquisition time
A
B
two have been assigned to a C-S stretch and CH2bend respectively while the latter has been attributed to either a P-C stretch or an O-C-C stretch Although the 745 cm-1 band may alternatively be assigned to a C-S stretch based on the previous measurements of CEES and HD The relatively intense bands at 890 1106 1218 1445 and 1465 cm-1 also match predicted bands at 880 1108 1216 1440 and 1464 cm-1 that are assigned to a C-C stretch CH3 rock N-C3 stretch various C-H3 bends and C-H bends respectively Both the computer generated and the measured spectra contain numerous other less intense bands One is worth mentioning A unique band appears at 370 cm-1 that is predicted at 368 cm-1 and corresponds to an O-P=O bend The surface-enhanced Raman spectrum of VX is also rich with spectral features It has the unique low frequency band at 370 cm-1 as well as a second band at 380 cm-1 that is assigned tothe S-P-O bend predicted in the normal Raman spectrum at 388 cm-1 Based on the measured and predicted normal Raman spectra the following SERS assignments are given 460 cm-1 to the CH3-P=O bend 544 cm-1 to the PO2CS wag 738 cm-1 to a C-S stretch (based on arguments above) 890 cm-1
to a C-C stretch 1101 cm-1 to a CH3 rock and 1456
cm-1 to a C-H bend The ability of SERS to measure chemical agents in water containing real-world chemical interferents was tested by using the generic tap water described in Table 1 The ability to reproduce measurements was accomplished by preparing three separate water stock solutions which were used to prepare three sample solutions each of 1 mgmL KCN in the generic tap water The samples defined as vials 1-9 were then measured at 4 points per vial in a semi-random fashion such that errors associated with stock solution preparation and errors associated with instrument drift could be identified No trends were apparent that signified such systematic errors Each spectrum collected consisted of 20 averaged scans taking 16 seconds at 8 cm-1 resolution The laser power at the sample was measured periodically during the day and it ranged from 102 to 105 mW spacccc
spot Vial 1 Vial 2 Vial 41 13754 13004 128192 13519 12692 12909
1 3 13541 12721 126394 13462 12648 12651
ave 13569 12766 12755Vial 6 Vial 3 Vial 5
1 11586 14068 10757 2 11236 14402 115122 3 11376 14546 11559
4 10894 11763 11214ave 11273 13695 11261
Vial 7 Vial8 Vial 91 11198 15804 11023
2 11243 15716 88753 3 11562 15216 11218
4 11655 15084 11077ave 11415 15455 10548
CN AVG STDEV ERR1 mgmL pk ht 12526 1560 1245
stock solution
SPIE-2003-5269 21
The CN measurements consisting of 36 data points produced an average height of 125 for the 2100 cm-1 peak with a standard deviation of 156 or 125 (Table 2) The HD and VX measurements were performed precisely the same way (Tables 3 and 4) For HD the 624 cm-1 peak was used for analysis and it had an average height of 53 with a standard deviation of 068 or 129 while for VX the 544 cm-1 peak was used for analysis and it had an average height of 1051 with a standard deviation of 308 or 293 The greater error in the VX measurements can be somewhat attributed to Vial 4 which produced lower SER signal intensities But removing this vial from the data set changes the standard deviation to 233 only a modest improvement
spot Vial 1 Vial 2 Vial 41 525 609 4682 45 675 484
1 3 527 644 5094 559 807 669
ave 51525 68375 5325Vial 6 Vial 3 Vial 5
1 521 574 575 2 536 53 4492 3 509 508 451
4 565 594 379ave 53275 5515 4635
Vial 7 Vial 8 Vial 91 457 589 413
2 583 497 4273 3 544 505 4
4 528 5 509ave 528 52275 43725
HD AVG STDEV ERR1 mgmL pk ht 530 068 1291
stock solution
spot Vial 1 Vial 2 Vial 41 1464 1034 4672 1485 99 61
1 3 1491 989 5684 1041 777 553
ave 137025 9475 5495Vial 6 Vial 3 Vial 5
1 1058 942 1293 2 697 121 965
2 3 727 1206 11124 689 1254 746
ave 79275 1153 1029Vial 7 Vial 8 Vial 9
1 1127 783 163 2 1358 812 1614
3 3 1371 908 15124 1097 875 1397
ave 123825 8445 153825
VX AVG STDEV ERR1 mgmL pk ht 1051 308 2925
stock solution
4 CONCLUSIONS In this paper we examined the ability of surface-enhanced Raman spectroscopy to reproducibly measure CN HD and VX in tap water without chemical interference Both normal and surface-enhanced Raman spectra were examined to select unique bands suitable to identify and quantify these chemical agents For SER measurements the 2100 cm-1 C-N stretch was used for CN the 624 cm-1 C-Cl stretch was used for HD and the 544 cm-1 PO2CS wag was used for VX It was determined that 1 mgmL samples of each of these chemicals measured 36 times in glass vials coated with a silver-doped sol-gel reproduced measurements with standard deviations of 125 129 and 293 It was further found that the 10 chemicals added to simulate generic tap water did not interfere with or alter the SER spectra It should be noted that the concentrations used in this study were considerably greater than those required by the JSAWM program Current work involves improving SER sensitivity and designing sampling systems with better reproducibility This includes the development of fractal silver and gold structures within the sol-gel matrix and the development of chemically selective sol-gels
ACKNOWLEDGMENTS The authors are grateful for the support of the US Army (DAAD13-02-C-0015 Joint Service Agent Water Monitor program) The authors would also like to thank Janet Jensen Ronald Crosier and Kristina Gonser for helpful discussions
Table 4 Measured SER peak heights for the VX band at 544 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
Table 3 Measured SER peak heights for the HD band at 624 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
SPIE-2003-5269 22
REFERENCES 1 Jensen JL NC Wong W R Loerop SPIE 4775-03 (2001) 2 JSAWM Requirements at wwwsbccomapgeaarmymilRDAecbcrtPRODSERJSAWMjsawmhtml 3 Erickson B ldquoThe Chemical Weapons Convention Redefines Analytical Challengerdquo Anal Chem 70 397A (1998) 4 Johnston RL Hoefler CM Fargo JC and Moberley B ldquoThe Defense Nuclear Agencyrsquos
ChemicalBiochemical Weapons Agreements Verification Technology Research Development Test and Evaluation Program and its Requirements for On-Site Analysisrdquo AT-ONSITE 5-8 (1994)
5 Black RM Clarke RJ Read RW and Reid MT ldquoApplication of GC-MS and GC-tandem MS to the analysis of chemical warfare samples found to contain residues of the nerve agent sarin sulphur mustard and their degradation productsrdquo J Chrom 662 301-321 (1994)
6 USA Army Center for Health Promotion and Preventive Medicine (USACHPPM) Technical Guideline 230A 7 Hoffland LD Piffath RJ Bouck JB rdquoSpectral signatures of chemical agents and simulantsrdquo Optical
Engineering 24 982-984 (1985) 8 Braue EHJ Pannella MGrdquoCIRCLE CELL FT-IR Analysis of Chemical Warfare Agents in Aqueous Solutionsrdquo
App Spectrosc 44 1513-1520 (1990) 9 Tseng C-H Mann CK and Vickers TJ ldquoDetermination of Organics on Metal Surfaces by Raman
Spectroscopyrdquo App Spectrosc 47 1767-1771 (1993) 10 Staff Reporter ldquoUS Army Tests Lidar to Detect Biological Toxinsrdquo Photonic Spectra pg 50 December 1998 11 Christesen SD Raman cross sections of chemical agents and simulants App Spectrosc 42 318-321 (1988) 12 Weaver MJ Farquharson S Tadayyoni MA ldquoSurface-enhancement factors for Raman scattering at silver electrodesrdquo J Chem Phys 82 4867-4874 (1985) 13 Alak AM and Vo-Dinh T ldquoSERS of Organophosphorous Chemical Agentsrdquo Anal Chem 59 2149-2153 (1987) 14 Lee Y Farquharson S ldquoRapid chemical agent identification by SERSrdquo SPIE 4378 21-26 (2001) 15 Farquharson S Maksymiuk P Ong K Christesen S ldquoChemical agent identification by surface-enhanced Raman spectroscopyrdquo SPIE 4577 166-173 (2001) 16 Spencer KM Sylvia J Clauson S and Janni J ldquoSurface Enhanced Raman as a Water Monitor for Warfare
Agents in Waterrdquo SPIE 4577 158-165 (2001) 17 Tessier P Christesen S Ong K Clemente E Lenhoff A Kaler E Velev O ldquoOn-line spectroscopic characterization of sodium cyanide with nanostructured gold surface-enhanced Raman spectroscopy substratesrdquo
App Spectrosc 56 1524-1530 (2002) 18 Farquharson S W W Smith S Elliott and J F Sperry Rapid biological agent identification by surface- enhanced Raman spectroscopy SPIE 3855110-116 (1999) 19 Farquharson S WW Smith YH Lee S Elliott and J F Sperry Detection of biological signatures A
comparison of electrolytic and metal-doped sol-gel surface-enhanced Raman media SPIE 4575 62-72 (2002) 20 Guzelian AA Sylvia J Janni J Clauson S and Spencer KM ldquoSERS of whole cell bacteria and trace levels of biological moleculesrdquo SPIE 4577 182-192 (2001) 21 Shende C Inscore F Gift A Maksymiuk P Farquharson S ldquoRapid dipicolinic acid extraction from Bacillus spores detected by surface-enhanced Raman spectroscopyrdquo App Spectrosc 58 accepted 22 Lee Y and Farquharson S ldquoSERS Sample Vials Based on Sol-Gel Process for Trace Pesticide Analysisrdquo SPIE
4206 140-146 (2000) 23 Farquharson S and Lee Y ldquoTrace Drug Analysis by SERSrdquo SPIE 4200-16 (2000) 24 Lee Y Farquharson S and Rainey P M Surface-Enhanced Raman Sensor for Trace Chemical Detection in
Water SPIE 3857 76-84 (1999) 25 Lee Y Farquharson S Kwong H and Shahriari M ldquoSol-Gel Chemical Sensor for Surface-Enhanced Raman
Spectroscopyrdquo SPIE 3537 252-260 (1998) 26 Farquharson S A Gift P Maksymiuk F Inscore and W Smith ldquopH dependence of methyl phosphonic acid
dipicolinic acid and cyanide by surface-enhanced Raman spectroscopyrdquo SPIE 5269-19 (2003) 27 Lide DR Ed Handbook of Chemistry and Physics CRC Press 77th Ed (1996-7) 28 Sosa C RJ Bartlett K KuBulat and WB Person ldquoA theoretical study of harmonic vibrational frequencies and
infrared intensities of XCH2CH2SCH2CH2X and XCH2CH2SH (= H Cl)rdquo J Phys Chem 93 577-588 (1993) 29 Hameka HF and JO Jenson ldquoComputer-generated predictions of the structure and the IR and Raman spectra of
VXrdquo ERDEC-TR-065 May 1993
- Chemagents Appendicespdf
-
- SPIE2001-4575-Bioagent-SERSapdf
-
- Detection of bioagent signatures A comparison of electrolytic and metal-doped sol-gel surface-enhanced Raman media
- Stuart Farquharson Wayne Smith and Yuan Lee
-
- Real-Time Analyzers 87 Church Street East Hartford CT 06108
-
- ABSTRACT
- 4 CONCLUSIONS
- 5 ACKNOWLEDGEMENTS
- 6 REFERENCES
-
SPIE 2001-4575
64
demonstrated reversible and reproducible measurements of p-aminobenzoic acid (PABA) in a flow through system Here we investigate the ability of the sol-gel SERS (SG-SERS) to measure the NA bases adenosine monophosphate and RNA extracted from E coli B subtilis and S aureus The measurements are compared to those obtained by E-SERS
2 EXPERIMENTAL The inorganic chemicals and solvents used to prepare samples were spectroscopic grade and purchased from Aldrich (Milwaukee WI) Fisher (Pittsburgh PA) or Pfaltz amp Bauer (Waterbury CT) The nucleic acid bases and dipicolinic acid were purchased from Sigma (St Louis MO) Normal Raman samples were measured to establish enhancement factors In each case 1cm3 of sample was placed into a 1x1 cm glass cuvette weighed and measured Unpacked densities were typically 6-7 gcm3 For all SER measurements including RNA samples were prepared as ~01mgmL (see Figure captions for exact concentrations) in 01M KCl and buffered to a pH of 92 with Na2B4O7bullH2O Adenine pH dependence measurements used pH buffer solutions at 4 (potassium acid phthalate) 69 (potassium phosphate monobasicsodium phosphate dibasic) 92 (Na2B4O7bullH2O) and 104 (tris-hydroxymethyl amino methane) Escherichia coli Bacillus subtilis and Staphylococcus aureus cultures (250ml per 1000mL Erlenmeyer flask) were grown overnight in a Trypticase soy broth (TSB) medium containing 1 glucose in a shaking water bath at 37 oC The bacteria were harvested by centrifugation for 10 minutes at 8000 rpm in a GSA rotor at 5degC then washed once in 085 saline The gram-positive bacteria were concentrated to 20 ml and passed through a French pressure cell twice at 15000 psi to break open the cells RNA was extracted according to Protocol 44139 to ensure pristine samples for initial measurements Since this method takes approximately 4 hours a streamlined method was developed For vegetative bacteria the specimen was boiled for 30 sec in 1 ml of distilled water to lyse the cells and release the RNA For bacterial spores the specimen was first incubated in 1 ml of saline solution containing 02 mg lysozyme and phosphate-buffered to pH of 624 for 1 hr at 37 oC This solution was then boiled for 2-3 minutes in 4X loading buffer to release the RNA For both specimens RNA STAT-60TM was added to the supernatant which was centrifuged at 12000 g for 5 minutes to precipitate the ~15 water-soluble proteins This procedure allowed extracting RNA for SER analysis in ~ 10 minutes Electrophoresis shows high purity while the existence of chemicals that could interfere with the SER measurements is still under investigation The electrolytic sample cell has been described previously36 Briefly a three electrode design is incorporated into a Plexiglas structure containing a well for the reference electrode (a saturated calomel electrode Cole Parmer Vernon Hills IL) and a 5mL sample well containing the silver working electrode and platinum wire counter electrode (05 mm wire Alfa Ward Hill MA) A channel connecting the two wells contained a 2 mm diameter semi-porous membrane (10-20 micron pore Ace Glass) The silver electrode was made from a 3 mm length of 2 mm diameter silver wire (Alfa) soldered to a copper wire lead encased in a 4 mm diameter Pyrex tube A cap containing the silver electrode platinum wire and nitrogen purge and vent lines fixed the silver electrode surface 1 mm from a 1 mm thick glass plate attached to the bottom of the sample well The potentiostat used to control the three electrodes was built in-house and has been described in detail elsewhere36 A multifuntional analog digital and timing inputoutput interface card (DAQCard-1200 National Instruments) is used to both drive the electrolytic cell as well as read the current generated in the cell A LabVIEW software program is used set the oxidation potential reduction potential SER measurement potential hold times and sweep rates The amount of charge passed was plotted as a cyclic voltammogram For all spectra presented five oxidation-reduction cycles (ORCs) stepping from -03 VSCE to 03 VSCE and back to -3 VSCE at 50 mVsec were used The SG-SER measurements were accomplished by simply placing the identical samples prepared above into Simple SERS Sample VialsTM (RTA) These 2-mL glass vials are internally coated with ~ 01 micron thick silver-doped sol-gel A prototype Fourier transform Raman spectrometer (RTA) was used for these measurements40 The system consisted of a NdYAG laser (Brimrose or Spectra Physics) for excitation at 1064 nm an interferometer built by On-Line Technologies (OLT East Hartford CT) for frequency separation an uncooled InGaAs detector for signal detection (RTA) and an Intel 400 MHz Pentium II based laptop computer (Dell Round Rock TX) for interferometric control data acquisition (OLT) and analysis (LabVIEW by National Instruments Austin TX) Additional components included a Notch filter (Kaiser Ann Arbor MI) and interferometer entrance and exit optics (Edmund Scientific Barrington NJ) Fiber optics were used to deliver the excitation beam to the sample and the scattered radiation to the interferometer (2 meter lengths of 200 and 365 micron core diameter respectively Spectran Avon CT) A second Notch filter (Kaiser) was used as a beam splitter to direct the excitation beam along the same axis as the collected radiation A microscope object (20x magnification 04 numeric aperture Newport Irvine CA) was used to focus the beam into the sample and to collect the scattered radiation back along the same axis In this co-axial backscattering arrangement the excitation beam passed through the glass plate onto the silver
SPIE 2001-4575
65
electrode surface for E-SERS through the vial glass wall and into the silver-doped sol-gel film for SG-SERS or through the glass wall of the cuvette and into the solid sample for normal Raman spectroscopy All E-SERS and normal Raman spectra were obtained with 750 mW of laser power at the sample while all SG-SERS spectra were obtained with 75 mW of laser power at the system Incident powers above 200 mW in some cases degraded the sol-gel
3 RESULTS AND DISCUSSION The generation of surface-enhanced Raman scattering at electrode surfaces has been extensively researched and the optimum sample conditions are well developed2729 Several researches incorporated electrodes into flowing systems and demonstrated that continuous monitoring of chemicals is possible18 These successes suggested investigated the capability of E-SERS to measure the NA bases and RNA The E-SERS measurements also provided a benchmark to compare and evaluate SG-SERS measurements The molecular structure of adenine (as well as the other base pairs) which includes an aromatic nitrogen-containing heterocycle is ideally suited to interact with the surface plasmons and contribute substantially to the chemical component of the SER effect1119 Even with excitation at 1064 nm a 3-minute scan of 18x10-5M adenine yields high signal-to-noise (SN) E-SER spectra and all of the bands are revealed with clarity (Figure 1 Table 1) Spectra of similar quality were obtained by SG-SERS and the principal spectral bands are easily observed The identical 18x10-5M adenine sample was measured in the same 3-minute time frame but with 110th the laser power The lower power appears to reduce the SN The amount of adenine responsible for the SER spectra as well as enhancement factors for the two SER media can be determined The molecules producing the E-SERS spectrum are those on the electrode surface within the illumination area of the laser (The solution concentration only determines the number of molecules available to adsorb to the electrode surface) For the current experiments the laser illuminates an area of 28x10-7m2 or 56 x10-7m2 if we assume the ORCs increase the surface area by a factor of two Furthermore if we assume monolayer coverage on the electrode and each 3x5 angstrom molecule (lying flat) occupies 15x10-19m2 then there are ca 4x1012 molecules contributing to the Raman scattering This is ca twice the number of molecules measured at electrode surfaces using either differential capacitance-potential curve measurements or rapid linear sweep voltammetry (eg 3x1018 moleculesm2 for pyridine and pyrazine)29 Thus the adenine spectrum in Figure 1 is due to 87x10-10g (6x10-12 moles) A detection limit defined as a SN of 3 can also be calculated The SN for a 3-minute scan is 844 for the 735 cm-1 band suggesting a mass detection limit of 3x10-12g (2x10-14 moles) This is consistent with previous estimates for adenine by others of 25 x10-14 moles1530 However sub-monolayer concentrations must be measured to verify this The root-mean-squared (RMS) noise is measured between 4400-4600 cm-1 Since noise is distributed evenly throughout the spectrum when transformed this region does not have any
Wavenumbers (∆cm-1) Figure 1 A) Normal Raman spectrum of pure adenine powder B) E-SERS and C) SG-SERS of 18x10-5M adenine at pH 92 All spectra 8 cm-1 resolution 200 scans (3 min) and 1064 nm excitation A) and B) 750 mW C) 75 mW B) measurement potential of 11VSCE
A
B
C
A
B
C SG-SERS
pH 10
D
725
735
735
pH 4
Wavenumbers (∆cm-1) Figure 2 A) and C) E-SERS and B) and D) SG-SERS of adenine at A) and B) pH 104 and C) and D) pH 40 Note consistent appearance of bands at 1270 and 1375 cm-1 as the pH is changed to 10 for both SER media E-SERS used 750 mW SG-SERS used 75 mW of 1064 nm excitation
SPIE 2001-4575
66
contributions from signals or baseline offsets The measurement error is given as SplusmnRMS and for adenine this equals 234 The number of molecules contributing to the SG-SERS are those on the silver particles that are embedded in the sol-gel The total silver surface area can be determined from the average particle size (40 nm diameter) concentration (073 by weight based on molar conc and measured sol-gel density) and the scattering volume (a cylinder defined by the laser area 28x10-7m2 and sol-gel thickness10-4m) The 61x109 silver particles in this volume have a collective area of 31x10-5m2 However it may reasonably be assumed that at least half of the silver surface is in contact with the sol-gel matrix and unavailable for analyte interaction Then approximately 10x1014 molecules or 22x10-8g of adenine contribute to the SG-SER spectrum The slightly lower SN of 207 suggests a mass detection limit of 32x10-10g Determination of the enhancement factors for the two SER media requires estimating the number of adenine molecules contributing to the normal Raman spectrum Here a cylindrical scattering volume is assumed again based on the laser area (28x10-7m2) and the penetration depth (1x10-3 m)41 The density of the sample was measured at 064 gcm3 indicating that 18x10-4g (13x10-6 moles) of adenine produced the normal Raman signal The enhancement factor EF is defined by the following equation
EF = (ISERSINR)bull(MNRMSERS) bull(PNRPSERS) bull(TNRTSERS)12
where I is the spectral band intensity (here 735 cm-1) M is the sample mass P is the incident laser power and T is the measurement time (or number of scans) For the E-SERS measurement the enhancement factor is 22x105 (01780184) bull (18x10-487x10-10)) while the SG-SERS enhancement factor is 10x105 (0160184) bull(18x10-422x10-8) bull(75075) bull(315)12) The lower enhancement for the SG-SERS may be real or the available surface of the silver embedded in the sol-gel may have been overestimated In addition to enhancing the Raman scattering efficiency to an extent similar to E-SERS the SG-SER medium also yields an identical shift of the adenine ring-breathing mode from 725 cm-1 in the normal Raman to 735 cm-1 Furthermore in the course of optimizing the E-SERS sample conditions it was found that pH influenced the adenine interaction with the silver surface (Figure 2) In particular the relative band intensities of the pyrimidine ring skeletal vibrations at 1270 and 1375 cm-1 and the imidazol ring skeletal vibration at1335 cm-1 change At pH 4 adenine is protonated presumably the imidazol ring since the band at 1335 cm-1 increases in intensity while the pyrimidine bands are virtually absent Conversely at pH 10 the imidazol band decreases in intensity while the pyrimidine bands appear It is worth noting that the ring-breathing mode at 735 cm-1 changes little between pH 4 and 10 suggesting that the skeletal changes are more a function of molecule-plasmon interactions than reorientation of the molecule on the surface Measurements of the identical pH series of adenine samples by SG-SERS yielded virtually identical spectral changes This suggests that the sol-gel does not influence the measurement This is critical to reproducing measurements and performing quantitative analysis Next the remaining NA bases were measured by both E-SERS and SG-SERS and compared Previously we examined the optimum pH and electrode potentials for E-SERS measurements to determine if a common pH could be used that yielded good sensitivity for all the bases and if variations in potential could be used to provide an added degree of selectivity between the bases Primarily it was found that high quality spectra were obtained between pH 7 and 95 and that cytosine and uracil were best enhanced at potentials positive of the potential-of-zero charge (pzc ca 065VSCE for Ag) guanine and thymine near the pzc and adenine negative of the pzc In all cases the ring-breathing modes were the most intense and in general could be used to identify the NA bases (Figure 3 Table 1) Specifically adenine has an intense band at 735 cm-1 cytosine at 797 cm-1 guanine at 653 cm-1 thymine at 784 cm-1 and uracil at 800 cm-1 The adenine cytosine guanine and thymine bands are sufficiently separated that their contributions to DNA should be determinable Although adenine and guanine contributions to RNA should also be determinable cytosine and uracil are highly overlapped and unfortunately share the same potential dependence Alternate unique bands at 1183 cm-1 for cytosine and 1275 cm-1 for uracil might be suitable for calculating contributions The SG-SER spectra of the remaining NA bases faithfully reproduced the E-SER spectra In particular the primary identifying bands occur at virtually the same wavenumbers (see Table 1) However the spectra for both cytosine and thymine contain an intense band at ca 1040 cm-1 Initially this was attributed to the pH buffer but samples prepared without either the buffer or the 01M KCl electrolyte yielded identical spectra containing this band In fact the E-SER and SG-SER spectra of thymine are virtually identical except for this band Also the SG-SERS of guanine contains an intense band at 1551 cm-1 that is not observed in the E-SER spectrum This band may be due to a moderately intense band at 1553cm-1 in the normal Raman spectrum that is SG-SER active It was also found that the SG-SERS of cytosine was considerably better than the E-SERS while uracil showed the opposite relationship It is also worth noting that all of the SG-SERS were obtained with 110th the laser power Most importantly the primary ring-breathing modes in the SG-SER spectra are sufficiently intense and unique to be used in determining contributions to DNA and RNA as outlined above
SPIE 2001-4575
67
Table 1 Comparison of E-SER and SG-SER Spectral Band Positions for the NA Bases and Adenosine Monophosphate
Adenine Cytosine Guanine Thymine Uracil AMP E-SERS SG-SERS E-SERS SG-SERS E-SERS SG-SERS E-SERS SG-SERS E-SERS SG-SERS E-SERS SG-SERS1647 1638 1634 1656 1655 1630 1587 1585 1510 1580 1551 1539 1456 1456 1465 1460 1462 1480 1453 1459 1394 1398 1425 1431 1435 1399 1404 1392 1374 1375 1373 1383 1370 1335 1332 1311 1307 1333 1331 1353 1348 1331 1329 1265 1273 1280 1292 1278 1276 1275 1279 1271 1183 1195 1222 1232 1221 1219 1204 1205 1180 1144 1097 1033 1029 1038 1040 1035 1051 1037 1041 1035 963 963 957 1001 1000 961 944 884 819 817 859 866 735 737 797 799 784 782 800 800 72738 742 630 630 653 664 667 684 603 602 590 611 466 561
Bands unique to SG-SERS The next chemical to be analyzed by both E-SERS and SG-SERS was adenosine monophosphate (AMP) The E-SER spectrum yields bands due to the adenine chemical functionality at 727 961 1233 1279 1331 1381 and 1486 cm-1 In addition phosphate bands are observed at 860 1097 1453 1587 and 1705 cm-1 (Figure 5) Other researchers have noted that the ribose component does not appear to contribute to the spectrum19 The AMP spectrum also changes as a function of potential As the electrode is swept more positive (here from -09 to -03VSCE) the phosphate bands at 860 1097 1453 and 1587 cm-1 increase in intensity compared to the adenine bands while a band at 1705 cm-1 appears The adenine bands at 1233 1381 and 1486 cm-1 virtually disappear These potential dependent spectral changes are consistent with earlier studies that show that phosphate is attracted to silver at potentials positive of the pzc but repelled at potentials negative of the pzc19
Wavenumbers (∆cm-1) Figure 3 E-SERS of A) 21x10-3M cytosine at -03VSCE 1000 scans B) ~10x10-5M guanine at -06VSCE 500 scans C) 23x10-3M thymine at -06VSCE 500 scans and D) 12x10-3M uracil at -093VSCE 500 scans All spectra at pH 92 750 mW 1064 nm at 8 cm-1
Wavenumbers (∆cm-1) Figure 4 SG-SERS of A) 21x10-3M cytosine 200 scans B) ~10x10-5M guanine 200 scans C) 23x10-3M thymine 200 scans and D) 12x10-3M uracil 500 scans All spectra at pH 92 75 mW 1064 nm at 8 cm-1
A
B
C
A
B
C
D D
SPIE 2001-4575
68
The SG-SER spectrum of AMP is considerably different The adenine bands virtually disappear except for the two primary bands which shift to742 and 1329 cm-1 While the phosphate band at 1459 cm-1 has gained considerable intensity In addition two new intense bands appear at 684 and 1539 cm-1 as well as a moderately intense band at 1180 cm-1 The SG-SER spectrum has greater similarity to the E-SER spectrum at -03VSCE and suggests that the silver particles embedded in the sol-gel behave as if at a potential positive of the pzc RNA samples extracted from E coli B subtilis and S aureus were next examined by both E-SERS and SG-SERS E-SER spectra of these samples yielded quality spectra in 10 minutes in which all of the major features can be identified (Figures 6 and 7) This includes guanine at 650 cm-1 adenine at 791 cm-1 cytosine and uracil combining at 790 cm-1 and phosphate at 1100 1335 (in combination with adenine and guanine) 1465 and 1570 cm-1 Surprisingly adenine which demonstrated the greatest surface-enhanced Raman effect does not dominate the ring-breathing mode portion of the spectrum The intensities of the other base-pairs bands are of the same order of magnitude This suggests that when the base-pairs are linked together as in RNA they are enhanced in concert In fact the relative intensities are very similar to a normal Raman spectrum of E coli RNA which shows the combined cytosine and uracil band at ca twice the intensity of the adenine band and ca four times the intensity of the guanine band Unfortunately this means that the independent enhancement factors for the NA bases can not be used to estimate relative concentrations For example the relative 791 and 734 cm-1 bands for B subtilis would indicate that the cytosine andor uracil concentration was at least 20 times the adenine concentration whereas each of the four RNA bases are known to contribute 15-35 Nevertheless it is worth noting that the three RNA samples yield different relative band intensities that were reproduced in numerous measurements Although the relative concentrations of the NA bases for these samples have not been determined these differences can be quantified If it is assumed that the 650 cm-1 band represents 25 guanine the 791 cm-1 band represents 25 adenine and the 790 cm-1 50 cytosine plus uracil in the E coli RNA spectrum then the relative concentrations can be estimated for the other RNA samples To aid this calculation the three spectra were normalized to the phosphate band at 1100 cm-1 which has been shown to correlate to the total phosphate concentration and can be used as an internal standard In addition a simple baseline correction was applied (Figure 7) This yields 15 adenine 30 guanine and 55 cytosine plus uracil for B subtilis RNA and 18 adenine 25 guanine and 57 cytosine plus uracil for S aureus RNA The average SN of these measurements was 26 with an average error of 8 of the value (SplusmnN) It is also worth noting that the three RNA spectra show a marked shift in a band near 825 cm-1 This band is assigned to the symmetric stretch of the O-P-O ester linkage9 The band appears at 815 cm-1 for S aureus shifting to 820 cm-1 for B subtilis and 830 cm-1 for E coli Others have used the normal Raman intensity of the band at 815 cm-1 as a direct indication of the amount of A-class helix present while the intensity of the band at 830 cm-1 has been used as a direct indication of the amount of B-class helix present However the latter is more associated with DNA than RNA
Wavenumbers (∆cm-1) Figure 5 E-SER spectra of 020 mgmL adenosine monophosphate at A) -03 and B) -09VSCE and C) SG-SER spectra Conditions sample in 01M KCl buffered to pH 92 A) and B) 750 mW C) 75 mW of 1064 64 scans (1-min) at 8 cm-1
Wavenumbers (∆cm-1) Figure 6 E-SERS of 01 mgmL RNA from E coli 02 mgmL RNA from B subtilis and 02 mgmL RNA from S aureus Conditions 01M KCl pH 92 -03VSCE 750 mW of 1064 nm 640 scans (10 min) at 8 cm-1
AMP RNA
E coli A
B
C
B subtilis
S aureus
SPIE 2001-4575
69
SG-SER spectra of reasonable quality were also obtained for E coli and B subtilis especially the latter (Figure 8) However the spectra differ substantially from the E-SERS of the same samples Both SG-SER spectra are dominated by adenine at 735 cm-1 and a band at 1030 cm-1 Although unassigned the latter does appear in the RNA E-SER spectra Bands at 1105 and 1565 cm-1 are likely due to phosphate while bands at 1320 and 1470 cm-1 are less confidently assigned to phosphate They are significantly less intense and somewhat shifted from their SG-SERS counterparts (1335 and 1455 cm-1) A band at 670 cm-1 may be due to guanine which was observed at 664 cm-1 for SG-SERS of the pure sample However the SG-SER spectrum of AMP also had an intense 667 cm-1 band A number of other bands occur at 890 1070 1165 1245 1290 1420 1505 cm-1 and remain unassigned The SG-SER spectra are somewhat disappointing in that only adenine and guanine contributions can be positively identified This limits the ability to determine relative NA base concentrations and distinguish bacterial RNA However several of the unassigned bands may be due to the bases (eg 1030 and 1420 cm-1 due to cytosine) Further experiments will be required to clarify this point
Ecoli
B subtilis
S aureas
P G
A C+U
OPO
A B
Wavenumbers (∆cm-1) Wavenumbers (∆cm-1) Figure 7 SER spectra of RNA from A) B subtilis with contributions indicated and B) E coli B subtilis and S aureus with baseline correction and peak positions used to calculate contributions indicated G = guanine A = adenine C+U = cytosine plus uracil P = phosphate (backbone) OPO = phosphate ester linkage (A- vs B-class helix)
Wavenumbers (∆cm-1) Wavenumbers (∆cm-1) Figure 9 E-SER (-03VSCE) and SG-SER spectra of RNA from A) E coli and B) B subtilis Sample conditions as in Figure 6 E-SER spectra at 750 mW SG-SERS at 75 mW
A B E-SERS
SG-SERS
E-SERS
SG-SERS
SPIE 2001-4575
70
A final comparison was made between the two SER methods by measuring dipicolinic acid (DPA) This chemical may be invaluable as a test for spore forming bacteria specifically B anthracis 50 to 90 of B anthracis sporilates During spore formation dipicolinic acid is synthesized and once completed 10-15 of the dry spore weight is composed of the Ca2+ complex located in the spore core42 Heating in water can be used to initiate germination at which point the exosporium breaks and releases the Ca dipicolinate which becomes dipicolinic acid in water The structure of this chemical strongly
4 CONCLUSIONS Towards the goal of developing a method and technology to rapidly detect and identify bioagents we have been investigating surface-enhanced Raman spectroscopy as a tool to measure relative concentrations of nucleic acid bases in RNA and determine bioagent taxonomy Initially we investigated E-SERS since this method has been extensively researched and the optimum sample conditions are well developed However this method requires a three-electrode sample cell and electrolyte solution Incorporation of an electrolytic cell into sample systems used to analyze for BWA as aerosols in water or on surfaces can be designed using flow injection analysis technologies but cross-contamination and plugging of sample lines seems inevitable For this reason we also investigated metal-doped sol-gels as a SER-active medium Previous studies have shown this material to be active in all solvents particularly water capable of continuous measurements in flowing systems and reproducible (quantitative) between coated sample vials Here we compared SG-SER spectra to traditional E-SER spectra of the nucleic acid base pairs adenosine monophosphate and RNA High quality spectra of adenine cytosine guanine thymine and uracil were obtained by both E-SERS and SG-SERS Both methods yielded very similar spectra for the NA bases including a pH dependent study of adenine Enhancement factors and detection limits for adenine were determined as 2x105 and 16x10-11g and 1x105 and 12x10-10g for E-SERS and SG-SERS respectively (normalized to 75 mW and 10-min acquisition time) Fifty percent of the silver particle surface area in the sol-gel matrix was assumed covered by adenine which may have been overestimated yielding a lower EF and higher detection limit It should also be realized that each E-SER spectrum required several attempts to optimize the measurement conditions (pH electrode potential etc) While each SG-SER spectrum involved no sample preparation and often represents the first and only attempt to make the measurement Quality spectra of RNA extracted from Escherichia coli Bacillus subtilis and Staphylococcus aureus were obtained by E-SERS that were easily interpreted Bands due to adenine guanine cytosine plus uracil and phosphate were identified The SER band intensity of the NA bases in the RNA samples were of the same order of magnitude suggesting that their interaction with the silver surface is concerted as is their Raman enhancement Interestingly the relative SER band
Wavenumbers (∆cm-1) Figure 10 A) Raman spectrum of solid dipicolinic acid B) Ax20 C) electrolytic SERS of 6x10-3 M dipicolinic acid in 01 M KCl at a potential of +07VSCE and pH of 4 and D) sol-gel SERS of 6x10-3 M dipicolinic acid Conditions for A and C as in Figure 1 C) 100 mW of 1064 nm 50 scans 8 cm-1
A B (Ax20)
C
D
suggested that it would be SER active However the E-SER spectrum was unstable and varied considerably as a function of potential A consistent spectrum was obtained at +06VSCE (Figure 10) This potential is not recommended for measurement because the surface is actively dissolving in solution The SG-SER spectrum was considerably more stable of higher quality and easily reproduced Bands at 660 825 1010 1390 1430 1570 1590 and 3075 cm-1 were observed Enhancement factors were determined for the two media using the symmetric ring stretching mode at 995 and 1010 cm-1 for the normal Raman and SER spectra respectively E-SERS yielded an EF of 5x103 while SG-SERS yielded an EF of 2x105 for DPA The SN of the latter suggests a detection limit of 20x10-10g (based on adenine coverage 75 mW and 10-min) The differences in SER activity for these two media may be attributed to the combined electrolytic potential of the solution chemical and metal15 Again the E-SERS suggests that the SG-SERS is at a potential positive of the pzc While the instability in the E-SERS may also be associated with surface interactions of two carboxylic acid groups of dipicolinic acid during the ORCs
DPA
SPIE 2001-4575
71
intensities for RNA extracted from E coli are very similar to those measured by normal Raman spectroscopy Although the relative percent that each of the NA bases contributed to each RNA sample was not determined reproducible band intensities allowed noting the following trends The percent adenine decreases while the combined percent cytosine and guanine increase for both B subtilis and S aureus compared to E coli Quality spectra were also obtained for the RNA samples by SG-SERS but only a few bands were readily identified Calculations of NA base concentrations by SG-SERS will require further research In light of recent events we summarize the capabilities of these SERS media in terms of rapid detection of B anthracis and dipicolinic acid However these capabilities must be qualified First and foremost the level to which SERS can distinguish bacteria or viruses has not yet been determined Development of a database of both DNA and RNA base concentrations for BWAs and common bacteria to establish the level of taxonomic identification is ongoing Second rapid collection of aerosol water or surface samples is being addressed by others who report trapping particles on filters from 100 liters of air per minute Third although not presented here we have developed methods to extract RNA or DNA from cells and spores for SER analysis within 10 minutes Finally we assume a detection limit of 3600 spores per 100 liters of air is required although a 50 lethal dosage of anthrax has not been established With these qualifications a mass detection limit for RNA using SERS is estimated as follows A single measurement is performed in ca 20 minutes (140 liters collected in 14 min RNA extracted in 8 min spectral acquisition and analysis in 10 min) The average human breaths 7 liters per minute therefore the analyzer must at the very minimum detect 5000 spores in 140 liters of air One spore is approximately 2x10-18m3 (1x1x2 microm) and if a density of 075 gcm3 is assumed this corresponds to a mass of 15x10-12g Each spore contains 4-12 RNA or 12x10-13g RNA for 8 If we assume 23 of the RNA can be isolated for analysis during lysis then the proposed instrument must be able to detect 4x10-10g RNA from 5000 spores per 70 liters of air within 10 minutes As noted above the mass detection limits for adenine were estimated at 16x10-11g and 12x10-10g for E-SERS and SG-SERS respectively Although these detection limits suggests that RNA from 5000 spores is detectable with the current instrumentation it is highly likely that only a portion of an RNA segment (eg 120-nucleotide 5S rRNA) is in contact with the metal surface and will contribute to the SER effect The SN for the RNA spectra were 110th of the average SN for the four individual RNA bases suggesting a 10 contribution Furthermore effective taxonomy will likely require knowing the NA base concentrations to 1 of the value (eg 25plusmn025) Again the average measurement error for the bases is 12 These values suggest that the E-SERS is within a factor of 4 of the required detection limit whereas the SG-SERS detection limit must be improved by 25 times The same arguments can be applied to the detection of dipicolinic acid If we assume a spore releases 10 by weight DPA during germination then the proposed instrument must be able to detect 75x10-10g DPA from 5000 spores per 70 liters of air within 10 minutes The detection limit for SG-SERS was estimated at 20x10-10g and suggest that the vials are suitable to perform a rapid screen for anthrax A series of concentration dependent measurements are currently being performed to verify this assertion Finally we note that the measurements performed here employed an FT-Raman spectrometer This instrumentation was chosen over dispersive Raman instrumentation for the high wavelength accuracy afforded by the HeNe reference laser (Connes Advantage43) which would allow reliable spectral subtraction matching of observed spectra to stored library spectra and confident use of chemometric approaches Such data analysis is likely to be required to enhance BWA identification However this instrumentation sacrifices sensitivity due to a less efficient detector (InGaAs vs Si) less efficient Raman scattering and less efficient generation of plasmon modes Substantial improvements can be obtained using visible excitation and Si detection and these measurements are underway
5 ACKNOWLEDGEMENTS The authors are grateful to Drs D Cookmeyer and S Tove of the US Army Research Office (Contract Number DAAH04-96-C-0078) for their interest and support of this research The authors would also like top acknowledge Dr R Yin and J Jensen for supporting development of the metal-doped sol-gels (Contract Number DAAD13-01-C-0019) They also thank Dr Wilfred H Nelson for assistance in spectral interpretations
SPIE 2001-4575
72
6 REFERENCES 1 Roberts WL and Rainey PM Clin Chem 39 1872-1877 (1993) 2 Pasechnik VA CC Shone and P Hambleton Bioseperations 3 267-283 (1993) 3 Jackson PJ ME Hugh-Jones DM Adair G Green KK Hill CR Kuske LM Grinberg FA Abramova and P
Keim Proc Natl Acad Sci 95 1224-1229 (1998) 4 Jensen JL NC Wong W R Loerop SPIE 4775-03 (2001) 5 Snyder A Pet al SPIE 3853-15 (1999) 6 Danosky T R and McFadden P N in press (1997) 7 Woodruff WH Farquharson S Science 201 831-833 (1978) 8 Lord RC and Thomas GJJr Spectrochemica Acta 23A 2551-2591 (1967) 9 Thomas GJJr Biochim Biophys Acta 213 417-423 (1970) 10 Placzek G Handbuch der Radiologie 2 EMarx ed Akademische Verlagagescellschatt Liepzig 1934 UCRL
Trans No 526 (1959) 11 Jeanmaire DL and RP Van Duyne J Electroanalytical Chem 84 1-20 (1977) 12 Chada S Manoharan R Moenne-Loccoz P Nelson WH Peticolas WL and Sperry JF Applied Spectroscopy
47 38-43 (1993) 13 Manoharan R Ghiamati E Chada S Nelson WH and Sperry JF Applied Spectroscopy 47 2145-2150 (1993) 14 Todd EA Morris MD Applied Spectroscopy 48 545-548 (1994) 15 Wentrup-Byrne E Sarinas S and Fredericks PM Applied Spectroscopy 47 1192-1197 (1993) 16 Weldon MK VR Zhelyaskov and MD Morris Applied Spectroscopy 52 265-269 (1998) 17 Kneipp K and J Fleming J Mol Structure 145 173-179 (1986) 18 Pothier NJ and Force RK Applied Spectroscopy 46 147-151 (1992) 19 Ervin KM E Koglin JM Sequaris P Valenta and HW Nurnberg J Electroanal Chem 114 179-194 (1980) 20 Kim SK TH Joo SW Suh and MS Kim J Raman Spectrosc 17 381-386 (1986) 21 Pothier NJ and Force RK Analytical Chemistry 62 678-680 (1990) 22 Kneipp K Y Wang RR Dasari and MS Feld Applied Spectroscopy 49 780-784 (1995) 23 Nie S and Emory SR Science 275 1102 (1997) 24 Graham D WE Smith AMT Linacre CH Munro ND Watson and PC White Analytical Chemistry 69 4703-
4707 (1997) 25 Fleischmann M PJ Hendra and AJ McQuillan Chem Phys Lett 26 163-166 (1974) 26 Pettinger B J Chemical Phys 85 7442-7451 (1986) 27 Surface-Enhanced Raman Scattering Section Four Theory SPIE MS 10 M Kerker and B Thompson Eds (1990) 28 Wang D-S and Kerker M SPIE (M Kerker and B Thompson Eds) MS 10 417-429 (1990) 29 Weaver MJ Farquharson S Tadayyoni MA J Chem Phys 82 4867-4874 (1985) 30 Norrod KL Sudnik LM Rousell D and Rowlen KL Applied Spectroscopy 51 994-1001 (1997) 31 Farquharson S Weaver WJ Lay PA Magnuson RH and Taube H J Am Chem Soc 105 3350-3351 (1983) 32 Lee PC and Meisel D ldquoAdsorption and Surface-Enhanced Raman of Dyes on Silver and gold Solsrdquo J Phys Chem
86 3391-3395 (1982) 33 Li Y-S and Wang Y Applied Spectroscopy 46 142-146 (1992) 34 Bello JM DL Stokes and T Vo-Dinh Analytical Chemistry 61 1779-1783 (1989) 35 van Duyne RP J C Hulteen D A Treichel M T Smith M L Duval and T R Jensen J Phys Chem B1033854-
3863 (1999) 36 Farquharson S and W W Smith W H Nelson and J F Sperry SPIE 3533-27 207-214 (1998) 37 Lee Y H W Smith S Farquharson H C Kwon M R Shahriari and P M Rainey SPIE 3537 252-260 (1998) 38 Lee Y-H S Farquharson and P M Rainey SPIE 3857 76-84 (1999) 39 Current Protocols in Molecular Biology Wiley Interscience1003-1006 (1987) 40 Farquharson S Smith W Carangelo R C and Brouillette C SPIE 3859 14-23 (1999) 41 Chase D B and JF Rabolt Fourier Transform Raman Spectroscopy Acad Press Ch1 p 131 (1994) 42 Brock TD MT Madigan JM Martinko and J Parker Biology of Microorganisms 7th Ed Prentice Hall p 76-80
(1994) 43 Connes J Rev Opt Theor Instrum 40 45 (1961)
166
Chemical agent identification by surface-enhanced Raman spectroscopy
Stuart Farquharson and Paul Maksymiuk
Real-Time Analyzers 87 Church Street East Hartford CT 06108
Kate Ong and Steven D Christesen US Army SBCCOM Aberdeen Proving Ground MD 21010
ABSTRACT The recent distribution of anthrax through the US postal system and the subsequent infection and death of several postal and national media employees amplifies the need for methods to rapidly detect identify and quantify this and other chemical and biological warfare agents The US military has also identified water supplies as a likely method of warfare agent deployment and is funding the development of a Joint Service Agent Water Monitor (JSAWM) In an effort to aid military personnel and the public at large we are developing a portable analyzer capable of identifying and quantifying chemical agents rapidly either on-demand or continuously The approach is based on the ability of Raman spectroscopy to identify molecular structure through the abundant vibration information provided in spectra and the ability of surface-enhanced Raman spectroscopy (SERS) to detect extremely low concentrations (eg part-per-billion) through the enhancement of Raman scattering by six orders of magnitude or more A key element to the analyzer design is a new SER active medium that is capable of quantitative reversible measurements The medium consists of silver or gold nanoparticles incorporated into a sol-gel matrix The porous silica network offers a unique environment for stabilizing SER active metals and the high surface area increases the interaction between the analyte and metal particles Here we present the use of new sol-gels that also selectively enhance chemicals based on polarity and charge Base-line measurements of chemical agents and their hydrolysis products are presented and compared to the JSAWM goal of 30 micrograms per liter detection Keywords Chemical warfare agent hydrolysis product SERS Raman spectroscopy sol-gel nanoparticle
1 INTRODUCTION Since September 11 2001 the threat of terrorist attacks and biological warfare within US borders has become a sobering reality The simplicity in manufacturing ease of deployment and the relatively low cost of chemical warfare agents (CWAs) raises public concern that they may also be used by terrorists Indeed terrorists released sarin (GB) in the Tokyo subway in 19951 Countering terrorism and terrorist attacks requires recognizing likely deployment scenarios and having the required technology to rapidly detect the deployment event One method of deployment has been long identified by the US military distribution through water supplies To counter this threat the Department of Defense is funding or monitoring the capabilities of 200 technologies with the goal of developing a Joint Service Agent Water Monitor (JSAWM) that is field portable2 The JSAWM must be capable of identifying the chemical agents with no false-positives and quantifying the agents at microgram per liter concentrations within 10 minutes (Table 1)3 This includes the analysis of drinking water supplies distribution and storage systems as well as potable water supplies The earliest technologies developed for CWA detection were based on electrochemical ionization or colorimetric analysis (eg phosgene tape) Although these analyzers were easy to use they were not generally agent specific and suffered from false-positives4 More traditional laboratory methods have also been investigated and in particular combined gas chromatography and mass spectrometry (GCMS) has been very successful at eliminating false-positives56 However GCMS requires extraction repeated calibration and long analysis times (typically 20 to 60 minutes)6 making it labor intensive and less than desirable for field use More rapid analysis of agents in the solid liquid and gas phase has been To whom correspondence should be addressed emailfarqureal-time-analyzerscom
Vibrational Spectroscopy-based Sensor Systems Steven D Christesen Arthur J Sedlacek III EditorsProceedings of SPIE Vol 4557 (2002) copy 2002 SPIE middot0277-786X02$1500
167
Table 1 Chemical Agent Structures Hydrolysis Half-lives and JSAWM Thresholds Agent Short-Hand Chemical Structure Hydrolysis
Half-Life JSAWM
Thresholds
Sarin (GB) F-[O=P-CH3]-O-CH(CH3)2 213 hours 32 microgL Soman (GD) F-[O=P-CH3]-O-CH(CH3)-(C-(CH3)3) 23 hours 32 microgL Tabun (GA) (CH3)2-N-[O=P-CN]-O-C2H5 41 hours 32 microgL VX C2H5O-[O=P-CH3]-S-(CH2)2-N-(CH(CH2)2)2 821 hours 32 microgL EA2192 HO-[O=P-CH3]-S-(CH2)2-N-(CH(CH2)2)2 gt9 years 32 microgL Mustard (H) ClCH2CH2-S-CH2CH2Cl encapsulates 47 microgL Lewisite (L) ClCH=CH-As-Cl2 rapid 27 microgL HCN HCN rapid 20 mgL BZ C7NH12-O-[C=O]-COH(C6H5)2 23 microgL T-2 Toxin 87 microgL at pH 7 to 75 and 20 to 25 oC
demonstrated by vibrational spectroscopy7-10 Hoffland et al7 reported infrared absorbance spectra and absolute Raman cross sections for several chemical agents while Christesen measured Raman cross sections for sarin tabun mustard gas and VX (ethyl S-2-diisopropylamino ethyl methylphosphonothioate)11 Again however these techniques also have limitations Raman spectroscopy is simply not a very sensitive technique and detection limits are typically 01 (1000 ppm) While infrared spectroscopy would have limited value in analyzing poisoned water since the very strong infrared absorption of water would obscure most other chemicals present Nevertheless efforts to overcome these limitations have been demonstrated Braue and Pannella8 quantified the G-series nerve agents (tabun sarin and soman) in terms of infrared attenuated total reflectance using a circle-cell And Alak and Vo-Dinh demonstrated the possibility of surface-enhanced Raman spectroscopy (SERS) to identify CWAs by measuring several organophosphonates that simulate the nerve agents12 However quantitative measurements have not been demonstrated for the SER-active material used (silver coated on alumina particles) or other SER-active media13 Recently we developed silver-doped sol-gels to promote the SER effect14-17 The porous silica network of the sol-gel matrix offers a unique environment for stabilizing SER-active metal particles and the sol-gel provides a high surface area that effectively increases the number of molecules observed within the Raman scattering volume The silver-doped sol-gels have been coated on the internal walls of standard glass vials to allow rapid SER analysis of small sample quantities (lt 01 mL) without preparation We have used p-aminobenzoic acid (PABA) as a test chemical to demonstrate surface enhancements greater than 106 reversible measurements in a flowing system reproducible measurements from vial-to-vial and batch-to-batch and measurements in multiple solvents including water14-17 Recently we used these vials to measure Tabun (GB) and Sarin and several hydrolysis products pinacolyl methyl phosphonate (PMP from Soman) and methyl phosphonic acid (MPA from all G-agents Figure 1) Although a number of unique vibrational bands are observed (eg C-N stretch doublet and P-C stretch) the G-agents were only observed for 5 concentrations and all spectra required baseline corrections Figure 1 Surface enhanced Raman spectra of ~5 vv A) Tabun and B) Sarin C) 1 vv PMP and D) 10 ppm MPA using sol-gel sample vials 785 nm excitation 1-min scan and CCD detection Performed at Aberdeen Proving Ground
C A
Wavenumber (∆cm-1) Wavenumber (∆cm-1)
D B
790
545 C-N
2135 2190 P-C 770
760 1290
Proc SPIE Vol 4577
168
Nevertheless MPA was readily observed for a 10 ppm sample with an estimated detection limit of 04 ppm (based on a signal-to-noise ratio of 3 for the 760 cm-1 band intensity) This measurement provides encouragement in that SERS may satisfy the needs of a JSAWM Furthermore MPA is also a hydrolysis product of VX and V-gas and EA2192 (Figure 2) and may prove a valuable indicator of agent usage Figure 2 Hydrolysis of Sarin to form hydrofluoric acid (HF) methylphosphonic acid 1-methylethyl ester (MPAMME) methyl phosphonic acid (MPA) and 2-propanol With this initial albeit modest success we began analyzing chemicals with various sol-gel compositions that we have been developing Here we describe four sol-gel compositions that select for 1) polar-positive 2) polar-negative 3) weakly polar-positive and 4) weakly polar-negative chemical species The ability of these sol-gels to select and enhance Raman scattering is described for several test chemicals and MPA
2 EXPERIMENTAL The chemicals analyzed as well as all chemicals used to prepare the metal-doped sol-gels were obtained at their purest commercially available grade from Aldrich (Milwaukee WI) The sol-gel designed to select for polar-negative species was prepared from a silver amine complex tetramethyl orthosilicate (TMOS) and methanol After mixing 02 mL of the sol-gel solution was transferred into a glass vial (2 mL) dried and heated The incorporated silver ions were then reduced using dilute sodium borohydride The vials were washed and dried prior to the addition of a sample solution In a similar manner the sol-gel designed to select for polar-positive species was prepared from a gold salt TMOS and methanol The sol-gel designed to select for weakly polar-negative species was prepared from a silver amine complex tetraethyl orthosilicate (TEOS) and methanol And the last sol-gel designed to select for weakly polar-positive species was prepared from a gold salt TEOS and methanol All samples were prepared in a chemical hood and transferred into plain or SER-active vials for analysis Normal Raman spectral measurements employed 1-mL pure samples that were placed in a 1-cm3 cuvette and weighed This yielded a powder density that allowed accurate calculation of molecules in the optical collection field SERS measurements employed 1-mg sample per mL water concentrations unless otherwise stated Once prepared a 01 mL sample was placed into one of the four selective sample vials which in turn was placed into the sample compartment of a Raman spectrometer for analysis A prototype Fourier transform Raman spectrometer (RTA) was used for these measurements18 The system consisted of a NdYAG laser (Brimrose) for excitation at 1064 nm an interferometer built by On-Line Technologies (OLT East Hartford CT) for frequency separation an uncooled InGaAs detector for signal detection (RTA) and an Intel 400 MHz Pentium II based laptop computer (Dell Round Rock TX) for interferometric control data acquisition (OLT) and analysis (LabVIEW by National Instruments Austin TX) Additional components included a Notch filter (Kaiser Ann Arbor MI) and interferometer entrance and exit optics (Edmund Scientific Barrington NJ) Fiber optics were used to deliver the excitation beam to the sample and the scattered radiation to the interferometer (1 meter lengths of 200 and 365 micron core diameter respectively Spectran Avon CT) A second Notch filter (Kaiser) was used as a beam splitter to direct the excitation beam along the same axis as the collected radiation A microscope object (20x04 Newport Irvine CA) was used to focus the beam into the sample and to collect the scattered radiation back along the same axis In this co-axial backscattering arrangement the excitation beam was passed through the outside of a glass vial and focused onto the silver-doped sol-gel film (01-03 mm thickness) containing the sample
3 RESULTS AND DISCUSSION p-aminobenzoic acid (PABA) and phenyl acetylene (PA) and were used to refine the selectivity and SER-activity of the four different metal-doped sol-gels PABA is a popular chemical used to evaluate the performance of SER-active media Here the polar end groups can be used to test selectivity of the polar-negative and polar-positive sol-gels PA is essentially non-
2O+ H HF + +OH
OH3H C
OP
OF
CH
CH3
3
3H C
OCP
OOH
CH
CH3
3
3H C
O
CPHO
CH
CH
3
3
C
Sarin MPAMME MPA 2-propanol
Proc SPIE Vol 4577
169
polar but a high electron density in the cylindrical π cloud around the carbon-carbon triple bond allows testing the selectivity of the weakly polar-negative and weakly polar-positive sol-gels As Figure 3 illustrates PABA passes through the polar sol-gel and is enhanced by either the silver or gold particles At 1 mgml the concentration of neutral PABA is ca 20 times that of the ionized form (pKa = 48) For electropositive silver the PABA anion is expected to interact through the carboxylate group and the associated vibrational modes are expected to dominate the spectrum Conversely for electronegative gold either form of PABA is expected to interact through the amine group The clear differences in our spectra support this expectation Furthermore bands at 840 and 1405 cm-1 assigned to a COO- bend and stretch respectively are significantly more intense for silver than gold Additional bands at 1140 and 1195 cm-1 are assigned to CH bending modes while bands at 1450 1500 and 1605 cm-1 are assigned to ring vibrational modes A very similar SER spectrum for PABA on a silver-coated alumina substrate has previously been reported with similar assignments19 For the gold-doped sol-gel new bands appear at 690 1355 and 1585 cm-1 The first band is assigned to a ring-H bending mode the second band to a ring-N- stretching mode and the third band to a possible NH2 scissors mode or ring mode The second band is not observed in the normal Raman spectrum but infrared bands occur at this frequency for aromatic ring-secondary amine stretching modes The scissors mode occurs at this frequency in Raman spectra for several chemicals but is absent in the PABA Raman spectrum Alternatively this mode may be the1600 cm-1 ring mode that has been shifted by the gold interaction Again a very similar SER spectrum of PABA has been reported but surprisingly using silver (colloids)2021 not gold as the enhancement medium These researchers also assumed the primary interaction of PABA with silver was through the carboxylate anion and made assignments accordingly For example they assigned the 1359 cm-1 to a COO- stretch not to the amine group as we have They also favor the ring stretching mode assignment for the 1582 cm-1 band Finally it should be said that other researchers have argued that the most dominant band in the SER spectra at 1450 cm-1 a ring vibration mode suggests that PABA lies flat on the surface and the π-orbitals dominate the surface interaction22
Figure 3 SER spectra of A) PABA using polar-negative and B) polar-positive sol-gels and C) PA using weakly polar-negative and D) weakly polar-positive sol-gels PABA is 1 mgmL PA is 1 vv Spectral conditions 75 mw 1064 nm 100 scans (15 min) 8 cm-1 resolution Non-polar PA passes through the non-polar sol-gels and is also enhanced by both metals The spectra are easily understood For electropositive silver PA interacts through the cylindrical triple bond π electron cloud and a -CequivC- doublet occurs near 2000 cm-1 The interaction is reasonably strong since this band appears at 2112 cm-1 in the normal Raman spectrum For electronegative gold this interaction is unlikely and only very weak bands occur near 2000 cm-1 The remaining bands are at 1000 cm-1 1200 cm-1 doublet and 1595 cm-1 all appear in the normal Raman spectra at virtually the same frequencies and are assigned to the symmetric ring-breathing mode CH bending modes and the trigonal ring-breathing mode respectively The polarnon-polar selectivity of the polar-negative and weakly polar-negative sol-gels was tested by adding a 11 molar mixture of PABA and PA The selective enhancement is quite good (Figure 4) The spectrum obtained using the polar sol-gel represents 78 PABA and 22 PA while the spectrum obtained using the weakly polar sol-gel represents 9 PABA and 91 PA The band peak intensities at 2000 cm-1 for PA and 1450 cm-1 for PABA were used for these calculations and are expanded in Figure 4 for clarity
C A
Wavenumber (∆cm-1) Wavenumber (∆cm-1)
2NH COOH
D B
C CH
Proc SPIE Vol 4577
170
Figure 4 SERS of 11 MM of PABA and PA in A) polar-negative and C) weakly polar-negative sol-gels The lower traces compare the pure chemicals B) 1 mgml PABA in polar-negative sol-gel and D) 1 PA in weakly polar-negative sol-gel while the insets magnify the minority species for clarity (x5 in A and x10 in B) Spectral conditions as in Figure 3 Following this development of selective sol-gels that maintained SER activity we measured cyanide and MPA (Figure 5) Not surprisingly the best sensitivity for both hydrolysis products was obtained using the polar-negative sol-gel The interaction of the cyanide anion with the silver surface is sufficient to shift the CequivN stretch observed at 2080 cm-1 in the normal Raman spectrum to 2145 cm-1 in the surface-enhanced Raman spectrum Furthermore the band is substantially broadened This anion has been extensively studied by electrolytic SERS and this shift and broadening have been attributed to the formation of a tetrahedral Ag(CN)3
2- surface structure23 Figure 5 Surface-enhanced (upper traces) and normal Raman spectra (bottom traces) of A) CN- and B) MPA in silver-doped TMOS SERS conditions as in Figure 3 and 1 mgmL Note MPA yields two distinct spectra for neutral (top) and acidic pH (middle) The normal Raman spectra employed pure powders 500 scans and 900 mW of 1064 nm SER measurements of MPA with the polar-negative sol-gel yielded two unique spectral signatures that depended on solution pH (Figure 5) For more neutral solutions the P-C stretch of MPA at 762 cm-1 dominates and the CH2 stretch at 2922 cm-1 appears The SN is sufficiently high that the anti-Stokes Raman shift at -762 cm-1 is observed For deprotonated MPA an oxygen-surface mode appears at 325 cm-1 (as well as its anti-Stokes complement) suggesting a strong interaction This results in substantial enhancement of the P-O-C mode at 1051 cm-1 (upper trace) Others report that this mode dominates the infrared spectra of nerve agents measured in water8 Comparison of the two spectra suggests the following molecule-to-surface orientations The appearance of the oxygen-surface and P-O-C modes in the upper spectrum of Figure 5B indicates that the tetrahedral molecule interacts with the silver surface through the deprotonated oxygen and is oriented end-on The
A B
Wavenumber (∆cm-1)
C A
Wavenumber (∆cm-1) Wavenumber (∆cm-1)
D B
Wavenumber (∆cm-1)
Proc SPIE Vol 4577
171
dominance of the P-C and the CH2 stretches and the disappearance of the P-O-C mode in the upper spectrum suggest the molecule is oriented with the methyl-hydroxyl-hydroxyl face towards the surface However considerably more research must be performed to verify these points Both chemical forms were measured as a function of concentration to determine the onset of monolayer coverage and estimate expected detection limits (Figure 6) Below monolayer coverage the signal to concentration dependence should be linear and the SN of any spectral measurement in this range can be used to predict the detection limit In the spectra presented here the peak height was used as the signal while the noise as root-mean-squared (RMS) was measured between 4400-4600 cm-1 Since noise is distributed evenly throughout the spectrum when transformed this region was used since it does not have any contributions from signals or baseline offsets Figure 6 shows a series of spectra for MPA along with a plot of the 1050 cm-1 band intensity (the noise was identical for this series) as a function of sample concentration A clear discontinuity appears in the vicinity of 01 mgmL (19 ppm) indicating the onset of monolayer coverage A detection limit defined as a SN of 3 was calculated for the 01 and 005gmL samples at 24x10-4 and 25x10-4 gL respectively A more modest detection limit of 101x10-4 gL was obtained using the 760 cm-1 band in the second series of concentration measurements These detection limits correspond to 46 and 190 ppb for a 3-min scan and 75 mW of incident laser power
Figure 6 A) Concentration dependence of MPA SERS measured in silver-doped TMOS) B) Concentrations are 001 005 01 05 1 gL (188 94 188 94 188 ppm) I760 series (bull) and I1050 series (∆)
Enhancement factors for cyanide and methyl phosphonic acid can be determined by estimating the number of molecules contributing to the surface-enhanced and normal Raman spectra The enhancement factor EF can be defined by the following equation
EF = (ISERSINR)bull(MNRMSERS) bull(PNRPSERS) bull(TNRTSERS)12
where I is the spectral band intensity M is the sample mass P is the incident laser power and T is the measurement time (or number of scans) for the two measurements For the normal Raman spectra a cylindrical scattering volume is assumed based on the laser area (28x10-7m2 6x10-4m diameter spot) and the penetration depth (1x10-3 m)24 The density of KCN and MPA as powders were measured at 0572 and 0516 gcm3 indicating that 16x10-4 and 144x10-4 g produced the normal Raman signals in Figure 5 respectively The number of molecules contributing to the SER spectra are those on the silver particles embedded in the sol-gel The total silver surface area can be determined from the average particle size concentration and the scattering volume Previous scanning electron micrographs showed the average silver particle size to be 40 nm in diameter (335x10-23m3)17 The silver concentration is 012M based on the reactant molar concentrations and dilution factors And the scattering volume is 76x10-
11m3 again based on a cylindrical scattering volume defined by a laser area of 28x10-7m2 and a sol-gel thickness of 27x10-
4m This volume contains 123x10-6g of silver equivalent to 35x109 silver particles with a collective surface area of 18x10-
5m2 However it may reasonably be assumed that at least half of the silver surface is in contact with the sol-gel matrix and unavailable for analyte interaction If we assume monolayer coverage and that each CN molecule occupies 15x10-20m2 then approximately 62x1014 molecules or 27x10-8g of CN contribute to the SER spectrum (20x10-19m2 46x1013 molecules 74x10-9g for MPA) Accordingly the EF for cyanide equals 48x104 ((180599) bull(16x10-427x10-8) bull(90075) bull(500100)12) The EF for MPA is considerably higher at 87x106 ((60326) bull(144x10-474x10-9) bull(90075) bull(500200)12)
0
20
40
60
80
100
120
140
0 02 04 06 08 1 12[MPA] (mgmL)
I (76
0)
0
100
200
300
400
500
600
I (10
50)
Wavenumber (∆cm-1)
A B
Proc SPIE Vol 4577
172
4 CONCLUSIONS Here we present for the first time surface-enhanced Raman spectra of Tabun and Sarin obtained using silver-doped sol-gels However the inferior enhancement suggested employing SER-active sol-gels varying in composition to improve both detection limits and selectivity of the target analytes To this end we successfully demonstrated the capabilities of four sol-gels that select for 1) polar-positive 2) polar-negative 3) weakly polar-positive and 4) weakly polar-negative chemical species p-aminobenzoic acid was used to show that silver could be used to attract polar-negative chemicals or functional groups (carboxylate anion) and that gold could be used to attract polar-positive chemicals (amine) while a mixture of p-aminobenzoic acid and phenylacetylene was used to show that tetramethyl orthosilicate preferentially solvates polar chemicals and that tetraethyl orthosilicate preferentially solvates weakly polar chemicals This increased sample control was applied to cyanide and methyl phosphonic acid two hydrolysis products of chemical warfare agents Exceptional results were obtained for methyl phosphonic acid allowing measurement of 1x10-2 gL for a 3-min scan and 75 mW of incident laser power with an estimated detection limit of 245x10-4 gL and an enhancement factor of 87x106 However this detection limit is 76 times less sensitive than required for the JSAWM (32x10-6gL for the G-agents) Finally we note that the measurements performed here employed an FT-Raman spectrometer This instrumentation was chosen over dispersive Raman instrumentation for the high wavelength accuracy afforded by the HeNe reference laser (Connes Advantage25) which would allow reliable spectral subtraction matching of observed spectra to stored library spectra and confident use of chemometric approaches Such data analysis is likely to be required to identify the chemical agents as well as distinguish hydrolysis products However this instrumentation which employs 1064 nm excitation and InGaAs detection sacrifices sensitivity We believe that the measurement sensitivity can be improved by at least two orders of magnitude by using shorter laser excitation wavelengths (eg 532 nm) This would provide more efficient Raman scattering (fourth power dependence on laser excitation wavelength) more efficient generation of plasmon modes and allow using more efficient detector material (Si vs InGaAs) These modifications are underway
5 ACKNOWLEDGEMENTS The authors would like to thank Dr R Yin and J Jensen of the US Army for supporting this work (Contract Number DAAD13-01-C-0019) They would also like to thank Advanced Fuel Research for making their laboratory facilities available
6 REFERENCES 1 Tu Anthony ldquoOverview of Sarin Terrorist Incidents in Japan in 1994 and 1995rdquo 6th CBW Protection Symposium
Stockholm Sweden 10-15 May 1998 2 Jensen JL NC Wong W R Loerop SPIE 4775-03 (2001) 3 JSAWM Requirements at wwwsbccomapgeaarmymilRDAecbcrtPRODSERJSAWMjsawmhtml 4 ldquoThe Chemical Weapons Convention Redefines Analytical Challengerdquo Analytical Chemistry News amp Features June 1
397A (1998) 5 Johnston RL Hoefler CM Fargo JC and Moberley B ldquoThe Defense Nuclear Agencyrsquos ChemicalBiochemical
Weapons Agreements Verification Technology Research Development Test and Evaluation Program and its Requirements for On-Site Analysisrdquo AT-ONSITE 5-8 (1994)
6 Black RM Clarke RJ Read RW and Reid MT ldquoApplication of gas chromatography-mass spectrometry and gas chromatography-tandem mass spectrometry to the analysis of chemical warfare samples found to contain residues of the nerve agent sarin sulphur mustard and their degradation productsrdquo J Chromatography 662 301-321 (1994)
7 Hoffland LD Piffath RJ Bouck JBrdquoSpectral signatures of chemical agents and simulantsrdquo Optical Engineering 24 982-984 (1985)
8 Braue EHJ Pannella MGrdquoCIRCLE CELL FT-IR Analysis of Chemical Warfare Agents in Aqueous Solutionsrdquo Applied Spectroscopy 44 1513-1520 (1990)
9 Tseng C-H Mann CK and Vickers TJ ldquoDetermination of Organics on Metal Surfaces by Raman Spectroscopyrdquo Applied Spectroscopy 47 1767-1771 (1993)
10 Staff Reporter ldquoUS Army Tests Lidar to Detect Biological Toxinsrdquo Photonic Spectra pg 50 December 1998
Proc SPIE Vol 4577
Proc SPIE Vol 4577
173
11 Christesen SD Raman cross sections of chemical agents and simulants Applied Spectroscopy 42 318-321 (1988) 12 Alak AM and Vo-Dinh T ldquoSERS of Organophosphorous Chemical Agentsrdquo Analytical Chemistry 59 2149-2153
(1987) 13 Norrod KL Sudnik LM Rousell D and Rowlen KL ldquoQuantitative Comparison of Five SERS Substrates
Sensitivity and Detection Limitrdquo Applied Spectroscopy 51 994-1001 (1997) 14 Lee Y and Farquharson S ldquoSERS Sample Vials Based on Sol-Gel Process for Trace Pesticide Analysisrdquo SPIE 4206
140-146 (2000) 15 Farquharson S and Lee Y ldquoTrace Drug Analysis by Surface-Enhanced Raman Spectroscopyrdquo SPIE 4200-16 (2000) 16 Lee Y Farquharson S and Rainey P M Surface-Enhanced Raman Sensor for Trace Chemical Detection in Water
SPIE 3857 76-84 (1999) 17 Lee Y Farquharson S Kwong H and Shahriari M ldquoSurface-Enhanced Raman Sensor for Surface-Enhanced Raman
Spectroscopyrdquo SPIE 3537 252-260 (1998) 18 Farquharson S Smith W Carangelo R C and Brouillette C ldquoIndustrial Raman Providing Easy Immediate Cost
Effective Chemical Analysis Anywhererdquo SPIE 3859 14-23 (1999) 19 Narayanan VA JM Bello JD Stokes and T Vo-Dinh Analusis 19 307-310 (1991) 20 Laserna JJ E L Torres and JD Winefordner Analytica Chemica Acta 469-480 (1987) 21 Torres EL and JD Winefordner Analytical Chemistry 59 1626-1632 (1987) 22 Suh JS DP DiLella M Moskovits J Phys Chem 87 1540-1544 (1983) 23 Benner RE R Dornhaus R Chang and BL Laube Correlations in the Raman spectra of cyanide complexes adsorbed
at silver electrodes with voltammograms Surface Science 101 341 (1980) 24 Chase D B and JF Rabolt Fourier Transform Raman Spectroscopy Acad Press Ch1 p 131 (1994) 25 Connes J Rev Opt Theor Instrum 40 45 (1961)
APPLIED SPECTROSCOPY 351
focusing the 488 nm laser beam 2 mm deep into thebulk of the crystals to avoid contributions from potentialdepletion layers As shown in Fig 1b the dependence ofthe Raman shift on the degree of deuteration is almostperfectly linear and ts very well with D 5 22684cmR1 24526 where D is the degree of deuteration (in )and R is the spectral mean of the PO4 vibration in cm21A linear correlation coef cient of 0998 indicates an ex-cellent linear dependence of the Raman peak shift withdegree of deuteration This result shows that the shift ofthe PO 4 peak is simply caused by the linear increase inatomic mass due to isotope substitution which decreasesthe length of hydrogen-like bonds
This excellent linear dependence allows us to map thepro le of the DH exchange layer at the surface of DKDPcrystals by acquiring Raman spectra and determining theposition of the PO 4 peak for various depths This methodis preferable over other methods such as determining thestrength of the OD vibration directly (eg at 715 cm21)because the position of the most intense peak in the Ra-man spectrum can be measured more precisely than theintensity of some of the weakest peaks in the spectrumThis is demonstrated in Fig 2 where depth-dependentRaman spectra (Fig 2a) and the resulting exchange layerpro les for two DKDP crystals are shown (Fig 2b) Thespectra in Fig 2a were obtained from a depth scan of aDKDP crystal with 75 degree of deuteration in thebulk grown at 45 8C The spectra start out as DKDP with30 deuteration close to the surface and approach thebulk DKDP spectrum within a few micrometers of depthThe fact that the relative degree of deuteration does notextend to 0 D is due to the limited depth resolution ofthe Raman microprobe which averages over 4 mm indepth Figure 2b depicts the resulting DH exchange layerpro les for this and a second crystal grown at 63 8Crespectively Both crystals had the same exposure to am-bient conditions and their main difference is the temper-ature at which they were grown The different exchangelayer pro les indicate that crystals grown at differenttemperatures have differen t proton conductiv ities 13
which leads to a difference in their rate of deuteriumdepletion The parameters controlling this behavior arecurrently the objective of a detailed study the results ofwhich will be reported elsewhere
CONCLUSION
In conclusion we have shown that the shift of the to-tally symmetric PO 4 stretch mode in the Raman spectrumof DKDP crystals scales linearly with degree of deuter-ation This allows us to correlate Raman peak positionsto deuteration levels in these crystals We have presenteda new technique to determine DH diffusion pro les inDKDP frequency conversion crystals based on micro-Ra-man spectroscopy This technique is fast inexpensiveand works under various environmental conditionswhich will allow us to better understand and control deu-terium depletion in DKDP crystals
ACKNOWLEDGMENTS
We would like to thank M Runkel for rst discovering DKDP crack-ing R Floyd for providing DKDP crystals and L Chase and A Burn-ham for their support and helpful discussions This work was performedunder the auspices of the US Department of Energy by the University
of California Lawrence Livermore National Laboratory under contract W-7405-Eng-48
1 J J De Yoreo A K Burnham and P K Whitman Int Mat Rev47 113 (2002)
2 C E Barker R A Sacks B M Van Wonterghern J A Caird JR Murray J H Campbell K Kyle R B Ehrlich and N DNielsen Proc SPIE-Int Soc Opt Eng 2633 501 (1995)
3 T Suratwala paper to be published4 Cleveland Crystals Inc httpwwwclevelandcrystalscom
KDPshtmltable5 E A Popova I T Savatinova and I A Velichko Sov Phys Solid
State 12 1543 (1971)6 I P Kaminow R C C Leite and S P S Porto J Phys Chem
Solids 26 2085 (1965)7 J A Subramony B J Marquardt J W Macklin and B Kahr
Chem Mat 11 1312 (1999)8 H Tanaka M Tokunaga and I Tatsuzaki Solid State Commun
49 153 (1984)9 R J Nelmes G M Meyer and J E Tibballs J Phys C 15 59
(1982)10 M A Yakshin D W Kim Y S Kim Y Y Broslavets O E
Sidoryuk and S Goldstein Laser Physics 7 941 (1997)11 I Takenaga Y Tominaga S Endo and M Kobayashi Solid State
Commun 84 931 (1992)12 C Krenn personal communication13 M Sharon and A K Kalia J Solid State Chem 21 171 (1977)
Rapid Dipicolinic Acid Extractionfrom Bacillus Spores Detectedby Surface-EnhancedRaman Spectroscopy
STUART FARQUHARSON ALAND GIFT PAUL MAKSYMIUK andFRANK E INSCOREReal-Time Analyzers Inc East Hartford Connecticut06108
Index Headings Dipicolinic acid Bacillus spores Anthrax Surface-enhanced Raman spectroscopy
INTRODUCTION
The anxiety caused by the distribution of anthrax en-dospores through the US postal system in October 2001was exacerbated by the long time required for positiveidenti cation of the Bacillus anthracis spores and the un-known extent of their distribution Since that time manymethods capable of rapid eld analysis have been inves-tigated to augment or replace the laboratory method ofgrowing microorganisms in culture media which takesdays to perform12 Prominent among these approachesare polymerase chain reactions (PCR)3 immunoassaysand detection of calcium dipicolinate as a biochemicalsignature PCR employs primers to separate organism-
Received 10 October 2003 accepted 14 November 2003 Author to whom correspondence should be sent
352 Volume 58 Number 3 2004
speci c nucleic acid sequences (eg capsular protein en-coding gene for Bacillus anthracis)4 and polymerases toamplify the segment until it is detectable Recently am-pli cation times have been substantially reduced andcomplete analysis can now be performed in an hour orless Immunoassay methods are also being developed thatuse competitive binding of the bioagent (as an antigen)and its labeled conjugate for a limited number of anti-bodies Although analyses can be performed in under 30minutes a well-de ned anthrax antigen has not yet beenidenti ed5ndash7 and consequently the false-positive rate isunacceptably high8
A number of other methods are being developed witha focus on the detection of calcium dipicolinate (CaDPA)and its derivatives as a B anthracis signature This is sobecause only spore-forming bacteria contain CaDPA andthe most common potentially interfering spores such aspollen and mold spores do not Relatively fast methodshave been developed to chemically extract CaDPA andthen detect it directly by uorescence9 or indirectly byluminescence1011 In the latter case hot dodecylamine(DDA) has been used to extract dipicolinic acid (DPA)and terbium has been utilized to form a highly lumines-cent DPA complex11 Although measurements have beenperformed in as little as ve minutes it was found thatas many as three concentration-dependent complexes canform each with different lifetimes This coupled withthe fact that the Tb31 cation produces the same lumines-cence spectrum makes determinations of low spore con-centrations problematic
It has been long known that Raman spectra of Bacillispores are dominated by bands associated with CaDPA12
and that these spectra may provide a suitable anthrax sig-nature at the genus level13 Since that time considerableimprovements in Raman instrumentation have led to lab-oratory measurements of single Bacilli spores14 and to eld measurements of spores captured from a mail-sort-ing system15 However the single spore measurementsrequired complex instrumentation that is not ruggedwhile the eld measurements required milligrams of sam-ple Furthermore the Raman spectra of both measure-ments contained uorescence contributions that would in-crease uncertainty in quanti cation
In related research we demonstrated that nanogramquantities of DPA could be detected by uorescence-freesurface-enhanced Raman spectroscopy (SERS)16 We alsodemonstrated that microliter volumes of chemicals canbe detected by SERS using metal-doped sol-gel-packedglass capillaries17 Towards the goal of developing a rap-id eld SERS-based anthrax spore detector we havecombined our previous research and we now report thatDPA can be extracted from a 10 mg B cereus spore sam-ple using DDA in 1 minute and can be detected by SERSin an additional 1 minute
EXPERIMENTAL
Dipicolinic acid (26-pyridinedicarboxylic acid DPA)and dodecylamine (DDA) were used as received fromSigma-Aldrich (Milwaukee WI) Lyophilized B cereusspores prepared according to the literature13 were sup-plied by the University of Rhode Island and used as re-ceived Multiple particles approximately 01 mm3 each
were separated and weighed at 5 to 15 mg representing05 to 15 million spores The sample masses were con-sistent with a previous determination of spore density at0081 gmL that indicated a high degree of entrained air
All chemicals used to prepare the silver-doped sol-gelcoated capillaries were also obtained and used as receivedfrom Sigma-Aldrich According to previously publishedprocedures17 two precursor solutions were preparedmixed and then drawn into 1-mm-diameter glass capil-laries The silver amine precursor consisted of a 51 vvratio of 1 N AgNO 3 to 28 NH3OH while the alkoxideprecursor consisted of a 21 vv ratio of methanol to te-tramethyl orthosilicate The alkoxide precursors weremixed with silver amine precursor in an 81 vv ratioApproximately 015 mL was drawn into the capillarycoating a 15-mm length After sol-gel formation the in-corporated silver ions were reduced with dilute sodiumborohydride which was followed by a water wash to re-move residual reducing agent
A 100 mL drop of a 50 mM DDA solution in ethanolpre-heated to 78 8C was added to each of the B cereusparticles to digest the spore coat After 1 minute the re-sultant solution was drawn into a SER-active capillarythat was immediately xed horizontally to an XY posi-tioning stage (Conix Research Spring eld OR) just in-side the focal point of an f 07 aspheric lens The lensfocused the beam into the sample and collected the scat-tered radiation back along the same axis A dichroic lter(Omega Optical Brattleborough VT) was used to re ectthe excitation laser to the lens and pass the Raman scat-tered radiation collected by the lens An f 2 achromatwas used to collimate the laser beam exiting a 200-mm-core-diameter source ber optic while a second f 2 ach-romat was used to focus the scattered radiation into a 365mm ber optic (Spectran Avon CT) A short-pass lterwas placed in the excitation beam path to block the sil-icon Raman scattering generated in the source ber fromre ecting off sampling optics and reaching the detectorA long-pass lter was placed in the collection beam pathto block the sample Rayleigh scattering from reachingthe detector A 785 nm diode laser (Process InstrumentsInc model 785-600 Salt Lake City UT) was used todeliver 100 to 150 mW of power to the sample A Fouriertransform Raman spectrometer (Real-Time Analyzersmodel IRA-785 East Hartford CT) and a silicon photo-avalanche detector (Perkin Elmer model C30902S Stam-ford CT) were used to acquire the SER spectra
RESULTS AND DISCUSSION
As an initial experiment the SER spectrum of 1 gLof DPA in water was measured using the newly devel-oped silver-doped sol-gel-coated capillaries (Fig 1A) Atthis concentration a high signal-to-noise ratio (SN) isobtained in 1 min In fact a reasonable spectrum is ob-tained in the same time frame for 1 mgL (Fig 1B) TheSER spectra are reasonably similar to the normal Raman(NR) spectrum obtained for a saturated solution of DPAin 1 N KOH (Fig 1C) and the following band shifts areobserved (NR to SER) 647 to 657 cm21 817 to 815cm21 998 to 1008 cm21 1384 to 1382 cm21 1434 to1428 cm21 and 1569 to 1567 cm21 Many of these bandshave been previously assigned1213 such as 998 cm21 to
APPLIED SPECTROSCOPY 353
FIG 1 SERS of DPA in water using silver-doped sol-gel-coated glasscapillary for (A) 1 gL and (B) 1 mgL (C ) NR of saturated DPA in 1N KOH in a glass capillary Spectral conditions (A) and (B) 150 mWof 785 nm 1-min acquisition time (C) 450 mW of 785 nm 5-minacquisition time both 8 cm 21 resolution
FIG 2 SERS of DPA extracted from 10 mg B cereus particle using100 mL of 50 mM hot DDA acquired in (A) 1 minute and (B) 2 seconds(C ) Attempted SERS of 50 mM hot DDA in ethanol using silver-dopedsol-gel-coated glass capillary acquired in 1 min Spectral conditions150 mW of 785 nm 8 cm21 resolution
the symmetric ring stretch 1384 cm21 to the OndashCndashOsymmetric stretch 1428 cm21 to the symmetric ring CndashH bend and 1569 cm21 to the asymmetric OndashCndashOstretch
The rst B cereus samples consisted of 2 mg of sporesin 2 mL of 5 mM hot DDA The samples were main-tained at 78 8C for 40 min and while hot approximately10 mL was drawn into a SER-active capillary Since spec-tra of DPA were obtained for these initial samples small-er spore masses higher DDA concentrations and shorterheating periods were examined In due course it wasfound that 10 mg of spores could be digested by 100 mLof 50 mM hot DDA in one minute and detected (Fig2A) In fact the signal was suf ciently intense that it canbe observed in as little as two seconds (Fig 2B) Theamount of DPA that was extracted was estimated to bebetween 5 and 10 mgL by comparing the signal intensityof the 1008 cm21 band to that measured for DPA in waterThis is consistent with previous research that found thatthe majority of the DPA is extracted from spores usingDDA11 and that B cereus spores contain approximately10 DPA by weight18 The SN of 127 for the 1008 cm21
band in the 1-minute SER spectrum suggests a limit ofdetection of approximately 250 ng of B cereus sporesbased on a SN of 3 Finally it should be noted that DDAdid not produce a detectable SER spectrum as shown inFig 2C
CONCLUSION
We have demonstrated that by combining rapid extrac-tion of dipicolinic acid from Bacillus cereus spores withchemical identi cation by surface-enhanced Raman spec-troscopy as little as 10 mg of spores can be detected Infact the entire measurement from the time of adding hotdodecylamine to the spores to the time when the dipi-colinic acid SER spectrum is acquired and analyzedcould be performed in less than two minutes The abilityof this method to distinguish between spore-forming bac-teria such as Bacillus anthracis and non-DPA containingpowders could help prevent costly shutdowns associated
with the appearance of suspicious material or intentionalmailing of common substances as an anthrax hoax Thismethod could also prove useful in detecting the locationof anthrax endospores in mail distribution facilities if an-other veri ed attack should occur
Research continues to fully characterize the surface-enhanced Raman spectroscopy signal intensities as afunction of sample concentration and to explore otherextractants that do not require the use of elevated tem-perature
ACKNOWLEDGMENTS
The authors are grateful for the support of the National Science Foun-dation (DMI-0296116 and DMI-0215819) and the US Army(DAAD13-02-C-0015 Joint Service Agent Water Monitor program)The authors are indebted to Chetan Shende for preparing the sol-gelcapillaries The authors also thank James Gillespie Nicholas Fell andAugustus Fountain for providing important background informationMark Farquharson for laboratory support and Professor Jay Sperry ofthe University of Rhode Island for supplying B cereus spores
1 V A Pasechnik C C Shone and P Hambleton Bioseparations 3267 (1993)
2 P J Jackson M E Hugh-Jones D M Adair G Green K K HillC R Kuske L M Grinberg F A Abramova and P Keim ProcNatl Acad Sci USA 95 1224 (1998)
3 B R Glick and J J Pasternak Molecular Biology Principles andApplications of Recombinant DNA (ASM Press Washington DC1994)
4 C A Bell J R Uhl T L Had eld J C David R F Meyer TF Smith and F R Cockerill III J Clin Microbiol 40 2897(2002)
5 D L Gatto-Menking H Yu J G Bruno M T Goode M Millerand A W Zulich Biosens Bioelectron 10 501 (1995)
6 J J Quinlan and P M Foegeding J Rapid Methods AutomationMicrobiol 6 1 (1998)
7 A A Hindle and E A H Hall Analyst (Cambridge UK) 1241599 (1999)
8 M S Ascher US Department of Health amp Human Services(httpwwwhhsgovophppresentationsAscherdoc)
9 R Nudelman B V Bronk and S Efrima Appl Spectrosc 54 445(2000)
10 D L Rosen C Sharpless and L B McBrown Anal Chem 691082 (1997)
354 Volume 58 Number 3 2004
11 P M Pellegrino N F Fell Jr and J B Gillespie Anal ChimActa 455 167 (2002)
12 W H Woodruff T G Spiro and C Gilvarg Biochem BiophysRes Commun 58 197 (1974)
13 E Ghiamati R S Manoharan W H Nelson and J F SperryAppl Spectrosc 46 357 (1992)
14 A P Esposito C E Talley T Huser C W Hollars C M Schal-dach and S M Lane Appl Spectrosc 57 868 (2003)
15 S Farquharson L Grigely V Khitrov W W Smith J F Sperryand G Fenerty J Raman Spectrosc paper accep ted (2003)
16 S Farquharson W W Smith S Elliott and J F Sperry SPIE-IntSoc Opt Eng 3855 110 (1999)
17 S Farquharson and P Maksymiuk Appl Spectrosc 57 479(2003)
18 F W Janssen A J Lund and L E Anderson Science (Washing-ton DC) 127 26 (1958)
SPIE -2003-5269 117
pH dependence of methyl phosphonic acid dipicolinic acid and cyanide by surface-enhanced Raman spectroscopy
Stuart Farquharson Alan Gift Paul Maksymiuk Frank Inscore and Wayne Smith
Real-Time Analyzers 87 Church Street East Hartford CT 06108
ABSTRACT US and Coalition forces fighting terrorism in Afghanistan and Iraq must consider a wide range of attack scenarios in addition to car bombings Among these is the intentional poisoning of water supplies to obstruct military operations To counter such attacks the military is developing portable analyzers that can identify and quantify potential chemical agents in water supplies at microgram per liter concentrations within 10 minutes To aid this effort we have been investigating the value of a surface-enhanced Raman spectroscopy based portable analyzer In particular we have been developing silver-doped sol-gels to generate SER spectra of chemical agents and their hydrolysis products Here we present SER spectra of methyl phosphonic acid and cyanide as a function of pH an important factor affecting quantitation measurements which to our knowledge has not been examined In addition dipicolinic acid a chemical signature associated with anthrax-causing spores is also presented Keywords Chemical warfare agents agent detection agent hydrolysis SERS Raman spectroscopy homeland security
1 INTRODUCTION In the past decade the Unites States and its allies have been challenged by a different kind of warfare exemplified by the terrorist attacks of September 11 2001 Suicide bombings and the use of chemical agents are the norm and military personnel must consider a wide range of attack scenarios Among these is the intentional poisoning of water supplies to obstruct military operations in Afghanistan and Iraq Currently colorimetric paper is used to detect agents on-site while gas chromatography combined with mass spectrometry (GCMS) is used in mobile support laboratories However both methods have severe drawbacks The paper changes color in response to contact with many chemicals besides CWAs causing a high incidence of false positives1 while GCMS although very chemically specific requires hours to perform and constant re-calibration234 Military operations would be greatly aided by a portable analyzer that can identify and quantify potential chemical agents at concentrations that impact safety This includes the analysis of drinking water supplies distribution and storage systems To meet this goal the Department of Defense has been investigating numerous approaches under the auspices of the Joint Service Agent Water Monitor (JSAWM) program5 The JSAWM must be capable of identifying the chemical agents with no false-positives and quantifying the agents at microgram per liter concentrations within 10 minutes678 Recently we and others have been investigating the ability of surface-enhanced Raman spectroscopy (SERS) to measure chemical agents9-12 bioagents13-17 and their hydrolysis products in water SERS employs the interaction of surface plasmon modes of metal particles with target analytes to increase scattering efficiency by as much as 1 million times18 In 1987 the potential of this approach to measure CWAs was demonstrated by measuring a series of pesticides19 Several of these organophosphonates have chemical structures similar to CWAs in particular P=O functional groups In our studies we have been employing silver-doped sol-gels to promote the SER effect The porous silica network of the sol-gel matrix offers a unique environment for immobilizing and stabilizing SER-active metal particles20-23 The sol-gels have been coated on the internal walls of standard glass vials to allow rapid SER analysis of small sample quantities without preparation We have measured over 100 chemicals with enhancements of 104 to 106 demonstrated reversible measurements in a flowing system reproducible measurements from vial-to-vial and measurements in multiple solvents including water20-23 Previously we used these vials to perform preliminary measurements of cyanide (CN) methylphosphonic acid (MPA) and dipicolinic acid (DPA) MPA is a hydrolysis product of the nerve agents (eg sarin Reaction 1) and may be a valuable indicator of nerve agent usage particularly since the alkyl methylphosphonic acids are relatively more stable than their corresponding parent complexes24 DPA is
SPIE -2003-5269 118
a chemical signature of spore forming bacteria such as Bacillus anthracis And in light of the inability to rapidly detect the anthrax spores distributed through the US mail in October 2001 a number of methods are being developed to extract and analyze this signature Reaction 1 Stepwise hydrolysis of Sarin to form hydrofluoric acid (HF) isopropyl methylphosphonic acid (IMPA) then methyl phosphonic acid (MPA) and 2-propanol In our previous SERS investigations MPA and DPA were measured at 50 and 100 mgL respectively In both cases limits of detection (LOD) were estimated at 100 microgL providing encouragement in that SERS may satisfy the needs of the JSAWM Since it has been shown that pH can substantially influence the intensity of SER bands25 which would clearly influence quantitative analysis we undertook the present study to determine the severity of these effects for cyanide methyl phosphonic acid and dipicolinic acid Furthermore we previously observed a band at 1050 cm-1 for MPA6 possibly due to an anion formed at basic pH Here we investigate the source of this spectral anomaly
2 EXPERIMENTAL All chemicals including potassium cyanide methyl phosphonic acid dipicolinic acid and those used to prepare the silver-doped and gold-doped sol-gel coated vials were obtained and used as received from Sigma-Aldrich All samples were prepared in HPLC grade water (Fischer Scientific Fair Lawn NJ) for SERS measurements The pH of these samples was adjusted using dilute nitric acid or potassium hydroxide and verified using a pH electrode (Corning Inc Corning NY) that had been calibrated with pH 400 700 and 1000 buffered standards from Fischer Scientific Once prepared the samples were transferred into the silver-doped sol-gel vials (Real-Time Analyzers Inc East Hartford CT) The vials were coated in a manner similar to that previously reported by adding ammonium hydroxide to a solution of silver nitrate tetramethyl orthosilicate (TMOS) and methanol20 Gold-vials were coated by adding nitric acid to a solution of gold tetrachloride TMOS and methanol The two precursor solutions were prepared mixed and transferred to 2-mL glass vials dried and heated After sol-gel formation the incorporated metal ions were reduced with dilute sodium borohydride (1mgmL) which was followed by a water wash to remove residual reducing agent After the resultant analyte solution was introduced the SER-active vial was immediately fixed horizontally to an XY positioning stage (Conix Research Springfield OR) just inside the focal point of an f07 aspheric lens The lens focused the beam into the sample and collected the scattered radiation back along the same axis A dichroic filter (Omega Optical Brattleborough VT) was used to reflect the excitation laser to the lens and pass the Raman scattered radiation collected by the lens An f2 achromat was used to collimate the laser beam exiting a 200 microm core diameter source fiber optic while a second f2 achromat was used to focus the scattered radiation into a 365 microm fiber optic (Spectran Avon CT) A short pass filter was placed in the excitation beam path to block the silicon Raman scattering generated in the source fiber from reflecting off sampling optics and reaching the detector A long pass filter was placed in the collection beam path to block the sample Rayleigh scattering from reaching the detector A 785 nm diode laser (Process Instruments Inc model 785-600 Salt Lake City UT) was used to deliver 100 to 150 mW of power to the sample A Fourier transform Raman spectrometer (Real-Time Analyzers model IRA-785 East Hartford CT) and a silicon photo-avalanche detector (Perkin Elmer model C30902S Stamford CT) were used to acquire the SER spectra
3 RESULTS AND DISCUSSION In a previous study of MPA6 aimed at developing a concentration calibration curve and determining limits of detection (LOD) we observed an anomaly at 1050 cm-1 Since it was found that the band intensity changed as a function of concentration the band must be associated with a sample parameter Two possible parameters photon flux and pH are examined here The first parameter was investigated by irradiating a 1mgmL MPA sample in a SER-active vial with laser powers of 200 mW and above and monitoring spectral changes It was immediately found that the 1050 cm-1 band
2O+ H HF + +OH
OH3H C
OP
OF
CH
CH3
3
3H C
OCP
OOH
CH
CH3
3
3H C
O
CPHO
CH
CH
3
3
C
Sarin IMPA MPA 2-propanol
SPIE -2003-5269 119
appeared and grew as a function of time and that the higher the laser power the faster the growth Figure 1 shows the growth of the 1050 cm-1 band over the course of 30 minutes when using 150 mW of 785 nm excitation while Figure 2A shows that the growth can be fit with a first order exponential equation namely I1050 = 03+05e-013t Furthermore the 760 cm-1 band could be fit with a first order decay equation with an identical exponential rate constant ie I760 = 08-
08e-013t The rates represent classical first order kinetics and their correspondence allows one to conclude that MPA is being transformed one-for-one into a photo-generated product At this time the photoproduct has not been positively identified but phosphonic (phosphorous) acid and phosphonate are likely candidates since the symmetric P(OH)3 stretch occurs at ~1050cm-1 Our studies show that a reduction in laser power to 100 mW at the sample for MPA essentially eliminates this degradation process This laser power or lower was used for further measurements
Methyl phosphonic acid is a diprotic acid that stepwise dissociates into two anions MPA- and MPA= according to the following reactions26
MPA MPA- + H+ pKa1 = 212 Reaction 2
MPA- MPA= + H+ pKa2 = 729 Reaction 3 The relative concentrations of MPA MPA- and MPA= can be determined at any pH by expressing [MPA] and [MPA=] in terms of [MPA-] using Reactions 2 and 3 and summing all three to equal the total starting concentration here 2 mgmL (0021M MW = 9602) viz
[MPA] + [MPA-] + [MPA=] = 0021M Equation 1 substituting from Reactions 1 and 2
([H+][MPA-])K1a + [MPA-] + (K2a[MPA-])[H+] = 0021M Equation 2 rearranging [MPA-] = 0021M(1+[H+]K1a + K2a[H+]) Equation 3 The relative concentrations of MPA MPA- and MPA= as a function of pH are shown in Figure 3 It is worth noting that near neutral pH both MPA- and MPA= will be present To confirm that the SER signal followed this pH dependence a starting solution consisting of 20 mg of MPA in 10 mL HPLC grade water was prepared and brought to pH of 20 using dilute nitric acid From this solution 2 mL were added to a SER-active vial and the SER spectrum recorded At this pH a peak at 760 cm-1 was barely discernable The 2 mL solution was returned to the starting solution and the pH was re-measured to correct for any changes that the silver-doped sol-gel vials might cause In most cases the change was less than 02 pH units and the pH is reported as the before and after average Next the pH of the
Figure 2 A) Exponential growth of 1050 cm-1 band and B) exponential decay of 760 cm-1 band for spectral series in Figure 1
Figure 1 Growth of 1050 cm-1 band as a function of time due to exposure to 150 mW of 785 nm Spectra are 5 sec each collected every 100-sec from 0 to 30-min
B
570 770 970 1170 1370Raman Shift (cm-1)
Arbit
rary
Unit
s
MPA Photodegradation
600 800 1000 1200 1400 Wavenumber (cm-1)
A
0 10 20 30 0 10 20 30 time (min) time (min)
Ram
an In
tens
ity (r
elat
ive)
30
min
0
SPIE -2003-5269 120
starting solution was adjusted to 325 using dilute KOH Again 2 mL were added to a vial and the SER spectrum recorded At this pH a reasonably strong 760 cm-1 band was observed This process was repeated as spectra were recorded at pHs of 70 74 75 79 85 and 100 A total of 1 mL of KOH was added diluting the total concentration by 10 Next the pH of the starting solution was made acidic by adding dilute nitric acid dropwise This time spectra were recorded at pHs of 72 69 64 and 37 Figure 4 shows the SER spectra for representative pHs (spectra were left out to simplify the figure) while Figure 3 shows the 760 cm-1 peak intensities as a function of pH (The band intensities were adjusted to compensate for dilution effects caused by the addition of HNO3 and KOH then normalized to 0021 M for the most intense band observed at pH 37) It is clear from Figure 3 that the 760 cm-1 band follows the MPA- concentration as a function of pH and must be assigned to this anion No bands were observed that corresponded to MPA or MPA= The lack of an MPA SER spectrum may be due to the absence of an attraction between the neutral analyte and the electropositive silver surface The same reasoning suggests that a strong SER spectrum should be observed for MPA= but it is not and a satisfactory explanation has not been found
0000
0005
0010
0015
0020
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14pH
Con
cent
ratio
n [M
]
pK1 = 212 pK2 = 729
MPA- MPA=
MPA
Figure 4 SER spectra of 002M MPA as a function of pH Conditions 100 mW of 785 nm 36 scans (1 min) 8 cm-1 recorded 2 min after sample introduction pH 19 and 20 not apparent on this scale pH 69 and 74 near identical to 70 and 75 and not shown for clarity
00
02
04
06
08
10
12
14
16
18
0 5 10 15 20 25 30Measurement Number
Ram
an In
tens
ity (7
60 c
m-1
)
Figure 3 Concentration dependence of MPA MPA- and MPA= as a function of pH for a 002M sample Intensity of 760 cm-1
band from Figure 6 as a function of pH ( for increasing basic adjustment for increasing acidic adjustment error was measured at ~10 for pH 64)
Figure 5 SER spectra of 002M MPA at pH 64 measured around a vial at three heights (9 points per height) Conditions as in Figure 4 but 10-sec scans
Since these measurements involved the removal and replacement of the SER-active vial in the sample holder to remove and add sample variation in the intensity as a function of vial position was minimized by illuminating the exact same height along the vial wall But this does not account for variability of the SERS response of the sol-gel coating around the vial To analyze this effect a vial containing MPA at pH 64 was rotated at ~ 40o
intervals at the original height and 18rdquo above and below this value Figure 5 shows the intensity of the 760 cm-1
band for the 27 positions It was found that the average value was 137plusmn014 an RSD of 10 overall and 5 for each height An error bar is included in Figure 4 for the pH 64 measurement
SPIE -2003-5269 121
With the above analyses in mind a preliminary investigation of the SER spectral response for MPA (as MPA-) as a function of concentration was performed A single vial was used for these measurements beginning with 1 mgL followed by measurements of 10 100 and 1000 mgL In all cases the pH was ~7 and 3 positions around the vial were measured per concentration Since the 760 cm-1 band was not observed for concentrations of 1 or 10 mgL using 100 mW of 785 nm the laser power at the sample was raised to 200 mW beginning with the 10 mgL concentration Photo-degradation was largely avoided (and not observed) by exposing the sample for only 33 seconds per spectral acquisition Representative spectra for 10 100 and 1000 mgL are shown in Figure 6 while a plot of the 760 cm-1 band intensity as a function of concentration is shown in Figure 7 These values were also used to estimate limits of detection based on the signal-to-noise ratio (SN) of the 760 cm-1 band using 100 mW of 785 nm laser excitation and a 10-min acquisition time As summarized in the Figure 7 inset the lower the measured concentration the lower the predicted LOD Even if as estimated 210 microgL could be measured using the silver-doped sol-gel vials an improvement of a factor of 70 is still required to meet JSAWM goals of measuring 3 microgL in 10-minutes Similar to MPA DPA is a diprotic acid (pKa1 = 216 and pKa2 = 692) and variations in pH will effect the relative concentrations of DPA DPA- and DPA= and possibly the SER spectra and band intensities This could prove significant if an acid or base is used to denature anthrax spores with the goal of extracting and analyzing DPA The MPA pH study described above was mimicked for DPA except that the starting solution consisted of 20 mg of DPA in 20 mL HPLC grade water (60x10-3M MW = 1671) The initial solution had a pH of 245 which was made basic by dilute KOH to pHs of 355 433 487 559 1069 and 1166 SER spectra were recorded at each pH using 100 mW of 785 nm and a 44-sec acquisition time Next one drop of concentrated nitric acid was used to remake the solution acidic at a pH of 200 Again sequential pH measurements were performed at 383 510 735 and 822 The solution pH was made acidic a third time but to pH 219 171 then 135 Throughout this process no more than 20 drops of acid or base were added and therefore the concentration was diluted by no more than 10 Most of the spectral bands showed a minor decrease in intensity as a function of increasing pH values However the bands at 525 and 795 cm-1 showed the most dramatic changes which occurred at acid pH Figure 8 shows the SER spectra of DPA for the spectral region and pH range of interest The identity of the DPA species was determined by plotting the normalized peak intensities with the lowest value set to 0 and the highest to 0006 M as a function of pH and overlaying these values on a plot of the relative concentrations for DPA DPA- and DPA= as previously done for MPA (Figure 9) As can be seen the 525 cm-1 band clearly corresponds to DPA The correspondence of the 795 cm-1 band to this species is less clear as the band retains intensity until pH of 55 This can be attributed to contribution to the overlapping band at 810 cm-1 which does not change as a function of pH The fact that most bands are observed at all pHs suggest that the primary interaction with silver is through the ring nitrogen This is supported by the fact that the most intense band occurs at 1008 cm-1 attributed to a symmetric ring breathing mode and that this interaction has been characterized for pyridine in numerous papers27
0
1
2
3
4
5
6
7
0 200 400 600 800 1000 1200
MPA Concentration (mgL)
760
Ban
d In
tens
ity
Figure 6 SER spectra of MPA in water at A) 1000 B) 100 and C) 10 mgL Conditions pH of 7 silver-doped sol-gel coated vial 200 mW of 785 nm 33-sec 8 cm-1 resolution
Figure 7 Plot of SER intensity of 760 cm-1 band of MPA as a function of concentration using 200 mW of 785 nm Inset table includes average intensity LOD standard deviation and percent deviation for each concentration but for 100 mW and 10 min
conc (mgL) LOD ave stddev dev10 021 002 1021100 072 010 14301000 312 040 1280
A
B C
SPIE -2003-5269 122
The ability to detect dipicolinic acid (DPA) by SERS is demonstrated in Figure 10 Even at 1 mgL the primary bands are visible The intensity of the dominant ring breathing mode observed at 1008 cm-1 for DPA in water (pH = 25 to 55) is plotted as a function of concentration in Figure 11 These values were also used to estimate limits of detection based on the SN of the 1008 cm-1 band using 100 mW of 785 nm laser excitation and a 10-min acquisition time Again the lower the measured concentration the lower the predicted LOD (see Figure 11 inset) and detection of 160 microgL is possible Gastrointestinal anthrax requires significant more spores than inhalation anthrax28 and a limit of detection might be placed at 1 million spores in 1 liter of water or 10 microgL Since each spore contains ~10 CaDPA by weight29 a goal for DPA might be 1 microgL indicating that the present measurements must be improved by nearly two orders of magnitude Due to the increased hazards of handling HCN gas KCN salt was used for these experiments Nevertheless all sample preparations were performed in a chemical hood KCN completely dissolves in water but its conjugate acid HCN is formed and has a Ka of 615x10-1030 viz
HCN CN- + H+ pKa = 921 Reaction 4
Figure 8 SER spectra of 1 mgmL DPA as a function of pH Conditions 100 mW 785 nm 100 scans (44-sec)
Figure 9 Concentration dependence of DPA DPA- and DPA= as a function of pH for a 0006M sample Intensity of 525 () and 795 (diams) cm-1 bands from Fig 8 as a function of pH
pH
135 171 219 383
0
0001
0002
0003
0004
0005
0006
0007
0 2 4 6 8 10 12 14pH
Con
cent
ratio
n [M
]
DPADPA-DPA=795525
DPA DPA=DPA-
pK2 = 692pK1 = 216
795 525
0
05
1
15
2
25
0 200 400 600 800 1000 1200
DPA Concentration (mgL)
1008
Ban
d In
tens
ity
conc (mgL) lod-10min-100mw1 017
10 016100 103
1000 355
Figure 10 SER spectra of DPA in water at A) 1000 B) 100 C) 10 and D) 1 mgL Conditions pH of 25-55 silver-doped sol-gel coated vial 175 mW of 785 nm 1-min 8 cm-1 D) has been multiplied by x10 to make bands visible
A
B C
D
Figure 11 Plot of SER intensity of 1008 cm-1 band of DPA as a function of concentration using 175 mW of 785 nm Inset table includes LOD in mgL for each concentration but for 100 mW and 10 min
SPIE -2003-5269 123
Consequently the cyanide concentration must be determined for each initial KCN concentration Specifically the samples prepared with concentrations of 01 1 10 100 and 1000 mgL of KCN produced CN- concentrations of 63x10-3 033 69 89 and 964 mgL at pHs of 816 90 967 102 and 107 respectively The pH dependence for the HCN and CN- concentrations are shown in Figure 12 Thus as the amount of KCN added to the solution decreases so does the pH of the solution (becomes less basic) and according to Reaction 4 the relative amount of CN- to HCN also decreases For example in the preparation of a 01 mgL solution of KCN the pH is shifted from 7 for pure water to only 816 and only 63 of the starting material becomes CN- or 63x10-3 mgL In comparison for a solution of 1000 mgL the pH is shifted from 7 to 107 and 96 of the starting material becomes CN- This is significant since the cyanide ion is better able to adsorb onto the silver particles and become SERS active SER spectra of 10 100 and 1000 mgL of KCN samples are shown in Figure 13 and a corresponding plot of the 2100 cm-1 band intensity as a function of concentration is shown in Figure 14
000102030405060708091011
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14pH
Con
cent
ratio
n [m
gm
L]
pKa = 921
CN -HCN
The SER spectra of cyanide are dominated by the single CN stretch at 2100 cm-1 which occurs in normal Raman spectra of solutions at 2080 cm-1 However a low frequency mode occurs at 135 cm-1 due to a Ag-CN stretch (not shown) indicative of a strong surface interaction It is also observed that as the concentration decreases the CN stretch shifts to 2140 cm-1 This shift has been attribute to the formation of a tetrahedral Ag(CN)3
2- surface structure31 as well
Figure 14 Concentration dependence of KCN SERS measured under conditions in Fig 11 Concentrations are 1 01 and 001 mgml Intensities are measured for the CN stretch at 2100 cm-1 Inset table includes LOD in mgL for each concentration in Figs 13 and 15 but for 100 mW and 10 min
Figure 13 SER spectra of KCN in water at A) 1000 B) 100 and C) 10 mgL Conditions pHs of 107 102 and 97 silver-doped sol-gel coated vial 100 mW of 785 nm 1-min 8 cm-1 C) has been multiplied by x10 to make band visible
Figure 12 Concentration dependence of HCN and CN- as a function of pH for a 1 mgmL sample Calculated intensity of 2100 () cm-1 band for a 1 mgml sample at pHs of 816 90 967 102 and 107
A
B
C
0
20
40
60
80
100
120
140
0 200 400 600 800 1000 1200CN Concentration (mgL)
2100
Ban
d In
tens
ity
Figure 15 SER spectra of KCN in water at A) 10 B) 1 and C) 01 mgL Conditions pHs of 97 12 and 12 gold-doped sol-gel coated vial 100 mW of 785 nm 1-min 8 cm-1 C) has been multiplied by x10 to make band visible
A
B
C
conc (mgL) lod-10min-100mw Condition01 001 Au-pH 121 007 Au-pH 12
69 003 Au-pH 9769 016 Ag-pH 9789 022 Ag-pH 102946 113 Ag-pH 107
SPIE -2003-5269 124
as to CN adsorbed to two different surface sites32 Alternatively the 2140 cm-1 band could be attributed to HCN since this species dominates at lower concentrations However it is unlikely that this species would be attracted to the electropositive silver surface Further both peaks should be present at pHs between 85 and 105 but this is not observed It has also been suggested that at concentrations near and above monolayer coverage the CN- species is forced to adsorb end-on due to crowding and at lower concentrations the molecule can reorient to lie flat33 This suggests that the 2100 and 2140 cm-1 bands correspond to the end-on and flat orientations respectively As Figures 13 and 14 show the intensity of the CN stretch for the 89 mgL sample is nearly as intense as the 964 mgmL band This suggests that the Raman signal for the flat orientation is more enhanced However more extensive measurements are required to verify this point Since resent research has suggested that cyanide may be more effectively detected on gold measurements of KCN solutions were also performed using gold-doped sol-gel vials Preliminary measurements are shown in Figure 15 for samples prepared from 01 1 and 10 mgL KCN Since the pHs are 816 90 967 the resultant CN- concentrations are 63x10-3 033 and 69 mgL Initially only the highest concentration was observed and the signal intensity was significantly better than the equivalent concentration measured using silver In an effort to shift Reaction 4 to the left transforming HCN to CN- (Le Chatelierrsquos principle) KOH was added to the lower concentration samples producing solutions with pH 12 This effectively forces all of the cyanide in solution to be CN- or 01 and 10 mgL respectively More importantly the CN stretch is now observed in the SER spectra The band appears at 2125 cm-1 as has been previously reported for gold12 As calculated for MPA and DPA LODs can be estimated from this data For the three concentrations of cyanide on silver the LODs are 016 to 11 mgL for 100 mW of 785 nm laser excitation and a 10-min acquisition time For gold pH adjusted the LODs are10 to 70 microgL an improvement of more than 10 times silver Nevertheless either substrate is sufficient to meet the JSAWM goals of measuring 3 mgL in 10-minutes as the requirements form cyanide are much less stringent than the nerve agents
4 CONCLUSIONS Here we examined the surface-enhanced Raman spectral response for methyl phosphonic acid dipicolinic acid and cyanide as a function of pH It was determined that the most prominent peak at 760 cm-1 reaches a maximum intensity between pH 3 and 7 and corresponds to the MPA- species Neither the MPA nor MPA= species appear to generate a SER spectrum and consequently no spectra were observed below pH 2 or above pH 8 In this study we also found that higher laser powers could cause photodegradation of MPA signified by the exponential growth of a band at 1050 cm-1 which is tentatively assigned to phosphorous acid Unlike MPA DPA was observed at all pHs This is attributed to the dominant interaction of the pyridine functional group with silver Minor spectral changes were observed at acid pHs and were assigned to neutral DPA Like MPA SER spectra of cyanide were pH dependent No spectra were observed for the HCN species while CN- was best observed at pHs more basic than 8 Preliminary concentration studies for the three analytes allowed estimating limits of detection for MPA DPA and CN using 100 mW of 785 nm and a 10-min acquisition time of 210 165 and 70 microgL respectively Although the latter value suggests that cyanide can be measured at sufficiently low concentrations to meet JSAWM goal improvements by 100 to 200 times are required for MPA and DPA It is clear from this study that pH of the sample is important and must be taken into account when developing concentration calibration curves as well as fieldable analyzers Future work will include tailoring the sol-gel with specific functional groups to dictate sample pH and thereby optimize sensitivity
ACKNOWLEDGMENTS The authors are grateful for the support of the US Army (DAAD13-02-C-0015 Joint Service Agent Water Monitor program) The authors would also like to thank Dr Steve Christensen of the US Army SBCCOM for helpful discussions and Mr Chetan Shende of Real-Time Analyzers for assistance in development of the gold-doped sol-gels
REFERENCES 1 Erickson B Analytical Chemistry News amp Features June 1 397A (1998)
SPIE -2003-5269 125
2 Johnston RL Hoefler CM Fargo JC and Moberley B AT-ONSITE 5-8 (1994) 3 Black RM RJ Clarke RW Read and MT Reid J Chromatography 662 301-321 (1994) 4 USA Army Center for Health Promotion and Preventive Medicine (USACHPPM) Technical Guideline 230A 5 Jensen JL NC Wong W R Loerop SPIE 4775-03 (2001) 6 Hoenig SL Handbook of Chemical Warfare and Terrorism Greenwood Press Wesport CT (2002) 7 Munro NB SS Talmage GD Griffin LC Waters AP Watson JF King and V Hauschild Env Health
Persp 107 933-974 (1999) 8 Holstege CP Kirk M Sidell FR Crit Care Clin13 923-42 (1997) 9 Farquharson S P Maksymiuk K Ong and S Christesen SPIE 4577 166-173 (2001) 10 Lee Y and S Farquharson SPIE 4378 21-26 (2001) 11 Spencer KM J Sylvia S Clauson and J Janni SPIE 4577 158-165 (2001) 12 Tessier P S Christesen K Ong E Clemente A Lenhoff E Kaler and O Velev Applied Spectroscopy 56
1524-1530 (2002) 13 Farquharson S WW Smith S Elliott and JF Sperry SPIE 3533 207-214 (1998) 14 Farquharson S WW Smith S Elliott and JF Sperry SPIE 3855110-116 (1999) 15 Farquharson S WW Smith YH Lee S Elliott and JF Sperry SPIE 4575 62-72 (2002) 16 Guzelian AA J Sylvia J Janni S Clauson and KM Spencer SPIE 4577 182-192 (2001) 17 Shende C F Inscore A Gift P Maksymiuk and S Farquharson in press 18 Weaver MJ S Farquharson and MA Tadayyoni J Chem Phys 82 4867-4874 (1985) 19 Alak AM and T Vo-Dinh Analytical Chemistry 59 2149-2153 (1987) 20 Lee Y and S Farquharson SPIE 4206 140-146 (2000) 21 Farquharson S and Y Lee SPIE 4200-16 (2000) 22 Lee Y S Farquharson and P M Rainey SPIE 3857 76-84 (1999) 23 Lee Y S Farquharson H Kwong and M Shahriari SPIE 3537 252-260 (1998) 24 Wang J M Pumera G Collins and A Mulchandani Analytical Chemistry 74 6121-6125 (2002) 25 Dou X YM Jung Z-Q Cao and Y Ozaki Applied Spectroscopy 53 1440-1447 (1999) 26 Data supplied by S Christesen and K Ewing 27 Kerker M and B Thompson Eds SPIE MS 10 (1990) 28 Inglesby TV DA Henderson JG Bartlett JAMA 287 2236 (2002) 29 FW Janssen AJ Lund and LE Anderson Science 127 26 (1958) 30 Lide DR Ed Handbook of Chemistry and Physics CRC Press 77th Ed (1996-7) 31 Billmann J G Kovacs and A Otto Surf Sci 92 153 (1980) 32 Murray CA and S Bodoff Phys Rev B 32 671 (1985) 33 Kellogg D and J Pemberton J Phys Chem 91 1120 (1987)
SPIE-2003-5269 16
Chemical agent detection by surface-enhanced Raman spectroscopy
Stuart Farquharson Alan Gift Paul Maksymiuk Frank Inscore and Wayne Smith
Real-Time Analyzers 87 Church Street East Hartford CT 06108
Kevin Morrisey and Steven D Christesen US Army SBCCOM Aberdeen Proving Ground MD 21010
ABSTRACT
In the past decade the Unites States and its allies have been challenged by a different kind of warfare exemplified by the terrorist attacks of September 11 2001 Although suicide bombings are the most often used form of terror military personnel must consider a wide range of attack scenarios Among these is the intentional poisoning of water supplies to obstruct military operations in Afghanistan and Iraq To counter such attacks the military is developing portable analyzers that can identify and quantify potential chemical agents in water supplies at microgram per liter concentrations within 10 minutes To aid this effort we have been investigating the value of a surface-enhanced Raman spectroscopy based portable analyzer In particular we have been developing silver-doped sol-gels to generate SER spectra of chemical agents and their hydrolysis products Here we present SER spectra of several chemical agents measured in a generic tap water Repeat measurements were performed to establish statistical error associated with SERS obtained using the sol-gel coated vials Keywords Chemical agents chemical agent detection SERS Raman spectroscopy
1 INTRODUCTION
In the autumn of 2001 terrorism within US borders became a sobering reality While extensive efforts are being implemented to secure the homeland US and Coalition forces in Afghanistan and Iraq are constantly faced with terrorist attacks In addition to car-bombings the military has identified several non-traditional attack scenarios including poisoning of water supplies by chemical warfare agents (CWAs) To counter this threat the Department of Defense has been investigating numerous approaches to detect such attacks under the auspices of the Joint Service Agent Water Monitor (JSAWM) program1 The JSAWM must be capable of identifying the chemical agents with no false-positives and quantifying the agents at microgram per liter concentrations within 10 minutes2 This includes the analysis of drinking water supplies distribution and storage systems Currently colorimetric paper is used to detect agents on-site while gas chromatography combined with mass spectrometry (GCMS) is used in mobile support laboratories However both methods have severe drawbacks The paper changes color in response to contact with many chemicals besides CWAs causing a high incidence of false positives3 while GCMS although very chemically specific requires up to an hour to perform and regular re-calibration456 Vibrational spectroscopy has also been investigated as a potential method of rapidly detecting CWAs7-
11 as early as 1985 in the case of infrared spectroscopy8 and 1988 in the case of Raman spectroscopy11 Again however these techniques also have limitations when it comes to measuring trace poisons in water Infrared spectra would be dominated by the very strong absorption of water which would obscure absorptions by most other chemicals present Whereas Raman spectroscopy is simply not a very sensitive technique and detection limits are typically grams per liter Surface-enhanced Raman spectroscopy (SERS) offers several advantages over conventional vibrational methods and that may provide the necessary sensitivity required for detecting trace quantities of chemical agents in water SERS employs the interaction of surface plasmon modes of metal particles with target analytes to increase scattering efficiency by as much as 1 million times12 In 1987 the potential of this approach to measure CWAs was demonstrated by measuring a series of pesticides13 Several of these organophosphonates have chemical structures similar to CWAs in particular P=O functional groups In the past few years we and others have further explored the ability of SERS to detect CWAs14-17 and even bioagents 18-21 We have been employing silver-doped sol-gels to promote the SER effect
SPIE-2003-5269 17
in these studies The porous silica network of the sol-gel matrix offers a unique environment for immobilizing and stabilizing SER-active metal particles22-25 The silver-doped sol-gels have been coated on the internal walls of standard glass vials to allow rapid SER analysis of small sample quantities without preparation We have measured over 100 chemicals with enhancements of 104 to 106 demonstrated reversible measurements in a flowing system reproducible measurements from vial-to-vial and measurements in multiple solvents including water21-25 Previously we used these vials to perform preliminary measurements of cyanide (CN) and methylphosphonic acid (MPA) Most of the nerve agents form MPA during hydrolysis while Tabun forms CN a chemical agent in its own right In another paper including in these proceedings we examined the limits of detection (LOD) for MPA by measuring a series of concentrations down to 50 mgL and estimated a limit of detection of 100 microgL26 These measurements provide encouragement in that SERS may satisfy the needs of the JSAWM To further establish the viability of SERS in particular silver-doped sol-gels here we present analysis of cyanide mustard and VX in tap water The measurements performed at the US Armyrsquos Edgewood Chemical Biological Center Aberdeen MD also included numerous repeat measurements to establish reproducibility
2 EXPERIMENTAL 2a General All chemicals including potassium cyanide 2-chloroethylethyl sulfide and those used to prepare the silver-doped and gold-doped sol-gel coated vials were obtained and used as received from Sigma-Aldrich All samples were prepared in a chemical hood using HPLC grade water unless otherwise noted (Fischer Scientific Fair Lawn NJ) for SERS measurements Once prepared the samples were transferred into the silver-doped sol-gel vials (Real-Time Analyzers Inc East Hartford CT) The vials were coated in a manner similar to that previously reported by adding ammonium hydroxide to a solution of silver nitrate tetramethyl orthosilicate (TMOS) and methanol22 The two precursor solutions were prepared mixed and transferred to 2-mL glass vials dried and heated After sol-gel formation the incorporated metal ions were reduced with dilute sodium borohydride (1mgmL) which is followed by a water wash to remove residual reducing agent After the resultant analyte solution was introduced the SER-active vial was fixed horizontally to an XY positioning stage (Conix Research Springfield OR) just inside the focal point of an f07 aspheric lens The lens focused the beam into the sample and collected the scattered radiation back along the same axis A dichroic filter (Omega Optical Brattleborough VT) was used to reflect the excitation laser to the lens and pass the Raman scattered radiation collected by the lens An f2 achromat was used to collimate the laser beam exiting a 200 microm core diameter source fiber optic while a second f2 achromat was used to focus the scattered radiation into a 365 microm fiber optic (Spectran Avon CT) A short pass filter was placed in the excitation beam path to block the silicon Raman scattering generated in the source fiber from reflecting off sampling optics and reaching the detector A long pass filter was placed in the collection beam path to block the sample Rayleigh scattering from reaching the detector A 785 nm diode laser (Process Instruments Inc model 785-600 Salt Lake City UT) was used to deliver 100 to 150 mW of power to the sample A Fourier transform Raman spectrometer (Real-Time Analyzers model IRA-785 East Hartford CT) and a silicon photo-avalanche detector (Perkin Elmer model C30902S Stamford CT) were used to acquire the SER spectra
2b Edgewood Chemical Biological Center The surface-enhanced Raman spectral measurements at the US Armyrsquos Chemical Biological center presented here were all performed on September 12 2003 To expedite measurements a plate was machined to hold up to 12 SER-active sample vials (Figure 1) The plate fit a standard XY plate reader that could be programmed Pure KCN bis-(2-chloroethyl)sulfide (distilled mustard gas HD) and ethyl S-2-diisopropylamino ethyl methylphosphonothioate (VX) were obtained on-site and used to prepare 1 gL tap water solutions in a chemical hood with appropriate safety equipment Simulated tap water was prepared by adding 10 chemicals most often found in tap water at appropriate concentrations to distilled water (Table 1) SER measurements were also performed in a chemical hood For added safety the FT-Raman instrument was placed outside the laboratory and 30 foot fiber optic and electrical cables were used to allow remote SERS measurements and plate manipulation For each experiment 1gL samples were prepared and added to 9 individual vials which were then loaded on the plate In some cases a tenth vial was included as a blank
Table 1 Generic Tap Water Composition
Compound mgL NaHCO3 100 CaSO4 27 MgSO4bull7H2O 67 NaNO3 1 Fulvic Acid 1 K2HPO4 07 KH2PO4 03 (NH4)2HSO4 001 NaCl 001 FeSO4 0001 pH 76-78
SPIE-2003-5269 18
A software program was written that allowed selecting the sequence that the vials were measured the number of positions along the length of the vials to measure (1 to 5) and the number of scans to co-add During sample analysis the program displayed the vial being analyzed the point being analyzed and the spectrum as it was being acquired Once all the data was collected a second software program was written to rapidly analyze the data The spectra collected for all the vials on a plate could be loaded at one time and then the spectra for each point could be displayed simultaneously or separately The user could then select the Raman peak to analyze in terms of peak height or area This was accomplished by selecting points on either side of the peak to define a baseline of zero The peak height or area could then be computed for all of the spectra loaded and then exported to a spreadsheet for statistical analysis
Figure 1 A) Vial Holder 6 slots to hold 2 vials each end-to-end B) Measurement Configuration Program user selects vials to measure sequence number of points per vial (1 to 5) and number of scans per point C) Spectral Acquisition Program shows spectrum being collected which vial and position D) Spectra Analysis Program user selects spectra to analyze by plate vial and point (s) as well as two wavenumbers defining the peak and the baseline to subtract The image is of 5 repeat measurements of 10 mgL KCN in generic tap water 16 sec each 100 mW of 785 nm
3 RESULTS AND DISCUSSION Raman and surface-enhanced Raman spectra were obtained for potassium cyanide bis-(2-chloroethyl)sulfide and ethyl S-2-diisopropylamino ethyl methylphosphonothioate representing three classes of chemical agents cyanides mustards and nerve agents respectively Spectra were also obtained for 2-chloroethyl ethyl sulfide (CEES) a structural analogue to HD which was included in the study to aid in assigning spectral bands KCN salt was used for cyanide experiments to avoid the increased hazards of handling HCN gas KCN completely dissolves in water forming its conjugate acid HCN according to its Ka of 615x10-1027 and at a concentration of 1 mgmL results in a pH 107 solution This is significant in that only CN- appears to interact sufficiently with silver to produce a SER spectrum and no spectral signal is observed below pH 726 Figure 2 shows the SER and normal Raman spectra for KCN The SERS spectrum of 1mgml KCN in water shows a single intense somewhat broad feature at
A B
C D
SPIE-2003-5269 19
2100 cm-1 assigned to the single CequivN stretch The band is much sharper in the normal Raman spectra of the solid KCN salt at 2074 cm-1 This band does broaden and shift to 2080 cm-1 in solution (not shown) However the observed SERS frequency is attributed to interaction with silver and low frequency mode at 135 cm-1 attributed to a Ag-CN stretch (not shown) supports this conclusion
cm-1 band to a C-S stretch but the authors concede that it is in fact more likely a C-Cl stretch28 It appears that the most intense Raman bands at 648 692 and 747 cm-1 shift to 620 660 and 720 cm-1 in the SER spectra and are tentatively assigned as above The width of these bands suggests that they overlap underlying spectral features Additional bands in the Raman spectra occur at 972 1034 1049 1263 1286 1423 1442 2865 2935 and 2960 cm-1 Corresponding bands occur in the SER spectra at 964 1015 1054 1286 1410 1447 2865 and 2935 cm-1 Most of these bands are associated with alkane modes specifically the bands at approximately 1040 cm-1 to a C-C stretch 1290 cm-1 to a CH2 in-phase twist 1440 cm-1 to a CH2 wag 2865 cm-1 to a symmetric CH2 stretch and 2965 cm-1 to an asymmetric CH2 stretch The Raman and SER spectra of sulfur mustard were measured at the Edgewood center (Figure 4) Both spectra are largely similar to CEES The C-Cl and C-S bands in the Raman spectrum of HD now occur at 640 655 700 739 and 760 cm-1 and are more resolved possibly due to the increased molecular symmetry Theoretical calculations indicate that the first three bands are due to C-Cl stretching modes and the latter two to C-S stretching modes28 Only the C-Cl bands maintain significant intensity in the SER spectra occurring at 624 and 643 cm-1 which is attributed to the
Prior to measurements of HD CEES was examined by Raman and SER spectroscopy (Figure 3) CEES also known as half-mustard is essentially identical to HD except one of the chlorine end atoms is replaced by a hydrogen atom Again although not as toxic as HD CEES is a blister agent and dilute aqueous samples were prepared using appropriate safety equipment Both the Raman and SER spectra of CEES are similar and dominated by bands between 600 and 800 cm-1 These are associated with C-Cl and C-S stretching modes which are tentatively assigned to 648 and 747 cm-1 in the Raman spectra respectively The shoulder at 630 cm-1 the overlapped band at 660 cm-1 and the strong band at 692 cm-1 could also be due to these modes or their asymmetric counterparts It is worth noting that theoretical calculations assign the 692
Figure 2 A) SER and B) NR spectra of KCN Conditions A) 1 mgml in tap water 100 mW of 785 nm at sample 1-min acquisition time B) solid 300 mW of 785 nm 5-min All spectra are 8 cm-1 resolution
A
B
Figure 3 A) SER and B) NR spectra of CEES Conditions A) 1 vv (10 mgml) in MeOH 100 mW of 785 nm 1-min acquisition time B) neat 300 mW of 785 nm 5-min
A
B
Cl-CH2-CH2-S-CH2-CH3
A
B
Figure 4 A) SER and B) NR of HD Conditions A) 1mgml in tap water B) pure both 100 mW of 785 nm 1-min
Cl-CH2-CH2-S-CH2-CH2-Cl
SPIE-2003-5269 20
expected strong interaction between chlorine and silver and adds support to the assignment of this band to a C-Cl stretch Weaker overlapping bands occur at 670 692 and 724 cm-1 the latter possibly due to C-S stretching modes Again the alkane modes are apparent in the normal Raman spectra of HD but only a broad feature at 1300 to 1450 cm-1 suggests CH2 contributions in the SER spectrum Although the observed bands in the VX spectrum have not been assigned (Figure 4) a computer generated Raman spectrum29 predicts many of the same features with surprising accuracy and are used here Two intense bands at 460 and 530 cm-1 closely match predicted bands at 463 and 546 cm-1 assigned to a CH3-P=O bend and a PO2CS wag Three highly overlapped bands occur at 694 745 and 771 cm-1 matching predicted bands at 713 730 and 760 cm-1 The first
Table 2 Measured SER peak heights for the CN stretch at 2100 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
Figure 5 A) SERS and B) NR spectra of VX Conditions A) 1 vv (10 mgml) in MeOH B) pure sample both 100 mW of 785 nm 1-min acquisition time
A
B
two have been assigned to a C-S stretch and CH2bend respectively while the latter has been attributed to either a P-C stretch or an O-C-C stretch Although the 745 cm-1 band may alternatively be assigned to a C-S stretch based on the previous measurements of CEES and HD The relatively intense bands at 890 1106 1218 1445 and 1465 cm-1 also match predicted bands at 880 1108 1216 1440 and 1464 cm-1 that are assigned to a C-C stretch CH3 rock N-C3 stretch various C-H3 bends and C-H bends respectively Both the computer generated and the measured spectra contain numerous other less intense bands One is worth mentioning A unique band appears at 370 cm-1 that is predicted at 368 cm-1 and corresponds to an O-P=O bend The surface-enhanced Raman spectrum of VX is also rich with spectral features It has the unique low frequency band at 370 cm-1 as well as a second band at 380 cm-1 that is assigned tothe S-P-O bend predicted in the normal Raman spectrum at 388 cm-1 Based on the measured and predicted normal Raman spectra the following SERS assignments are given 460 cm-1 to the CH3-P=O bend 544 cm-1 to the PO2CS wag 738 cm-1 to a C-S stretch (based on arguments above) 890 cm-1
to a C-C stretch 1101 cm-1 to a CH3 rock and 1456
cm-1 to a C-H bend The ability of SERS to measure chemical agents in water containing real-world chemical interferents was tested by using the generic tap water described in Table 1 The ability to reproduce measurements was accomplished by preparing three separate water stock solutions which were used to prepare three sample solutions each of 1 mgmL KCN in the generic tap water The samples defined as vials 1-9 were then measured at 4 points per vial in a semi-random fashion such that errors associated with stock solution preparation and errors associated with instrument drift could be identified No trends were apparent that signified such systematic errors Each spectrum collected consisted of 20 averaged scans taking 16 seconds at 8 cm-1 resolution The laser power at the sample was measured periodically during the day and it ranged from 102 to 105 mW spacccc
spot Vial 1 Vial 2 Vial 41 13754 13004 128192 13519 12692 12909
1 3 13541 12721 126394 13462 12648 12651
ave 13569 12766 12755Vial 6 Vial 3 Vial 5
1 11586 14068 10757 2 11236 14402 115122 3 11376 14546 11559
4 10894 11763 11214ave 11273 13695 11261
Vial 7 Vial8 Vial 91 11198 15804 11023
2 11243 15716 88753 3 11562 15216 11218
4 11655 15084 11077ave 11415 15455 10548
CN AVG STDEV ERR1 mgmL pk ht 12526 1560 1245
stock solution
SPIE-2003-5269 21
The CN measurements consisting of 36 data points produced an average height of 125 for the 2100 cm-1 peak with a standard deviation of 156 or 125 (Table 2) The HD and VX measurements were performed precisely the same way (Tables 3 and 4) For HD the 624 cm-1 peak was used for analysis and it had an average height of 53 with a standard deviation of 068 or 129 while for VX the 544 cm-1 peak was used for analysis and it had an average height of 1051 with a standard deviation of 308 or 293 The greater error in the VX measurements can be somewhat attributed to Vial 4 which produced lower SER signal intensities But removing this vial from the data set changes the standard deviation to 233 only a modest improvement
spot Vial 1 Vial 2 Vial 41 525 609 4682 45 675 484
1 3 527 644 5094 559 807 669
ave 51525 68375 5325Vial 6 Vial 3 Vial 5
1 521 574 575 2 536 53 4492 3 509 508 451
4 565 594 379ave 53275 5515 4635
Vial 7 Vial 8 Vial 91 457 589 413
2 583 497 4273 3 544 505 4
4 528 5 509ave 528 52275 43725
HD AVG STDEV ERR1 mgmL pk ht 530 068 1291
stock solution
spot Vial 1 Vial 2 Vial 41 1464 1034 4672 1485 99 61
1 3 1491 989 5684 1041 777 553
ave 137025 9475 5495Vial 6 Vial 3 Vial 5
1 1058 942 1293 2 697 121 965
2 3 727 1206 11124 689 1254 746
ave 79275 1153 1029Vial 7 Vial 8 Vial 9
1 1127 783 163 2 1358 812 1614
3 3 1371 908 15124 1097 875 1397
ave 123825 8445 153825
VX AVG STDEV ERR1 mgmL pk ht 1051 308 2925
stock solution
4 CONCLUSIONS In this paper we examined the ability of surface-enhanced Raman spectroscopy to reproducibly measure CN HD and VX in tap water without chemical interference Both normal and surface-enhanced Raman spectra were examined to select unique bands suitable to identify and quantify these chemical agents For SER measurements the 2100 cm-1 C-N stretch was used for CN the 624 cm-1 C-Cl stretch was used for HD and the 544 cm-1 PO2CS wag was used for VX It was determined that 1 mgmL samples of each of these chemicals measured 36 times in glass vials coated with a silver-doped sol-gel reproduced measurements with standard deviations of 125 129 and 293 It was further found that the 10 chemicals added to simulate generic tap water did not interfere with or alter the SER spectra It should be noted that the concentrations used in this study were considerably greater than those required by the JSAWM program Current work involves improving SER sensitivity and designing sampling systems with better reproducibility This includes the development of fractal silver and gold structures within the sol-gel matrix and the development of chemically selective sol-gels
ACKNOWLEDGMENTS The authors are grateful for the support of the US Army (DAAD13-02-C-0015 Joint Service Agent Water Monitor program) The authors would also like to thank Janet Jensen Ronald Crosier and Kristina Gonser for helpful discussions
Table 4 Measured SER peak heights for the VX band at 544 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
Table 3 Measured SER peak heights for the HD band at 624 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
SPIE-2003-5269 22
REFERENCES 1 Jensen JL NC Wong W R Loerop SPIE 4775-03 (2001) 2 JSAWM Requirements at wwwsbccomapgeaarmymilRDAecbcrtPRODSERJSAWMjsawmhtml 3 Erickson B ldquoThe Chemical Weapons Convention Redefines Analytical Challengerdquo Anal Chem 70 397A (1998) 4 Johnston RL Hoefler CM Fargo JC and Moberley B ldquoThe Defense Nuclear Agencyrsquos
ChemicalBiochemical Weapons Agreements Verification Technology Research Development Test and Evaluation Program and its Requirements for On-Site Analysisrdquo AT-ONSITE 5-8 (1994)
5 Black RM Clarke RJ Read RW and Reid MT ldquoApplication of GC-MS and GC-tandem MS to the analysis of chemical warfare samples found to contain residues of the nerve agent sarin sulphur mustard and their degradation productsrdquo J Chrom 662 301-321 (1994)
6 USA Army Center for Health Promotion and Preventive Medicine (USACHPPM) Technical Guideline 230A 7 Hoffland LD Piffath RJ Bouck JB rdquoSpectral signatures of chemical agents and simulantsrdquo Optical
Engineering 24 982-984 (1985) 8 Braue EHJ Pannella MGrdquoCIRCLE CELL FT-IR Analysis of Chemical Warfare Agents in Aqueous Solutionsrdquo
App Spectrosc 44 1513-1520 (1990) 9 Tseng C-H Mann CK and Vickers TJ ldquoDetermination of Organics on Metal Surfaces by Raman
Spectroscopyrdquo App Spectrosc 47 1767-1771 (1993) 10 Staff Reporter ldquoUS Army Tests Lidar to Detect Biological Toxinsrdquo Photonic Spectra pg 50 December 1998 11 Christesen SD Raman cross sections of chemical agents and simulants App Spectrosc 42 318-321 (1988) 12 Weaver MJ Farquharson S Tadayyoni MA ldquoSurface-enhancement factors for Raman scattering at silver electrodesrdquo J Chem Phys 82 4867-4874 (1985) 13 Alak AM and Vo-Dinh T ldquoSERS of Organophosphorous Chemical Agentsrdquo Anal Chem 59 2149-2153 (1987) 14 Lee Y Farquharson S ldquoRapid chemical agent identification by SERSrdquo SPIE 4378 21-26 (2001) 15 Farquharson S Maksymiuk P Ong K Christesen S ldquoChemical agent identification by surface-enhanced Raman spectroscopyrdquo SPIE 4577 166-173 (2001) 16 Spencer KM Sylvia J Clauson S and Janni J ldquoSurface Enhanced Raman as a Water Monitor for Warfare
Agents in Waterrdquo SPIE 4577 158-165 (2001) 17 Tessier P Christesen S Ong K Clemente E Lenhoff A Kaler E Velev O ldquoOn-line spectroscopic characterization of sodium cyanide with nanostructured gold surface-enhanced Raman spectroscopy substratesrdquo
App Spectrosc 56 1524-1530 (2002) 18 Farquharson S W W Smith S Elliott and J F Sperry Rapid biological agent identification by surface- enhanced Raman spectroscopy SPIE 3855110-116 (1999) 19 Farquharson S WW Smith YH Lee S Elliott and J F Sperry Detection of biological signatures A
comparison of electrolytic and metal-doped sol-gel surface-enhanced Raman media SPIE 4575 62-72 (2002) 20 Guzelian AA Sylvia J Janni J Clauson S and Spencer KM ldquoSERS of whole cell bacteria and trace levels of biological moleculesrdquo SPIE 4577 182-192 (2001) 21 Shende C Inscore F Gift A Maksymiuk P Farquharson S ldquoRapid dipicolinic acid extraction from Bacillus spores detected by surface-enhanced Raman spectroscopyrdquo App Spectrosc 58 accepted 22 Lee Y and Farquharson S ldquoSERS Sample Vials Based on Sol-Gel Process for Trace Pesticide Analysisrdquo SPIE
4206 140-146 (2000) 23 Farquharson S and Lee Y ldquoTrace Drug Analysis by SERSrdquo SPIE 4200-16 (2000) 24 Lee Y Farquharson S and Rainey P M Surface-Enhanced Raman Sensor for Trace Chemical Detection in
Water SPIE 3857 76-84 (1999) 25 Lee Y Farquharson S Kwong H and Shahriari M ldquoSol-Gel Chemical Sensor for Surface-Enhanced Raman
Spectroscopyrdquo SPIE 3537 252-260 (1998) 26 Farquharson S A Gift P Maksymiuk F Inscore and W Smith ldquopH dependence of methyl phosphonic acid
dipicolinic acid and cyanide by surface-enhanced Raman spectroscopyrdquo SPIE 5269-19 (2003) 27 Lide DR Ed Handbook of Chemistry and Physics CRC Press 77th Ed (1996-7) 28 Sosa C RJ Bartlett K KuBulat and WB Person ldquoA theoretical study of harmonic vibrational frequencies and
infrared intensities of XCH2CH2SCH2CH2X and XCH2CH2SH (= H Cl)rdquo J Phys Chem 93 577-588 (1993) 29 Hameka HF and JO Jenson ldquoComputer-generated predictions of the structure and the IR and Raman spectra of
VXrdquo ERDEC-TR-065 May 1993
- Chemagents Appendicespdf
-
- SPIE2001-4575-Bioagent-SERSapdf
-
- Detection of bioagent signatures A comparison of electrolytic and metal-doped sol-gel surface-enhanced Raman media
- Stuart Farquharson Wayne Smith and Yuan Lee
-
- Real-Time Analyzers 87 Church Street East Hartford CT 06108
-
- ABSTRACT
- 4 CONCLUSIONS
- 5 ACKNOWLEDGEMENTS
- 6 REFERENCES
-
SPIE 2001-4575
65
electrode surface for E-SERS through the vial glass wall and into the silver-doped sol-gel film for SG-SERS or through the glass wall of the cuvette and into the solid sample for normal Raman spectroscopy All E-SERS and normal Raman spectra were obtained with 750 mW of laser power at the sample while all SG-SERS spectra were obtained with 75 mW of laser power at the system Incident powers above 200 mW in some cases degraded the sol-gel
3 RESULTS AND DISCUSSION The generation of surface-enhanced Raman scattering at electrode surfaces has been extensively researched and the optimum sample conditions are well developed2729 Several researches incorporated electrodes into flowing systems and demonstrated that continuous monitoring of chemicals is possible18 These successes suggested investigated the capability of E-SERS to measure the NA bases and RNA The E-SERS measurements also provided a benchmark to compare and evaluate SG-SERS measurements The molecular structure of adenine (as well as the other base pairs) which includes an aromatic nitrogen-containing heterocycle is ideally suited to interact with the surface plasmons and contribute substantially to the chemical component of the SER effect1119 Even with excitation at 1064 nm a 3-minute scan of 18x10-5M adenine yields high signal-to-noise (SN) E-SER spectra and all of the bands are revealed with clarity (Figure 1 Table 1) Spectra of similar quality were obtained by SG-SERS and the principal spectral bands are easily observed The identical 18x10-5M adenine sample was measured in the same 3-minute time frame but with 110th the laser power The lower power appears to reduce the SN The amount of adenine responsible for the SER spectra as well as enhancement factors for the two SER media can be determined The molecules producing the E-SERS spectrum are those on the electrode surface within the illumination area of the laser (The solution concentration only determines the number of molecules available to adsorb to the electrode surface) For the current experiments the laser illuminates an area of 28x10-7m2 or 56 x10-7m2 if we assume the ORCs increase the surface area by a factor of two Furthermore if we assume monolayer coverage on the electrode and each 3x5 angstrom molecule (lying flat) occupies 15x10-19m2 then there are ca 4x1012 molecules contributing to the Raman scattering This is ca twice the number of molecules measured at electrode surfaces using either differential capacitance-potential curve measurements or rapid linear sweep voltammetry (eg 3x1018 moleculesm2 for pyridine and pyrazine)29 Thus the adenine spectrum in Figure 1 is due to 87x10-10g (6x10-12 moles) A detection limit defined as a SN of 3 can also be calculated The SN for a 3-minute scan is 844 for the 735 cm-1 band suggesting a mass detection limit of 3x10-12g (2x10-14 moles) This is consistent with previous estimates for adenine by others of 25 x10-14 moles1530 However sub-monolayer concentrations must be measured to verify this The root-mean-squared (RMS) noise is measured between 4400-4600 cm-1 Since noise is distributed evenly throughout the spectrum when transformed this region does not have any
Wavenumbers (∆cm-1) Figure 1 A) Normal Raman spectrum of pure adenine powder B) E-SERS and C) SG-SERS of 18x10-5M adenine at pH 92 All spectra 8 cm-1 resolution 200 scans (3 min) and 1064 nm excitation A) and B) 750 mW C) 75 mW B) measurement potential of 11VSCE
A
B
C
A
B
C SG-SERS
pH 10
D
725
735
735
pH 4
Wavenumbers (∆cm-1) Figure 2 A) and C) E-SERS and B) and D) SG-SERS of adenine at A) and B) pH 104 and C) and D) pH 40 Note consistent appearance of bands at 1270 and 1375 cm-1 as the pH is changed to 10 for both SER media E-SERS used 750 mW SG-SERS used 75 mW of 1064 nm excitation
SPIE 2001-4575
66
contributions from signals or baseline offsets The measurement error is given as SplusmnRMS and for adenine this equals 234 The number of molecules contributing to the SG-SERS are those on the silver particles that are embedded in the sol-gel The total silver surface area can be determined from the average particle size (40 nm diameter) concentration (073 by weight based on molar conc and measured sol-gel density) and the scattering volume (a cylinder defined by the laser area 28x10-7m2 and sol-gel thickness10-4m) The 61x109 silver particles in this volume have a collective area of 31x10-5m2 However it may reasonably be assumed that at least half of the silver surface is in contact with the sol-gel matrix and unavailable for analyte interaction Then approximately 10x1014 molecules or 22x10-8g of adenine contribute to the SG-SER spectrum The slightly lower SN of 207 suggests a mass detection limit of 32x10-10g Determination of the enhancement factors for the two SER media requires estimating the number of adenine molecules contributing to the normal Raman spectrum Here a cylindrical scattering volume is assumed again based on the laser area (28x10-7m2) and the penetration depth (1x10-3 m)41 The density of the sample was measured at 064 gcm3 indicating that 18x10-4g (13x10-6 moles) of adenine produced the normal Raman signal The enhancement factor EF is defined by the following equation
EF = (ISERSINR)bull(MNRMSERS) bull(PNRPSERS) bull(TNRTSERS)12
where I is the spectral band intensity (here 735 cm-1) M is the sample mass P is the incident laser power and T is the measurement time (or number of scans) For the E-SERS measurement the enhancement factor is 22x105 (01780184) bull (18x10-487x10-10)) while the SG-SERS enhancement factor is 10x105 (0160184) bull(18x10-422x10-8) bull(75075) bull(315)12) The lower enhancement for the SG-SERS may be real or the available surface of the silver embedded in the sol-gel may have been overestimated In addition to enhancing the Raman scattering efficiency to an extent similar to E-SERS the SG-SER medium also yields an identical shift of the adenine ring-breathing mode from 725 cm-1 in the normal Raman to 735 cm-1 Furthermore in the course of optimizing the E-SERS sample conditions it was found that pH influenced the adenine interaction with the silver surface (Figure 2) In particular the relative band intensities of the pyrimidine ring skeletal vibrations at 1270 and 1375 cm-1 and the imidazol ring skeletal vibration at1335 cm-1 change At pH 4 adenine is protonated presumably the imidazol ring since the band at 1335 cm-1 increases in intensity while the pyrimidine bands are virtually absent Conversely at pH 10 the imidazol band decreases in intensity while the pyrimidine bands appear It is worth noting that the ring-breathing mode at 735 cm-1 changes little between pH 4 and 10 suggesting that the skeletal changes are more a function of molecule-plasmon interactions than reorientation of the molecule on the surface Measurements of the identical pH series of adenine samples by SG-SERS yielded virtually identical spectral changes This suggests that the sol-gel does not influence the measurement This is critical to reproducing measurements and performing quantitative analysis Next the remaining NA bases were measured by both E-SERS and SG-SERS and compared Previously we examined the optimum pH and electrode potentials for E-SERS measurements to determine if a common pH could be used that yielded good sensitivity for all the bases and if variations in potential could be used to provide an added degree of selectivity between the bases Primarily it was found that high quality spectra were obtained between pH 7 and 95 and that cytosine and uracil were best enhanced at potentials positive of the potential-of-zero charge (pzc ca 065VSCE for Ag) guanine and thymine near the pzc and adenine negative of the pzc In all cases the ring-breathing modes were the most intense and in general could be used to identify the NA bases (Figure 3 Table 1) Specifically adenine has an intense band at 735 cm-1 cytosine at 797 cm-1 guanine at 653 cm-1 thymine at 784 cm-1 and uracil at 800 cm-1 The adenine cytosine guanine and thymine bands are sufficiently separated that their contributions to DNA should be determinable Although adenine and guanine contributions to RNA should also be determinable cytosine and uracil are highly overlapped and unfortunately share the same potential dependence Alternate unique bands at 1183 cm-1 for cytosine and 1275 cm-1 for uracil might be suitable for calculating contributions The SG-SER spectra of the remaining NA bases faithfully reproduced the E-SER spectra In particular the primary identifying bands occur at virtually the same wavenumbers (see Table 1) However the spectra for both cytosine and thymine contain an intense band at ca 1040 cm-1 Initially this was attributed to the pH buffer but samples prepared without either the buffer or the 01M KCl electrolyte yielded identical spectra containing this band In fact the E-SER and SG-SER spectra of thymine are virtually identical except for this band Also the SG-SERS of guanine contains an intense band at 1551 cm-1 that is not observed in the E-SER spectrum This band may be due to a moderately intense band at 1553cm-1 in the normal Raman spectrum that is SG-SER active It was also found that the SG-SERS of cytosine was considerably better than the E-SERS while uracil showed the opposite relationship It is also worth noting that all of the SG-SERS were obtained with 110th the laser power Most importantly the primary ring-breathing modes in the SG-SER spectra are sufficiently intense and unique to be used in determining contributions to DNA and RNA as outlined above
SPIE 2001-4575
67
Table 1 Comparison of E-SER and SG-SER Spectral Band Positions for the NA Bases and Adenosine Monophosphate
Adenine Cytosine Guanine Thymine Uracil AMP E-SERS SG-SERS E-SERS SG-SERS E-SERS SG-SERS E-SERS SG-SERS E-SERS SG-SERS E-SERS SG-SERS1647 1638 1634 1656 1655 1630 1587 1585 1510 1580 1551 1539 1456 1456 1465 1460 1462 1480 1453 1459 1394 1398 1425 1431 1435 1399 1404 1392 1374 1375 1373 1383 1370 1335 1332 1311 1307 1333 1331 1353 1348 1331 1329 1265 1273 1280 1292 1278 1276 1275 1279 1271 1183 1195 1222 1232 1221 1219 1204 1205 1180 1144 1097 1033 1029 1038 1040 1035 1051 1037 1041 1035 963 963 957 1001 1000 961 944 884 819 817 859 866 735 737 797 799 784 782 800 800 72738 742 630 630 653 664 667 684 603 602 590 611 466 561
Bands unique to SG-SERS The next chemical to be analyzed by both E-SERS and SG-SERS was adenosine monophosphate (AMP) The E-SER spectrum yields bands due to the adenine chemical functionality at 727 961 1233 1279 1331 1381 and 1486 cm-1 In addition phosphate bands are observed at 860 1097 1453 1587 and 1705 cm-1 (Figure 5) Other researchers have noted that the ribose component does not appear to contribute to the spectrum19 The AMP spectrum also changes as a function of potential As the electrode is swept more positive (here from -09 to -03VSCE) the phosphate bands at 860 1097 1453 and 1587 cm-1 increase in intensity compared to the adenine bands while a band at 1705 cm-1 appears The adenine bands at 1233 1381 and 1486 cm-1 virtually disappear These potential dependent spectral changes are consistent with earlier studies that show that phosphate is attracted to silver at potentials positive of the pzc but repelled at potentials negative of the pzc19
Wavenumbers (∆cm-1) Figure 3 E-SERS of A) 21x10-3M cytosine at -03VSCE 1000 scans B) ~10x10-5M guanine at -06VSCE 500 scans C) 23x10-3M thymine at -06VSCE 500 scans and D) 12x10-3M uracil at -093VSCE 500 scans All spectra at pH 92 750 mW 1064 nm at 8 cm-1
Wavenumbers (∆cm-1) Figure 4 SG-SERS of A) 21x10-3M cytosine 200 scans B) ~10x10-5M guanine 200 scans C) 23x10-3M thymine 200 scans and D) 12x10-3M uracil 500 scans All spectra at pH 92 75 mW 1064 nm at 8 cm-1
A
B
C
A
B
C
D D
SPIE 2001-4575
68
The SG-SER spectrum of AMP is considerably different The adenine bands virtually disappear except for the two primary bands which shift to742 and 1329 cm-1 While the phosphate band at 1459 cm-1 has gained considerable intensity In addition two new intense bands appear at 684 and 1539 cm-1 as well as a moderately intense band at 1180 cm-1 The SG-SER spectrum has greater similarity to the E-SER spectrum at -03VSCE and suggests that the silver particles embedded in the sol-gel behave as if at a potential positive of the pzc RNA samples extracted from E coli B subtilis and S aureus were next examined by both E-SERS and SG-SERS E-SER spectra of these samples yielded quality spectra in 10 minutes in which all of the major features can be identified (Figures 6 and 7) This includes guanine at 650 cm-1 adenine at 791 cm-1 cytosine and uracil combining at 790 cm-1 and phosphate at 1100 1335 (in combination with adenine and guanine) 1465 and 1570 cm-1 Surprisingly adenine which demonstrated the greatest surface-enhanced Raman effect does not dominate the ring-breathing mode portion of the spectrum The intensities of the other base-pairs bands are of the same order of magnitude This suggests that when the base-pairs are linked together as in RNA they are enhanced in concert In fact the relative intensities are very similar to a normal Raman spectrum of E coli RNA which shows the combined cytosine and uracil band at ca twice the intensity of the adenine band and ca four times the intensity of the guanine band Unfortunately this means that the independent enhancement factors for the NA bases can not be used to estimate relative concentrations For example the relative 791 and 734 cm-1 bands for B subtilis would indicate that the cytosine andor uracil concentration was at least 20 times the adenine concentration whereas each of the four RNA bases are known to contribute 15-35 Nevertheless it is worth noting that the three RNA samples yield different relative band intensities that were reproduced in numerous measurements Although the relative concentrations of the NA bases for these samples have not been determined these differences can be quantified If it is assumed that the 650 cm-1 band represents 25 guanine the 791 cm-1 band represents 25 adenine and the 790 cm-1 50 cytosine plus uracil in the E coli RNA spectrum then the relative concentrations can be estimated for the other RNA samples To aid this calculation the three spectra were normalized to the phosphate band at 1100 cm-1 which has been shown to correlate to the total phosphate concentration and can be used as an internal standard In addition a simple baseline correction was applied (Figure 7) This yields 15 adenine 30 guanine and 55 cytosine plus uracil for B subtilis RNA and 18 adenine 25 guanine and 57 cytosine plus uracil for S aureus RNA The average SN of these measurements was 26 with an average error of 8 of the value (SplusmnN) It is also worth noting that the three RNA spectra show a marked shift in a band near 825 cm-1 This band is assigned to the symmetric stretch of the O-P-O ester linkage9 The band appears at 815 cm-1 for S aureus shifting to 820 cm-1 for B subtilis and 830 cm-1 for E coli Others have used the normal Raman intensity of the band at 815 cm-1 as a direct indication of the amount of A-class helix present while the intensity of the band at 830 cm-1 has been used as a direct indication of the amount of B-class helix present However the latter is more associated with DNA than RNA
Wavenumbers (∆cm-1) Figure 5 E-SER spectra of 020 mgmL adenosine monophosphate at A) -03 and B) -09VSCE and C) SG-SER spectra Conditions sample in 01M KCl buffered to pH 92 A) and B) 750 mW C) 75 mW of 1064 64 scans (1-min) at 8 cm-1
Wavenumbers (∆cm-1) Figure 6 E-SERS of 01 mgmL RNA from E coli 02 mgmL RNA from B subtilis and 02 mgmL RNA from S aureus Conditions 01M KCl pH 92 -03VSCE 750 mW of 1064 nm 640 scans (10 min) at 8 cm-1
AMP RNA
E coli A
B
C
B subtilis
S aureus
SPIE 2001-4575
69
SG-SER spectra of reasonable quality were also obtained for E coli and B subtilis especially the latter (Figure 8) However the spectra differ substantially from the E-SERS of the same samples Both SG-SER spectra are dominated by adenine at 735 cm-1 and a band at 1030 cm-1 Although unassigned the latter does appear in the RNA E-SER spectra Bands at 1105 and 1565 cm-1 are likely due to phosphate while bands at 1320 and 1470 cm-1 are less confidently assigned to phosphate They are significantly less intense and somewhat shifted from their SG-SERS counterparts (1335 and 1455 cm-1) A band at 670 cm-1 may be due to guanine which was observed at 664 cm-1 for SG-SERS of the pure sample However the SG-SER spectrum of AMP also had an intense 667 cm-1 band A number of other bands occur at 890 1070 1165 1245 1290 1420 1505 cm-1 and remain unassigned The SG-SER spectra are somewhat disappointing in that only adenine and guanine contributions can be positively identified This limits the ability to determine relative NA base concentrations and distinguish bacterial RNA However several of the unassigned bands may be due to the bases (eg 1030 and 1420 cm-1 due to cytosine) Further experiments will be required to clarify this point
Ecoli
B subtilis
S aureas
P G
A C+U
OPO
A B
Wavenumbers (∆cm-1) Wavenumbers (∆cm-1) Figure 7 SER spectra of RNA from A) B subtilis with contributions indicated and B) E coli B subtilis and S aureus with baseline correction and peak positions used to calculate contributions indicated G = guanine A = adenine C+U = cytosine plus uracil P = phosphate (backbone) OPO = phosphate ester linkage (A- vs B-class helix)
Wavenumbers (∆cm-1) Wavenumbers (∆cm-1) Figure 9 E-SER (-03VSCE) and SG-SER spectra of RNA from A) E coli and B) B subtilis Sample conditions as in Figure 6 E-SER spectra at 750 mW SG-SERS at 75 mW
A B E-SERS
SG-SERS
E-SERS
SG-SERS
SPIE 2001-4575
70
A final comparison was made between the two SER methods by measuring dipicolinic acid (DPA) This chemical may be invaluable as a test for spore forming bacteria specifically B anthracis 50 to 90 of B anthracis sporilates During spore formation dipicolinic acid is synthesized and once completed 10-15 of the dry spore weight is composed of the Ca2+ complex located in the spore core42 Heating in water can be used to initiate germination at which point the exosporium breaks and releases the Ca dipicolinate which becomes dipicolinic acid in water The structure of this chemical strongly
4 CONCLUSIONS Towards the goal of developing a method and technology to rapidly detect and identify bioagents we have been investigating surface-enhanced Raman spectroscopy as a tool to measure relative concentrations of nucleic acid bases in RNA and determine bioagent taxonomy Initially we investigated E-SERS since this method has been extensively researched and the optimum sample conditions are well developed However this method requires a three-electrode sample cell and electrolyte solution Incorporation of an electrolytic cell into sample systems used to analyze for BWA as aerosols in water or on surfaces can be designed using flow injection analysis technologies but cross-contamination and plugging of sample lines seems inevitable For this reason we also investigated metal-doped sol-gels as a SER-active medium Previous studies have shown this material to be active in all solvents particularly water capable of continuous measurements in flowing systems and reproducible (quantitative) between coated sample vials Here we compared SG-SER spectra to traditional E-SER spectra of the nucleic acid base pairs adenosine monophosphate and RNA High quality spectra of adenine cytosine guanine thymine and uracil were obtained by both E-SERS and SG-SERS Both methods yielded very similar spectra for the NA bases including a pH dependent study of adenine Enhancement factors and detection limits for adenine were determined as 2x105 and 16x10-11g and 1x105 and 12x10-10g for E-SERS and SG-SERS respectively (normalized to 75 mW and 10-min acquisition time) Fifty percent of the silver particle surface area in the sol-gel matrix was assumed covered by adenine which may have been overestimated yielding a lower EF and higher detection limit It should also be realized that each E-SER spectrum required several attempts to optimize the measurement conditions (pH electrode potential etc) While each SG-SER spectrum involved no sample preparation and often represents the first and only attempt to make the measurement Quality spectra of RNA extracted from Escherichia coli Bacillus subtilis and Staphylococcus aureus were obtained by E-SERS that were easily interpreted Bands due to adenine guanine cytosine plus uracil and phosphate were identified The SER band intensity of the NA bases in the RNA samples were of the same order of magnitude suggesting that their interaction with the silver surface is concerted as is their Raman enhancement Interestingly the relative SER band
Wavenumbers (∆cm-1) Figure 10 A) Raman spectrum of solid dipicolinic acid B) Ax20 C) electrolytic SERS of 6x10-3 M dipicolinic acid in 01 M KCl at a potential of +07VSCE and pH of 4 and D) sol-gel SERS of 6x10-3 M dipicolinic acid Conditions for A and C as in Figure 1 C) 100 mW of 1064 nm 50 scans 8 cm-1
A B (Ax20)
C
D
suggested that it would be SER active However the E-SER spectrum was unstable and varied considerably as a function of potential A consistent spectrum was obtained at +06VSCE (Figure 10) This potential is not recommended for measurement because the surface is actively dissolving in solution The SG-SER spectrum was considerably more stable of higher quality and easily reproduced Bands at 660 825 1010 1390 1430 1570 1590 and 3075 cm-1 were observed Enhancement factors were determined for the two media using the symmetric ring stretching mode at 995 and 1010 cm-1 for the normal Raman and SER spectra respectively E-SERS yielded an EF of 5x103 while SG-SERS yielded an EF of 2x105 for DPA The SN of the latter suggests a detection limit of 20x10-10g (based on adenine coverage 75 mW and 10-min) The differences in SER activity for these two media may be attributed to the combined electrolytic potential of the solution chemical and metal15 Again the E-SERS suggests that the SG-SERS is at a potential positive of the pzc While the instability in the E-SERS may also be associated with surface interactions of two carboxylic acid groups of dipicolinic acid during the ORCs
DPA
SPIE 2001-4575
71
intensities for RNA extracted from E coli are very similar to those measured by normal Raman spectroscopy Although the relative percent that each of the NA bases contributed to each RNA sample was not determined reproducible band intensities allowed noting the following trends The percent adenine decreases while the combined percent cytosine and guanine increase for both B subtilis and S aureus compared to E coli Quality spectra were also obtained for the RNA samples by SG-SERS but only a few bands were readily identified Calculations of NA base concentrations by SG-SERS will require further research In light of recent events we summarize the capabilities of these SERS media in terms of rapid detection of B anthracis and dipicolinic acid However these capabilities must be qualified First and foremost the level to which SERS can distinguish bacteria or viruses has not yet been determined Development of a database of both DNA and RNA base concentrations for BWAs and common bacteria to establish the level of taxonomic identification is ongoing Second rapid collection of aerosol water or surface samples is being addressed by others who report trapping particles on filters from 100 liters of air per minute Third although not presented here we have developed methods to extract RNA or DNA from cells and spores for SER analysis within 10 minutes Finally we assume a detection limit of 3600 spores per 100 liters of air is required although a 50 lethal dosage of anthrax has not been established With these qualifications a mass detection limit for RNA using SERS is estimated as follows A single measurement is performed in ca 20 minutes (140 liters collected in 14 min RNA extracted in 8 min spectral acquisition and analysis in 10 min) The average human breaths 7 liters per minute therefore the analyzer must at the very minimum detect 5000 spores in 140 liters of air One spore is approximately 2x10-18m3 (1x1x2 microm) and if a density of 075 gcm3 is assumed this corresponds to a mass of 15x10-12g Each spore contains 4-12 RNA or 12x10-13g RNA for 8 If we assume 23 of the RNA can be isolated for analysis during lysis then the proposed instrument must be able to detect 4x10-10g RNA from 5000 spores per 70 liters of air within 10 minutes As noted above the mass detection limits for adenine were estimated at 16x10-11g and 12x10-10g for E-SERS and SG-SERS respectively Although these detection limits suggests that RNA from 5000 spores is detectable with the current instrumentation it is highly likely that only a portion of an RNA segment (eg 120-nucleotide 5S rRNA) is in contact with the metal surface and will contribute to the SER effect The SN for the RNA spectra were 110th of the average SN for the four individual RNA bases suggesting a 10 contribution Furthermore effective taxonomy will likely require knowing the NA base concentrations to 1 of the value (eg 25plusmn025) Again the average measurement error for the bases is 12 These values suggest that the E-SERS is within a factor of 4 of the required detection limit whereas the SG-SERS detection limit must be improved by 25 times The same arguments can be applied to the detection of dipicolinic acid If we assume a spore releases 10 by weight DPA during germination then the proposed instrument must be able to detect 75x10-10g DPA from 5000 spores per 70 liters of air within 10 minutes The detection limit for SG-SERS was estimated at 20x10-10g and suggest that the vials are suitable to perform a rapid screen for anthrax A series of concentration dependent measurements are currently being performed to verify this assertion Finally we note that the measurements performed here employed an FT-Raman spectrometer This instrumentation was chosen over dispersive Raman instrumentation for the high wavelength accuracy afforded by the HeNe reference laser (Connes Advantage43) which would allow reliable spectral subtraction matching of observed spectra to stored library spectra and confident use of chemometric approaches Such data analysis is likely to be required to enhance BWA identification However this instrumentation sacrifices sensitivity due to a less efficient detector (InGaAs vs Si) less efficient Raman scattering and less efficient generation of plasmon modes Substantial improvements can be obtained using visible excitation and Si detection and these measurements are underway
5 ACKNOWLEDGEMENTS The authors are grateful to Drs D Cookmeyer and S Tove of the US Army Research Office (Contract Number DAAH04-96-C-0078) for their interest and support of this research The authors would also like top acknowledge Dr R Yin and J Jensen for supporting development of the metal-doped sol-gels (Contract Number DAAD13-01-C-0019) They also thank Dr Wilfred H Nelson for assistance in spectral interpretations
SPIE 2001-4575
72
6 REFERENCES 1 Roberts WL and Rainey PM Clin Chem 39 1872-1877 (1993) 2 Pasechnik VA CC Shone and P Hambleton Bioseperations 3 267-283 (1993) 3 Jackson PJ ME Hugh-Jones DM Adair G Green KK Hill CR Kuske LM Grinberg FA Abramova and P
Keim Proc Natl Acad Sci 95 1224-1229 (1998) 4 Jensen JL NC Wong W R Loerop SPIE 4775-03 (2001) 5 Snyder A Pet al SPIE 3853-15 (1999) 6 Danosky T R and McFadden P N in press (1997) 7 Woodruff WH Farquharson S Science 201 831-833 (1978) 8 Lord RC and Thomas GJJr Spectrochemica Acta 23A 2551-2591 (1967) 9 Thomas GJJr Biochim Biophys Acta 213 417-423 (1970) 10 Placzek G Handbuch der Radiologie 2 EMarx ed Akademische Verlagagescellschatt Liepzig 1934 UCRL
Trans No 526 (1959) 11 Jeanmaire DL and RP Van Duyne J Electroanalytical Chem 84 1-20 (1977) 12 Chada S Manoharan R Moenne-Loccoz P Nelson WH Peticolas WL and Sperry JF Applied Spectroscopy
47 38-43 (1993) 13 Manoharan R Ghiamati E Chada S Nelson WH and Sperry JF Applied Spectroscopy 47 2145-2150 (1993) 14 Todd EA Morris MD Applied Spectroscopy 48 545-548 (1994) 15 Wentrup-Byrne E Sarinas S and Fredericks PM Applied Spectroscopy 47 1192-1197 (1993) 16 Weldon MK VR Zhelyaskov and MD Morris Applied Spectroscopy 52 265-269 (1998) 17 Kneipp K and J Fleming J Mol Structure 145 173-179 (1986) 18 Pothier NJ and Force RK Applied Spectroscopy 46 147-151 (1992) 19 Ervin KM E Koglin JM Sequaris P Valenta and HW Nurnberg J Electroanal Chem 114 179-194 (1980) 20 Kim SK TH Joo SW Suh and MS Kim J Raman Spectrosc 17 381-386 (1986) 21 Pothier NJ and Force RK Analytical Chemistry 62 678-680 (1990) 22 Kneipp K Y Wang RR Dasari and MS Feld Applied Spectroscopy 49 780-784 (1995) 23 Nie S and Emory SR Science 275 1102 (1997) 24 Graham D WE Smith AMT Linacre CH Munro ND Watson and PC White Analytical Chemistry 69 4703-
4707 (1997) 25 Fleischmann M PJ Hendra and AJ McQuillan Chem Phys Lett 26 163-166 (1974) 26 Pettinger B J Chemical Phys 85 7442-7451 (1986) 27 Surface-Enhanced Raman Scattering Section Four Theory SPIE MS 10 M Kerker and B Thompson Eds (1990) 28 Wang D-S and Kerker M SPIE (M Kerker and B Thompson Eds) MS 10 417-429 (1990) 29 Weaver MJ Farquharson S Tadayyoni MA J Chem Phys 82 4867-4874 (1985) 30 Norrod KL Sudnik LM Rousell D and Rowlen KL Applied Spectroscopy 51 994-1001 (1997) 31 Farquharson S Weaver WJ Lay PA Magnuson RH and Taube H J Am Chem Soc 105 3350-3351 (1983) 32 Lee PC and Meisel D ldquoAdsorption and Surface-Enhanced Raman of Dyes on Silver and gold Solsrdquo J Phys Chem
86 3391-3395 (1982) 33 Li Y-S and Wang Y Applied Spectroscopy 46 142-146 (1992) 34 Bello JM DL Stokes and T Vo-Dinh Analytical Chemistry 61 1779-1783 (1989) 35 van Duyne RP J C Hulteen D A Treichel M T Smith M L Duval and T R Jensen J Phys Chem B1033854-
3863 (1999) 36 Farquharson S and W W Smith W H Nelson and J F Sperry SPIE 3533-27 207-214 (1998) 37 Lee Y H W Smith S Farquharson H C Kwon M R Shahriari and P M Rainey SPIE 3537 252-260 (1998) 38 Lee Y-H S Farquharson and P M Rainey SPIE 3857 76-84 (1999) 39 Current Protocols in Molecular Biology Wiley Interscience1003-1006 (1987) 40 Farquharson S Smith W Carangelo R C and Brouillette C SPIE 3859 14-23 (1999) 41 Chase D B and JF Rabolt Fourier Transform Raman Spectroscopy Acad Press Ch1 p 131 (1994) 42 Brock TD MT Madigan JM Martinko and J Parker Biology of Microorganisms 7th Ed Prentice Hall p 76-80
(1994) 43 Connes J Rev Opt Theor Instrum 40 45 (1961)
166
Chemical agent identification by surface-enhanced Raman spectroscopy
Stuart Farquharson and Paul Maksymiuk
Real-Time Analyzers 87 Church Street East Hartford CT 06108
Kate Ong and Steven D Christesen US Army SBCCOM Aberdeen Proving Ground MD 21010
ABSTRACT The recent distribution of anthrax through the US postal system and the subsequent infection and death of several postal and national media employees amplifies the need for methods to rapidly detect identify and quantify this and other chemical and biological warfare agents The US military has also identified water supplies as a likely method of warfare agent deployment and is funding the development of a Joint Service Agent Water Monitor (JSAWM) In an effort to aid military personnel and the public at large we are developing a portable analyzer capable of identifying and quantifying chemical agents rapidly either on-demand or continuously The approach is based on the ability of Raman spectroscopy to identify molecular structure through the abundant vibration information provided in spectra and the ability of surface-enhanced Raman spectroscopy (SERS) to detect extremely low concentrations (eg part-per-billion) through the enhancement of Raman scattering by six orders of magnitude or more A key element to the analyzer design is a new SER active medium that is capable of quantitative reversible measurements The medium consists of silver or gold nanoparticles incorporated into a sol-gel matrix The porous silica network offers a unique environment for stabilizing SER active metals and the high surface area increases the interaction between the analyte and metal particles Here we present the use of new sol-gels that also selectively enhance chemicals based on polarity and charge Base-line measurements of chemical agents and their hydrolysis products are presented and compared to the JSAWM goal of 30 micrograms per liter detection Keywords Chemical warfare agent hydrolysis product SERS Raman spectroscopy sol-gel nanoparticle
1 INTRODUCTION Since September 11 2001 the threat of terrorist attacks and biological warfare within US borders has become a sobering reality The simplicity in manufacturing ease of deployment and the relatively low cost of chemical warfare agents (CWAs) raises public concern that they may also be used by terrorists Indeed terrorists released sarin (GB) in the Tokyo subway in 19951 Countering terrorism and terrorist attacks requires recognizing likely deployment scenarios and having the required technology to rapidly detect the deployment event One method of deployment has been long identified by the US military distribution through water supplies To counter this threat the Department of Defense is funding or monitoring the capabilities of 200 technologies with the goal of developing a Joint Service Agent Water Monitor (JSAWM) that is field portable2 The JSAWM must be capable of identifying the chemical agents with no false-positives and quantifying the agents at microgram per liter concentrations within 10 minutes (Table 1)3 This includes the analysis of drinking water supplies distribution and storage systems as well as potable water supplies The earliest technologies developed for CWA detection were based on electrochemical ionization or colorimetric analysis (eg phosgene tape) Although these analyzers were easy to use they were not generally agent specific and suffered from false-positives4 More traditional laboratory methods have also been investigated and in particular combined gas chromatography and mass spectrometry (GCMS) has been very successful at eliminating false-positives56 However GCMS requires extraction repeated calibration and long analysis times (typically 20 to 60 minutes)6 making it labor intensive and less than desirable for field use More rapid analysis of agents in the solid liquid and gas phase has been To whom correspondence should be addressed emailfarqureal-time-analyzerscom
Vibrational Spectroscopy-based Sensor Systems Steven D Christesen Arthur J Sedlacek III EditorsProceedings of SPIE Vol 4557 (2002) copy 2002 SPIE middot0277-786X02$1500
167
Table 1 Chemical Agent Structures Hydrolysis Half-lives and JSAWM Thresholds Agent Short-Hand Chemical Structure Hydrolysis
Half-Life JSAWM
Thresholds
Sarin (GB) F-[O=P-CH3]-O-CH(CH3)2 213 hours 32 microgL Soman (GD) F-[O=P-CH3]-O-CH(CH3)-(C-(CH3)3) 23 hours 32 microgL Tabun (GA) (CH3)2-N-[O=P-CN]-O-C2H5 41 hours 32 microgL VX C2H5O-[O=P-CH3]-S-(CH2)2-N-(CH(CH2)2)2 821 hours 32 microgL EA2192 HO-[O=P-CH3]-S-(CH2)2-N-(CH(CH2)2)2 gt9 years 32 microgL Mustard (H) ClCH2CH2-S-CH2CH2Cl encapsulates 47 microgL Lewisite (L) ClCH=CH-As-Cl2 rapid 27 microgL HCN HCN rapid 20 mgL BZ C7NH12-O-[C=O]-COH(C6H5)2 23 microgL T-2 Toxin 87 microgL at pH 7 to 75 and 20 to 25 oC
demonstrated by vibrational spectroscopy7-10 Hoffland et al7 reported infrared absorbance spectra and absolute Raman cross sections for several chemical agents while Christesen measured Raman cross sections for sarin tabun mustard gas and VX (ethyl S-2-diisopropylamino ethyl methylphosphonothioate)11 Again however these techniques also have limitations Raman spectroscopy is simply not a very sensitive technique and detection limits are typically 01 (1000 ppm) While infrared spectroscopy would have limited value in analyzing poisoned water since the very strong infrared absorption of water would obscure most other chemicals present Nevertheless efforts to overcome these limitations have been demonstrated Braue and Pannella8 quantified the G-series nerve agents (tabun sarin and soman) in terms of infrared attenuated total reflectance using a circle-cell And Alak and Vo-Dinh demonstrated the possibility of surface-enhanced Raman spectroscopy (SERS) to identify CWAs by measuring several organophosphonates that simulate the nerve agents12 However quantitative measurements have not been demonstrated for the SER-active material used (silver coated on alumina particles) or other SER-active media13 Recently we developed silver-doped sol-gels to promote the SER effect14-17 The porous silica network of the sol-gel matrix offers a unique environment for stabilizing SER-active metal particles and the sol-gel provides a high surface area that effectively increases the number of molecules observed within the Raman scattering volume The silver-doped sol-gels have been coated on the internal walls of standard glass vials to allow rapid SER analysis of small sample quantities (lt 01 mL) without preparation We have used p-aminobenzoic acid (PABA) as a test chemical to demonstrate surface enhancements greater than 106 reversible measurements in a flowing system reproducible measurements from vial-to-vial and batch-to-batch and measurements in multiple solvents including water14-17 Recently we used these vials to measure Tabun (GB) and Sarin and several hydrolysis products pinacolyl methyl phosphonate (PMP from Soman) and methyl phosphonic acid (MPA from all G-agents Figure 1) Although a number of unique vibrational bands are observed (eg C-N stretch doublet and P-C stretch) the G-agents were only observed for 5 concentrations and all spectra required baseline corrections Figure 1 Surface enhanced Raman spectra of ~5 vv A) Tabun and B) Sarin C) 1 vv PMP and D) 10 ppm MPA using sol-gel sample vials 785 nm excitation 1-min scan and CCD detection Performed at Aberdeen Proving Ground
C A
Wavenumber (∆cm-1) Wavenumber (∆cm-1)
D B
790
545 C-N
2135 2190 P-C 770
760 1290
Proc SPIE Vol 4577
168
Nevertheless MPA was readily observed for a 10 ppm sample with an estimated detection limit of 04 ppm (based on a signal-to-noise ratio of 3 for the 760 cm-1 band intensity) This measurement provides encouragement in that SERS may satisfy the needs of a JSAWM Furthermore MPA is also a hydrolysis product of VX and V-gas and EA2192 (Figure 2) and may prove a valuable indicator of agent usage Figure 2 Hydrolysis of Sarin to form hydrofluoric acid (HF) methylphosphonic acid 1-methylethyl ester (MPAMME) methyl phosphonic acid (MPA) and 2-propanol With this initial albeit modest success we began analyzing chemicals with various sol-gel compositions that we have been developing Here we describe four sol-gel compositions that select for 1) polar-positive 2) polar-negative 3) weakly polar-positive and 4) weakly polar-negative chemical species The ability of these sol-gels to select and enhance Raman scattering is described for several test chemicals and MPA
2 EXPERIMENTAL The chemicals analyzed as well as all chemicals used to prepare the metal-doped sol-gels were obtained at their purest commercially available grade from Aldrich (Milwaukee WI) The sol-gel designed to select for polar-negative species was prepared from a silver amine complex tetramethyl orthosilicate (TMOS) and methanol After mixing 02 mL of the sol-gel solution was transferred into a glass vial (2 mL) dried and heated The incorporated silver ions were then reduced using dilute sodium borohydride The vials were washed and dried prior to the addition of a sample solution In a similar manner the sol-gel designed to select for polar-positive species was prepared from a gold salt TMOS and methanol The sol-gel designed to select for weakly polar-negative species was prepared from a silver amine complex tetraethyl orthosilicate (TEOS) and methanol And the last sol-gel designed to select for weakly polar-positive species was prepared from a gold salt TEOS and methanol All samples were prepared in a chemical hood and transferred into plain or SER-active vials for analysis Normal Raman spectral measurements employed 1-mL pure samples that were placed in a 1-cm3 cuvette and weighed This yielded a powder density that allowed accurate calculation of molecules in the optical collection field SERS measurements employed 1-mg sample per mL water concentrations unless otherwise stated Once prepared a 01 mL sample was placed into one of the four selective sample vials which in turn was placed into the sample compartment of a Raman spectrometer for analysis A prototype Fourier transform Raman spectrometer (RTA) was used for these measurements18 The system consisted of a NdYAG laser (Brimrose) for excitation at 1064 nm an interferometer built by On-Line Technologies (OLT East Hartford CT) for frequency separation an uncooled InGaAs detector for signal detection (RTA) and an Intel 400 MHz Pentium II based laptop computer (Dell Round Rock TX) for interferometric control data acquisition (OLT) and analysis (LabVIEW by National Instruments Austin TX) Additional components included a Notch filter (Kaiser Ann Arbor MI) and interferometer entrance and exit optics (Edmund Scientific Barrington NJ) Fiber optics were used to deliver the excitation beam to the sample and the scattered radiation to the interferometer (1 meter lengths of 200 and 365 micron core diameter respectively Spectran Avon CT) A second Notch filter (Kaiser) was used as a beam splitter to direct the excitation beam along the same axis as the collected radiation A microscope object (20x04 Newport Irvine CA) was used to focus the beam into the sample and to collect the scattered radiation back along the same axis In this co-axial backscattering arrangement the excitation beam was passed through the outside of a glass vial and focused onto the silver-doped sol-gel film (01-03 mm thickness) containing the sample
3 RESULTS AND DISCUSSION p-aminobenzoic acid (PABA) and phenyl acetylene (PA) and were used to refine the selectivity and SER-activity of the four different metal-doped sol-gels PABA is a popular chemical used to evaluate the performance of SER-active media Here the polar end groups can be used to test selectivity of the polar-negative and polar-positive sol-gels PA is essentially non-
2O+ H HF + +OH
OH3H C
OP
OF
CH
CH3
3
3H C
OCP
OOH
CH
CH3
3
3H C
O
CPHO
CH
CH
3
3
C
Sarin MPAMME MPA 2-propanol
Proc SPIE Vol 4577
169
polar but a high electron density in the cylindrical π cloud around the carbon-carbon triple bond allows testing the selectivity of the weakly polar-negative and weakly polar-positive sol-gels As Figure 3 illustrates PABA passes through the polar sol-gel and is enhanced by either the silver or gold particles At 1 mgml the concentration of neutral PABA is ca 20 times that of the ionized form (pKa = 48) For electropositive silver the PABA anion is expected to interact through the carboxylate group and the associated vibrational modes are expected to dominate the spectrum Conversely for electronegative gold either form of PABA is expected to interact through the amine group The clear differences in our spectra support this expectation Furthermore bands at 840 and 1405 cm-1 assigned to a COO- bend and stretch respectively are significantly more intense for silver than gold Additional bands at 1140 and 1195 cm-1 are assigned to CH bending modes while bands at 1450 1500 and 1605 cm-1 are assigned to ring vibrational modes A very similar SER spectrum for PABA on a silver-coated alumina substrate has previously been reported with similar assignments19 For the gold-doped sol-gel new bands appear at 690 1355 and 1585 cm-1 The first band is assigned to a ring-H bending mode the second band to a ring-N- stretching mode and the third band to a possible NH2 scissors mode or ring mode The second band is not observed in the normal Raman spectrum but infrared bands occur at this frequency for aromatic ring-secondary amine stretching modes The scissors mode occurs at this frequency in Raman spectra for several chemicals but is absent in the PABA Raman spectrum Alternatively this mode may be the1600 cm-1 ring mode that has been shifted by the gold interaction Again a very similar SER spectrum of PABA has been reported but surprisingly using silver (colloids)2021 not gold as the enhancement medium These researchers also assumed the primary interaction of PABA with silver was through the carboxylate anion and made assignments accordingly For example they assigned the 1359 cm-1 to a COO- stretch not to the amine group as we have They also favor the ring stretching mode assignment for the 1582 cm-1 band Finally it should be said that other researchers have argued that the most dominant band in the SER spectra at 1450 cm-1 a ring vibration mode suggests that PABA lies flat on the surface and the π-orbitals dominate the surface interaction22
Figure 3 SER spectra of A) PABA using polar-negative and B) polar-positive sol-gels and C) PA using weakly polar-negative and D) weakly polar-positive sol-gels PABA is 1 mgmL PA is 1 vv Spectral conditions 75 mw 1064 nm 100 scans (15 min) 8 cm-1 resolution Non-polar PA passes through the non-polar sol-gels and is also enhanced by both metals The spectra are easily understood For electropositive silver PA interacts through the cylindrical triple bond π electron cloud and a -CequivC- doublet occurs near 2000 cm-1 The interaction is reasonably strong since this band appears at 2112 cm-1 in the normal Raman spectrum For electronegative gold this interaction is unlikely and only very weak bands occur near 2000 cm-1 The remaining bands are at 1000 cm-1 1200 cm-1 doublet and 1595 cm-1 all appear in the normal Raman spectra at virtually the same frequencies and are assigned to the symmetric ring-breathing mode CH bending modes and the trigonal ring-breathing mode respectively The polarnon-polar selectivity of the polar-negative and weakly polar-negative sol-gels was tested by adding a 11 molar mixture of PABA and PA The selective enhancement is quite good (Figure 4) The spectrum obtained using the polar sol-gel represents 78 PABA and 22 PA while the spectrum obtained using the weakly polar sol-gel represents 9 PABA and 91 PA The band peak intensities at 2000 cm-1 for PA and 1450 cm-1 for PABA were used for these calculations and are expanded in Figure 4 for clarity
C A
Wavenumber (∆cm-1) Wavenumber (∆cm-1)
2NH COOH
D B
C CH
Proc SPIE Vol 4577
170
Figure 4 SERS of 11 MM of PABA and PA in A) polar-negative and C) weakly polar-negative sol-gels The lower traces compare the pure chemicals B) 1 mgml PABA in polar-negative sol-gel and D) 1 PA in weakly polar-negative sol-gel while the insets magnify the minority species for clarity (x5 in A and x10 in B) Spectral conditions as in Figure 3 Following this development of selective sol-gels that maintained SER activity we measured cyanide and MPA (Figure 5) Not surprisingly the best sensitivity for both hydrolysis products was obtained using the polar-negative sol-gel The interaction of the cyanide anion with the silver surface is sufficient to shift the CequivN stretch observed at 2080 cm-1 in the normal Raman spectrum to 2145 cm-1 in the surface-enhanced Raman spectrum Furthermore the band is substantially broadened This anion has been extensively studied by electrolytic SERS and this shift and broadening have been attributed to the formation of a tetrahedral Ag(CN)3
2- surface structure23 Figure 5 Surface-enhanced (upper traces) and normal Raman spectra (bottom traces) of A) CN- and B) MPA in silver-doped TMOS SERS conditions as in Figure 3 and 1 mgmL Note MPA yields two distinct spectra for neutral (top) and acidic pH (middle) The normal Raman spectra employed pure powders 500 scans and 900 mW of 1064 nm SER measurements of MPA with the polar-negative sol-gel yielded two unique spectral signatures that depended on solution pH (Figure 5) For more neutral solutions the P-C stretch of MPA at 762 cm-1 dominates and the CH2 stretch at 2922 cm-1 appears The SN is sufficiently high that the anti-Stokes Raman shift at -762 cm-1 is observed For deprotonated MPA an oxygen-surface mode appears at 325 cm-1 (as well as its anti-Stokes complement) suggesting a strong interaction This results in substantial enhancement of the P-O-C mode at 1051 cm-1 (upper trace) Others report that this mode dominates the infrared spectra of nerve agents measured in water8 Comparison of the two spectra suggests the following molecule-to-surface orientations The appearance of the oxygen-surface and P-O-C modes in the upper spectrum of Figure 5B indicates that the tetrahedral molecule interacts with the silver surface through the deprotonated oxygen and is oriented end-on The
A B
Wavenumber (∆cm-1)
C A
Wavenumber (∆cm-1) Wavenumber (∆cm-1)
D B
Wavenumber (∆cm-1)
Proc SPIE Vol 4577
171
dominance of the P-C and the CH2 stretches and the disappearance of the P-O-C mode in the upper spectrum suggest the molecule is oriented with the methyl-hydroxyl-hydroxyl face towards the surface However considerably more research must be performed to verify these points Both chemical forms were measured as a function of concentration to determine the onset of monolayer coverage and estimate expected detection limits (Figure 6) Below monolayer coverage the signal to concentration dependence should be linear and the SN of any spectral measurement in this range can be used to predict the detection limit In the spectra presented here the peak height was used as the signal while the noise as root-mean-squared (RMS) was measured between 4400-4600 cm-1 Since noise is distributed evenly throughout the spectrum when transformed this region was used since it does not have any contributions from signals or baseline offsets Figure 6 shows a series of spectra for MPA along with a plot of the 1050 cm-1 band intensity (the noise was identical for this series) as a function of sample concentration A clear discontinuity appears in the vicinity of 01 mgmL (19 ppm) indicating the onset of monolayer coverage A detection limit defined as a SN of 3 was calculated for the 01 and 005gmL samples at 24x10-4 and 25x10-4 gL respectively A more modest detection limit of 101x10-4 gL was obtained using the 760 cm-1 band in the second series of concentration measurements These detection limits correspond to 46 and 190 ppb for a 3-min scan and 75 mW of incident laser power
Figure 6 A) Concentration dependence of MPA SERS measured in silver-doped TMOS) B) Concentrations are 001 005 01 05 1 gL (188 94 188 94 188 ppm) I760 series (bull) and I1050 series (∆)
Enhancement factors for cyanide and methyl phosphonic acid can be determined by estimating the number of molecules contributing to the surface-enhanced and normal Raman spectra The enhancement factor EF can be defined by the following equation
EF = (ISERSINR)bull(MNRMSERS) bull(PNRPSERS) bull(TNRTSERS)12
where I is the spectral band intensity M is the sample mass P is the incident laser power and T is the measurement time (or number of scans) for the two measurements For the normal Raman spectra a cylindrical scattering volume is assumed based on the laser area (28x10-7m2 6x10-4m diameter spot) and the penetration depth (1x10-3 m)24 The density of KCN and MPA as powders were measured at 0572 and 0516 gcm3 indicating that 16x10-4 and 144x10-4 g produced the normal Raman signals in Figure 5 respectively The number of molecules contributing to the SER spectra are those on the silver particles embedded in the sol-gel The total silver surface area can be determined from the average particle size concentration and the scattering volume Previous scanning electron micrographs showed the average silver particle size to be 40 nm in diameter (335x10-23m3)17 The silver concentration is 012M based on the reactant molar concentrations and dilution factors And the scattering volume is 76x10-
11m3 again based on a cylindrical scattering volume defined by a laser area of 28x10-7m2 and a sol-gel thickness of 27x10-
4m This volume contains 123x10-6g of silver equivalent to 35x109 silver particles with a collective surface area of 18x10-
5m2 However it may reasonably be assumed that at least half of the silver surface is in contact with the sol-gel matrix and unavailable for analyte interaction If we assume monolayer coverage and that each CN molecule occupies 15x10-20m2 then approximately 62x1014 molecules or 27x10-8g of CN contribute to the SER spectrum (20x10-19m2 46x1013 molecules 74x10-9g for MPA) Accordingly the EF for cyanide equals 48x104 ((180599) bull(16x10-427x10-8) bull(90075) bull(500100)12) The EF for MPA is considerably higher at 87x106 ((60326) bull(144x10-474x10-9) bull(90075) bull(500200)12)
0
20
40
60
80
100
120
140
0 02 04 06 08 1 12[MPA] (mgmL)
I (76
0)
0
100
200
300
400
500
600
I (10
50)
Wavenumber (∆cm-1)
A B
Proc SPIE Vol 4577
172
4 CONCLUSIONS Here we present for the first time surface-enhanced Raman spectra of Tabun and Sarin obtained using silver-doped sol-gels However the inferior enhancement suggested employing SER-active sol-gels varying in composition to improve both detection limits and selectivity of the target analytes To this end we successfully demonstrated the capabilities of four sol-gels that select for 1) polar-positive 2) polar-negative 3) weakly polar-positive and 4) weakly polar-negative chemical species p-aminobenzoic acid was used to show that silver could be used to attract polar-negative chemicals or functional groups (carboxylate anion) and that gold could be used to attract polar-positive chemicals (amine) while a mixture of p-aminobenzoic acid and phenylacetylene was used to show that tetramethyl orthosilicate preferentially solvates polar chemicals and that tetraethyl orthosilicate preferentially solvates weakly polar chemicals This increased sample control was applied to cyanide and methyl phosphonic acid two hydrolysis products of chemical warfare agents Exceptional results were obtained for methyl phosphonic acid allowing measurement of 1x10-2 gL for a 3-min scan and 75 mW of incident laser power with an estimated detection limit of 245x10-4 gL and an enhancement factor of 87x106 However this detection limit is 76 times less sensitive than required for the JSAWM (32x10-6gL for the G-agents) Finally we note that the measurements performed here employed an FT-Raman spectrometer This instrumentation was chosen over dispersive Raman instrumentation for the high wavelength accuracy afforded by the HeNe reference laser (Connes Advantage25) which would allow reliable spectral subtraction matching of observed spectra to stored library spectra and confident use of chemometric approaches Such data analysis is likely to be required to identify the chemical agents as well as distinguish hydrolysis products However this instrumentation which employs 1064 nm excitation and InGaAs detection sacrifices sensitivity We believe that the measurement sensitivity can be improved by at least two orders of magnitude by using shorter laser excitation wavelengths (eg 532 nm) This would provide more efficient Raman scattering (fourth power dependence on laser excitation wavelength) more efficient generation of plasmon modes and allow using more efficient detector material (Si vs InGaAs) These modifications are underway
5 ACKNOWLEDGEMENTS The authors would like to thank Dr R Yin and J Jensen of the US Army for supporting this work (Contract Number DAAD13-01-C-0019) They would also like to thank Advanced Fuel Research for making their laboratory facilities available
6 REFERENCES 1 Tu Anthony ldquoOverview of Sarin Terrorist Incidents in Japan in 1994 and 1995rdquo 6th CBW Protection Symposium
Stockholm Sweden 10-15 May 1998 2 Jensen JL NC Wong W R Loerop SPIE 4775-03 (2001) 3 JSAWM Requirements at wwwsbccomapgeaarmymilRDAecbcrtPRODSERJSAWMjsawmhtml 4 ldquoThe Chemical Weapons Convention Redefines Analytical Challengerdquo Analytical Chemistry News amp Features June 1
397A (1998) 5 Johnston RL Hoefler CM Fargo JC and Moberley B ldquoThe Defense Nuclear Agencyrsquos ChemicalBiochemical
Weapons Agreements Verification Technology Research Development Test and Evaluation Program and its Requirements for On-Site Analysisrdquo AT-ONSITE 5-8 (1994)
6 Black RM Clarke RJ Read RW and Reid MT ldquoApplication of gas chromatography-mass spectrometry and gas chromatography-tandem mass spectrometry to the analysis of chemical warfare samples found to contain residues of the nerve agent sarin sulphur mustard and their degradation productsrdquo J Chromatography 662 301-321 (1994)
7 Hoffland LD Piffath RJ Bouck JBrdquoSpectral signatures of chemical agents and simulantsrdquo Optical Engineering 24 982-984 (1985)
8 Braue EHJ Pannella MGrdquoCIRCLE CELL FT-IR Analysis of Chemical Warfare Agents in Aqueous Solutionsrdquo Applied Spectroscopy 44 1513-1520 (1990)
9 Tseng C-H Mann CK and Vickers TJ ldquoDetermination of Organics on Metal Surfaces by Raman Spectroscopyrdquo Applied Spectroscopy 47 1767-1771 (1993)
10 Staff Reporter ldquoUS Army Tests Lidar to Detect Biological Toxinsrdquo Photonic Spectra pg 50 December 1998
Proc SPIE Vol 4577
Proc SPIE Vol 4577
173
11 Christesen SD Raman cross sections of chemical agents and simulants Applied Spectroscopy 42 318-321 (1988) 12 Alak AM and Vo-Dinh T ldquoSERS of Organophosphorous Chemical Agentsrdquo Analytical Chemistry 59 2149-2153
(1987) 13 Norrod KL Sudnik LM Rousell D and Rowlen KL ldquoQuantitative Comparison of Five SERS Substrates
Sensitivity and Detection Limitrdquo Applied Spectroscopy 51 994-1001 (1997) 14 Lee Y and Farquharson S ldquoSERS Sample Vials Based on Sol-Gel Process for Trace Pesticide Analysisrdquo SPIE 4206
140-146 (2000) 15 Farquharson S and Lee Y ldquoTrace Drug Analysis by Surface-Enhanced Raman Spectroscopyrdquo SPIE 4200-16 (2000) 16 Lee Y Farquharson S and Rainey P M Surface-Enhanced Raman Sensor for Trace Chemical Detection in Water
SPIE 3857 76-84 (1999) 17 Lee Y Farquharson S Kwong H and Shahriari M ldquoSurface-Enhanced Raman Sensor for Surface-Enhanced Raman
Spectroscopyrdquo SPIE 3537 252-260 (1998) 18 Farquharson S Smith W Carangelo R C and Brouillette C ldquoIndustrial Raman Providing Easy Immediate Cost
Effective Chemical Analysis Anywhererdquo SPIE 3859 14-23 (1999) 19 Narayanan VA JM Bello JD Stokes and T Vo-Dinh Analusis 19 307-310 (1991) 20 Laserna JJ E L Torres and JD Winefordner Analytica Chemica Acta 469-480 (1987) 21 Torres EL and JD Winefordner Analytical Chemistry 59 1626-1632 (1987) 22 Suh JS DP DiLella M Moskovits J Phys Chem 87 1540-1544 (1983) 23 Benner RE R Dornhaus R Chang and BL Laube Correlations in the Raman spectra of cyanide complexes adsorbed
at silver electrodes with voltammograms Surface Science 101 341 (1980) 24 Chase D B and JF Rabolt Fourier Transform Raman Spectroscopy Acad Press Ch1 p 131 (1994) 25 Connes J Rev Opt Theor Instrum 40 45 (1961)
APPLIED SPECTROSCOPY 351
focusing the 488 nm laser beam 2 mm deep into thebulk of the crystals to avoid contributions from potentialdepletion layers As shown in Fig 1b the dependence ofthe Raman shift on the degree of deuteration is almostperfectly linear and ts very well with D 5 22684cmR1 24526 where D is the degree of deuteration (in )and R is the spectral mean of the PO4 vibration in cm21A linear correlation coef cient of 0998 indicates an ex-cellent linear dependence of the Raman peak shift withdegree of deuteration This result shows that the shift ofthe PO 4 peak is simply caused by the linear increase inatomic mass due to isotope substitution which decreasesthe length of hydrogen-like bonds
This excellent linear dependence allows us to map thepro le of the DH exchange layer at the surface of DKDPcrystals by acquiring Raman spectra and determining theposition of the PO 4 peak for various depths This methodis preferable over other methods such as determining thestrength of the OD vibration directly (eg at 715 cm21)because the position of the most intense peak in the Ra-man spectrum can be measured more precisely than theintensity of some of the weakest peaks in the spectrumThis is demonstrated in Fig 2 where depth-dependentRaman spectra (Fig 2a) and the resulting exchange layerpro les for two DKDP crystals are shown (Fig 2b) Thespectra in Fig 2a were obtained from a depth scan of aDKDP crystal with 75 degree of deuteration in thebulk grown at 45 8C The spectra start out as DKDP with30 deuteration close to the surface and approach thebulk DKDP spectrum within a few micrometers of depthThe fact that the relative degree of deuteration does notextend to 0 D is due to the limited depth resolution ofthe Raman microprobe which averages over 4 mm indepth Figure 2b depicts the resulting DH exchange layerpro les for this and a second crystal grown at 63 8Crespectively Both crystals had the same exposure to am-bient conditions and their main difference is the temper-ature at which they were grown The different exchangelayer pro les indicate that crystals grown at differenttemperatures have differen t proton conductiv ities 13
which leads to a difference in their rate of deuteriumdepletion The parameters controlling this behavior arecurrently the objective of a detailed study the results ofwhich will be reported elsewhere
CONCLUSION
In conclusion we have shown that the shift of the to-tally symmetric PO 4 stretch mode in the Raman spectrumof DKDP crystals scales linearly with degree of deuter-ation This allows us to correlate Raman peak positionsto deuteration levels in these crystals We have presenteda new technique to determine DH diffusion pro les inDKDP frequency conversion crystals based on micro-Ra-man spectroscopy This technique is fast inexpensiveand works under various environmental conditionswhich will allow us to better understand and control deu-terium depletion in DKDP crystals
ACKNOWLEDGMENTS
We would like to thank M Runkel for rst discovering DKDP crack-ing R Floyd for providing DKDP crystals and L Chase and A Burn-ham for their support and helpful discussions This work was performedunder the auspices of the US Department of Energy by the University
of California Lawrence Livermore National Laboratory under contract W-7405-Eng-48
1 J J De Yoreo A K Burnham and P K Whitman Int Mat Rev47 113 (2002)
2 C E Barker R A Sacks B M Van Wonterghern J A Caird JR Murray J H Campbell K Kyle R B Ehrlich and N DNielsen Proc SPIE-Int Soc Opt Eng 2633 501 (1995)
3 T Suratwala paper to be published4 Cleveland Crystals Inc httpwwwclevelandcrystalscom
KDPshtmltable5 E A Popova I T Savatinova and I A Velichko Sov Phys Solid
State 12 1543 (1971)6 I P Kaminow R C C Leite and S P S Porto J Phys Chem
Solids 26 2085 (1965)7 J A Subramony B J Marquardt J W Macklin and B Kahr
Chem Mat 11 1312 (1999)8 H Tanaka M Tokunaga and I Tatsuzaki Solid State Commun
49 153 (1984)9 R J Nelmes G M Meyer and J E Tibballs J Phys C 15 59
(1982)10 M A Yakshin D W Kim Y S Kim Y Y Broslavets O E
Sidoryuk and S Goldstein Laser Physics 7 941 (1997)11 I Takenaga Y Tominaga S Endo and M Kobayashi Solid State
Commun 84 931 (1992)12 C Krenn personal communication13 M Sharon and A K Kalia J Solid State Chem 21 171 (1977)
Rapid Dipicolinic Acid Extractionfrom Bacillus Spores Detectedby Surface-EnhancedRaman Spectroscopy
STUART FARQUHARSON ALAND GIFT PAUL MAKSYMIUK andFRANK E INSCOREReal-Time Analyzers Inc East Hartford Connecticut06108
Index Headings Dipicolinic acid Bacillus spores Anthrax Surface-enhanced Raman spectroscopy
INTRODUCTION
The anxiety caused by the distribution of anthrax en-dospores through the US postal system in October 2001was exacerbated by the long time required for positiveidenti cation of the Bacillus anthracis spores and the un-known extent of their distribution Since that time manymethods capable of rapid eld analysis have been inves-tigated to augment or replace the laboratory method ofgrowing microorganisms in culture media which takesdays to perform12 Prominent among these approachesare polymerase chain reactions (PCR)3 immunoassaysand detection of calcium dipicolinate as a biochemicalsignature PCR employs primers to separate organism-
Received 10 October 2003 accepted 14 November 2003 Author to whom correspondence should be sent
352 Volume 58 Number 3 2004
speci c nucleic acid sequences (eg capsular protein en-coding gene for Bacillus anthracis)4 and polymerases toamplify the segment until it is detectable Recently am-pli cation times have been substantially reduced andcomplete analysis can now be performed in an hour orless Immunoassay methods are also being developed thatuse competitive binding of the bioagent (as an antigen)and its labeled conjugate for a limited number of anti-bodies Although analyses can be performed in under 30minutes a well-de ned anthrax antigen has not yet beenidenti ed5ndash7 and consequently the false-positive rate isunacceptably high8
A number of other methods are being developed witha focus on the detection of calcium dipicolinate (CaDPA)and its derivatives as a B anthracis signature This is sobecause only spore-forming bacteria contain CaDPA andthe most common potentially interfering spores such aspollen and mold spores do not Relatively fast methodshave been developed to chemically extract CaDPA andthen detect it directly by uorescence9 or indirectly byluminescence1011 In the latter case hot dodecylamine(DDA) has been used to extract dipicolinic acid (DPA)and terbium has been utilized to form a highly lumines-cent DPA complex11 Although measurements have beenperformed in as little as ve minutes it was found thatas many as three concentration-dependent complexes canform each with different lifetimes This coupled withthe fact that the Tb31 cation produces the same lumines-cence spectrum makes determinations of low spore con-centrations problematic
It has been long known that Raman spectra of Bacillispores are dominated by bands associated with CaDPA12
and that these spectra may provide a suitable anthrax sig-nature at the genus level13 Since that time considerableimprovements in Raman instrumentation have led to lab-oratory measurements of single Bacilli spores14 and to eld measurements of spores captured from a mail-sort-ing system15 However the single spore measurementsrequired complex instrumentation that is not ruggedwhile the eld measurements required milligrams of sam-ple Furthermore the Raman spectra of both measure-ments contained uorescence contributions that would in-crease uncertainty in quanti cation
In related research we demonstrated that nanogramquantities of DPA could be detected by uorescence-freesurface-enhanced Raman spectroscopy (SERS)16 We alsodemonstrated that microliter volumes of chemicals canbe detected by SERS using metal-doped sol-gel-packedglass capillaries17 Towards the goal of developing a rap-id eld SERS-based anthrax spore detector we havecombined our previous research and we now report thatDPA can be extracted from a 10 mg B cereus spore sam-ple using DDA in 1 minute and can be detected by SERSin an additional 1 minute
EXPERIMENTAL
Dipicolinic acid (26-pyridinedicarboxylic acid DPA)and dodecylamine (DDA) were used as received fromSigma-Aldrich (Milwaukee WI) Lyophilized B cereusspores prepared according to the literature13 were sup-plied by the University of Rhode Island and used as re-ceived Multiple particles approximately 01 mm3 each
were separated and weighed at 5 to 15 mg representing05 to 15 million spores The sample masses were con-sistent with a previous determination of spore density at0081 gmL that indicated a high degree of entrained air
All chemicals used to prepare the silver-doped sol-gelcoated capillaries were also obtained and used as receivedfrom Sigma-Aldrich According to previously publishedprocedures17 two precursor solutions were preparedmixed and then drawn into 1-mm-diameter glass capil-laries The silver amine precursor consisted of a 51 vvratio of 1 N AgNO 3 to 28 NH3OH while the alkoxideprecursor consisted of a 21 vv ratio of methanol to te-tramethyl orthosilicate The alkoxide precursors weremixed with silver amine precursor in an 81 vv ratioApproximately 015 mL was drawn into the capillarycoating a 15-mm length After sol-gel formation the in-corporated silver ions were reduced with dilute sodiumborohydride which was followed by a water wash to re-move residual reducing agent
A 100 mL drop of a 50 mM DDA solution in ethanolpre-heated to 78 8C was added to each of the B cereusparticles to digest the spore coat After 1 minute the re-sultant solution was drawn into a SER-active capillarythat was immediately xed horizontally to an XY posi-tioning stage (Conix Research Spring eld OR) just in-side the focal point of an f 07 aspheric lens The lensfocused the beam into the sample and collected the scat-tered radiation back along the same axis A dichroic lter(Omega Optical Brattleborough VT) was used to re ectthe excitation laser to the lens and pass the Raman scat-tered radiation collected by the lens An f 2 achromatwas used to collimate the laser beam exiting a 200-mm-core-diameter source ber optic while a second f 2 ach-romat was used to focus the scattered radiation into a 365mm ber optic (Spectran Avon CT) A short-pass lterwas placed in the excitation beam path to block the sil-icon Raman scattering generated in the source ber fromre ecting off sampling optics and reaching the detectorA long-pass lter was placed in the collection beam pathto block the sample Rayleigh scattering from reachingthe detector A 785 nm diode laser (Process InstrumentsInc model 785-600 Salt Lake City UT) was used todeliver 100 to 150 mW of power to the sample A Fouriertransform Raman spectrometer (Real-Time Analyzersmodel IRA-785 East Hartford CT) and a silicon photo-avalanche detector (Perkin Elmer model C30902S Stam-ford CT) were used to acquire the SER spectra
RESULTS AND DISCUSSION
As an initial experiment the SER spectrum of 1 gLof DPA in water was measured using the newly devel-oped silver-doped sol-gel-coated capillaries (Fig 1A) Atthis concentration a high signal-to-noise ratio (SN) isobtained in 1 min In fact a reasonable spectrum is ob-tained in the same time frame for 1 mgL (Fig 1B) TheSER spectra are reasonably similar to the normal Raman(NR) spectrum obtained for a saturated solution of DPAin 1 N KOH (Fig 1C) and the following band shifts areobserved (NR to SER) 647 to 657 cm21 817 to 815cm21 998 to 1008 cm21 1384 to 1382 cm21 1434 to1428 cm21 and 1569 to 1567 cm21 Many of these bandshave been previously assigned1213 such as 998 cm21 to
APPLIED SPECTROSCOPY 353
FIG 1 SERS of DPA in water using silver-doped sol-gel-coated glasscapillary for (A) 1 gL and (B) 1 mgL (C ) NR of saturated DPA in 1N KOH in a glass capillary Spectral conditions (A) and (B) 150 mWof 785 nm 1-min acquisition time (C) 450 mW of 785 nm 5-minacquisition time both 8 cm 21 resolution
FIG 2 SERS of DPA extracted from 10 mg B cereus particle using100 mL of 50 mM hot DDA acquired in (A) 1 minute and (B) 2 seconds(C ) Attempted SERS of 50 mM hot DDA in ethanol using silver-dopedsol-gel-coated glass capillary acquired in 1 min Spectral conditions150 mW of 785 nm 8 cm21 resolution
the symmetric ring stretch 1384 cm21 to the OndashCndashOsymmetric stretch 1428 cm21 to the symmetric ring CndashH bend and 1569 cm21 to the asymmetric OndashCndashOstretch
The rst B cereus samples consisted of 2 mg of sporesin 2 mL of 5 mM hot DDA The samples were main-tained at 78 8C for 40 min and while hot approximately10 mL was drawn into a SER-active capillary Since spec-tra of DPA were obtained for these initial samples small-er spore masses higher DDA concentrations and shorterheating periods were examined In due course it wasfound that 10 mg of spores could be digested by 100 mLof 50 mM hot DDA in one minute and detected (Fig2A) In fact the signal was suf ciently intense that it canbe observed in as little as two seconds (Fig 2B) Theamount of DPA that was extracted was estimated to bebetween 5 and 10 mgL by comparing the signal intensityof the 1008 cm21 band to that measured for DPA in waterThis is consistent with previous research that found thatthe majority of the DPA is extracted from spores usingDDA11 and that B cereus spores contain approximately10 DPA by weight18 The SN of 127 for the 1008 cm21
band in the 1-minute SER spectrum suggests a limit ofdetection of approximately 250 ng of B cereus sporesbased on a SN of 3 Finally it should be noted that DDAdid not produce a detectable SER spectrum as shown inFig 2C
CONCLUSION
We have demonstrated that by combining rapid extrac-tion of dipicolinic acid from Bacillus cereus spores withchemical identi cation by surface-enhanced Raman spec-troscopy as little as 10 mg of spores can be detected Infact the entire measurement from the time of adding hotdodecylamine to the spores to the time when the dipi-colinic acid SER spectrum is acquired and analyzedcould be performed in less than two minutes The abilityof this method to distinguish between spore-forming bac-teria such as Bacillus anthracis and non-DPA containingpowders could help prevent costly shutdowns associated
with the appearance of suspicious material or intentionalmailing of common substances as an anthrax hoax Thismethod could also prove useful in detecting the locationof anthrax endospores in mail distribution facilities if an-other veri ed attack should occur
Research continues to fully characterize the surface-enhanced Raman spectroscopy signal intensities as afunction of sample concentration and to explore otherextractants that do not require the use of elevated tem-perature
ACKNOWLEDGMENTS
The authors are grateful for the support of the National Science Foun-dation (DMI-0296116 and DMI-0215819) and the US Army(DAAD13-02-C-0015 Joint Service Agent Water Monitor program)The authors are indebted to Chetan Shende for preparing the sol-gelcapillaries The authors also thank James Gillespie Nicholas Fell andAugustus Fountain for providing important background informationMark Farquharson for laboratory support and Professor Jay Sperry ofthe University of Rhode Island for supplying B cereus spores
1 V A Pasechnik C C Shone and P Hambleton Bioseparations 3267 (1993)
2 P J Jackson M E Hugh-Jones D M Adair G Green K K HillC R Kuske L M Grinberg F A Abramova and P Keim ProcNatl Acad Sci USA 95 1224 (1998)
3 B R Glick and J J Pasternak Molecular Biology Principles andApplications of Recombinant DNA (ASM Press Washington DC1994)
4 C A Bell J R Uhl T L Had eld J C David R F Meyer TF Smith and F R Cockerill III J Clin Microbiol 40 2897(2002)
5 D L Gatto-Menking H Yu J G Bruno M T Goode M Millerand A W Zulich Biosens Bioelectron 10 501 (1995)
6 J J Quinlan and P M Foegeding J Rapid Methods AutomationMicrobiol 6 1 (1998)
7 A A Hindle and E A H Hall Analyst (Cambridge UK) 1241599 (1999)
8 M S Ascher US Department of Health amp Human Services(httpwwwhhsgovophppresentationsAscherdoc)
9 R Nudelman B V Bronk and S Efrima Appl Spectrosc 54 445(2000)
10 D L Rosen C Sharpless and L B McBrown Anal Chem 691082 (1997)
354 Volume 58 Number 3 2004
11 P M Pellegrino N F Fell Jr and J B Gillespie Anal ChimActa 455 167 (2002)
12 W H Woodruff T G Spiro and C Gilvarg Biochem BiophysRes Commun 58 197 (1974)
13 E Ghiamati R S Manoharan W H Nelson and J F SperryAppl Spectrosc 46 357 (1992)
14 A P Esposito C E Talley T Huser C W Hollars C M Schal-dach and S M Lane Appl Spectrosc 57 868 (2003)
15 S Farquharson L Grigely V Khitrov W W Smith J F Sperryand G Fenerty J Raman Spectrosc paper accep ted (2003)
16 S Farquharson W W Smith S Elliott and J F Sperry SPIE-IntSoc Opt Eng 3855 110 (1999)
17 S Farquharson and P Maksymiuk Appl Spectrosc 57 479(2003)
18 F W Janssen A J Lund and L E Anderson Science (Washing-ton DC) 127 26 (1958)
SPIE -2003-5269 117
pH dependence of methyl phosphonic acid dipicolinic acid and cyanide by surface-enhanced Raman spectroscopy
Stuart Farquharson Alan Gift Paul Maksymiuk Frank Inscore and Wayne Smith
Real-Time Analyzers 87 Church Street East Hartford CT 06108
ABSTRACT US and Coalition forces fighting terrorism in Afghanistan and Iraq must consider a wide range of attack scenarios in addition to car bombings Among these is the intentional poisoning of water supplies to obstruct military operations To counter such attacks the military is developing portable analyzers that can identify and quantify potential chemical agents in water supplies at microgram per liter concentrations within 10 minutes To aid this effort we have been investigating the value of a surface-enhanced Raman spectroscopy based portable analyzer In particular we have been developing silver-doped sol-gels to generate SER spectra of chemical agents and their hydrolysis products Here we present SER spectra of methyl phosphonic acid and cyanide as a function of pH an important factor affecting quantitation measurements which to our knowledge has not been examined In addition dipicolinic acid a chemical signature associated with anthrax-causing spores is also presented Keywords Chemical warfare agents agent detection agent hydrolysis SERS Raman spectroscopy homeland security
1 INTRODUCTION In the past decade the Unites States and its allies have been challenged by a different kind of warfare exemplified by the terrorist attacks of September 11 2001 Suicide bombings and the use of chemical agents are the norm and military personnel must consider a wide range of attack scenarios Among these is the intentional poisoning of water supplies to obstruct military operations in Afghanistan and Iraq Currently colorimetric paper is used to detect agents on-site while gas chromatography combined with mass spectrometry (GCMS) is used in mobile support laboratories However both methods have severe drawbacks The paper changes color in response to contact with many chemicals besides CWAs causing a high incidence of false positives1 while GCMS although very chemically specific requires hours to perform and constant re-calibration234 Military operations would be greatly aided by a portable analyzer that can identify and quantify potential chemical agents at concentrations that impact safety This includes the analysis of drinking water supplies distribution and storage systems To meet this goal the Department of Defense has been investigating numerous approaches under the auspices of the Joint Service Agent Water Monitor (JSAWM) program5 The JSAWM must be capable of identifying the chemical agents with no false-positives and quantifying the agents at microgram per liter concentrations within 10 minutes678 Recently we and others have been investigating the ability of surface-enhanced Raman spectroscopy (SERS) to measure chemical agents9-12 bioagents13-17 and their hydrolysis products in water SERS employs the interaction of surface plasmon modes of metal particles with target analytes to increase scattering efficiency by as much as 1 million times18 In 1987 the potential of this approach to measure CWAs was demonstrated by measuring a series of pesticides19 Several of these organophosphonates have chemical structures similar to CWAs in particular P=O functional groups In our studies we have been employing silver-doped sol-gels to promote the SER effect The porous silica network of the sol-gel matrix offers a unique environment for immobilizing and stabilizing SER-active metal particles20-23 The sol-gels have been coated on the internal walls of standard glass vials to allow rapid SER analysis of small sample quantities without preparation We have measured over 100 chemicals with enhancements of 104 to 106 demonstrated reversible measurements in a flowing system reproducible measurements from vial-to-vial and measurements in multiple solvents including water20-23 Previously we used these vials to perform preliminary measurements of cyanide (CN) methylphosphonic acid (MPA) and dipicolinic acid (DPA) MPA is a hydrolysis product of the nerve agents (eg sarin Reaction 1) and may be a valuable indicator of nerve agent usage particularly since the alkyl methylphosphonic acids are relatively more stable than their corresponding parent complexes24 DPA is
SPIE -2003-5269 118
a chemical signature of spore forming bacteria such as Bacillus anthracis And in light of the inability to rapidly detect the anthrax spores distributed through the US mail in October 2001 a number of methods are being developed to extract and analyze this signature Reaction 1 Stepwise hydrolysis of Sarin to form hydrofluoric acid (HF) isopropyl methylphosphonic acid (IMPA) then methyl phosphonic acid (MPA) and 2-propanol In our previous SERS investigations MPA and DPA were measured at 50 and 100 mgL respectively In both cases limits of detection (LOD) were estimated at 100 microgL providing encouragement in that SERS may satisfy the needs of the JSAWM Since it has been shown that pH can substantially influence the intensity of SER bands25 which would clearly influence quantitative analysis we undertook the present study to determine the severity of these effects for cyanide methyl phosphonic acid and dipicolinic acid Furthermore we previously observed a band at 1050 cm-1 for MPA6 possibly due to an anion formed at basic pH Here we investigate the source of this spectral anomaly
2 EXPERIMENTAL All chemicals including potassium cyanide methyl phosphonic acid dipicolinic acid and those used to prepare the silver-doped and gold-doped sol-gel coated vials were obtained and used as received from Sigma-Aldrich All samples were prepared in HPLC grade water (Fischer Scientific Fair Lawn NJ) for SERS measurements The pH of these samples was adjusted using dilute nitric acid or potassium hydroxide and verified using a pH electrode (Corning Inc Corning NY) that had been calibrated with pH 400 700 and 1000 buffered standards from Fischer Scientific Once prepared the samples were transferred into the silver-doped sol-gel vials (Real-Time Analyzers Inc East Hartford CT) The vials were coated in a manner similar to that previously reported by adding ammonium hydroxide to a solution of silver nitrate tetramethyl orthosilicate (TMOS) and methanol20 Gold-vials were coated by adding nitric acid to a solution of gold tetrachloride TMOS and methanol The two precursor solutions were prepared mixed and transferred to 2-mL glass vials dried and heated After sol-gel formation the incorporated metal ions were reduced with dilute sodium borohydride (1mgmL) which was followed by a water wash to remove residual reducing agent After the resultant analyte solution was introduced the SER-active vial was immediately fixed horizontally to an XY positioning stage (Conix Research Springfield OR) just inside the focal point of an f07 aspheric lens The lens focused the beam into the sample and collected the scattered radiation back along the same axis A dichroic filter (Omega Optical Brattleborough VT) was used to reflect the excitation laser to the lens and pass the Raman scattered radiation collected by the lens An f2 achromat was used to collimate the laser beam exiting a 200 microm core diameter source fiber optic while a second f2 achromat was used to focus the scattered radiation into a 365 microm fiber optic (Spectran Avon CT) A short pass filter was placed in the excitation beam path to block the silicon Raman scattering generated in the source fiber from reflecting off sampling optics and reaching the detector A long pass filter was placed in the collection beam path to block the sample Rayleigh scattering from reaching the detector A 785 nm diode laser (Process Instruments Inc model 785-600 Salt Lake City UT) was used to deliver 100 to 150 mW of power to the sample A Fourier transform Raman spectrometer (Real-Time Analyzers model IRA-785 East Hartford CT) and a silicon photo-avalanche detector (Perkin Elmer model C30902S Stamford CT) were used to acquire the SER spectra
3 RESULTS AND DISCUSSION In a previous study of MPA6 aimed at developing a concentration calibration curve and determining limits of detection (LOD) we observed an anomaly at 1050 cm-1 Since it was found that the band intensity changed as a function of concentration the band must be associated with a sample parameter Two possible parameters photon flux and pH are examined here The first parameter was investigated by irradiating a 1mgmL MPA sample in a SER-active vial with laser powers of 200 mW and above and monitoring spectral changes It was immediately found that the 1050 cm-1 band
2O+ H HF + +OH
OH3H C
OP
OF
CH
CH3
3
3H C
OCP
OOH
CH
CH3
3
3H C
O
CPHO
CH
CH
3
3
C
Sarin IMPA MPA 2-propanol
SPIE -2003-5269 119
appeared and grew as a function of time and that the higher the laser power the faster the growth Figure 1 shows the growth of the 1050 cm-1 band over the course of 30 minutes when using 150 mW of 785 nm excitation while Figure 2A shows that the growth can be fit with a first order exponential equation namely I1050 = 03+05e-013t Furthermore the 760 cm-1 band could be fit with a first order decay equation with an identical exponential rate constant ie I760 = 08-
08e-013t The rates represent classical first order kinetics and their correspondence allows one to conclude that MPA is being transformed one-for-one into a photo-generated product At this time the photoproduct has not been positively identified but phosphonic (phosphorous) acid and phosphonate are likely candidates since the symmetric P(OH)3 stretch occurs at ~1050cm-1 Our studies show that a reduction in laser power to 100 mW at the sample for MPA essentially eliminates this degradation process This laser power or lower was used for further measurements
Methyl phosphonic acid is a diprotic acid that stepwise dissociates into two anions MPA- and MPA= according to the following reactions26
MPA MPA- + H+ pKa1 = 212 Reaction 2
MPA- MPA= + H+ pKa2 = 729 Reaction 3 The relative concentrations of MPA MPA- and MPA= can be determined at any pH by expressing [MPA] and [MPA=] in terms of [MPA-] using Reactions 2 and 3 and summing all three to equal the total starting concentration here 2 mgmL (0021M MW = 9602) viz
[MPA] + [MPA-] + [MPA=] = 0021M Equation 1 substituting from Reactions 1 and 2
([H+][MPA-])K1a + [MPA-] + (K2a[MPA-])[H+] = 0021M Equation 2 rearranging [MPA-] = 0021M(1+[H+]K1a + K2a[H+]) Equation 3 The relative concentrations of MPA MPA- and MPA= as a function of pH are shown in Figure 3 It is worth noting that near neutral pH both MPA- and MPA= will be present To confirm that the SER signal followed this pH dependence a starting solution consisting of 20 mg of MPA in 10 mL HPLC grade water was prepared and brought to pH of 20 using dilute nitric acid From this solution 2 mL were added to a SER-active vial and the SER spectrum recorded At this pH a peak at 760 cm-1 was barely discernable The 2 mL solution was returned to the starting solution and the pH was re-measured to correct for any changes that the silver-doped sol-gel vials might cause In most cases the change was less than 02 pH units and the pH is reported as the before and after average Next the pH of the
Figure 2 A) Exponential growth of 1050 cm-1 band and B) exponential decay of 760 cm-1 band for spectral series in Figure 1
Figure 1 Growth of 1050 cm-1 band as a function of time due to exposure to 150 mW of 785 nm Spectra are 5 sec each collected every 100-sec from 0 to 30-min
B
570 770 970 1170 1370Raman Shift (cm-1)
Arbit
rary
Unit
s
MPA Photodegradation
600 800 1000 1200 1400 Wavenumber (cm-1)
A
0 10 20 30 0 10 20 30 time (min) time (min)
Ram
an In
tens
ity (r
elat
ive)
30
min
0
SPIE -2003-5269 120
starting solution was adjusted to 325 using dilute KOH Again 2 mL were added to a vial and the SER spectrum recorded At this pH a reasonably strong 760 cm-1 band was observed This process was repeated as spectra were recorded at pHs of 70 74 75 79 85 and 100 A total of 1 mL of KOH was added diluting the total concentration by 10 Next the pH of the starting solution was made acidic by adding dilute nitric acid dropwise This time spectra were recorded at pHs of 72 69 64 and 37 Figure 4 shows the SER spectra for representative pHs (spectra were left out to simplify the figure) while Figure 3 shows the 760 cm-1 peak intensities as a function of pH (The band intensities were adjusted to compensate for dilution effects caused by the addition of HNO3 and KOH then normalized to 0021 M for the most intense band observed at pH 37) It is clear from Figure 3 that the 760 cm-1 band follows the MPA- concentration as a function of pH and must be assigned to this anion No bands were observed that corresponded to MPA or MPA= The lack of an MPA SER spectrum may be due to the absence of an attraction between the neutral analyte and the electropositive silver surface The same reasoning suggests that a strong SER spectrum should be observed for MPA= but it is not and a satisfactory explanation has not been found
0000
0005
0010
0015
0020
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14pH
Con
cent
ratio
n [M
]
pK1 = 212 pK2 = 729
MPA- MPA=
MPA
Figure 4 SER spectra of 002M MPA as a function of pH Conditions 100 mW of 785 nm 36 scans (1 min) 8 cm-1 recorded 2 min after sample introduction pH 19 and 20 not apparent on this scale pH 69 and 74 near identical to 70 and 75 and not shown for clarity
00
02
04
06
08
10
12
14
16
18
0 5 10 15 20 25 30Measurement Number
Ram
an In
tens
ity (7
60 c
m-1
)
Figure 3 Concentration dependence of MPA MPA- and MPA= as a function of pH for a 002M sample Intensity of 760 cm-1
band from Figure 6 as a function of pH ( for increasing basic adjustment for increasing acidic adjustment error was measured at ~10 for pH 64)
Figure 5 SER spectra of 002M MPA at pH 64 measured around a vial at three heights (9 points per height) Conditions as in Figure 4 but 10-sec scans
Since these measurements involved the removal and replacement of the SER-active vial in the sample holder to remove and add sample variation in the intensity as a function of vial position was minimized by illuminating the exact same height along the vial wall But this does not account for variability of the SERS response of the sol-gel coating around the vial To analyze this effect a vial containing MPA at pH 64 was rotated at ~ 40o
intervals at the original height and 18rdquo above and below this value Figure 5 shows the intensity of the 760 cm-1
band for the 27 positions It was found that the average value was 137plusmn014 an RSD of 10 overall and 5 for each height An error bar is included in Figure 4 for the pH 64 measurement
SPIE -2003-5269 121
With the above analyses in mind a preliminary investigation of the SER spectral response for MPA (as MPA-) as a function of concentration was performed A single vial was used for these measurements beginning with 1 mgL followed by measurements of 10 100 and 1000 mgL In all cases the pH was ~7 and 3 positions around the vial were measured per concentration Since the 760 cm-1 band was not observed for concentrations of 1 or 10 mgL using 100 mW of 785 nm the laser power at the sample was raised to 200 mW beginning with the 10 mgL concentration Photo-degradation was largely avoided (and not observed) by exposing the sample for only 33 seconds per spectral acquisition Representative spectra for 10 100 and 1000 mgL are shown in Figure 6 while a plot of the 760 cm-1 band intensity as a function of concentration is shown in Figure 7 These values were also used to estimate limits of detection based on the signal-to-noise ratio (SN) of the 760 cm-1 band using 100 mW of 785 nm laser excitation and a 10-min acquisition time As summarized in the Figure 7 inset the lower the measured concentration the lower the predicted LOD Even if as estimated 210 microgL could be measured using the silver-doped sol-gel vials an improvement of a factor of 70 is still required to meet JSAWM goals of measuring 3 microgL in 10-minutes Similar to MPA DPA is a diprotic acid (pKa1 = 216 and pKa2 = 692) and variations in pH will effect the relative concentrations of DPA DPA- and DPA= and possibly the SER spectra and band intensities This could prove significant if an acid or base is used to denature anthrax spores with the goal of extracting and analyzing DPA The MPA pH study described above was mimicked for DPA except that the starting solution consisted of 20 mg of DPA in 20 mL HPLC grade water (60x10-3M MW = 1671) The initial solution had a pH of 245 which was made basic by dilute KOH to pHs of 355 433 487 559 1069 and 1166 SER spectra were recorded at each pH using 100 mW of 785 nm and a 44-sec acquisition time Next one drop of concentrated nitric acid was used to remake the solution acidic at a pH of 200 Again sequential pH measurements were performed at 383 510 735 and 822 The solution pH was made acidic a third time but to pH 219 171 then 135 Throughout this process no more than 20 drops of acid or base were added and therefore the concentration was diluted by no more than 10 Most of the spectral bands showed a minor decrease in intensity as a function of increasing pH values However the bands at 525 and 795 cm-1 showed the most dramatic changes which occurred at acid pH Figure 8 shows the SER spectra of DPA for the spectral region and pH range of interest The identity of the DPA species was determined by plotting the normalized peak intensities with the lowest value set to 0 and the highest to 0006 M as a function of pH and overlaying these values on a plot of the relative concentrations for DPA DPA- and DPA= as previously done for MPA (Figure 9) As can be seen the 525 cm-1 band clearly corresponds to DPA The correspondence of the 795 cm-1 band to this species is less clear as the band retains intensity until pH of 55 This can be attributed to contribution to the overlapping band at 810 cm-1 which does not change as a function of pH The fact that most bands are observed at all pHs suggest that the primary interaction with silver is through the ring nitrogen This is supported by the fact that the most intense band occurs at 1008 cm-1 attributed to a symmetric ring breathing mode and that this interaction has been characterized for pyridine in numerous papers27
0
1
2
3
4
5
6
7
0 200 400 600 800 1000 1200
MPA Concentration (mgL)
760
Ban
d In
tens
ity
Figure 6 SER spectra of MPA in water at A) 1000 B) 100 and C) 10 mgL Conditions pH of 7 silver-doped sol-gel coated vial 200 mW of 785 nm 33-sec 8 cm-1 resolution
Figure 7 Plot of SER intensity of 760 cm-1 band of MPA as a function of concentration using 200 mW of 785 nm Inset table includes average intensity LOD standard deviation and percent deviation for each concentration but for 100 mW and 10 min
conc (mgL) LOD ave stddev dev10 021 002 1021100 072 010 14301000 312 040 1280
A
B C
SPIE -2003-5269 122
The ability to detect dipicolinic acid (DPA) by SERS is demonstrated in Figure 10 Even at 1 mgL the primary bands are visible The intensity of the dominant ring breathing mode observed at 1008 cm-1 for DPA in water (pH = 25 to 55) is plotted as a function of concentration in Figure 11 These values were also used to estimate limits of detection based on the SN of the 1008 cm-1 band using 100 mW of 785 nm laser excitation and a 10-min acquisition time Again the lower the measured concentration the lower the predicted LOD (see Figure 11 inset) and detection of 160 microgL is possible Gastrointestinal anthrax requires significant more spores than inhalation anthrax28 and a limit of detection might be placed at 1 million spores in 1 liter of water or 10 microgL Since each spore contains ~10 CaDPA by weight29 a goal for DPA might be 1 microgL indicating that the present measurements must be improved by nearly two orders of magnitude Due to the increased hazards of handling HCN gas KCN salt was used for these experiments Nevertheless all sample preparations were performed in a chemical hood KCN completely dissolves in water but its conjugate acid HCN is formed and has a Ka of 615x10-1030 viz
HCN CN- + H+ pKa = 921 Reaction 4
Figure 8 SER spectra of 1 mgmL DPA as a function of pH Conditions 100 mW 785 nm 100 scans (44-sec)
Figure 9 Concentration dependence of DPA DPA- and DPA= as a function of pH for a 0006M sample Intensity of 525 () and 795 (diams) cm-1 bands from Fig 8 as a function of pH
pH
135 171 219 383
0
0001
0002
0003
0004
0005
0006
0007
0 2 4 6 8 10 12 14pH
Con
cent
ratio
n [M
]
DPADPA-DPA=795525
DPA DPA=DPA-
pK2 = 692pK1 = 216
795 525
0
05
1
15
2
25
0 200 400 600 800 1000 1200
DPA Concentration (mgL)
1008
Ban
d In
tens
ity
conc (mgL) lod-10min-100mw1 017
10 016100 103
1000 355
Figure 10 SER spectra of DPA in water at A) 1000 B) 100 C) 10 and D) 1 mgL Conditions pH of 25-55 silver-doped sol-gel coated vial 175 mW of 785 nm 1-min 8 cm-1 D) has been multiplied by x10 to make bands visible
A
B C
D
Figure 11 Plot of SER intensity of 1008 cm-1 band of DPA as a function of concentration using 175 mW of 785 nm Inset table includes LOD in mgL for each concentration but for 100 mW and 10 min
SPIE -2003-5269 123
Consequently the cyanide concentration must be determined for each initial KCN concentration Specifically the samples prepared with concentrations of 01 1 10 100 and 1000 mgL of KCN produced CN- concentrations of 63x10-3 033 69 89 and 964 mgL at pHs of 816 90 967 102 and 107 respectively The pH dependence for the HCN and CN- concentrations are shown in Figure 12 Thus as the amount of KCN added to the solution decreases so does the pH of the solution (becomes less basic) and according to Reaction 4 the relative amount of CN- to HCN also decreases For example in the preparation of a 01 mgL solution of KCN the pH is shifted from 7 for pure water to only 816 and only 63 of the starting material becomes CN- or 63x10-3 mgL In comparison for a solution of 1000 mgL the pH is shifted from 7 to 107 and 96 of the starting material becomes CN- This is significant since the cyanide ion is better able to adsorb onto the silver particles and become SERS active SER spectra of 10 100 and 1000 mgL of KCN samples are shown in Figure 13 and a corresponding plot of the 2100 cm-1 band intensity as a function of concentration is shown in Figure 14
000102030405060708091011
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14pH
Con
cent
ratio
n [m
gm
L]
pKa = 921
CN -HCN
The SER spectra of cyanide are dominated by the single CN stretch at 2100 cm-1 which occurs in normal Raman spectra of solutions at 2080 cm-1 However a low frequency mode occurs at 135 cm-1 due to a Ag-CN stretch (not shown) indicative of a strong surface interaction It is also observed that as the concentration decreases the CN stretch shifts to 2140 cm-1 This shift has been attribute to the formation of a tetrahedral Ag(CN)3
2- surface structure31 as well
Figure 14 Concentration dependence of KCN SERS measured under conditions in Fig 11 Concentrations are 1 01 and 001 mgml Intensities are measured for the CN stretch at 2100 cm-1 Inset table includes LOD in mgL for each concentration in Figs 13 and 15 but for 100 mW and 10 min
Figure 13 SER spectra of KCN in water at A) 1000 B) 100 and C) 10 mgL Conditions pHs of 107 102 and 97 silver-doped sol-gel coated vial 100 mW of 785 nm 1-min 8 cm-1 C) has been multiplied by x10 to make band visible
Figure 12 Concentration dependence of HCN and CN- as a function of pH for a 1 mgmL sample Calculated intensity of 2100 () cm-1 band for a 1 mgml sample at pHs of 816 90 967 102 and 107
A
B
C
0
20
40
60
80
100
120
140
0 200 400 600 800 1000 1200CN Concentration (mgL)
2100
Ban
d In
tens
ity
Figure 15 SER spectra of KCN in water at A) 10 B) 1 and C) 01 mgL Conditions pHs of 97 12 and 12 gold-doped sol-gel coated vial 100 mW of 785 nm 1-min 8 cm-1 C) has been multiplied by x10 to make band visible
A
B
C
conc (mgL) lod-10min-100mw Condition01 001 Au-pH 121 007 Au-pH 12
69 003 Au-pH 9769 016 Ag-pH 9789 022 Ag-pH 102946 113 Ag-pH 107
SPIE -2003-5269 124
as to CN adsorbed to two different surface sites32 Alternatively the 2140 cm-1 band could be attributed to HCN since this species dominates at lower concentrations However it is unlikely that this species would be attracted to the electropositive silver surface Further both peaks should be present at pHs between 85 and 105 but this is not observed It has also been suggested that at concentrations near and above monolayer coverage the CN- species is forced to adsorb end-on due to crowding and at lower concentrations the molecule can reorient to lie flat33 This suggests that the 2100 and 2140 cm-1 bands correspond to the end-on and flat orientations respectively As Figures 13 and 14 show the intensity of the CN stretch for the 89 mgL sample is nearly as intense as the 964 mgmL band This suggests that the Raman signal for the flat orientation is more enhanced However more extensive measurements are required to verify this point Since resent research has suggested that cyanide may be more effectively detected on gold measurements of KCN solutions were also performed using gold-doped sol-gel vials Preliminary measurements are shown in Figure 15 for samples prepared from 01 1 and 10 mgL KCN Since the pHs are 816 90 967 the resultant CN- concentrations are 63x10-3 033 and 69 mgL Initially only the highest concentration was observed and the signal intensity was significantly better than the equivalent concentration measured using silver In an effort to shift Reaction 4 to the left transforming HCN to CN- (Le Chatelierrsquos principle) KOH was added to the lower concentration samples producing solutions with pH 12 This effectively forces all of the cyanide in solution to be CN- or 01 and 10 mgL respectively More importantly the CN stretch is now observed in the SER spectra The band appears at 2125 cm-1 as has been previously reported for gold12 As calculated for MPA and DPA LODs can be estimated from this data For the three concentrations of cyanide on silver the LODs are 016 to 11 mgL for 100 mW of 785 nm laser excitation and a 10-min acquisition time For gold pH adjusted the LODs are10 to 70 microgL an improvement of more than 10 times silver Nevertheless either substrate is sufficient to meet the JSAWM goals of measuring 3 mgL in 10-minutes as the requirements form cyanide are much less stringent than the nerve agents
4 CONCLUSIONS Here we examined the surface-enhanced Raman spectral response for methyl phosphonic acid dipicolinic acid and cyanide as a function of pH It was determined that the most prominent peak at 760 cm-1 reaches a maximum intensity between pH 3 and 7 and corresponds to the MPA- species Neither the MPA nor MPA= species appear to generate a SER spectrum and consequently no spectra were observed below pH 2 or above pH 8 In this study we also found that higher laser powers could cause photodegradation of MPA signified by the exponential growth of a band at 1050 cm-1 which is tentatively assigned to phosphorous acid Unlike MPA DPA was observed at all pHs This is attributed to the dominant interaction of the pyridine functional group with silver Minor spectral changes were observed at acid pHs and were assigned to neutral DPA Like MPA SER spectra of cyanide were pH dependent No spectra were observed for the HCN species while CN- was best observed at pHs more basic than 8 Preliminary concentration studies for the three analytes allowed estimating limits of detection for MPA DPA and CN using 100 mW of 785 nm and a 10-min acquisition time of 210 165 and 70 microgL respectively Although the latter value suggests that cyanide can be measured at sufficiently low concentrations to meet JSAWM goal improvements by 100 to 200 times are required for MPA and DPA It is clear from this study that pH of the sample is important and must be taken into account when developing concentration calibration curves as well as fieldable analyzers Future work will include tailoring the sol-gel with specific functional groups to dictate sample pH and thereby optimize sensitivity
ACKNOWLEDGMENTS The authors are grateful for the support of the US Army (DAAD13-02-C-0015 Joint Service Agent Water Monitor program) The authors would also like to thank Dr Steve Christensen of the US Army SBCCOM for helpful discussions and Mr Chetan Shende of Real-Time Analyzers for assistance in development of the gold-doped sol-gels
REFERENCES 1 Erickson B Analytical Chemistry News amp Features June 1 397A (1998)
SPIE -2003-5269 125
2 Johnston RL Hoefler CM Fargo JC and Moberley B AT-ONSITE 5-8 (1994) 3 Black RM RJ Clarke RW Read and MT Reid J Chromatography 662 301-321 (1994) 4 USA Army Center for Health Promotion and Preventive Medicine (USACHPPM) Technical Guideline 230A 5 Jensen JL NC Wong W R Loerop SPIE 4775-03 (2001) 6 Hoenig SL Handbook of Chemical Warfare and Terrorism Greenwood Press Wesport CT (2002) 7 Munro NB SS Talmage GD Griffin LC Waters AP Watson JF King and V Hauschild Env Health
Persp 107 933-974 (1999) 8 Holstege CP Kirk M Sidell FR Crit Care Clin13 923-42 (1997) 9 Farquharson S P Maksymiuk K Ong and S Christesen SPIE 4577 166-173 (2001) 10 Lee Y and S Farquharson SPIE 4378 21-26 (2001) 11 Spencer KM J Sylvia S Clauson and J Janni SPIE 4577 158-165 (2001) 12 Tessier P S Christesen K Ong E Clemente A Lenhoff E Kaler and O Velev Applied Spectroscopy 56
1524-1530 (2002) 13 Farquharson S WW Smith S Elliott and JF Sperry SPIE 3533 207-214 (1998) 14 Farquharson S WW Smith S Elliott and JF Sperry SPIE 3855110-116 (1999) 15 Farquharson S WW Smith YH Lee S Elliott and JF Sperry SPIE 4575 62-72 (2002) 16 Guzelian AA J Sylvia J Janni S Clauson and KM Spencer SPIE 4577 182-192 (2001) 17 Shende C F Inscore A Gift P Maksymiuk and S Farquharson in press 18 Weaver MJ S Farquharson and MA Tadayyoni J Chem Phys 82 4867-4874 (1985) 19 Alak AM and T Vo-Dinh Analytical Chemistry 59 2149-2153 (1987) 20 Lee Y and S Farquharson SPIE 4206 140-146 (2000) 21 Farquharson S and Y Lee SPIE 4200-16 (2000) 22 Lee Y S Farquharson and P M Rainey SPIE 3857 76-84 (1999) 23 Lee Y S Farquharson H Kwong and M Shahriari SPIE 3537 252-260 (1998) 24 Wang J M Pumera G Collins and A Mulchandani Analytical Chemistry 74 6121-6125 (2002) 25 Dou X YM Jung Z-Q Cao and Y Ozaki Applied Spectroscopy 53 1440-1447 (1999) 26 Data supplied by S Christesen and K Ewing 27 Kerker M and B Thompson Eds SPIE MS 10 (1990) 28 Inglesby TV DA Henderson JG Bartlett JAMA 287 2236 (2002) 29 FW Janssen AJ Lund and LE Anderson Science 127 26 (1958) 30 Lide DR Ed Handbook of Chemistry and Physics CRC Press 77th Ed (1996-7) 31 Billmann J G Kovacs and A Otto Surf Sci 92 153 (1980) 32 Murray CA and S Bodoff Phys Rev B 32 671 (1985) 33 Kellogg D and J Pemberton J Phys Chem 91 1120 (1987)
SPIE-2003-5269 16
Chemical agent detection by surface-enhanced Raman spectroscopy
Stuart Farquharson Alan Gift Paul Maksymiuk Frank Inscore and Wayne Smith
Real-Time Analyzers 87 Church Street East Hartford CT 06108
Kevin Morrisey and Steven D Christesen US Army SBCCOM Aberdeen Proving Ground MD 21010
ABSTRACT
In the past decade the Unites States and its allies have been challenged by a different kind of warfare exemplified by the terrorist attacks of September 11 2001 Although suicide bombings are the most often used form of terror military personnel must consider a wide range of attack scenarios Among these is the intentional poisoning of water supplies to obstruct military operations in Afghanistan and Iraq To counter such attacks the military is developing portable analyzers that can identify and quantify potential chemical agents in water supplies at microgram per liter concentrations within 10 minutes To aid this effort we have been investigating the value of a surface-enhanced Raman spectroscopy based portable analyzer In particular we have been developing silver-doped sol-gels to generate SER spectra of chemical agents and their hydrolysis products Here we present SER spectra of several chemical agents measured in a generic tap water Repeat measurements were performed to establish statistical error associated with SERS obtained using the sol-gel coated vials Keywords Chemical agents chemical agent detection SERS Raman spectroscopy
1 INTRODUCTION
In the autumn of 2001 terrorism within US borders became a sobering reality While extensive efforts are being implemented to secure the homeland US and Coalition forces in Afghanistan and Iraq are constantly faced with terrorist attacks In addition to car-bombings the military has identified several non-traditional attack scenarios including poisoning of water supplies by chemical warfare agents (CWAs) To counter this threat the Department of Defense has been investigating numerous approaches to detect such attacks under the auspices of the Joint Service Agent Water Monitor (JSAWM) program1 The JSAWM must be capable of identifying the chemical agents with no false-positives and quantifying the agents at microgram per liter concentrations within 10 minutes2 This includes the analysis of drinking water supplies distribution and storage systems Currently colorimetric paper is used to detect agents on-site while gas chromatography combined with mass spectrometry (GCMS) is used in mobile support laboratories However both methods have severe drawbacks The paper changes color in response to contact with many chemicals besides CWAs causing a high incidence of false positives3 while GCMS although very chemically specific requires up to an hour to perform and regular re-calibration456 Vibrational spectroscopy has also been investigated as a potential method of rapidly detecting CWAs7-
11 as early as 1985 in the case of infrared spectroscopy8 and 1988 in the case of Raman spectroscopy11 Again however these techniques also have limitations when it comes to measuring trace poisons in water Infrared spectra would be dominated by the very strong absorption of water which would obscure absorptions by most other chemicals present Whereas Raman spectroscopy is simply not a very sensitive technique and detection limits are typically grams per liter Surface-enhanced Raman spectroscopy (SERS) offers several advantages over conventional vibrational methods and that may provide the necessary sensitivity required for detecting trace quantities of chemical agents in water SERS employs the interaction of surface plasmon modes of metal particles with target analytes to increase scattering efficiency by as much as 1 million times12 In 1987 the potential of this approach to measure CWAs was demonstrated by measuring a series of pesticides13 Several of these organophosphonates have chemical structures similar to CWAs in particular P=O functional groups In the past few years we and others have further explored the ability of SERS to detect CWAs14-17 and even bioagents 18-21 We have been employing silver-doped sol-gels to promote the SER effect
SPIE-2003-5269 17
in these studies The porous silica network of the sol-gel matrix offers a unique environment for immobilizing and stabilizing SER-active metal particles22-25 The silver-doped sol-gels have been coated on the internal walls of standard glass vials to allow rapid SER analysis of small sample quantities without preparation We have measured over 100 chemicals with enhancements of 104 to 106 demonstrated reversible measurements in a flowing system reproducible measurements from vial-to-vial and measurements in multiple solvents including water21-25 Previously we used these vials to perform preliminary measurements of cyanide (CN) and methylphosphonic acid (MPA) Most of the nerve agents form MPA during hydrolysis while Tabun forms CN a chemical agent in its own right In another paper including in these proceedings we examined the limits of detection (LOD) for MPA by measuring a series of concentrations down to 50 mgL and estimated a limit of detection of 100 microgL26 These measurements provide encouragement in that SERS may satisfy the needs of the JSAWM To further establish the viability of SERS in particular silver-doped sol-gels here we present analysis of cyanide mustard and VX in tap water The measurements performed at the US Armyrsquos Edgewood Chemical Biological Center Aberdeen MD also included numerous repeat measurements to establish reproducibility
2 EXPERIMENTAL 2a General All chemicals including potassium cyanide 2-chloroethylethyl sulfide and those used to prepare the silver-doped and gold-doped sol-gel coated vials were obtained and used as received from Sigma-Aldrich All samples were prepared in a chemical hood using HPLC grade water unless otherwise noted (Fischer Scientific Fair Lawn NJ) for SERS measurements Once prepared the samples were transferred into the silver-doped sol-gel vials (Real-Time Analyzers Inc East Hartford CT) The vials were coated in a manner similar to that previously reported by adding ammonium hydroxide to a solution of silver nitrate tetramethyl orthosilicate (TMOS) and methanol22 The two precursor solutions were prepared mixed and transferred to 2-mL glass vials dried and heated After sol-gel formation the incorporated metal ions were reduced with dilute sodium borohydride (1mgmL) which is followed by a water wash to remove residual reducing agent After the resultant analyte solution was introduced the SER-active vial was fixed horizontally to an XY positioning stage (Conix Research Springfield OR) just inside the focal point of an f07 aspheric lens The lens focused the beam into the sample and collected the scattered radiation back along the same axis A dichroic filter (Omega Optical Brattleborough VT) was used to reflect the excitation laser to the lens and pass the Raman scattered radiation collected by the lens An f2 achromat was used to collimate the laser beam exiting a 200 microm core diameter source fiber optic while a second f2 achromat was used to focus the scattered radiation into a 365 microm fiber optic (Spectran Avon CT) A short pass filter was placed in the excitation beam path to block the silicon Raman scattering generated in the source fiber from reflecting off sampling optics and reaching the detector A long pass filter was placed in the collection beam path to block the sample Rayleigh scattering from reaching the detector A 785 nm diode laser (Process Instruments Inc model 785-600 Salt Lake City UT) was used to deliver 100 to 150 mW of power to the sample A Fourier transform Raman spectrometer (Real-Time Analyzers model IRA-785 East Hartford CT) and a silicon photo-avalanche detector (Perkin Elmer model C30902S Stamford CT) were used to acquire the SER spectra
2b Edgewood Chemical Biological Center The surface-enhanced Raman spectral measurements at the US Armyrsquos Chemical Biological center presented here were all performed on September 12 2003 To expedite measurements a plate was machined to hold up to 12 SER-active sample vials (Figure 1) The plate fit a standard XY plate reader that could be programmed Pure KCN bis-(2-chloroethyl)sulfide (distilled mustard gas HD) and ethyl S-2-diisopropylamino ethyl methylphosphonothioate (VX) were obtained on-site and used to prepare 1 gL tap water solutions in a chemical hood with appropriate safety equipment Simulated tap water was prepared by adding 10 chemicals most often found in tap water at appropriate concentrations to distilled water (Table 1) SER measurements were also performed in a chemical hood For added safety the FT-Raman instrument was placed outside the laboratory and 30 foot fiber optic and electrical cables were used to allow remote SERS measurements and plate manipulation For each experiment 1gL samples were prepared and added to 9 individual vials which were then loaded on the plate In some cases a tenth vial was included as a blank
Table 1 Generic Tap Water Composition
Compound mgL NaHCO3 100 CaSO4 27 MgSO4bull7H2O 67 NaNO3 1 Fulvic Acid 1 K2HPO4 07 KH2PO4 03 (NH4)2HSO4 001 NaCl 001 FeSO4 0001 pH 76-78
SPIE-2003-5269 18
A software program was written that allowed selecting the sequence that the vials were measured the number of positions along the length of the vials to measure (1 to 5) and the number of scans to co-add During sample analysis the program displayed the vial being analyzed the point being analyzed and the spectrum as it was being acquired Once all the data was collected a second software program was written to rapidly analyze the data The spectra collected for all the vials on a plate could be loaded at one time and then the spectra for each point could be displayed simultaneously or separately The user could then select the Raman peak to analyze in terms of peak height or area This was accomplished by selecting points on either side of the peak to define a baseline of zero The peak height or area could then be computed for all of the spectra loaded and then exported to a spreadsheet for statistical analysis
Figure 1 A) Vial Holder 6 slots to hold 2 vials each end-to-end B) Measurement Configuration Program user selects vials to measure sequence number of points per vial (1 to 5) and number of scans per point C) Spectral Acquisition Program shows spectrum being collected which vial and position D) Spectra Analysis Program user selects spectra to analyze by plate vial and point (s) as well as two wavenumbers defining the peak and the baseline to subtract The image is of 5 repeat measurements of 10 mgL KCN in generic tap water 16 sec each 100 mW of 785 nm
3 RESULTS AND DISCUSSION Raman and surface-enhanced Raman spectra were obtained for potassium cyanide bis-(2-chloroethyl)sulfide and ethyl S-2-diisopropylamino ethyl methylphosphonothioate representing three classes of chemical agents cyanides mustards and nerve agents respectively Spectra were also obtained for 2-chloroethyl ethyl sulfide (CEES) a structural analogue to HD which was included in the study to aid in assigning spectral bands KCN salt was used for cyanide experiments to avoid the increased hazards of handling HCN gas KCN completely dissolves in water forming its conjugate acid HCN according to its Ka of 615x10-1027 and at a concentration of 1 mgmL results in a pH 107 solution This is significant in that only CN- appears to interact sufficiently with silver to produce a SER spectrum and no spectral signal is observed below pH 726 Figure 2 shows the SER and normal Raman spectra for KCN The SERS spectrum of 1mgml KCN in water shows a single intense somewhat broad feature at
A B
C D
SPIE-2003-5269 19
2100 cm-1 assigned to the single CequivN stretch The band is much sharper in the normal Raman spectra of the solid KCN salt at 2074 cm-1 This band does broaden and shift to 2080 cm-1 in solution (not shown) However the observed SERS frequency is attributed to interaction with silver and low frequency mode at 135 cm-1 attributed to a Ag-CN stretch (not shown) supports this conclusion
cm-1 band to a C-S stretch but the authors concede that it is in fact more likely a C-Cl stretch28 It appears that the most intense Raman bands at 648 692 and 747 cm-1 shift to 620 660 and 720 cm-1 in the SER spectra and are tentatively assigned as above The width of these bands suggests that they overlap underlying spectral features Additional bands in the Raman spectra occur at 972 1034 1049 1263 1286 1423 1442 2865 2935 and 2960 cm-1 Corresponding bands occur in the SER spectra at 964 1015 1054 1286 1410 1447 2865 and 2935 cm-1 Most of these bands are associated with alkane modes specifically the bands at approximately 1040 cm-1 to a C-C stretch 1290 cm-1 to a CH2 in-phase twist 1440 cm-1 to a CH2 wag 2865 cm-1 to a symmetric CH2 stretch and 2965 cm-1 to an asymmetric CH2 stretch The Raman and SER spectra of sulfur mustard were measured at the Edgewood center (Figure 4) Both spectra are largely similar to CEES The C-Cl and C-S bands in the Raman spectrum of HD now occur at 640 655 700 739 and 760 cm-1 and are more resolved possibly due to the increased molecular symmetry Theoretical calculations indicate that the first three bands are due to C-Cl stretching modes and the latter two to C-S stretching modes28 Only the C-Cl bands maintain significant intensity in the SER spectra occurring at 624 and 643 cm-1 which is attributed to the
Prior to measurements of HD CEES was examined by Raman and SER spectroscopy (Figure 3) CEES also known as half-mustard is essentially identical to HD except one of the chlorine end atoms is replaced by a hydrogen atom Again although not as toxic as HD CEES is a blister agent and dilute aqueous samples were prepared using appropriate safety equipment Both the Raman and SER spectra of CEES are similar and dominated by bands between 600 and 800 cm-1 These are associated with C-Cl and C-S stretching modes which are tentatively assigned to 648 and 747 cm-1 in the Raman spectra respectively The shoulder at 630 cm-1 the overlapped band at 660 cm-1 and the strong band at 692 cm-1 could also be due to these modes or their asymmetric counterparts It is worth noting that theoretical calculations assign the 692
Figure 2 A) SER and B) NR spectra of KCN Conditions A) 1 mgml in tap water 100 mW of 785 nm at sample 1-min acquisition time B) solid 300 mW of 785 nm 5-min All spectra are 8 cm-1 resolution
A
B
Figure 3 A) SER and B) NR spectra of CEES Conditions A) 1 vv (10 mgml) in MeOH 100 mW of 785 nm 1-min acquisition time B) neat 300 mW of 785 nm 5-min
A
B
Cl-CH2-CH2-S-CH2-CH3
A
B
Figure 4 A) SER and B) NR of HD Conditions A) 1mgml in tap water B) pure both 100 mW of 785 nm 1-min
Cl-CH2-CH2-S-CH2-CH2-Cl
SPIE-2003-5269 20
expected strong interaction between chlorine and silver and adds support to the assignment of this band to a C-Cl stretch Weaker overlapping bands occur at 670 692 and 724 cm-1 the latter possibly due to C-S stretching modes Again the alkane modes are apparent in the normal Raman spectra of HD but only a broad feature at 1300 to 1450 cm-1 suggests CH2 contributions in the SER spectrum Although the observed bands in the VX spectrum have not been assigned (Figure 4) a computer generated Raman spectrum29 predicts many of the same features with surprising accuracy and are used here Two intense bands at 460 and 530 cm-1 closely match predicted bands at 463 and 546 cm-1 assigned to a CH3-P=O bend and a PO2CS wag Three highly overlapped bands occur at 694 745 and 771 cm-1 matching predicted bands at 713 730 and 760 cm-1 The first
Table 2 Measured SER peak heights for the CN stretch at 2100 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
Figure 5 A) SERS and B) NR spectra of VX Conditions A) 1 vv (10 mgml) in MeOH B) pure sample both 100 mW of 785 nm 1-min acquisition time
A
B
two have been assigned to a C-S stretch and CH2bend respectively while the latter has been attributed to either a P-C stretch or an O-C-C stretch Although the 745 cm-1 band may alternatively be assigned to a C-S stretch based on the previous measurements of CEES and HD The relatively intense bands at 890 1106 1218 1445 and 1465 cm-1 also match predicted bands at 880 1108 1216 1440 and 1464 cm-1 that are assigned to a C-C stretch CH3 rock N-C3 stretch various C-H3 bends and C-H bends respectively Both the computer generated and the measured spectra contain numerous other less intense bands One is worth mentioning A unique band appears at 370 cm-1 that is predicted at 368 cm-1 and corresponds to an O-P=O bend The surface-enhanced Raman spectrum of VX is also rich with spectral features It has the unique low frequency band at 370 cm-1 as well as a second band at 380 cm-1 that is assigned tothe S-P-O bend predicted in the normal Raman spectrum at 388 cm-1 Based on the measured and predicted normal Raman spectra the following SERS assignments are given 460 cm-1 to the CH3-P=O bend 544 cm-1 to the PO2CS wag 738 cm-1 to a C-S stretch (based on arguments above) 890 cm-1
to a C-C stretch 1101 cm-1 to a CH3 rock and 1456
cm-1 to a C-H bend The ability of SERS to measure chemical agents in water containing real-world chemical interferents was tested by using the generic tap water described in Table 1 The ability to reproduce measurements was accomplished by preparing three separate water stock solutions which were used to prepare three sample solutions each of 1 mgmL KCN in the generic tap water The samples defined as vials 1-9 were then measured at 4 points per vial in a semi-random fashion such that errors associated with stock solution preparation and errors associated with instrument drift could be identified No trends were apparent that signified such systematic errors Each spectrum collected consisted of 20 averaged scans taking 16 seconds at 8 cm-1 resolution The laser power at the sample was measured periodically during the day and it ranged from 102 to 105 mW spacccc
spot Vial 1 Vial 2 Vial 41 13754 13004 128192 13519 12692 12909
1 3 13541 12721 126394 13462 12648 12651
ave 13569 12766 12755Vial 6 Vial 3 Vial 5
1 11586 14068 10757 2 11236 14402 115122 3 11376 14546 11559
4 10894 11763 11214ave 11273 13695 11261
Vial 7 Vial8 Vial 91 11198 15804 11023
2 11243 15716 88753 3 11562 15216 11218
4 11655 15084 11077ave 11415 15455 10548
CN AVG STDEV ERR1 mgmL pk ht 12526 1560 1245
stock solution
SPIE-2003-5269 21
The CN measurements consisting of 36 data points produced an average height of 125 for the 2100 cm-1 peak with a standard deviation of 156 or 125 (Table 2) The HD and VX measurements were performed precisely the same way (Tables 3 and 4) For HD the 624 cm-1 peak was used for analysis and it had an average height of 53 with a standard deviation of 068 or 129 while for VX the 544 cm-1 peak was used for analysis and it had an average height of 1051 with a standard deviation of 308 or 293 The greater error in the VX measurements can be somewhat attributed to Vial 4 which produced lower SER signal intensities But removing this vial from the data set changes the standard deviation to 233 only a modest improvement
spot Vial 1 Vial 2 Vial 41 525 609 4682 45 675 484
1 3 527 644 5094 559 807 669
ave 51525 68375 5325Vial 6 Vial 3 Vial 5
1 521 574 575 2 536 53 4492 3 509 508 451
4 565 594 379ave 53275 5515 4635
Vial 7 Vial 8 Vial 91 457 589 413
2 583 497 4273 3 544 505 4
4 528 5 509ave 528 52275 43725
HD AVG STDEV ERR1 mgmL pk ht 530 068 1291
stock solution
spot Vial 1 Vial 2 Vial 41 1464 1034 4672 1485 99 61
1 3 1491 989 5684 1041 777 553
ave 137025 9475 5495Vial 6 Vial 3 Vial 5
1 1058 942 1293 2 697 121 965
2 3 727 1206 11124 689 1254 746
ave 79275 1153 1029Vial 7 Vial 8 Vial 9
1 1127 783 163 2 1358 812 1614
3 3 1371 908 15124 1097 875 1397
ave 123825 8445 153825
VX AVG STDEV ERR1 mgmL pk ht 1051 308 2925
stock solution
4 CONCLUSIONS In this paper we examined the ability of surface-enhanced Raman spectroscopy to reproducibly measure CN HD and VX in tap water without chemical interference Both normal and surface-enhanced Raman spectra were examined to select unique bands suitable to identify and quantify these chemical agents For SER measurements the 2100 cm-1 C-N stretch was used for CN the 624 cm-1 C-Cl stretch was used for HD and the 544 cm-1 PO2CS wag was used for VX It was determined that 1 mgmL samples of each of these chemicals measured 36 times in glass vials coated with a silver-doped sol-gel reproduced measurements with standard deviations of 125 129 and 293 It was further found that the 10 chemicals added to simulate generic tap water did not interfere with or alter the SER spectra It should be noted that the concentrations used in this study were considerably greater than those required by the JSAWM program Current work involves improving SER sensitivity and designing sampling systems with better reproducibility This includes the development of fractal silver and gold structures within the sol-gel matrix and the development of chemically selective sol-gels
ACKNOWLEDGMENTS The authors are grateful for the support of the US Army (DAAD13-02-C-0015 Joint Service Agent Water Monitor program) The authors would also like to thank Janet Jensen Ronald Crosier and Kristina Gonser for helpful discussions
Table 4 Measured SER peak heights for the VX band at 544 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
Table 3 Measured SER peak heights for the HD band at 624 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
SPIE-2003-5269 22
REFERENCES 1 Jensen JL NC Wong W R Loerop SPIE 4775-03 (2001) 2 JSAWM Requirements at wwwsbccomapgeaarmymilRDAecbcrtPRODSERJSAWMjsawmhtml 3 Erickson B ldquoThe Chemical Weapons Convention Redefines Analytical Challengerdquo Anal Chem 70 397A (1998) 4 Johnston RL Hoefler CM Fargo JC and Moberley B ldquoThe Defense Nuclear Agencyrsquos
ChemicalBiochemical Weapons Agreements Verification Technology Research Development Test and Evaluation Program and its Requirements for On-Site Analysisrdquo AT-ONSITE 5-8 (1994)
5 Black RM Clarke RJ Read RW and Reid MT ldquoApplication of GC-MS and GC-tandem MS to the analysis of chemical warfare samples found to contain residues of the nerve agent sarin sulphur mustard and their degradation productsrdquo J Chrom 662 301-321 (1994)
6 USA Army Center for Health Promotion and Preventive Medicine (USACHPPM) Technical Guideline 230A 7 Hoffland LD Piffath RJ Bouck JB rdquoSpectral signatures of chemical agents and simulantsrdquo Optical
Engineering 24 982-984 (1985) 8 Braue EHJ Pannella MGrdquoCIRCLE CELL FT-IR Analysis of Chemical Warfare Agents in Aqueous Solutionsrdquo
App Spectrosc 44 1513-1520 (1990) 9 Tseng C-H Mann CK and Vickers TJ ldquoDetermination of Organics on Metal Surfaces by Raman
Spectroscopyrdquo App Spectrosc 47 1767-1771 (1993) 10 Staff Reporter ldquoUS Army Tests Lidar to Detect Biological Toxinsrdquo Photonic Spectra pg 50 December 1998 11 Christesen SD Raman cross sections of chemical agents and simulants App Spectrosc 42 318-321 (1988) 12 Weaver MJ Farquharson S Tadayyoni MA ldquoSurface-enhancement factors for Raman scattering at silver electrodesrdquo J Chem Phys 82 4867-4874 (1985) 13 Alak AM and Vo-Dinh T ldquoSERS of Organophosphorous Chemical Agentsrdquo Anal Chem 59 2149-2153 (1987) 14 Lee Y Farquharson S ldquoRapid chemical agent identification by SERSrdquo SPIE 4378 21-26 (2001) 15 Farquharson S Maksymiuk P Ong K Christesen S ldquoChemical agent identification by surface-enhanced Raman spectroscopyrdquo SPIE 4577 166-173 (2001) 16 Spencer KM Sylvia J Clauson S and Janni J ldquoSurface Enhanced Raman as a Water Monitor for Warfare
Agents in Waterrdquo SPIE 4577 158-165 (2001) 17 Tessier P Christesen S Ong K Clemente E Lenhoff A Kaler E Velev O ldquoOn-line spectroscopic characterization of sodium cyanide with nanostructured gold surface-enhanced Raman spectroscopy substratesrdquo
App Spectrosc 56 1524-1530 (2002) 18 Farquharson S W W Smith S Elliott and J F Sperry Rapid biological agent identification by surface- enhanced Raman spectroscopy SPIE 3855110-116 (1999) 19 Farquharson S WW Smith YH Lee S Elliott and J F Sperry Detection of biological signatures A
comparison of electrolytic and metal-doped sol-gel surface-enhanced Raman media SPIE 4575 62-72 (2002) 20 Guzelian AA Sylvia J Janni J Clauson S and Spencer KM ldquoSERS of whole cell bacteria and trace levels of biological moleculesrdquo SPIE 4577 182-192 (2001) 21 Shende C Inscore F Gift A Maksymiuk P Farquharson S ldquoRapid dipicolinic acid extraction from Bacillus spores detected by surface-enhanced Raman spectroscopyrdquo App Spectrosc 58 accepted 22 Lee Y and Farquharson S ldquoSERS Sample Vials Based on Sol-Gel Process for Trace Pesticide Analysisrdquo SPIE
4206 140-146 (2000) 23 Farquharson S and Lee Y ldquoTrace Drug Analysis by SERSrdquo SPIE 4200-16 (2000) 24 Lee Y Farquharson S and Rainey P M Surface-Enhanced Raman Sensor for Trace Chemical Detection in
Water SPIE 3857 76-84 (1999) 25 Lee Y Farquharson S Kwong H and Shahriari M ldquoSol-Gel Chemical Sensor for Surface-Enhanced Raman
Spectroscopyrdquo SPIE 3537 252-260 (1998) 26 Farquharson S A Gift P Maksymiuk F Inscore and W Smith ldquopH dependence of methyl phosphonic acid
dipicolinic acid and cyanide by surface-enhanced Raman spectroscopyrdquo SPIE 5269-19 (2003) 27 Lide DR Ed Handbook of Chemistry and Physics CRC Press 77th Ed (1996-7) 28 Sosa C RJ Bartlett K KuBulat and WB Person ldquoA theoretical study of harmonic vibrational frequencies and
infrared intensities of XCH2CH2SCH2CH2X and XCH2CH2SH (= H Cl)rdquo J Phys Chem 93 577-588 (1993) 29 Hameka HF and JO Jenson ldquoComputer-generated predictions of the structure and the IR and Raman spectra of
VXrdquo ERDEC-TR-065 May 1993
- Chemagents Appendicespdf
-
- SPIE2001-4575-Bioagent-SERSapdf
-
- Detection of bioagent signatures A comparison of electrolytic and metal-doped sol-gel surface-enhanced Raman media
- Stuart Farquharson Wayne Smith and Yuan Lee
-
- Real-Time Analyzers 87 Church Street East Hartford CT 06108
-
- ABSTRACT
- 4 CONCLUSIONS
- 5 ACKNOWLEDGEMENTS
- 6 REFERENCES
-
SPIE 2001-4575
66
contributions from signals or baseline offsets The measurement error is given as SplusmnRMS and for adenine this equals 234 The number of molecules contributing to the SG-SERS are those on the silver particles that are embedded in the sol-gel The total silver surface area can be determined from the average particle size (40 nm diameter) concentration (073 by weight based on molar conc and measured sol-gel density) and the scattering volume (a cylinder defined by the laser area 28x10-7m2 and sol-gel thickness10-4m) The 61x109 silver particles in this volume have a collective area of 31x10-5m2 However it may reasonably be assumed that at least half of the silver surface is in contact with the sol-gel matrix and unavailable for analyte interaction Then approximately 10x1014 molecules or 22x10-8g of adenine contribute to the SG-SER spectrum The slightly lower SN of 207 suggests a mass detection limit of 32x10-10g Determination of the enhancement factors for the two SER media requires estimating the number of adenine molecules contributing to the normal Raman spectrum Here a cylindrical scattering volume is assumed again based on the laser area (28x10-7m2) and the penetration depth (1x10-3 m)41 The density of the sample was measured at 064 gcm3 indicating that 18x10-4g (13x10-6 moles) of adenine produced the normal Raman signal The enhancement factor EF is defined by the following equation
EF = (ISERSINR)bull(MNRMSERS) bull(PNRPSERS) bull(TNRTSERS)12
where I is the spectral band intensity (here 735 cm-1) M is the sample mass P is the incident laser power and T is the measurement time (or number of scans) For the E-SERS measurement the enhancement factor is 22x105 (01780184) bull (18x10-487x10-10)) while the SG-SERS enhancement factor is 10x105 (0160184) bull(18x10-422x10-8) bull(75075) bull(315)12) The lower enhancement for the SG-SERS may be real or the available surface of the silver embedded in the sol-gel may have been overestimated In addition to enhancing the Raman scattering efficiency to an extent similar to E-SERS the SG-SER medium also yields an identical shift of the adenine ring-breathing mode from 725 cm-1 in the normal Raman to 735 cm-1 Furthermore in the course of optimizing the E-SERS sample conditions it was found that pH influenced the adenine interaction with the silver surface (Figure 2) In particular the relative band intensities of the pyrimidine ring skeletal vibrations at 1270 and 1375 cm-1 and the imidazol ring skeletal vibration at1335 cm-1 change At pH 4 adenine is protonated presumably the imidazol ring since the band at 1335 cm-1 increases in intensity while the pyrimidine bands are virtually absent Conversely at pH 10 the imidazol band decreases in intensity while the pyrimidine bands appear It is worth noting that the ring-breathing mode at 735 cm-1 changes little between pH 4 and 10 suggesting that the skeletal changes are more a function of molecule-plasmon interactions than reorientation of the molecule on the surface Measurements of the identical pH series of adenine samples by SG-SERS yielded virtually identical spectral changes This suggests that the sol-gel does not influence the measurement This is critical to reproducing measurements and performing quantitative analysis Next the remaining NA bases were measured by both E-SERS and SG-SERS and compared Previously we examined the optimum pH and electrode potentials for E-SERS measurements to determine if a common pH could be used that yielded good sensitivity for all the bases and if variations in potential could be used to provide an added degree of selectivity between the bases Primarily it was found that high quality spectra were obtained between pH 7 and 95 and that cytosine and uracil were best enhanced at potentials positive of the potential-of-zero charge (pzc ca 065VSCE for Ag) guanine and thymine near the pzc and adenine negative of the pzc In all cases the ring-breathing modes were the most intense and in general could be used to identify the NA bases (Figure 3 Table 1) Specifically adenine has an intense band at 735 cm-1 cytosine at 797 cm-1 guanine at 653 cm-1 thymine at 784 cm-1 and uracil at 800 cm-1 The adenine cytosine guanine and thymine bands are sufficiently separated that their contributions to DNA should be determinable Although adenine and guanine contributions to RNA should also be determinable cytosine and uracil are highly overlapped and unfortunately share the same potential dependence Alternate unique bands at 1183 cm-1 for cytosine and 1275 cm-1 for uracil might be suitable for calculating contributions The SG-SER spectra of the remaining NA bases faithfully reproduced the E-SER spectra In particular the primary identifying bands occur at virtually the same wavenumbers (see Table 1) However the spectra for both cytosine and thymine contain an intense band at ca 1040 cm-1 Initially this was attributed to the pH buffer but samples prepared without either the buffer or the 01M KCl electrolyte yielded identical spectra containing this band In fact the E-SER and SG-SER spectra of thymine are virtually identical except for this band Also the SG-SERS of guanine contains an intense band at 1551 cm-1 that is not observed in the E-SER spectrum This band may be due to a moderately intense band at 1553cm-1 in the normal Raman spectrum that is SG-SER active It was also found that the SG-SERS of cytosine was considerably better than the E-SERS while uracil showed the opposite relationship It is also worth noting that all of the SG-SERS were obtained with 110th the laser power Most importantly the primary ring-breathing modes in the SG-SER spectra are sufficiently intense and unique to be used in determining contributions to DNA and RNA as outlined above
SPIE 2001-4575
67
Table 1 Comparison of E-SER and SG-SER Spectral Band Positions for the NA Bases and Adenosine Monophosphate
Adenine Cytosine Guanine Thymine Uracil AMP E-SERS SG-SERS E-SERS SG-SERS E-SERS SG-SERS E-SERS SG-SERS E-SERS SG-SERS E-SERS SG-SERS1647 1638 1634 1656 1655 1630 1587 1585 1510 1580 1551 1539 1456 1456 1465 1460 1462 1480 1453 1459 1394 1398 1425 1431 1435 1399 1404 1392 1374 1375 1373 1383 1370 1335 1332 1311 1307 1333 1331 1353 1348 1331 1329 1265 1273 1280 1292 1278 1276 1275 1279 1271 1183 1195 1222 1232 1221 1219 1204 1205 1180 1144 1097 1033 1029 1038 1040 1035 1051 1037 1041 1035 963 963 957 1001 1000 961 944 884 819 817 859 866 735 737 797 799 784 782 800 800 72738 742 630 630 653 664 667 684 603 602 590 611 466 561
Bands unique to SG-SERS The next chemical to be analyzed by both E-SERS and SG-SERS was adenosine monophosphate (AMP) The E-SER spectrum yields bands due to the adenine chemical functionality at 727 961 1233 1279 1331 1381 and 1486 cm-1 In addition phosphate bands are observed at 860 1097 1453 1587 and 1705 cm-1 (Figure 5) Other researchers have noted that the ribose component does not appear to contribute to the spectrum19 The AMP spectrum also changes as a function of potential As the electrode is swept more positive (here from -09 to -03VSCE) the phosphate bands at 860 1097 1453 and 1587 cm-1 increase in intensity compared to the adenine bands while a band at 1705 cm-1 appears The adenine bands at 1233 1381 and 1486 cm-1 virtually disappear These potential dependent spectral changes are consistent with earlier studies that show that phosphate is attracted to silver at potentials positive of the pzc but repelled at potentials negative of the pzc19
Wavenumbers (∆cm-1) Figure 3 E-SERS of A) 21x10-3M cytosine at -03VSCE 1000 scans B) ~10x10-5M guanine at -06VSCE 500 scans C) 23x10-3M thymine at -06VSCE 500 scans and D) 12x10-3M uracil at -093VSCE 500 scans All spectra at pH 92 750 mW 1064 nm at 8 cm-1
Wavenumbers (∆cm-1) Figure 4 SG-SERS of A) 21x10-3M cytosine 200 scans B) ~10x10-5M guanine 200 scans C) 23x10-3M thymine 200 scans and D) 12x10-3M uracil 500 scans All spectra at pH 92 75 mW 1064 nm at 8 cm-1
A
B
C
A
B
C
D D
SPIE 2001-4575
68
The SG-SER spectrum of AMP is considerably different The adenine bands virtually disappear except for the two primary bands which shift to742 and 1329 cm-1 While the phosphate band at 1459 cm-1 has gained considerable intensity In addition two new intense bands appear at 684 and 1539 cm-1 as well as a moderately intense band at 1180 cm-1 The SG-SER spectrum has greater similarity to the E-SER spectrum at -03VSCE and suggests that the silver particles embedded in the sol-gel behave as if at a potential positive of the pzc RNA samples extracted from E coli B subtilis and S aureus were next examined by both E-SERS and SG-SERS E-SER spectra of these samples yielded quality spectra in 10 minutes in which all of the major features can be identified (Figures 6 and 7) This includes guanine at 650 cm-1 adenine at 791 cm-1 cytosine and uracil combining at 790 cm-1 and phosphate at 1100 1335 (in combination with adenine and guanine) 1465 and 1570 cm-1 Surprisingly adenine which demonstrated the greatest surface-enhanced Raman effect does not dominate the ring-breathing mode portion of the spectrum The intensities of the other base-pairs bands are of the same order of magnitude This suggests that when the base-pairs are linked together as in RNA they are enhanced in concert In fact the relative intensities are very similar to a normal Raman spectrum of E coli RNA which shows the combined cytosine and uracil band at ca twice the intensity of the adenine band and ca four times the intensity of the guanine band Unfortunately this means that the independent enhancement factors for the NA bases can not be used to estimate relative concentrations For example the relative 791 and 734 cm-1 bands for B subtilis would indicate that the cytosine andor uracil concentration was at least 20 times the adenine concentration whereas each of the four RNA bases are known to contribute 15-35 Nevertheless it is worth noting that the three RNA samples yield different relative band intensities that were reproduced in numerous measurements Although the relative concentrations of the NA bases for these samples have not been determined these differences can be quantified If it is assumed that the 650 cm-1 band represents 25 guanine the 791 cm-1 band represents 25 adenine and the 790 cm-1 50 cytosine plus uracil in the E coli RNA spectrum then the relative concentrations can be estimated for the other RNA samples To aid this calculation the three spectra were normalized to the phosphate band at 1100 cm-1 which has been shown to correlate to the total phosphate concentration and can be used as an internal standard In addition a simple baseline correction was applied (Figure 7) This yields 15 adenine 30 guanine and 55 cytosine plus uracil for B subtilis RNA and 18 adenine 25 guanine and 57 cytosine plus uracil for S aureus RNA The average SN of these measurements was 26 with an average error of 8 of the value (SplusmnN) It is also worth noting that the three RNA spectra show a marked shift in a band near 825 cm-1 This band is assigned to the symmetric stretch of the O-P-O ester linkage9 The band appears at 815 cm-1 for S aureus shifting to 820 cm-1 for B subtilis and 830 cm-1 for E coli Others have used the normal Raman intensity of the band at 815 cm-1 as a direct indication of the amount of A-class helix present while the intensity of the band at 830 cm-1 has been used as a direct indication of the amount of B-class helix present However the latter is more associated with DNA than RNA
Wavenumbers (∆cm-1) Figure 5 E-SER spectra of 020 mgmL adenosine monophosphate at A) -03 and B) -09VSCE and C) SG-SER spectra Conditions sample in 01M KCl buffered to pH 92 A) and B) 750 mW C) 75 mW of 1064 64 scans (1-min) at 8 cm-1
Wavenumbers (∆cm-1) Figure 6 E-SERS of 01 mgmL RNA from E coli 02 mgmL RNA from B subtilis and 02 mgmL RNA from S aureus Conditions 01M KCl pH 92 -03VSCE 750 mW of 1064 nm 640 scans (10 min) at 8 cm-1
AMP RNA
E coli A
B
C
B subtilis
S aureus
SPIE 2001-4575
69
SG-SER spectra of reasonable quality were also obtained for E coli and B subtilis especially the latter (Figure 8) However the spectra differ substantially from the E-SERS of the same samples Both SG-SER spectra are dominated by adenine at 735 cm-1 and a band at 1030 cm-1 Although unassigned the latter does appear in the RNA E-SER spectra Bands at 1105 and 1565 cm-1 are likely due to phosphate while bands at 1320 and 1470 cm-1 are less confidently assigned to phosphate They are significantly less intense and somewhat shifted from their SG-SERS counterparts (1335 and 1455 cm-1) A band at 670 cm-1 may be due to guanine which was observed at 664 cm-1 for SG-SERS of the pure sample However the SG-SER spectrum of AMP also had an intense 667 cm-1 band A number of other bands occur at 890 1070 1165 1245 1290 1420 1505 cm-1 and remain unassigned The SG-SER spectra are somewhat disappointing in that only adenine and guanine contributions can be positively identified This limits the ability to determine relative NA base concentrations and distinguish bacterial RNA However several of the unassigned bands may be due to the bases (eg 1030 and 1420 cm-1 due to cytosine) Further experiments will be required to clarify this point
Ecoli
B subtilis
S aureas
P G
A C+U
OPO
A B
Wavenumbers (∆cm-1) Wavenumbers (∆cm-1) Figure 7 SER spectra of RNA from A) B subtilis with contributions indicated and B) E coli B subtilis and S aureus with baseline correction and peak positions used to calculate contributions indicated G = guanine A = adenine C+U = cytosine plus uracil P = phosphate (backbone) OPO = phosphate ester linkage (A- vs B-class helix)
Wavenumbers (∆cm-1) Wavenumbers (∆cm-1) Figure 9 E-SER (-03VSCE) and SG-SER spectra of RNA from A) E coli and B) B subtilis Sample conditions as in Figure 6 E-SER spectra at 750 mW SG-SERS at 75 mW
A B E-SERS
SG-SERS
E-SERS
SG-SERS
SPIE 2001-4575
70
A final comparison was made between the two SER methods by measuring dipicolinic acid (DPA) This chemical may be invaluable as a test for spore forming bacteria specifically B anthracis 50 to 90 of B anthracis sporilates During spore formation dipicolinic acid is synthesized and once completed 10-15 of the dry spore weight is composed of the Ca2+ complex located in the spore core42 Heating in water can be used to initiate germination at which point the exosporium breaks and releases the Ca dipicolinate which becomes dipicolinic acid in water The structure of this chemical strongly
4 CONCLUSIONS Towards the goal of developing a method and technology to rapidly detect and identify bioagents we have been investigating surface-enhanced Raman spectroscopy as a tool to measure relative concentrations of nucleic acid bases in RNA and determine bioagent taxonomy Initially we investigated E-SERS since this method has been extensively researched and the optimum sample conditions are well developed However this method requires a three-electrode sample cell and electrolyte solution Incorporation of an electrolytic cell into sample systems used to analyze for BWA as aerosols in water or on surfaces can be designed using flow injection analysis technologies but cross-contamination and plugging of sample lines seems inevitable For this reason we also investigated metal-doped sol-gels as a SER-active medium Previous studies have shown this material to be active in all solvents particularly water capable of continuous measurements in flowing systems and reproducible (quantitative) between coated sample vials Here we compared SG-SER spectra to traditional E-SER spectra of the nucleic acid base pairs adenosine monophosphate and RNA High quality spectra of adenine cytosine guanine thymine and uracil were obtained by both E-SERS and SG-SERS Both methods yielded very similar spectra for the NA bases including a pH dependent study of adenine Enhancement factors and detection limits for adenine were determined as 2x105 and 16x10-11g and 1x105 and 12x10-10g for E-SERS and SG-SERS respectively (normalized to 75 mW and 10-min acquisition time) Fifty percent of the silver particle surface area in the sol-gel matrix was assumed covered by adenine which may have been overestimated yielding a lower EF and higher detection limit It should also be realized that each E-SER spectrum required several attempts to optimize the measurement conditions (pH electrode potential etc) While each SG-SER spectrum involved no sample preparation and often represents the first and only attempt to make the measurement Quality spectra of RNA extracted from Escherichia coli Bacillus subtilis and Staphylococcus aureus were obtained by E-SERS that were easily interpreted Bands due to adenine guanine cytosine plus uracil and phosphate were identified The SER band intensity of the NA bases in the RNA samples were of the same order of magnitude suggesting that their interaction with the silver surface is concerted as is their Raman enhancement Interestingly the relative SER band
Wavenumbers (∆cm-1) Figure 10 A) Raman spectrum of solid dipicolinic acid B) Ax20 C) electrolytic SERS of 6x10-3 M dipicolinic acid in 01 M KCl at a potential of +07VSCE and pH of 4 and D) sol-gel SERS of 6x10-3 M dipicolinic acid Conditions for A and C as in Figure 1 C) 100 mW of 1064 nm 50 scans 8 cm-1
A B (Ax20)
C
D
suggested that it would be SER active However the E-SER spectrum was unstable and varied considerably as a function of potential A consistent spectrum was obtained at +06VSCE (Figure 10) This potential is not recommended for measurement because the surface is actively dissolving in solution The SG-SER spectrum was considerably more stable of higher quality and easily reproduced Bands at 660 825 1010 1390 1430 1570 1590 and 3075 cm-1 were observed Enhancement factors were determined for the two media using the symmetric ring stretching mode at 995 and 1010 cm-1 for the normal Raman and SER spectra respectively E-SERS yielded an EF of 5x103 while SG-SERS yielded an EF of 2x105 for DPA The SN of the latter suggests a detection limit of 20x10-10g (based on adenine coverage 75 mW and 10-min) The differences in SER activity for these two media may be attributed to the combined electrolytic potential of the solution chemical and metal15 Again the E-SERS suggests that the SG-SERS is at a potential positive of the pzc While the instability in the E-SERS may also be associated with surface interactions of two carboxylic acid groups of dipicolinic acid during the ORCs
DPA
SPIE 2001-4575
71
intensities for RNA extracted from E coli are very similar to those measured by normal Raman spectroscopy Although the relative percent that each of the NA bases contributed to each RNA sample was not determined reproducible band intensities allowed noting the following trends The percent adenine decreases while the combined percent cytosine and guanine increase for both B subtilis and S aureus compared to E coli Quality spectra were also obtained for the RNA samples by SG-SERS but only a few bands were readily identified Calculations of NA base concentrations by SG-SERS will require further research In light of recent events we summarize the capabilities of these SERS media in terms of rapid detection of B anthracis and dipicolinic acid However these capabilities must be qualified First and foremost the level to which SERS can distinguish bacteria or viruses has not yet been determined Development of a database of both DNA and RNA base concentrations for BWAs and common bacteria to establish the level of taxonomic identification is ongoing Second rapid collection of aerosol water or surface samples is being addressed by others who report trapping particles on filters from 100 liters of air per minute Third although not presented here we have developed methods to extract RNA or DNA from cells and spores for SER analysis within 10 minutes Finally we assume a detection limit of 3600 spores per 100 liters of air is required although a 50 lethal dosage of anthrax has not been established With these qualifications a mass detection limit for RNA using SERS is estimated as follows A single measurement is performed in ca 20 minutes (140 liters collected in 14 min RNA extracted in 8 min spectral acquisition and analysis in 10 min) The average human breaths 7 liters per minute therefore the analyzer must at the very minimum detect 5000 spores in 140 liters of air One spore is approximately 2x10-18m3 (1x1x2 microm) and if a density of 075 gcm3 is assumed this corresponds to a mass of 15x10-12g Each spore contains 4-12 RNA or 12x10-13g RNA for 8 If we assume 23 of the RNA can be isolated for analysis during lysis then the proposed instrument must be able to detect 4x10-10g RNA from 5000 spores per 70 liters of air within 10 minutes As noted above the mass detection limits for adenine were estimated at 16x10-11g and 12x10-10g for E-SERS and SG-SERS respectively Although these detection limits suggests that RNA from 5000 spores is detectable with the current instrumentation it is highly likely that only a portion of an RNA segment (eg 120-nucleotide 5S rRNA) is in contact with the metal surface and will contribute to the SER effect The SN for the RNA spectra were 110th of the average SN for the four individual RNA bases suggesting a 10 contribution Furthermore effective taxonomy will likely require knowing the NA base concentrations to 1 of the value (eg 25plusmn025) Again the average measurement error for the bases is 12 These values suggest that the E-SERS is within a factor of 4 of the required detection limit whereas the SG-SERS detection limit must be improved by 25 times The same arguments can be applied to the detection of dipicolinic acid If we assume a spore releases 10 by weight DPA during germination then the proposed instrument must be able to detect 75x10-10g DPA from 5000 spores per 70 liters of air within 10 minutes The detection limit for SG-SERS was estimated at 20x10-10g and suggest that the vials are suitable to perform a rapid screen for anthrax A series of concentration dependent measurements are currently being performed to verify this assertion Finally we note that the measurements performed here employed an FT-Raman spectrometer This instrumentation was chosen over dispersive Raman instrumentation for the high wavelength accuracy afforded by the HeNe reference laser (Connes Advantage43) which would allow reliable spectral subtraction matching of observed spectra to stored library spectra and confident use of chemometric approaches Such data analysis is likely to be required to enhance BWA identification However this instrumentation sacrifices sensitivity due to a less efficient detector (InGaAs vs Si) less efficient Raman scattering and less efficient generation of plasmon modes Substantial improvements can be obtained using visible excitation and Si detection and these measurements are underway
5 ACKNOWLEDGEMENTS The authors are grateful to Drs D Cookmeyer and S Tove of the US Army Research Office (Contract Number DAAH04-96-C-0078) for their interest and support of this research The authors would also like top acknowledge Dr R Yin and J Jensen for supporting development of the metal-doped sol-gels (Contract Number DAAD13-01-C-0019) They also thank Dr Wilfred H Nelson for assistance in spectral interpretations
SPIE 2001-4575
72
6 REFERENCES 1 Roberts WL and Rainey PM Clin Chem 39 1872-1877 (1993) 2 Pasechnik VA CC Shone and P Hambleton Bioseperations 3 267-283 (1993) 3 Jackson PJ ME Hugh-Jones DM Adair G Green KK Hill CR Kuske LM Grinberg FA Abramova and P
Keim Proc Natl Acad Sci 95 1224-1229 (1998) 4 Jensen JL NC Wong W R Loerop SPIE 4775-03 (2001) 5 Snyder A Pet al SPIE 3853-15 (1999) 6 Danosky T R and McFadden P N in press (1997) 7 Woodruff WH Farquharson S Science 201 831-833 (1978) 8 Lord RC and Thomas GJJr Spectrochemica Acta 23A 2551-2591 (1967) 9 Thomas GJJr Biochim Biophys Acta 213 417-423 (1970) 10 Placzek G Handbuch der Radiologie 2 EMarx ed Akademische Verlagagescellschatt Liepzig 1934 UCRL
Trans No 526 (1959) 11 Jeanmaire DL and RP Van Duyne J Electroanalytical Chem 84 1-20 (1977) 12 Chada S Manoharan R Moenne-Loccoz P Nelson WH Peticolas WL and Sperry JF Applied Spectroscopy
47 38-43 (1993) 13 Manoharan R Ghiamati E Chada S Nelson WH and Sperry JF Applied Spectroscopy 47 2145-2150 (1993) 14 Todd EA Morris MD Applied Spectroscopy 48 545-548 (1994) 15 Wentrup-Byrne E Sarinas S and Fredericks PM Applied Spectroscopy 47 1192-1197 (1993) 16 Weldon MK VR Zhelyaskov and MD Morris Applied Spectroscopy 52 265-269 (1998) 17 Kneipp K and J Fleming J Mol Structure 145 173-179 (1986) 18 Pothier NJ and Force RK Applied Spectroscopy 46 147-151 (1992) 19 Ervin KM E Koglin JM Sequaris P Valenta and HW Nurnberg J Electroanal Chem 114 179-194 (1980) 20 Kim SK TH Joo SW Suh and MS Kim J Raman Spectrosc 17 381-386 (1986) 21 Pothier NJ and Force RK Analytical Chemistry 62 678-680 (1990) 22 Kneipp K Y Wang RR Dasari and MS Feld Applied Spectroscopy 49 780-784 (1995) 23 Nie S and Emory SR Science 275 1102 (1997) 24 Graham D WE Smith AMT Linacre CH Munro ND Watson and PC White Analytical Chemistry 69 4703-
4707 (1997) 25 Fleischmann M PJ Hendra and AJ McQuillan Chem Phys Lett 26 163-166 (1974) 26 Pettinger B J Chemical Phys 85 7442-7451 (1986) 27 Surface-Enhanced Raman Scattering Section Four Theory SPIE MS 10 M Kerker and B Thompson Eds (1990) 28 Wang D-S and Kerker M SPIE (M Kerker and B Thompson Eds) MS 10 417-429 (1990) 29 Weaver MJ Farquharson S Tadayyoni MA J Chem Phys 82 4867-4874 (1985) 30 Norrod KL Sudnik LM Rousell D and Rowlen KL Applied Spectroscopy 51 994-1001 (1997) 31 Farquharson S Weaver WJ Lay PA Magnuson RH and Taube H J Am Chem Soc 105 3350-3351 (1983) 32 Lee PC and Meisel D ldquoAdsorption and Surface-Enhanced Raman of Dyes on Silver and gold Solsrdquo J Phys Chem
86 3391-3395 (1982) 33 Li Y-S and Wang Y Applied Spectroscopy 46 142-146 (1992) 34 Bello JM DL Stokes and T Vo-Dinh Analytical Chemistry 61 1779-1783 (1989) 35 van Duyne RP J C Hulteen D A Treichel M T Smith M L Duval and T R Jensen J Phys Chem B1033854-
3863 (1999) 36 Farquharson S and W W Smith W H Nelson and J F Sperry SPIE 3533-27 207-214 (1998) 37 Lee Y H W Smith S Farquharson H C Kwon M R Shahriari and P M Rainey SPIE 3537 252-260 (1998) 38 Lee Y-H S Farquharson and P M Rainey SPIE 3857 76-84 (1999) 39 Current Protocols in Molecular Biology Wiley Interscience1003-1006 (1987) 40 Farquharson S Smith W Carangelo R C and Brouillette C SPIE 3859 14-23 (1999) 41 Chase D B and JF Rabolt Fourier Transform Raman Spectroscopy Acad Press Ch1 p 131 (1994) 42 Brock TD MT Madigan JM Martinko and J Parker Biology of Microorganisms 7th Ed Prentice Hall p 76-80
(1994) 43 Connes J Rev Opt Theor Instrum 40 45 (1961)
166
Chemical agent identification by surface-enhanced Raman spectroscopy
Stuart Farquharson and Paul Maksymiuk
Real-Time Analyzers 87 Church Street East Hartford CT 06108
Kate Ong and Steven D Christesen US Army SBCCOM Aberdeen Proving Ground MD 21010
ABSTRACT The recent distribution of anthrax through the US postal system and the subsequent infection and death of several postal and national media employees amplifies the need for methods to rapidly detect identify and quantify this and other chemical and biological warfare agents The US military has also identified water supplies as a likely method of warfare agent deployment and is funding the development of a Joint Service Agent Water Monitor (JSAWM) In an effort to aid military personnel and the public at large we are developing a portable analyzer capable of identifying and quantifying chemical agents rapidly either on-demand or continuously The approach is based on the ability of Raman spectroscopy to identify molecular structure through the abundant vibration information provided in spectra and the ability of surface-enhanced Raman spectroscopy (SERS) to detect extremely low concentrations (eg part-per-billion) through the enhancement of Raman scattering by six orders of magnitude or more A key element to the analyzer design is a new SER active medium that is capable of quantitative reversible measurements The medium consists of silver or gold nanoparticles incorporated into a sol-gel matrix The porous silica network offers a unique environment for stabilizing SER active metals and the high surface area increases the interaction between the analyte and metal particles Here we present the use of new sol-gels that also selectively enhance chemicals based on polarity and charge Base-line measurements of chemical agents and their hydrolysis products are presented and compared to the JSAWM goal of 30 micrograms per liter detection Keywords Chemical warfare agent hydrolysis product SERS Raman spectroscopy sol-gel nanoparticle
1 INTRODUCTION Since September 11 2001 the threat of terrorist attacks and biological warfare within US borders has become a sobering reality The simplicity in manufacturing ease of deployment and the relatively low cost of chemical warfare agents (CWAs) raises public concern that they may also be used by terrorists Indeed terrorists released sarin (GB) in the Tokyo subway in 19951 Countering terrorism and terrorist attacks requires recognizing likely deployment scenarios and having the required technology to rapidly detect the deployment event One method of deployment has been long identified by the US military distribution through water supplies To counter this threat the Department of Defense is funding or monitoring the capabilities of 200 technologies with the goal of developing a Joint Service Agent Water Monitor (JSAWM) that is field portable2 The JSAWM must be capable of identifying the chemical agents with no false-positives and quantifying the agents at microgram per liter concentrations within 10 minutes (Table 1)3 This includes the analysis of drinking water supplies distribution and storage systems as well as potable water supplies The earliest technologies developed for CWA detection were based on electrochemical ionization or colorimetric analysis (eg phosgene tape) Although these analyzers were easy to use they were not generally agent specific and suffered from false-positives4 More traditional laboratory methods have also been investigated and in particular combined gas chromatography and mass spectrometry (GCMS) has been very successful at eliminating false-positives56 However GCMS requires extraction repeated calibration and long analysis times (typically 20 to 60 minutes)6 making it labor intensive and less than desirable for field use More rapid analysis of agents in the solid liquid and gas phase has been To whom correspondence should be addressed emailfarqureal-time-analyzerscom
Vibrational Spectroscopy-based Sensor Systems Steven D Christesen Arthur J Sedlacek III EditorsProceedings of SPIE Vol 4557 (2002) copy 2002 SPIE middot0277-786X02$1500
167
Table 1 Chemical Agent Structures Hydrolysis Half-lives and JSAWM Thresholds Agent Short-Hand Chemical Structure Hydrolysis
Half-Life JSAWM
Thresholds
Sarin (GB) F-[O=P-CH3]-O-CH(CH3)2 213 hours 32 microgL Soman (GD) F-[O=P-CH3]-O-CH(CH3)-(C-(CH3)3) 23 hours 32 microgL Tabun (GA) (CH3)2-N-[O=P-CN]-O-C2H5 41 hours 32 microgL VX C2H5O-[O=P-CH3]-S-(CH2)2-N-(CH(CH2)2)2 821 hours 32 microgL EA2192 HO-[O=P-CH3]-S-(CH2)2-N-(CH(CH2)2)2 gt9 years 32 microgL Mustard (H) ClCH2CH2-S-CH2CH2Cl encapsulates 47 microgL Lewisite (L) ClCH=CH-As-Cl2 rapid 27 microgL HCN HCN rapid 20 mgL BZ C7NH12-O-[C=O]-COH(C6H5)2 23 microgL T-2 Toxin 87 microgL at pH 7 to 75 and 20 to 25 oC
demonstrated by vibrational spectroscopy7-10 Hoffland et al7 reported infrared absorbance spectra and absolute Raman cross sections for several chemical agents while Christesen measured Raman cross sections for sarin tabun mustard gas and VX (ethyl S-2-diisopropylamino ethyl methylphosphonothioate)11 Again however these techniques also have limitations Raman spectroscopy is simply not a very sensitive technique and detection limits are typically 01 (1000 ppm) While infrared spectroscopy would have limited value in analyzing poisoned water since the very strong infrared absorption of water would obscure most other chemicals present Nevertheless efforts to overcome these limitations have been demonstrated Braue and Pannella8 quantified the G-series nerve agents (tabun sarin and soman) in terms of infrared attenuated total reflectance using a circle-cell And Alak and Vo-Dinh demonstrated the possibility of surface-enhanced Raman spectroscopy (SERS) to identify CWAs by measuring several organophosphonates that simulate the nerve agents12 However quantitative measurements have not been demonstrated for the SER-active material used (silver coated on alumina particles) or other SER-active media13 Recently we developed silver-doped sol-gels to promote the SER effect14-17 The porous silica network of the sol-gel matrix offers a unique environment for stabilizing SER-active metal particles and the sol-gel provides a high surface area that effectively increases the number of molecules observed within the Raman scattering volume The silver-doped sol-gels have been coated on the internal walls of standard glass vials to allow rapid SER analysis of small sample quantities (lt 01 mL) without preparation We have used p-aminobenzoic acid (PABA) as a test chemical to demonstrate surface enhancements greater than 106 reversible measurements in a flowing system reproducible measurements from vial-to-vial and batch-to-batch and measurements in multiple solvents including water14-17 Recently we used these vials to measure Tabun (GB) and Sarin and several hydrolysis products pinacolyl methyl phosphonate (PMP from Soman) and methyl phosphonic acid (MPA from all G-agents Figure 1) Although a number of unique vibrational bands are observed (eg C-N stretch doublet and P-C stretch) the G-agents were only observed for 5 concentrations and all spectra required baseline corrections Figure 1 Surface enhanced Raman spectra of ~5 vv A) Tabun and B) Sarin C) 1 vv PMP and D) 10 ppm MPA using sol-gel sample vials 785 nm excitation 1-min scan and CCD detection Performed at Aberdeen Proving Ground
C A
Wavenumber (∆cm-1) Wavenumber (∆cm-1)
D B
790
545 C-N
2135 2190 P-C 770
760 1290
Proc SPIE Vol 4577
168
Nevertheless MPA was readily observed for a 10 ppm sample with an estimated detection limit of 04 ppm (based on a signal-to-noise ratio of 3 for the 760 cm-1 band intensity) This measurement provides encouragement in that SERS may satisfy the needs of a JSAWM Furthermore MPA is also a hydrolysis product of VX and V-gas and EA2192 (Figure 2) and may prove a valuable indicator of agent usage Figure 2 Hydrolysis of Sarin to form hydrofluoric acid (HF) methylphosphonic acid 1-methylethyl ester (MPAMME) methyl phosphonic acid (MPA) and 2-propanol With this initial albeit modest success we began analyzing chemicals with various sol-gel compositions that we have been developing Here we describe four sol-gel compositions that select for 1) polar-positive 2) polar-negative 3) weakly polar-positive and 4) weakly polar-negative chemical species The ability of these sol-gels to select and enhance Raman scattering is described for several test chemicals and MPA
2 EXPERIMENTAL The chemicals analyzed as well as all chemicals used to prepare the metal-doped sol-gels were obtained at their purest commercially available grade from Aldrich (Milwaukee WI) The sol-gel designed to select for polar-negative species was prepared from a silver amine complex tetramethyl orthosilicate (TMOS) and methanol After mixing 02 mL of the sol-gel solution was transferred into a glass vial (2 mL) dried and heated The incorporated silver ions were then reduced using dilute sodium borohydride The vials were washed and dried prior to the addition of a sample solution In a similar manner the sol-gel designed to select for polar-positive species was prepared from a gold salt TMOS and methanol The sol-gel designed to select for weakly polar-negative species was prepared from a silver amine complex tetraethyl orthosilicate (TEOS) and methanol And the last sol-gel designed to select for weakly polar-positive species was prepared from a gold salt TEOS and methanol All samples were prepared in a chemical hood and transferred into plain or SER-active vials for analysis Normal Raman spectral measurements employed 1-mL pure samples that were placed in a 1-cm3 cuvette and weighed This yielded a powder density that allowed accurate calculation of molecules in the optical collection field SERS measurements employed 1-mg sample per mL water concentrations unless otherwise stated Once prepared a 01 mL sample was placed into one of the four selective sample vials which in turn was placed into the sample compartment of a Raman spectrometer for analysis A prototype Fourier transform Raman spectrometer (RTA) was used for these measurements18 The system consisted of a NdYAG laser (Brimrose) for excitation at 1064 nm an interferometer built by On-Line Technologies (OLT East Hartford CT) for frequency separation an uncooled InGaAs detector for signal detection (RTA) and an Intel 400 MHz Pentium II based laptop computer (Dell Round Rock TX) for interferometric control data acquisition (OLT) and analysis (LabVIEW by National Instruments Austin TX) Additional components included a Notch filter (Kaiser Ann Arbor MI) and interferometer entrance and exit optics (Edmund Scientific Barrington NJ) Fiber optics were used to deliver the excitation beam to the sample and the scattered radiation to the interferometer (1 meter lengths of 200 and 365 micron core diameter respectively Spectran Avon CT) A second Notch filter (Kaiser) was used as a beam splitter to direct the excitation beam along the same axis as the collected radiation A microscope object (20x04 Newport Irvine CA) was used to focus the beam into the sample and to collect the scattered radiation back along the same axis In this co-axial backscattering arrangement the excitation beam was passed through the outside of a glass vial and focused onto the silver-doped sol-gel film (01-03 mm thickness) containing the sample
3 RESULTS AND DISCUSSION p-aminobenzoic acid (PABA) and phenyl acetylene (PA) and were used to refine the selectivity and SER-activity of the four different metal-doped sol-gels PABA is a popular chemical used to evaluate the performance of SER-active media Here the polar end groups can be used to test selectivity of the polar-negative and polar-positive sol-gels PA is essentially non-
2O+ H HF + +OH
OH3H C
OP
OF
CH
CH3
3
3H C
OCP
OOH
CH
CH3
3
3H C
O
CPHO
CH
CH
3
3
C
Sarin MPAMME MPA 2-propanol
Proc SPIE Vol 4577
169
polar but a high electron density in the cylindrical π cloud around the carbon-carbon triple bond allows testing the selectivity of the weakly polar-negative and weakly polar-positive sol-gels As Figure 3 illustrates PABA passes through the polar sol-gel and is enhanced by either the silver or gold particles At 1 mgml the concentration of neutral PABA is ca 20 times that of the ionized form (pKa = 48) For electropositive silver the PABA anion is expected to interact through the carboxylate group and the associated vibrational modes are expected to dominate the spectrum Conversely for electronegative gold either form of PABA is expected to interact through the amine group The clear differences in our spectra support this expectation Furthermore bands at 840 and 1405 cm-1 assigned to a COO- bend and stretch respectively are significantly more intense for silver than gold Additional bands at 1140 and 1195 cm-1 are assigned to CH bending modes while bands at 1450 1500 and 1605 cm-1 are assigned to ring vibrational modes A very similar SER spectrum for PABA on a silver-coated alumina substrate has previously been reported with similar assignments19 For the gold-doped sol-gel new bands appear at 690 1355 and 1585 cm-1 The first band is assigned to a ring-H bending mode the second band to a ring-N- stretching mode and the third band to a possible NH2 scissors mode or ring mode The second band is not observed in the normal Raman spectrum but infrared bands occur at this frequency for aromatic ring-secondary amine stretching modes The scissors mode occurs at this frequency in Raman spectra for several chemicals but is absent in the PABA Raman spectrum Alternatively this mode may be the1600 cm-1 ring mode that has been shifted by the gold interaction Again a very similar SER spectrum of PABA has been reported but surprisingly using silver (colloids)2021 not gold as the enhancement medium These researchers also assumed the primary interaction of PABA with silver was through the carboxylate anion and made assignments accordingly For example they assigned the 1359 cm-1 to a COO- stretch not to the amine group as we have They also favor the ring stretching mode assignment for the 1582 cm-1 band Finally it should be said that other researchers have argued that the most dominant band in the SER spectra at 1450 cm-1 a ring vibration mode suggests that PABA lies flat on the surface and the π-orbitals dominate the surface interaction22
Figure 3 SER spectra of A) PABA using polar-negative and B) polar-positive sol-gels and C) PA using weakly polar-negative and D) weakly polar-positive sol-gels PABA is 1 mgmL PA is 1 vv Spectral conditions 75 mw 1064 nm 100 scans (15 min) 8 cm-1 resolution Non-polar PA passes through the non-polar sol-gels and is also enhanced by both metals The spectra are easily understood For electropositive silver PA interacts through the cylindrical triple bond π electron cloud and a -CequivC- doublet occurs near 2000 cm-1 The interaction is reasonably strong since this band appears at 2112 cm-1 in the normal Raman spectrum For electronegative gold this interaction is unlikely and only very weak bands occur near 2000 cm-1 The remaining bands are at 1000 cm-1 1200 cm-1 doublet and 1595 cm-1 all appear in the normal Raman spectra at virtually the same frequencies and are assigned to the symmetric ring-breathing mode CH bending modes and the trigonal ring-breathing mode respectively The polarnon-polar selectivity of the polar-negative and weakly polar-negative sol-gels was tested by adding a 11 molar mixture of PABA and PA The selective enhancement is quite good (Figure 4) The spectrum obtained using the polar sol-gel represents 78 PABA and 22 PA while the spectrum obtained using the weakly polar sol-gel represents 9 PABA and 91 PA The band peak intensities at 2000 cm-1 for PA and 1450 cm-1 for PABA were used for these calculations and are expanded in Figure 4 for clarity
C A
Wavenumber (∆cm-1) Wavenumber (∆cm-1)
2NH COOH
D B
C CH
Proc SPIE Vol 4577
170
Figure 4 SERS of 11 MM of PABA and PA in A) polar-negative and C) weakly polar-negative sol-gels The lower traces compare the pure chemicals B) 1 mgml PABA in polar-negative sol-gel and D) 1 PA in weakly polar-negative sol-gel while the insets magnify the minority species for clarity (x5 in A and x10 in B) Spectral conditions as in Figure 3 Following this development of selective sol-gels that maintained SER activity we measured cyanide and MPA (Figure 5) Not surprisingly the best sensitivity for both hydrolysis products was obtained using the polar-negative sol-gel The interaction of the cyanide anion with the silver surface is sufficient to shift the CequivN stretch observed at 2080 cm-1 in the normal Raman spectrum to 2145 cm-1 in the surface-enhanced Raman spectrum Furthermore the band is substantially broadened This anion has been extensively studied by electrolytic SERS and this shift and broadening have been attributed to the formation of a tetrahedral Ag(CN)3
2- surface structure23 Figure 5 Surface-enhanced (upper traces) and normal Raman spectra (bottom traces) of A) CN- and B) MPA in silver-doped TMOS SERS conditions as in Figure 3 and 1 mgmL Note MPA yields two distinct spectra for neutral (top) and acidic pH (middle) The normal Raman spectra employed pure powders 500 scans and 900 mW of 1064 nm SER measurements of MPA with the polar-negative sol-gel yielded two unique spectral signatures that depended on solution pH (Figure 5) For more neutral solutions the P-C stretch of MPA at 762 cm-1 dominates and the CH2 stretch at 2922 cm-1 appears The SN is sufficiently high that the anti-Stokes Raman shift at -762 cm-1 is observed For deprotonated MPA an oxygen-surface mode appears at 325 cm-1 (as well as its anti-Stokes complement) suggesting a strong interaction This results in substantial enhancement of the P-O-C mode at 1051 cm-1 (upper trace) Others report that this mode dominates the infrared spectra of nerve agents measured in water8 Comparison of the two spectra suggests the following molecule-to-surface orientations The appearance of the oxygen-surface and P-O-C modes in the upper spectrum of Figure 5B indicates that the tetrahedral molecule interacts with the silver surface through the deprotonated oxygen and is oriented end-on The
A B
Wavenumber (∆cm-1)
C A
Wavenumber (∆cm-1) Wavenumber (∆cm-1)
D B
Wavenumber (∆cm-1)
Proc SPIE Vol 4577
171
dominance of the P-C and the CH2 stretches and the disappearance of the P-O-C mode in the upper spectrum suggest the molecule is oriented with the methyl-hydroxyl-hydroxyl face towards the surface However considerably more research must be performed to verify these points Both chemical forms were measured as a function of concentration to determine the onset of monolayer coverage and estimate expected detection limits (Figure 6) Below monolayer coverage the signal to concentration dependence should be linear and the SN of any spectral measurement in this range can be used to predict the detection limit In the spectra presented here the peak height was used as the signal while the noise as root-mean-squared (RMS) was measured between 4400-4600 cm-1 Since noise is distributed evenly throughout the spectrum when transformed this region was used since it does not have any contributions from signals or baseline offsets Figure 6 shows a series of spectra for MPA along with a plot of the 1050 cm-1 band intensity (the noise was identical for this series) as a function of sample concentration A clear discontinuity appears in the vicinity of 01 mgmL (19 ppm) indicating the onset of monolayer coverage A detection limit defined as a SN of 3 was calculated for the 01 and 005gmL samples at 24x10-4 and 25x10-4 gL respectively A more modest detection limit of 101x10-4 gL was obtained using the 760 cm-1 band in the second series of concentration measurements These detection limits correspond to 46 and 190 ppb for a 3-min scan and 75 mW of incident laser power
Figure 6 A) Concentration dependence of MPA SERS measured in silver-doped TMOS) B) Concentrations are 001 005 01 05 1 gL (188 94 188 94 188 ppm) I760 series (bull) and I1050 series (∆)
Enhancement factors for cyanide and methyl phosphonic acid can be determined by estimating the number of molecules contributing to the surface-enhanced and normal Raman spectra The enhancement factor EF can be defined by the following equation
EF = (ISERSINR)bull(MNRMSERS) bull(PNRPSERS) bull(TNRTSERS)12
where I is the spectral band intensity M is the sample mass P is the incident laser power and T is the measurement time (or number of scans) for the two measurements For the normal Raman spectra a cylindrical scattering volume is assumed based on the laser area (28x10-7m2 6x10-4m diameter spot) and the penetration depth (1x10-3 m)24 The density of KCN and MPA as powders were measured at 0572 and 0516 gcm3 indicating that 16x10-4 and 144x10-4 g produced the normal Raman signals in Figure 5 respectively The number of molecules contributing to the SER spectra are those on the silver particles embedded in the sol-gel The total silver surface area can be determined from the average particle size concentration and the scattering volume Previous scanning electron micrographs showed the average silver particle size to be 40 nm in diameter (335x10-23m3)17 The silver concentration is 012M based on the reactant molar concentrations and dilution factors And the scattering volume is 76x10-
11m3 again based on a cylindrical scattering volume defined by a laser area of 28x10-7m2 and a sol-gel thickness of 27x10-
4m This volume contains 123x10-6g of silver equivalent to 35x109 silver particles with a collective surface area of 18x10-
5m2 However it may reasonably be assumed that at least half of the silver surface is in contact with the sol-gel matrix and unavailable for analyte interaction If we assume monolayer coverage and that each CN molecule occupies 15x10-20m2 then approximately 62x1014 molecules or 27x10-8g of CN contribute to the SER spectrum (20x10-19m2 46x1013 molecules 74x10-9g for MPA) Accordingly the EF for cyanide equals 48x104 ((180599) bull(16x10-427x10-8) bull(90075) bull(500100)12) The EF for MPA is considerably higher at 87x106 ((60326) bull(144x10-474x10-9) bull(90075) bull(500200)12)
0
20
40
60
80
100
120
140
0 02 04 06 08 1 12[MPA] (mgmL)
I (76
0)
0
100
200
300
400
500
600
I (10
50)
Wavenumber (∆cm-1)
A B
Proc SPIE Vol 4577
172
4 CONCLUSIONS Here we present for the first time surface-enhanced Raman spectra of Tabun and Sarin obtained using silver-doped sol-gels However the inferior enhancement suggested employing SER-active sol-gels varying in composition to improve both detection limits and selectivity of the target analytes To this end we successfully demonstrated the capabilities of four sol-gels that select for 1) polar-positive 2) polar-negative 3) weakly polar-positive and 4) weakly polar-negative chemical species p-aminobenzoic acid was used to show that silver could be used to attract polar-negative chemicals or functional groups (carboxylate anion) and that gold could be used to attract polar-positive chemicals (amine) while a mixture of p-aminobenzoic acid and phenylacetylene was used to show that tetramethyl orthosilicate preferentially solvates polar chemicals and that tetraethyl orthosilicate preferentially solvates weakly polar chemicals This increased sample control was applied to cyanide and methyl phosphonic acid two hydrolysis products of chemical warfare agents Exceptional results were obtained for methyl phosphonic acid allowing measurement of 1x10-2 gL for a 3-min scan and 75 mW of incident laser power with an estimated detection limit of 245x10-4 gL and an enhancement factor of 87x106 However this detection limit is 76 times less sensitive than required for the JSAWM (32x10-6gL for the G-agents) Finally we note that the measurements performed here employed an FT-Raman spectrometer This instrumentation was chosen over dispersive Raman instrumentation for the high wavelength accuracy afforded by the HeNe reference laser (Connes Advantage25) which would allow reliable spectral subtraction matching of observed spectra to stored library spectra and confident use of chemometric approaches Such data analysis is likely to be required to identify the chemical agents as well as distinguish hydrolysis products However this instrumentation which employs 1064 nm excitation and InGaAs detection sacrifices sensitivity We believe that the measurement sensitivity can be improved by at least two orders of magnitude by using shorter laser excitation wavelengths (eg 532 nm) This would provide more efficient Raman scattering (fourth power dependence on laser excitation wavelength) more efficient generation of plasmon modes and allow using more efficient detector material (Si vs InGaAs) These modifications are underway
5 ACKNOWLEDGEMENTS The authors would like to thank Dr R Yin and J Jensen of the US Army for supporting this work (Contract Number DAAD13-01-C-0019) They would also like to thank Advanced Fuel Research for making their laboratory facilities available
6 REFERENCES 1 Tu Anthony ldquoOverview of Sarin Terrorist Incidents in Japan in 1994 and 1995rdquo 6th CBW Protection Symposium
Stockholm Sweden 10-15 May 1998 2 Jensen JL NC Wong W R Loerop SPIE 4775-03 (2001) 3 JSAWM Requirements at wwwsbccomapgeaarmymilRDAecbcrtPRODSERJSAWMjsawmhtml 4 ldquoThe Chemical Weapons Convention Redefines Analytical Challengerdquo Analytical Chemistry News amp Features June 1
397A (1998) 5 Johnston RL Hoefler CM Fargo JC and Moberley B ldquoThe Defense Nuclear Agencyrsquos ChemicalBiochemical
Weapons Agreements Verification Technology Research Development Test and Evaluation Program and its Requirements for On-Site Analysisrdquo AT-ONSITE 5-8 (1994)
6 Black RM Clarke RJ Read RW and Reid MT ldquoApplication of gas chromatography-mass spectrometry and gas chromatography-tandem mass spectrometry to the analysis of chemical warfare samples found to contain residues of the nerve agent sarin sulphur mustard and their degradation productsrdquo J Chromatography 662 301-321 (1994)
7 Hoffland LD Piffath RJ Bouck JBrdquoSpectral signatures of chemical agents and simulantsrdquo Optical Engineering 24 982-984 (1985)
8 Braue EHJ Pannella MGrdquoCIRCLE CELL FT-IR Analysis of Chemical Warfare Agents in Aqueous Solutionsrdquo Applied Spectroscopy 44 1513-1520 (1990)
9 Tseng C-H Mann CK and Vickers TJ ldquoDetermination of Organics on Metal Surfaces by Raman Spectroscopyrdquo Applied Spectroscopy 47 1767-1771 (1993)
10 Staff Reporter ldquoUS Army Tests Lidar to Detect Biological Toxinsrdquo Photonic Spectra pg 50 December 1998
Proc SPIE Vol 4577
Proc SPIE Vol 4577
173
11 Christesen SD Raman cross sections of chemical agents and simulants Applied Spectroscopy 42 318-321 (1988) 12 Alak AM and Vo-Dinh T ldquoSERS of Organophosphorous Chemical Agentsrdquo Analytical Chemistry 59 2149-2153
(1987) 13 Norrod KL Sudnik LM Rousell D and Rowlen KL ldquoQuantitative Comparison of Five SERS Substrates
Sensitivity and Detection Limitrdquo Applied Spectroscopy 51 994-1001 (1997) 14 Lee Y and Farquharson S ldquoSERS Sample Vials Based on Sol-Gel Process for Trace Pesticide Analysisrdquo SPIE 4206
140-146 (2000) 15 Farquharson S and Lee Y ldquoTrace Drug Analysis by Surface-Enhanced Raman Spectroscopyrdquo SPIE 4200-16 (2000) 16 Lee Y Farquharson S and Rainey P M Surface-Enhanced Raman Sensor for Trace Chemical Detection in Water
SPIE 3857 76-84 (1999) 17 Lee Y Farquharson S Kwong H and Shahriari M ldquoSurface-Enhanced Raman Sensor for Surface-Enhanced Raman
Spectroscopyrdquo SPIE 3537 252-260 (1998) 18 Farquharson S Smith W Carangelo R C and Brouillette C ldquoIndustrial Raman Providing Easy Immediate Cost
Effective Chemical Analysis Anywhererdquo SPIE 3859 14-23 (1999) 19 Narayanan VA JM Bello JD Stokes and T Vo-Dinh Analusis 19 307-310 (1991) 20 Laserna JJ E L Torres and JD Winefordner Analytica Chemica Acta 469-480 (1987) 21 Torres EL and JD Winefordner Analytical Chemistry 59 1626-1632 (1987) 22 Suh JS DP DiLella M Moskovits J Phys Chem 87 1540-1544 (1983) 23 Benner RE R Dornhaus R Chang and BL Laube Correlations in the Raman spectra of cyanide complexes adsorbed
at silver electrodes with voltammograms Surface Science 101 341 (1980) 24 Chase D B and JF Rabolt Fourier Transform Raman Spectroscopy Acad Press Ch1 p 131 (1994) 25 Connes J Rev Opt Theor Instrum 40 45 (1961)
APPLIED SPECTROSCOPY 351
focusing the 488 nm laser beam 2 mm deep into thebulk of the crystals to avoid contributions from potentialdepletion layers As shown in Fig 1b the dependence ofthe Raman shift on the degree of deuteration is almostperfectly linear and ts very well with D 5 22684cmR1 24526 where D is the degree of deuteration (in )and R is the spectral mean of the PO4 vibration in cm21A linear correlation coef cient of 0998 indicates an ex-cellent linear dependence of the Raman peak shift withdegree of deuteration This result shows that the shift ofthe PO 4 peak is simply caused by the linear increase inatomic mass due to isotope substitution which decreasesthe length of hydrogen-like bonds
This excellent linear dependence allows us to map thepro le of the DH exchange layer at the surface of DKDPcrystals by acquiring Raman spectra and determining theposition of the PO 4 peak for various depths This methodis preferable over other methods such as determining thestrength of the OD vibration directly (eg at 715 cm21)because the position of the most intense peak in the Ra-man spectrum can be measured more precisely than theintensity of some of the weakest peaks in the spectrumThis is demonstrated in Fig 2 where depth-dependentRaman spectra (Fig 2a) and the resulting exchange layerpro les for two DKDP crystals are shown (Fig 2b) Thespectra in Fig 2a were obtained from a depth scan of aDKDP crystal with 75 degree of deuteration in thebulk grown at 45 8C The spectra start out as DKDP with30 deuteration close to the surface and approach thebulk DKDP spectrum within a few micrometers of depthThe fact that the relative degree of deuteration does notextend to 0 D is due to the limited depth resolution ofthe Raman microprobe which averages over 4 mm indepth Figure 2b depicts the resulting DH exchange layerpro les for this and a second crystal grown at 63 8Crespectively Both crystals had the same exposure to am-bient conditions and their main difference is the temper-ature at which they were grown The different exchangelayer pro les indicate that crystals grown at differenttemperatures have differen t proton conductiv ities 13
which leads to a difference in their rate of deuteriumdepletion The parameters controlling this behavior arecurrently the objective of a detailed study the results ofwhich will be reported elsewhere
CONCLUSION
In conclusion we have shown that the shift of the to-tally symmetric PO 4 stretch mode in the Raman spectrumof DKDP crystals scales linearly with degree of deuter-ation This allows us to correlate Raman peak positionsto deuteration levels in these crystals We have presenteda new technique to determine DH diffusion pro les inDKDP frequency conversion crystals based on micro-Ra-man spectroscopy This technique is fast inexpensiveand works under various environmental conditionswhich will allow us to better understand and control deu-terium depletion in DKDP crystals
ACKNOWLEDGMENTS
We would like to thank M Runkel for rst discovering DKDP crack-ing R Floyd for providing DKDP crystals and L Chase and A Burn-ham for their support and helpful discussions This work was performedunder the auspices of the US Department of Energy by the University
of California Lawrence Livermore National Laboratory under contract W-7405-Eng-48
1 J J De Yoreo A K Burnham and P K Whitman Int Mat Rev47 113 (2002)
2 C E Barker R A Sacks B M Van Wonterghern J A Caird JR Murray J H Campbell K Kyle R B Ehrlich and N DNielsen Proc SPIE-Int Soc Opt Eng 2633 501 (1995)
3 T Suratwala paper to be published4 Cleveland Crystals Inc httpwwwclevelandcrystalscom
KDPshtmltable5 E A Popova I T Savatinova and I A Velichko Sov Phys Solid
State 12 1543 (1971)6 I P Kaminow R C C Leite and S P S Porto J Phys Chem
Solids 26 2085 (1965)7 J A Subramony B J Marquardt J W Macklin and B Kahr
Chem Mat 11 1312 (1999)8 H Tanaka M Tokunaga and I Tatsuzaki Solid State Commun
49 153 (1984)9 R J Nelmes G M Meyer and J E Tibballs J Phys C 15 59
(1982)10 M A Yakshin D W Kim Y S Kim Y Y Broslavets O E
Sidoryuk and S Goldstein Laser Physics 7 941 (1997)11 I Takenaga Y Tominaga S Endo and M Kobayashi Solid State
Commun 84 931 (1992)12 C Krenn personal communication13 M Sharon and A K Kalia J Solid State Chem 21 171 (1977)
Rapid Dipicolinic Acid Extractionfrom Bacillus Spores Detectedby Surface-EnhancedRaman Spectroscopy
STUART FARQUHARSON ALAND GIFT PAUL MAKSYMIUK andFRANK E INSCOREReal-Time Analyzers Inc East Hartford Connecticut06108
Index Headings Dipicolinic acid Bacillus spores Anthrax Surface-enhanced Raman spectroscopy
INTRODUCTION
The anxiety caused by the distribution of anthrax en-dospores through the US postal system in October 2001was exacerbated by the long time required for positiveidenti cation of the Bacillus anthracis spores and the un-known extent of their distribution Since that time manymethods capable of rapid eld analysis have been inves-tigated to augment or replace the laboratory method ofgrowing microorganisms in culture media which takesdays to perform12 Prominent among these approachesare polymerase chain reactions (PCR)3 immunoassaysand detection of calcium dipicolinate as a biochemicalsignature PCR employs primers to separate organism-
Received 10 October 2003 accepted 14 November 2003 Author to whom correspondence should be sent
352 Volume 58 Number 3 2004
speci c nucleic acid sequences (eg capsular protein en-coding gene for Bacillus anthracis)4 and polymerases toamplify the segment until it is detectable Recently am-pli cation times have been substantially reduced andcomplete analysis can now be performed in an hour orless Immunoassay methods are also being developed thatuse competitive binding of the bioagent (as an antigen)and its labeled conjugate for a limited number of anti-bodies Although analyses can be performed in under 30minutes a well-de ned anthrax antigen has not yet beenidenti ed5ndash7 and consequently the false-positive rate isunacceptably high8
A number of other methods are being developed witha focus on the detection of calcium dipicolinate (CaDPA)and its derivatives as a B anthracis signature This is sobecause only spore-forming bacteria contain CaDPA andthe most common potentially interfering spores such aspollen and mold spores do not Relatively fast methodshave been developed to chemically extract CaDPA andthen detect it directly by uorescence9 or indirectly byluminescence1011 In the latter case hot dodecylamine(DDA) has been used to extract dipicolinic acid (DPA)and terbium has been utilized to form a highly lumines-cent DPA complex11 Although measurements have beenperformed in as little as ve minutes it was found thatas many as three concentration-dependent complexes canform each with different lifetimes This coupled withthe fact that the Tb31 cation produces the same lumines-cence spectrum makes determinations of low spore con-centrations problematic
It has been long known that Raman spectra of Bacillispores are dominated by bands associated with CaDPA12
and that these spectra may provide a suitable anthrax sig-nature at the genus level13 Since that time considerableimprovements in Raman instrumentation have led to lab-oratory measurements of single Bacilli spores14 and to eld measurements of spores captured from a mail-sort-ing system15 However the single spore measurementsrequired complex instrumentation that is not ruggedwhile the eld measurements required milligrams of sam-ple Furthermore the Raman spectra of both measure-ments contained uorescence contributions that would in-crease uncertainty in quanti cation
In related research we demonstrated that nanogramquantities of DPA could be detected by uorescence-freesurface-enhanced Raman spectroscopy (SERS)16 We alsodemonstrated that microliter volumes of chemicals canbe detected by SERS using metal-doped sol-gel-packedglass capillaries17 Towards the goal of developing a rap-id eld SERS-based anthrax spore detector we havecombined our previous research and we now report thatDPA can be extracted from a 10 mg B cereus spore sam-ple using DDA in 1 minute and can be detected by SERSin an additional 1 minute
EXPERIMENTAL
Dipicolinic acid (26-pyridinedicarboxylic acid DPA)and dodecylamine (DDA) were used as received fromSigma-Aldrich (Milwaukee WI) Lyophilized B cereusspores prepared according to the literature13 were sup-plied by the University of Rhode Island and used as re-ceived Multiple particles approximately 01 mm3 each
were separated and weighed at 5 to 15 mg representing05 to 15 million spores The sample masses were con-sistent with a previous determination of spore density at0081 gmL that indicated a high degree of entrained air
All chemicals used to prepare the silver-doped sol-gelcoated capillaries were also obtained and used as receivedfrom Sigma-Aldrich According to previously publishedprocedures17 two precursor solutions were preparedmixed and then drawn into 1-mm-diameter glass capil-laries The silver amine precursor consisted of a 51 vvratio of 1 N AgNO 3 to 28 NH3OH while the alkoxideprecursor consisted of a 21 vv ratio of methanol to te-tramethyl orthosilicate The alkoxide precursors weremixed with silver amine precursor in an 81 vv ratioApproximately 015 mL was drawn into the capillarycoating a 15-mm length After sol-gel formation the in-corporated silver ions were reduced with dilute sodiumborohydride which was followed by a water wash to re-move residual reducing agent
A 100 mL drop of a 50 mM DDA solution in ethanolpre-heated to 78 8C was added to each of the B cereusparticles to digest the spore coat After 1 minute the re-sultant solution was drawn into a SER-active capillarythat was immediately xed horizontally to an XY posi-tioning stage (Conix Research Spring eld OR) just in-side the focal point of an f 07 aspheric lens The lensfocused the beam into the sample and collected the scat-tered radiation back along the same axis A dichroic lter(Omega Optical Brattleborough VT) was used to re ectthe excitation laser to the lens and pass the Raman scat-tered radiation collected by the lens An f 2 achromatwas used to collimate the laser beam exiting a 200-mm-core-diameter source ber optic while a second f 2 ach-romat was used to focus the scattered radiation into a 365mm ber optic (Spectran Avon CT) A short-pass lterwas placed in the excitation beam path to block the sil-icon Raman scattering generated in the source ber fromre ecting off sampling optics and reaching the detectorA long-pass lter was placed in the collection beam pathto block the sample Rayleigh scattering from reachingthe detector A 785 nm diode laser (Process InstrumentsInc model 785-600 Salt Lake City UT) was used todeliver 100 to 150 mW of power to the sample A Fouriertransform Raman spectrometer (Real-Time Analyzersmodel IRA-785 East Hartford CT) and a silicon photo-avalanche detector (Perkin Elmer model C30902S Stam-ford CT) were used to acquire the SER spectra
RESULTS AND DISCUSSION
As an initial experiment the SER spectrum of 1 gLof DPA in water was measured using the newly devel-oped silver-doped sol-gel-coated capillaries (Fig 1A) Atthis concentration a high signal-to-noise ratio (SN) isobtained in 1 min In fact a reasonable spectrum is ob-tained in the same time frame for 1 mgL (Fig 1B) TheSER spectra are reasonably similar to the normal Raman(NR) spectrum obtained for a saturated solution of DPAin 1 N KOH (Fig 1C) and the following band shifts areobserved (NR to SER) 647 to 657 cm21 817 to 815cm21 998 to 1008 cm21 1384 to 1382 cm21 1434 to1428 cm21 and 1569 to 1567 cm21 Many of these bandshave been previously assigned1213 such as 998 cm21 to
APPLIED SPECTROSCOPY 353
FIG 1 SERS of DPA in water using silver-doped sol-gel-coated glasscapillary for (A) 1 gL and (B) 1 mgL (C ) NR of saturated DPA in 1N KOH in a glass capillary Spectral conditions (A) and (B) 150 mWof 785 nm 1-min acquisition time (C) 450 mW of 785 nm 5-minacquisition time both 8 cm 21 resolution
FIG 2 SERS of DPA extracted from 10 mg B cereus particle using100 mL of 50 mM hot DDA acquired in (A) 1 minute and (B) 2 seconds(C ) Attempted SERS of 50 mM hot DDA in ethanol using silver-dopedsol-gel-coated glass capillary acquired in 1 min Spectral conditions150 mW of 785 nm 8 cm21 resolution
the symmetric ring stretch 1384 cm21 to the OndashCndashOsymmetric stretch 1428 cm21 to the symmetric ring CndashH bend and 1569 cm21 to the asymmetric OndashCndashOstretch
The rst B cereus samples consisted of 2 mg of sporesin 2 mL of 5 mM hot DDA The samples were main-tained at 78 8C for 40 min and while hot approximately10 mL was drawn into a SER-active capillary Since spec-tra of DPA were obtained for these initial samples small-er spore masses higher DDA concentrations and shorterheating periods were examined In due course it wasfound that 10 mg of spores could be digested by 100 mLof 50 mM hot DDA in one minute and detected (Fig2A) In fact the signal was suf ciently intense that it canbe observed in as little as two seconds (Fig 2B) Theamount of DPA that was extracted was estimated to bebetween 5 and 10 mgL by comparing the signal intensityof the 1008 cm21 band to that measured for DPA in waterThis is consistent with previous research that found thatthe majority of the DPA is extracted from spores usingDDA11 and that B cereus spores contain approximately10 DPA by weight18 The SN of 127 for the 1008 cm21
band in the 1-minute SER spectrum suggests a limit ofdetection of approximately 250 ng of B cereus sporesbased on a SN of 3 Finally it should be noted that DDAdid not produce a detectable SER spectrum as shown inFig 2C
CONCLUSION
We have demonstrated that by combining rapid extrac-tion of dipicolinic acid from Bacillus cereus spores withchemical identi cation by surface-enhanced Raman spec-troscopy as little as 10 mg of spores can be detected Infact the entire measurement from the time of adding hotdodecylamine to the spores to the time when the dipi-colinic acid SER spectrum is acquired and analyzedcould be performed in less than two minutes The abilityof this method to distinguish between spore-forming bac-teria such as Bacillus anthracis and non-DPA containingpowders could help prevent costly shutdowns associated
with the appearance of suspicious material or intentionalmailing of common substances as an anthrax hoax Thismethod could also prove useful in detecting the locationof anthrax endospores in mail distribution facilities if an-other veri ed attack should occur
Research continues to fully characterize the surface-enhanced Raman spectroscopy signal intensities as afunction of sample concentration and to explore otherextractants that do not require the use of elevated tem-perature
ACKNOWLEDGMENTS
The authors are grateful for the support of the National Science Foun-dation (DMI-0296116 and DMI-0215819) and the US Army(DAAD13-02-C-0015 Joint Service Agent Water Monitor program)The authors are indebted to Chetan Shende for preparing the sol-gelcapillaries The authors also thank James Gillespie Nicholas Fell andAugustus Fountain for providing important background informationMark Farquharson for laboratory support and Professor Jay Sperry ofthe University of Rhode Island for supplying B cereus spores
1 V A Pasechnik C C Shone and P Hambleton Bioseparations 3267 (1993)
2 P J Jackson M E Hugh-Jones D M Adair G Green K K HillC R Kuske L M Grinberg F A Abramova and P Keim ProcNatl Acad Sci USA 95 1224 (1998)
3 B R Glick and J J Pasternak Molecular Biology Principles andApplications of Recombinant DNA (ASM Press Washington DC1994)
4 C A Bell J R Uhl T L Had eld J C David R F Meyer TF Smith and F R Cockerill III J Clin Microbiol 40 2897(2002)
5 D L Gatto-Menking H Yu J G Bruno M T Goode M Millerand A W Zulich Biosens Bioelectron 10 501 (1995)
6 J J Quinlan and P M Foegeding J Rapid Methods AutomationMicrobiol 6 1 (1998)
7 A A Hindle and E A H Hall Analyst (Cambridge UK) 1241599 (1999)
8 M S Ascher US Department of Health amp Human Services(httpwwwhhsgovophppresentationsAscherdoc)
9 R Nudelman B V Bronk and S Efrima Appl Spectrosc 54 445(2000)
10 D L Rosen C Sharpless and L B McBrown Anal Chem 691082 (1997)
354 Volume 58 Number 3 2004
11 P M Pellegrino N F Fell Jr and J B Gillespie Anal ChimActa 455 167 (2002)
12 W H Woodruff T G Spiro and C Gilvarg Biochem BiophysRes Commun 58 197 (1974)
13 E Ghiamati R S Manoharan W H Nelson and J F SperryAppl Spectrosc 46 357 (1992)
14 A P Esposito C E Talley T Huser C W Hollars C M Schal-dach and S M Lane Appl Spectrosc 57 868 (2003)
15 S Farquharson L Grigely V Khitrov W W Smith J F Sperryand G Fenerty J Raman Spectrosc paper accep ted (2003)
16 S Farquharson W W Smith S Elliott and J F Sperry SPIE-IntSoc Opt Eng 3855 110 (1999)
17 S Farquharson and P Maksymiuk Appl Spectrosc 57 479(2003)
18 F W Janssen A J Lund and L E Anderson Science (Washing-ton DC) 127 26 (1958)
SPIE -2003-5269 117
pH dependence of methyl phosphonic acid dipicolinic acid and cyanide by surface-enhanced Raman spectroscopy
Stuart Farquharson Alan Gift Paul Maksymiuk Frank Inscore and Wayne Smith
Real-Time Analyzers 87 Church Street East Hartford CT 06108
ABSTRACT US and Coalition forces fighting terrorism in Afghanistan and Iraq must consider a wide range of attack scenarios in addition to car bombings Among these is the intentional poisoning of water supplies to obstruct military operations To counter such attacks the military is developing portable analyzers that can identify and quantify potential chemical agents in water supplies at microgram per liter concentrations within 10 minutes To aid this effort we have been investigating the value of a surface-enhanced Raman spectroscopy based portable analyzer In particular we have been developing silver-doped sol-gels to generate SER spectra of chemical agents and their hydrolysis products Here we present SER spectra of methyl phosphonic acid and cyanide as a function of pH an important factor affecting quantitation measurements which to our knowledge has not been examined In addition dipicolinic acid a chemical signature associated with anthrax-causing spores is also presented Keywords Chemical warfare agents agent detection agent hydrolysis SERS Raman spectroscopy homeland security
1 INTRODUCTION In the past decade the Unites States and its allies have been challenged by a different kind of warfare exemplified by the terrorist attacks of September 11 2001 Suicide bombings and the use of chemical agents are the norm and military personnel must consider a wide range of attack scenarios Among these is the intentional poisoning of water supplies to obstruct military operations in Afghanistan and Iraq Currently colorimetric paper is used to detect agents on-site while gas chromatography combined with mass spectrometry (GCMS) is used in mobile support laboratories However both methods have severe drawbacks The paper changes color in response to contact with many chemicals besides CWAs causing a high incidence of false positives1 while GCMS although very chemically specific requires hours to perform and constant re-calibration234 Military operations would be greatly aided by a portable analyzer that can identify and quantify potential chemical agents at concentrations that impact safety This includes the analysis of drinking water supplies distribution and storage systems To meet this goal the Department of Defense has been investigating numerous approaches under the auspices of the Joint Service Agent Water Monitor (JSAWM) program5 The JSAWM must be capable of identifying the chemical agents with no false-positives and quantifying the agents at microgram per liter concentrations within 10 minutes678 Recently we and others have been investigating the ability of surface-enhanced Raman spectroscopy (SERS) to measure chemical agents9-12 bioagents13-17 and their hydrolysis products in water SERS employs the interaction of surface plasmon modes of metal particles with target analytes to increase scattering efficiency by as much as 1 million times18 In 1987 the potential of this approach to measure CWAs was demonstrated by measuring a series of pesticides19 Several of these organophosphonates have chemical structures similar to CWAs in particular P=O functional groups In our studies we have been employing silver-doped sol-gels to promote the SER effect The porous silica network of the sol-gel matrix offers a unique environment for immobilizing and stabilizing SER-active metal particles20-23 The sol-gels have been coated on the internal walls of standard glass vials to allow rapid SER analysis of small sample quantities without preparation We have measured over 100 chemicals with enhancements of 104 to 106 demonstrated reversible measurements in a flowing system reproducible measurements from vial-to-vial and measurements in multiple solvents including water20-23 Previously we used these vials to perform preliminary measurements of cyanide (CN) methylphosphonic acid (MPA) and dipicolinic acid (DPA) MPA is a hydrolysis product of the nerve agents (eg sarin Reaction 1) and may be a valuable indicator of nerve agent usage particularly since the alkyl methylphosphonic acids are relatively more stable than their corresponding parent complexes24 DPA is
SPIE -2003-5269 118
a chemical signature of spore forming bacteria such as Bacillus anthracis And in light of the inability to rapidly detect the anthrax spores distributed through the US mail in October 2001 a number of methods are being developed to extract and analyze this signature Reaction 1 Stepwise hydrolysis of Sarin to form hydrofluoric acid (HF) isopropyl methylphosphonic acid (IMPA) then methyl phosphonic acid (MPA) and 2-propanol In our previous SERS investigations MPA and DPA were measured at 50 and 100 mgL respectively In both cases limits of detection (LOD) were estimated at 100 microgL providing encouragement in that SERS may satisfy the needs of the JSAWM Since it has been shown that pH can substantially influence the intensity of SER bands25 which would clearly influence quantitative analysis we undertook the present study to determine the severity of these effects for cyanide methyl phosphonic acid and dipicolinic acid Furthermore we previously observed a band at 1050 cm-1 for MPA6 possibly due to an anion formed at basic pH Here we investigate the source of this spectral anomaly
2 EXPERIMENTAL All chemicals including potassium cyanide methyl phosphonic acid dipicolinic acid and those used to prepare the silver-doped and gold-doped sol-gel coated vials were obtained and used as received from Sigma-Aldrich All samples were prepared in HPLC grade water (Fischer Scientific Fair Lawn NJ) for SERS measurements The pH of these samples was adjusted using dilute nitric acid or potassium hydroxide and verified using a pH electrode (Corning Inc Corning NY) that had been calibrated with pH 400 700 and 1000 buffered standards from Fischer Scientific Once prepared the samples were transferred into the silver-doped sol-gel vials (Real-Time Analyzers Inc East Hartford CT) The vials were coated in a manner similar to that previously reported by adding ammonium hydroxide to a solution of silver nitrate tetramethyl orthosilicate (TMOS) and methanol20 Gold-vials were coated by adding nitric acid to a solution of gold tetrachloride TMOS and methanol The two precursor solutions were prepared mixed and transferred to 2-mL glass vials dried and heated After sol-gel formation the incorporated metal ions were reduced with dilute sodium borohydride (1mgmL) which was followed by a water wash to remove residual reducing agent After the resultant analyte solution was introduced the SER-active vial was immediately fixed horizontally to an XY positioning stage (Conix Research Springfield OR) just inside the focal point of an f07 aspheric lens The lens focused the beam into the sample and collected the scattered radiation back along the same axis A dichroic filter (Omega Optical Brattleborough VT) was used to reflect the excitation laser to the lens and pass the Raman scattered radiation collected by the lens An f2 achromat was used to collimate the laser beam exiting a 200 microm core diameter source fiber optic while a second f2 achromat was used to focus the scattered radiation into a 365 microm fiber optic (Spectran Avon CT) A short pass filter was placed in the excitation beam path to block the silicon Raman scattering generated in the source fiber from reflecting off sampling optics and reaching the detector A long pass filter was placed in the collection beam path to block the sample Rayleigh scattering from reaching the detector A 785 nm diode laser (Process Instruments Inc model 785-600 Salt Lake City UT) was used to deliver 100 to 150 mW of power to the sample A Fourier transform Raman spectrometer (Real-Time Analyzers model IRA-785 East Hartford CT) and a silicon photo-avalanche detector (Perkin Elmer model C30902S Stamford CT) were used to acquire the SER spectra
3 RESULTS AND DISCUSSION In a previous study of MPA6 aimed at developing a concentration calibration curve and determining limits of detection (LOD) we observed an anomaly at 1050 cm-1 Since it was found that the band intensity changed as a function of concentration the band must be associated with a sample parameter Two possible parameters photon flux and pH are examined here The first parameter was investigated by irradiating a 1mgmL MPA sample in a SER-active vial with laser powers of 200 mW and above and monitoring spectral changes It was immediately found that the 1050 cm-1 band
2O+ H HF + +OH
OH3H C
OP
OF
CH
CH3
3
3H C
OCP
OOH
CH
CH3
3
3H C
O
CPHO
CH
CH
3
3
C
Sarin IMPA MPA 2-propanol
SPIE -2003-5269 119
appeared and grew as a function of time and that the higher the laser power the faster the growth Figure 1 shows the growth of the 1050 cm-1 band over the course of 30 minutes when using 150 mW of 785 nm excitation while Figure 2A shows that the growth can be fit with a first order exponential equation namely I1050 = 03+05e-013t Furthermore the 760 cm-1 band could be fit with a first order decay equation with an identical exponential rate constant ie I760 = 08-
08e-013t The rates represent classical first order kinetics and their correspondence allows one to conclude that MPA is being transformed one-for-one into a photo-generated product At this time the photoproduct has not been positively identified but phosphonic (phosphorous) acid and phosphonate are likely candidates since the symmetric P(OH)3 stretch occurs at ~1050cm-1 Our studies show that a reduction in laser power to 100 mW at the sample for MPA essentially eliminates this degradation process This laser power or lower was used for further measurements
Methyl phosphonic acid is a diprotic acid that stepwise dissociates into two anions MPA- and MPA= according to the following reactions26
MPA MPA- + H+ pKa1 = 212 Reaction 2
MPA- MPA= + H+ pKa2 = 729 Reaction 3 The relative concentrations of MPA MPA- and MPA= can be determined at any pH by expressing [MPA] and [MPA=] in terms of [MPA-] using Reactions 2 and 3 and summing all three to equal the total starting concentration here 2 mgmL (0021M MW = 9602) viz
[MPA] + [MPA-] + [MPA=] = 0021M Equation 1 substituting from Reactions 1 and 2
([H+][MPA-])K1a + [MPA-] + (K2a[MPA-])[H+] = 0021M Equation 2 rearranging [MPA-] = 0021M(1+[H+]K1a + K2a[H+]) Equation 3 The relative concentrations of MPA MPA- and MPA= as a function of pH are shown in Figure 3 It is worth noting that near neutral pH both MPA- and MPA= will be present To confirm that the SER signal followed this pH dependence a starting solution consisting of 20 mg of MPA in 10 mL HPLC grade water was prepared and brought to pH of 20 using dilute nitric acid From this solution 2 mL were added to a SER-active vial and the SER spectrum recorded At this pH a peak at 760 cm-1 was barely discernable The 2 mL solution was returned to the starting solution and the pH was re-measured to correct for any changes that the silver-doped sol-gel vials might cause In most cases the change was less than 02 pH units and the pH is reported as the before and after average Next the pH of the
Figure 2 A) Exponential growth of 1050 cm-1 band and B) exponential decay of 760 cm-1 band for spectral series in Figure 1
Figure 1 Growth of 1050 cm-1 band as a function of time due to exposure to 150 mW of 785 nm Spectra are 5 sec each collected every 100-sec from 0 to 30-min
B
570 770 970 1170 1370Raman Shift (cm-1)
Arbit
rary
Unit
s
MPA Photodegradation
600 800 1000 1200 1400 Wavenumber (cm-1)
A
0 10 20 30 0 10 20 30 time (min) time (min)
Ram
an In
tens
ity (r
elat
ive)
30
min
0
SPIE -2003-5269 120
starting solution was adjusted to 325 using dilute KOH Again 2 mL were added to a vial and the SER spectrum recorded At this pH a reasonably strong 760 cm-1 band was observed This process was repeated as spectra were recorded at pHs of 70 74 75 79 85 and 100 A total of 1 mL of KOH was added diluting the total concentration by 10 Next the pH of the starting solution was made acidic by adding dilute nitric acid dropwise This time spectra were recorded at pHs of 72 69 64 and 37 Figure 4 shows the SER spectra for representative pHs (spectra were left out to simplify the figure) while Figure 3 shows the 760 cm-1 peak intensities as a function of pH (The band intensities were adjusted to compensate for dilution effects caused by the addition of HNO3 and KOH then normalized to 0021 M for the most intense band observed at pH 37) It is clear from Figure 3 that the 760 cm-1 band follows the MPA- concentration as a function of pH and must be assigned to this anion No bands were observed that corresponded to MPA or MPA= The lack of an MPA SER spectrum may be due to the absence of an attraction between the neutral analyte and the electropositive silver surface The same reasoning suggests that a strong SER spectrum should be observed for MPA= but it is not and a satisfactory explanation has not been found
0000
0005
0010
0015
0020
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14pH
Con
cent
ratio
n [M
]
pK1 = 212 pK2 = 729
MPA- MPA=
MPA
Figure 4 SER spectra of 002M MPA as a function of pH Conditions 100 mW of 785 nm 36 scans (1 min) 8 cm-1 recorded 2 min after sample introduction pH 19 and 20 not apparent on this scale pH 69 and 74 near identical to 70 and 75 and not shown for clarity
00
02
04
06
08
10
12
14
16
18
0 5 10 15 20 25 30Measurement Number
Ram
an In
tens
ity (7
60 c
m-1
)
Figure 3 Concentration dependence of MPA MPA- and MPA= as a function of pH for a 002M sample Intensity of 760 cm-1
band from Figure 6 as a function of pH ( for increasing basic adjustment for increasing acidic adjustment error was measured at ~10 for pH 64)
Figure 5 SER spectra of 002M MPA at pH 64 measured around a vial at three heights (9 points per height) Conditions as in Figure 4 but 10-sec scans
Since these measurements involved the removal and replacement of the SER-active vial in the sample holder to remove and add sample variation in the intensity as a function of vial position was minimized by illuminating the exact same height along the vial wall But this does not account for variability of the SERS response of the sol-gel coating around the vial To analyze this effect a vial containing MPA at pH 64 was rotated at ~ 40o
intervals at the original height and 18rdquo above and below this value Figure 5 shows the intensity of the 760 cm-1
band for the 27 positions It was found that the average value was 137plusmn014 an RSD of 10 overall and 5 for each height An error bar is included in Figure 4 for the pH 64 measurement
SPIE -2003-5269 121
With the above analyses in mind a preliminary investigation of the SER spectral response for MPA (as MPA-) as a function of concentration was performed A single vial was used for these measurements beginning with 1 mgL followed by measurements of 10 100 and 1000 mgL In all cases the pH was ~7 and 3 positions around the vial were measured per concentration Since the 760 cm-1 band was not observed for concentrations of 1 or 10 mgL using 100 mW of 785 nm the laser power at the sample was raised to 200 mW beginning with the 10 mgL concentration Photo-degradation was largely avoided (and not observed) by exposing the sample for only 33 seconds per spectral acquisition Representative spectra for 10 100 and 1000 mgL are shown in Figure 6 while a plot of the 760 cm-1 band intensity as a function of concentration is shown in Figure 7 These values were also used to estimate limits of detection based on the signal-to-noise ratio (SN) of the 760 cm-1 band using 100 mW of 785 nm laser excitation and a 10-min acquisition time As summarized in the Figure 7 inset the lower the measured concentration the lower the predicted LOD Even if as estimated 210 microgL could be measured using the silver-doped sol-gel vials an improvement of a factor of 70 is still required to meet JSAWM goals of measuring 3 microgL in 10-minutes Similar to MPA DPA is a diprotic acid (pKa1 = 216 and pKa2 = 692) and variations in pH will effect the relative concentrations of DPA DPA- and DPA= and possibly the SER spectra and band intensities This could prove significant if an acid or base is used to denature anthrax spores with the goal of extracting and analyzing DPA The MPA pH study described above was mimicked for DPA except that the starting solution consisted of 20 mg of DPA in 20 mL HPLC grade water (60x10-3M MW = 1671) The initial solution had a pH of 245 which was made basic by dilute KOH to pHs of 355 433 487 559 1069 and 1166 SER spectra were recorded at each pH using 100 mW of 785 nm and a 44-sec acquisition time Next one drop of concentrated nitric acid was used to remake the solution acidic at a pH of 200 Again sequential pH measurements were performed at 383 510 735 and 822 The solution pH was made acidic a third time but to pH 219 171 then 135 Throughout this process no more than 20 drops of acid or base were added and therefore the concentration was diluted by no more than 10 Most of the spectral bands showed a minor decrease in intensity as a function of increasing pH values However the bands at 525 and 795 cm-1 showed the most dramatic changes which occurred at acid pH Figure 8 shows the SER spectra of DPA for the spectral region and pH range of interest The identity of the DPA species was determined by plotting the normalized peak intensities with the lowest value set to 0 and the highest to 0006 M as a function of pH and overlaying these values on a plot of the relative concentrations for DPA DPA- and DPA= as previously done for MPA (Figure 9) As can be seen the 525 cm-1 band clearly corresponds to DPA The correspondence of the 795 cm-1 band to this species is less clear as the band retains intensity until pH of 55 This can be attributed to contribution to the overlapping band at 810 cm-1 which does not change as a function of pH The fact that most bands are observed at all pHs suggest that the primary interaction with silver is through the ring nitrogen This is supported by the fact that the most intense band occurs at 1008 cm-1 attributed to a symmetric ring breathing mode and that this interaction has been characterized for pyridine in numerous papers27
0
1
2
3
4
5
6
7
0 200 400 600 800 1000 1200
MPA Concentration (mgL)
760
Ban
d In
tens
ity
Figure 6 SER spectra of MPA in water at A) 1000 B) 100 and C) 10 mgL Conditions pH of 7 silver-doped sol-gel coated vial 200 mW of 785 nm 33-sec 8 cm-1 resolution
Figure 7 Plot of SER intensity of 760 cm-1 band of MPA as a function of concentration using 200 mW of 785 nm Inset table includes average intensity LOD standard deviation and percent deviation for each concentration but for 100 mW and 10 min
conc (mgL) LOD ave stddev dev10 021 002 1021100 072 010 14301000 312 040 1280
A
B C
SPIE -2003-5269 122
The ability to detect dipicolinic acid (DPA) by SERS is demonstrated in Figure 10 Even at 1 mgL the primary bands are visible The intensity of the dominant ring breathing mode observed at 1008 cm-1 for DPA in water (pH = 25 to 55) is plotted as a function of concentration in Figure 11 These values were also used to estimate limits of detection based on the SN of the 1008 cm-1 band using 100 mW of 785 nm laser excitation and a 10-min acquisition time Again the lower the measured concentration the lower the predicted LOD (see Figure 11 inset) and detection of 160 microgL is possible Gastrointestinal anthrax requires significant more spores than inhalation anthrax28 and a limit of detection might be placed at 1 million spores in 1 liter of water or 10 microgL Since each spore contains ~10 CaDPA by weight29 a goal for DPA might be 1 microgL indicating that the present measurements must be improved by nearly two orders of magnitude Due to the increased hazards of handling HCN gas KCN salt was used for these experiments Nevertheless all sample preparations were performed in a chemical hood KCN completely dissolves in water but its conjugate acid HCN is formed and has a Ka of 615x10-1030 viz
HCN CN- + H+ pKa = 921 Reaction 4
Figure 8 SER spectra of 1 mgmL DPA as a function of pH Conditions 100 mW 785 nm 100 scans (44-sec)
Figure 9 Concentration dependence of DPA DPA- and DPA= as a function of pH for a 0006M sample Intensity of 525 () and 795 (diams) cm-1 bands from Fig 8 as a function of pH
pH
135 171 219 383
0
0001
0002
0003
0004
0005
0006
0007
0 2 4 6 8 10 12 14pH
Con
cent
ratio
n [M
]
DPADPA-DPA=795525
DPA DPA=DPA-
pK2 = 692pK1 = 216
795 525
0
05
1
15
2
25
0 200 400 600 800 1000 1200
DPA Concentration (mgL)
1008
Ban
d In
tens
ity
conc (mgL) lod-10min-100mw1 017
10 016100 103
1000 355
Figure 10 SER spectra of DPA in water at A) 1000 B) 100 C) 10 and D) 1 mgL Conditions pH of 25-55 silver-doped sol-gel coated vial 175 mW of 785 nm 1-min 8 cm-1 D) has been multiplied by x10 to make bands visible
A
B C
D
Figure 11 Plot of SER intensity of 1008 cm-1 band of DPA as a function of concentration using 175 mW of 785 nm Inset table includes LOD in mgL for each concentration but for 100 mW and 10 min
SPIE -2003-5269 123
Consequently the cyanide concentration must be determined for each initial KCN concentration Specifically the samples prepared with concentrations of 01 1 10 100 and 1000 mgL of KCN produced CN- concentrations of 63x10-3 033 69 89 and 964 mgL at pHs of 816 90 967 102 and 107 respectively The pH dependence for the HCN and CN- concentrations are shown in Figure 12 Thus as the amount of KCN added to the solution decreases so does the pH of the solution (becomes less basic) and according to Reaction 4 the relative amount of CN- to HCN also decreases For example in the preparation of a 01 mgL solution of KCN the pH is shifted from 7 for pure water to only 816 and only 63 of the starting material becomes CN- or 63x10-3 mgL In comparison for a solution of 1000 mgL the pH is shifted from 7 to 107 and 96 of the starting material becomes CN- This is significant since the cyanide ion is better able to adsorb onto the silver particles and become SERS active SER spectra of 10 100 and 1000 mgL of KCN samples are shown in Figure 13 and a corresponding plot of the 2100 cm-1 band intensity as a function of concentration is shown in Figure 14
000102030405060708091011
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14pH
Con
cent
ratio
n [m
gm
L]
pKa = 921
CN -HCN
The SER spectra of cyanide are dominated by the single CN stretch at 2100 cm-1 which occurs in normal Raman spectra of solutions at 2080 cm-1 However a low frequency mode occurs at 135 cm-1 due to a Ag-CN stretch (not shown) indicative of a strong surface interaction It is also observed that as the concentration decreases the CN stretch shifts to 2140 cm-1 This shift has been attribute to the formation of a tetrahedral Ag(CN)3
2- surface structure31 as well
Figure 14 Concentration dependence of KCN SERS measured under conditions in Fig 11 Concentrations are 1 01 and 001 mgml Intensities are measured for the CN stretch at 2100 cm-1 Inset table includes LOD in mgL for each concentration in Figs 13 and 15 but for 100 mW and 10 min
Figure 13 SER spectra of KCN in water at A) 1000 B) 100 and C) 10 mgL Conditions pHs of 107 102 and 97 silver-doped sol-gel coated vial 100 mW of 785 nm 1-min 8 cm-1 C) has been multiplied by x10 to make band visible
Figure 12 Concentration dependence of HCN and CN- as a function of pH for a 1 mgmL sample Calculated intensity of 2100 () cm-1 band for a 1 mgml sample at pHs of 816 90 967 102 and 107
A
B
C
0
20
40
60
80
100
120
140
0 200 400 600 800 1000 1200CN Concentration (mgL)
2100
Ban
d In
tens
ity
Figure 15 SER spectra of KCN in water at A) 10 B) 1 and C) 01 mgL Conditions pHs of 97 12 and 12 gold-doped sol-gel coated vial 100 mW of 785 nm 1-min 8 cm-1 C) has been multiplied by x10 to make band visible
A
B
C
conc (mgL) lod-10min-100mw Condition01 001 Au-pH 121 007 Au-pH 12
69 003 Au-pH 9769 016 Ag-pH 9789 022 Ag-pH 102946 113 Ag-pH 107
SPIE -2003-5269 124
as to CN adsorbed to two different surface sites32 Alternatively the 2140 cm-1 band could be attributed to HCN since this species dominates at lower concentrations However it is unlikely that this species would be attracted to the electropositive silver surface Further both peaks should be present at pHs between 85 and 105 but this is not observed It has also been suggested that at concentrations near and above monolayer coverage the CN- species is forced to adsorb end-on due to crowding and at lower concentrations the molecule can reorient to lie flat33 This suggests that the 2100 and 2140 cm-1 bands correspond to the end-on and flat orientations respectively As Figures 13 and 14 show the intensity of the CN stretch for the 89 mgL sample is nearly as intense as the 964 mgmL band This suggests that the Raman signal for the flat orientation is more enhanced However more extensive measurements are required to verify this point Since resent research has suggested that cyanide may be more effectively detected on gold measurements of KCN solutions were also performed using gold-doped sol-gel vials Preliminary measurements are shown in Figure 15 for samples prepared from 01 1 and 10 mgL KCN Since the pHs are 816 90 967 the resultant CN- concentrations are 63x10-3 033 and 69 mgL Initially only the highest concentration was observed and the signal intensity was significantly better than the equivalent concentration measured using silver In an effort to shift Reaction 4 to the left transforming HCN to CN- (Le Chatelierrsquos principle) KOH was added to the lower concentration samples producing solutions with pH 12 This effectively forces all of the cyanide in solution to be CN- or 01 and 10 mgL respectively More importantly the CN stretch is now observed in the SER spectra The band appears at 2125 cm-1 as has been previously reported for gold12 As calculated for MPA and DPA LODs can be estimated from this data For the three concentrations of cyanide on silver the LODs are 016 to 11 mgL for 100 mW of 785 nm laser excitation and a 10-min acquisition time For gold pH adjusted the LODs are10 to 70 microgL an improvement of more than 10 times silver Nevertheless either substrate is sufficient to meet the JSAWM goals of measuring 3 mgL in 10-minutes as the requirements form cyanide are much less stringent than the nerve agents
4 CONCLUSIONS Here we examined the surface-enhanced Raman spectral response for methyl phosphonic acid dipicolinic acid and cyanide as a function of pH It was determined that the most prominent peak at 760 cm-1 reaches a maximum intensity between pH 3 and 7 and corresponds to the MPA- species Neither the MPA nor MPA= species appear to generate a SER spectrum and consequently no spectra were observed below pH 2 or above pH 8 In this study we also found that higher laser powers could cause photodegradation of MPA signified by the exponential growth of a band at 1050 cm-1 which is tentatively assigned to phosphorous acid Unlike MPA DPA was observed at all pHs This is attributed to the dominant interaction of the pyridine functional group with silver Minor spectral changes were observed at acid pHs and were assigned to neutral DPA Like MPA SER spectra of cyanide were pH dependent No spectra were observed for the HCN species while CN- was best observed at pHs more basic than 8 Preliminary concentration studies for the three analytes allowed estimating limits of detection for MPA DPA and CN using 100 mW of 785 nm and a 10-min acquisition time of 210 165 and 70 microgL respectively Although the latter value suggests that cyanide can be measured at sufficiently low concentrations to meet JSAWM goal improvements by 100 to 200 times are required for MPA and DPA It is clear from this study that pH of the sample is important and must be taken into account when developing concentration calibration curves as well as fieldable analyzers Future work will include tailoring the sol-gel with specific functional groups to dictate sample pH and thereby optimize sensitivity
ACKNOWLEDGMENTS The authors are grateful for the support of the US Army (DAAD13-02-C-0015 Joint Service Agent Water Monitor program) The authors would also like to thank Dr Steve Christensen of the US Army SBCCOM for helpful discussions and Mr Chetan Shende of Real-Time Analyzers for assistance in development of the gold-doped sol-gels
REFERENCES 1 Erickson B Analytical Chemistry News amp Features June 1 397A (1998)
SPIE -2003-5269 125
2 Johnston RL Hoefler CM Fargo JC and Moberley B AT-ONSITE 5-8 (1994) 3 Black RM RJ Clarke RW Read and MT Reid J Chromatography 662 301-321 (1994) 4 USA Army Center for Health Promotion and Preventive Medicine (USACHPPM) Technical Guideline 230A 5 Jensen JL NC Wong W R Loerop SPIE 4775-03 (2001) 6 Hoenig SL Handbook of Chemical Warfare and Terrorism Greenwood Press Wesport CT (2002) 7 Munro NB SS Talmage GD Griffin LC Waters AP Watson JF King and V Hauschild Env Health
Persp 107 933-974 (1999) 8 Holstege CP Kirk M Sidell FR Crit Care Clin13 923-42 (1997) 9 Farquharson S P Maksymiuk K Ong and S Christesen SPIE 4577 166-173 (2001) 10 Lee Y and S Farquharson SPIE 4378 21-26 (2001) 11 Spencer KM J Sylvia S Clauson and J Janni SPIE 4577 158-165 (2001) 12 Tessier P S Christesen K Ong E Clemente A Lenhoff E Kaler and O Velev Applied Spectroscopy 56
1524-1530 (2002) 13 Farquharson S WW Smith S Elliott and JF Sperry SPIE 3533 207-214 (1998) 14 Farquharson S WW Smith S Elliott and JF Sperry SPIE 3855110-116 (1999) 15 Farquharson S WW Smith YH Lee S Elliott and JF Sperry SPIE 4575 62-72 (2002) 16 Guzelian AA J Sylvia J Janni S Clauson and KM Spencer SPIE 4577 182-192 (2001) 17 Shende C F Inscore A Gift P Maksymiuk and S Farquharson in press 18 Weaver MJ S Farquharson and MA Tadayyoni J Chem Phys 82 4867-4874 (1985) 19 Alak AM and T Vo-Dinh Analytical Chemistry 59 2149-2153 (1987) 20 Lee Y and S Farquharson SPIE 4206 140-146 (2000) 21 Farquharson S and Y Lee SPIE 4200-16 (2000) 22 Lee Y S Farquharson and P M Rainey SPIE 3857 76-84 (1999) 23 Lee Y S Farquharson H Kwong and M Shahriari SPIE 3537 252-260 (1998) 24 Wang J M Pumera G Collins and A Mulchandani Analytical Chemistry 74 6121-6125 (2002) 25 Dou X YM Jung Z-Q Cao and Y Ozaki Applied Spectroscopy 53 1440-1447 (1999) 26 Data supplied by S Christesen and K Ewing 27 Kerker M and B Thompson Eds SPIE MS 10 (1990) 28 Inglesby TV DA Henderson JG Bartlett JAMA 287 2236 (2002) 29 FW Janssen AJ Lund and LE Anderson Science 127 26 (1958) 30 Lide DR Ed Handbook of Chemistry and Physics CRC Press 77th Ed (1996-7) 31 Billmann J G Kovacs and A Otto Surf Sci 92 153 (1980) 32 Murray CA and S Bodoff Phys Rev B 32 671 (1985) 33 Kellogg D and J Pemberton J Phys Chem 91 1120 (1987)
SPIE-2003-5269 16
Chemical agent detection by surface-enhanced Raman spectroscopy
Stuart Farquharson Alan Gift Paul Maksymiuk Frank Inscore and Wayne Smith
Real-Time Analyzers 87 Church Street East Hartford CT 06108
Kevin Morrisey and Steven D Christesen US Army SBCCOM Aberdeen Proving Ground MD 21010
ABSTRACT
In the past decade the Unites States and its allies have been challenged by a different kind of warfare exemplified by the terrorist attacks of September 11 2001 Although suicide bombings are the most often used form of terror military personnel must consider a wide range of attack scenarios Among these is the intentional poisoning of water supplies to obstruct military operations in Afghanistan and Iraq To counter such attacks the military is developing portable analyzers that can identify and quantify potential chemical agents in water supplies at microgram per liter concentrations within 10 minutes To aid this effort we have been investigating the value of a surface-enhanced Raman spectroscopy based portable analyzer In particular we have been developing silver-doped sol-gels to generate SER spectra of chemical agents and their hydrolysis products Here we present SER spectra of several chemical agents measured in a generic tap water Repeat measurements were performed to establish statistical error associated with SERS obtained using the sol-gel coated vials Keywords Chemical agents chemical agent detection SERS Raman spectroscopy
1 INTRODUCTION
In the autumn of 2001 terrorism within US borders became a sobering reality While extensive efforts are being implemented to secure the homeland US and Coalition forces in Afghanistan and Iraq are constantly faced with terrorist attacks In addition to car-bombings the military has identified several non-traditional attack scenarios including poisoning of water supplies by chemical warfare agents (CWAs) To counter this threat the Department of Defense has been investigating numerous approaches to detect such attacks under the auspices of the Joint Service Agent Water Monitor (JSAWM) program1 The JSAWM must be capable of identifying the chemical agents with no false-positives and quantifying the agents at microgram per liter concentrations within 10 minutes2 This includes the analysis of drinking water supplies distribution and storage systems Currently colorimetric paper is used to detect agents on-site while gas chromatography combined with mass spectrometry (GCMS) is used in mobile support laboratories However both methods have severe drawbacks The paper changes color in response to contact with many chemicals besides CWAs causing a high incidence of false positives3 while GCMS although very chemically specific requires up to an hour to perform and regular re-calibration456 Vibrational spectroscopy has also been investigated as a potential method of rapidly detecting CWAs7-
11 as early as 1985 in the case of infrared spectroscopy8 and 1988 in the case of Raman spectroscopy11 Again however these techniques also have limitations when it comes to measuring trace poisons in water Infrared spectra would be dominated by the very strong absorption of water which would obscure absorptions by most other chemicals present Whereas Raman spectroscopy is simply not a very sensitive technique and detection limits are typically grams per liter Surface-enhanced Raman spectroscopy (SERS) offers several advantages over conventional vibrational methods and that may provide the necessary sensitivity required for detecting trace quantities of chemical agents in water SERS employs the interaction of surface plasmon modes of metal particles with target analytes to increase scattering efficiency by as much as 1 million times12 In 1987 the potential of this approach to measure CWAs was demonstrated by measuring a series of pesticides13 Several of these organophosphonates have chemical structures similar to CWAs in particular P=O functional groups In the past few years we and others have further explored the ability of SERS to detect CWAs14-17 and even bioagents 18-21 We have been employing silver-doped sol-gels to promote the SER effect
SPIE-2003-5269 17
in these studies The porous silica network of the sol-gel matrix offers a unique environment for immobilizing and stabilizing SER-active metal particles22-25 The silver-doped sol-gels have been coated on the internal walls of standard glass vials to allow rapid SER analysis of small sample quantities without preparation We have measured over 100 chemicals with enhancements of 104 to 106 demonstrated reversible measurements in a flowing system reproducible measurements from vial-to-vial and measurements in multiple solvents including water21-25 Previously we used these vials to perform preliminary measurements of cyanide (CN) and methylphosphonic acid (MPA) Most of the nerve agents form MPA during hydrolysis while Tabun forms CN a chemical agent in its own right In another paper including in these proceedings we examined the limits of detection (LOD) for MPA by measuring a series of concentrations down to 50 mgL and estimated a limit of detection of 100 microgL26 These measurements provide encouragement in that SERS may satisfy the needs of the JSAWM To further establish the viability of SERS in particular silver-doped sol-gels here we present analysis of cyanide mustard and VX in tap water The measurements performed at the US Armyrsquos Edgewood Chemical Biological Center Aberdeen MD also included numerous repeat measurements to establish reproducibility
2 EXPERIMENTAL 2a General All chemicals including potassium cyanide 2-chloroethylethyl sulfide and those used to prepare the silver-doped and gold-doped sol-gel coated vials were obtained and used as received from Sigma-Aldrich All samples were prepared in a chemical hood using HPLC grade water unless otherwise noted (Fischer Scientific Fair Lawn NJ) for SERS measurements Once prepared the samples were transferred into the silver-doped sol-gel vials (Real-Time Analyzers Inc East Hartford CT) The vials were coated in a manner similar to that previously reported by adding ammonium hydroxide to a solution of silver nitrate tetramethyl orthosilicate (TMOS) and methanol22 The two precursor solutions were prepared mixed and transferred to 2-mL glass vials dried and heated After sol-gel formation the incorporated metal ions were reduced with dilute sodium borohydride (1mgmL) which is followed by a water wash to remove residual reducing agent After the resultant analyte solution was introduced the SER-active vial was fixed horizontally to an XY positioning stage (Conix Research Springfield OR) just inside the focal point of an f07 aspheric lens The lens focused the beam into the sample and collected the scattered radiation back along the same axis A dichroic filter (Omega Optical Brattleborough VT) was used to reflect the excitation laser to the lens and pass the Raman scattered radiation collected by the lens An f2 achromat was used to collimate the laser beam exiting a 200 microm core diameter source fiber optic while a second f2 achromat was used to focus the scattered radiation into a 365 microm fiber optic (Spectran Avon CT) A short pass filter was placed in the excitation beam path to block the silicon Raman scattering generated in the source fiber from reflecting off sampling optics and reaching the detector A long pass filter was placed in the collection beam path to block the sample Rayleigh scattering from reaching the detector A 785 nm diode laser (Process Instruments Inc model 785-600 Salt Lake City UT) was used to deliver 100 to 150 mW of power to the sample A Fourier transform Raman spectrometer (Real-Time Analyzers model IRA-785 East Hartford CT) and a silicon photo-avalanche detector (Perkin Elmer model C30902S Stamford CT) were used to acquire the SER spectra
2b Edgewood Chemical Biological Center The surface-enhanced Raman spectral measurements at the US Armyrsquos Chemical Biological center presented here were all performed on September 12 2003 To expedite measurements a plate was machined to hold up to 12 SER-active sample vials (Figure 1) The plate fit a standard XY plate reader that could be programmed Pure KCN bis-(2-chloroethyl)sulfide (distilled mustard gas HD) and ethyl S-2-diisopropylamino ethyl methylphosphonothioate (VX) were obtained on-site and used to prepare 1 gL tap water solutions in a chemical hood with appropriate safety equipment Simulated tap water was prepared by adding 10 chemicals most often found in tap water at appropriate concentrations to distilled water (Table 1) SER measurements were also performed in a chemical hood For added safety the FT-Raman instrument was placed outside the laboratory and 30 foot fiber optic and electrical cables were used to allow remote SERS measurements and plate manipulation For each experiment 1gL samples were prepared and added to 9 individual vials which were then loaded on the plate In some cases a tenth vial was included as a blank
Table 1 Generic Tap Water Composition
Compound mgL NaHCO3 100 CaSO4 27 MgSO4bull7H2O 67 NaNO3 1 Fulvic Acid 1 K2HPO4 07 KH2PO4 03 (NH4)2HSO4 001 NaCl 001 FeSO4 0001 pH 76-78
SPIE-2003-5269 18
A software program was written that allowed selecting the sequence that the vials were measured the number of positions along the length of the vials to measure (1 to 5) and the number of scans to co-add During sample analysis the program displayed the vial being analyzed the point being analyzed and the spectrum as it was being acquired Once all the data was collected a second software program was written to rapidly analyze the data The spectra collected for all the vials on a plate could be loaded at one time and then the spectra for each point could be displayed simultaneously or separately The user could then select the Raman peak to analyze in terms of peak height or area This was accomplished by selecting points on either side of the peak to define a baseline of zero The peak height or area could then be computed for all of the spectra loaded and then exported to a spreadsheet for statistical analysis
Figure 1 A) Vial Holder 6 slots to hold 2 vials each end-to-end B) Measurement Configuration Program user selects vials to measure sequence number of points per vial (1 to 5) and number of scans per point C) Spectral Acquisition Program shows spectrum being collected which vial and position D) Spectra Analysis Program user selects spectra to analyze by plate vial and point (s) as well as two wavenumbers defining the peak and the baseline to subtract The image is of 5 repeat measurements of 10 mgL KCN in generic tap water 16 sec each 100 mW of 785 nm
3 RESULTS AND DISCUSSION Raman and surface-enhanced Raman spectra were obtained for potassium cyanide bis-(2-chloroethyl)sulfide and ethyl S-2-diisopropylamino ethyl methylphosphonothioate representing three classes of chemical agents cyanides mustards and nerve agents respectively Spectra were also obtained for 2-chloroethyl ethyl sulfide (CEES) a structural analogue to HD which was included in the study to aid in assigning spectral bands KCN salt was used for cyanide experiments to avoid the increased hazards of handling HCN gas KCN completely dissolves in water forming its conjugate acid HCN according to its Ka of 615x10-1027 and at a concentration of 1 mgmL results in a pH 107 solution This is significant in that only CN- appears to interact sufficiently with silver to produce a SER spectrum and no spectral signal is observed below pH 726 Figure 2 shows the SER and normal Raman spectra for KCN The SERS spectrum of 1mgml KCN in water shows a single intense somewhat broad feature at
A B
C D
SPIE-2003-5269 19
2100 cm-1 assigned to the single CequivN stretch The band is much sharper in the normal Raman spectra of the solid KCN salt at 2074 cm-1 This band does broaden and shift to 2080 cm-1 in solution (not shown) However the observed SERS frequency is attributed to interaction with silver and low frequency mode at 135 cm-1 attributed to a Ag-CN stretch (not shown) supports this conclusion
cm-1 band to a C-S stretch but the authors concede that it is in fact more likely a C-Cl stretch28 It appears that the most intense Raman bands at 648 692 and 747 cm-1 shift to 620 660 and 720 cm-1 in the SER spectra and are tentatively assigned as above The width of these bands suggests that they overlap underlying spectral features Additional bands in the Raman spectra occur at 972 1034 1049 1263 1286 1423 1442 2865 2935 and 2960 cm-1 Corresponding bands occur in the SER spectra at 964 1015 1054 1286 1410 1447 2865 and 2935 cm-1 Most of these bands are associated with alkane modes specifically the bands at approximately 1040 cm-1 to a C-C stretch 1290 cm-1 to a CH2 in-phase twist 1440 cm-1 to a CH2 wag 2865 cm-1 to a symmetric CH2 stretch and 2965 cm-1 to an asymmetric CH2 stretch The Raman and SER spectra of sulfur mustard were measured at the Edgewood center (Figure 4) Both spectra are largely similar to CEES The C-Cl and C-S bands in the Raman spectrum of HD now occur at 640 655 700 739 and 760 cm-1 and are more resolved possibly due to the increased molecular symmetry Theoretical calculations indicate that the first three bands are due to C-Cl stretching modes and the latter two to C-S stretching modes28 Only the C-Cl bands maintain significant intensity in the SER spectra occurring at 624 and 643 cm-1 which is attributed to the
Prior to measurements of HD CEES was examined by Raman and SER spectroscopy (Figure 3) CEES also known as half-mustard is essentially identical to HD except one of the chlorine end atoms is replaced by a hydrogen atom Again although not as toxic as HD CEES is a blister agent and dilute aqueous samples were prepared using appropriate safety equipment Both the Raman and SER spectra of CEES are similar and dominated by bands between 600 and 800 cm-1 These are associated with C-Cl and C-S stretching modes which are tentatively assigned to 648 and 747 cm-1 in the Raman spectra respectively The shoulder at 630 cm-1 the overlapped band at 660 cm-1 and the strong band at 692 cm-1 could also be due to these modes or their asymmetric counterparts It is worth noting that theoretical calculations assign the 692
Figure 2 A) SER and B) NR spectra of KCN Conditions A) 1 mgml in tap water 100 mW of 785 nm at sample 1-min acquisition time B) solid 300 mW of 785 nm 5-min All spectra are 8 cm-1 resolution
A
B
Figure 3 A) SER and B) NR spectra of CEES Conditions A) 1 vv (10 mgml) in MeOH 100 mW of 785 nm 1-min acquisition time B) neat 300 mW of 785 nm 5-min
A
B
Cl-CH2-CH2-S-CH2-CH3
A
B
Figure 4 A) SER and B) NR of HD Conditions A) 1mgml in tap water B) pure both 100 mW of 785 nm 1-min
Cl-CH2-CH2-S-CH2-CH2-Cl
SPIE-2003-5269 20
expected strong interaction between chlorine and silver and adds support to the assignment of this band to a C-Cl stretch Weaker overlapping bands occur at 670 692 and 724 cm-1 the latter possibly due to C-S stretching modes Again the alkane modes are apparent in the normal Raman spectra of HD but only a broad feature at 1300 to 1450 cm-1 suggests CH2 contributions in the SER spectrum Although the observed bands in the VX spectrum have not been assigned (Figure 4) a computer generated Raman spectrum29 predicts many of the same features with surprising accuracy and are used here Two intense bands at 460 and 530 cm-1 closely match predicted bands at 463 and 546 cm-1 assigned to a CH3-P=O bend and a PO2CS wag Three highly overlapped bands occur at 694 745 and 771 cm-1 matching predicted bands at 713 730 and 760 cm-1 The first
Table 2 Measured SER peak heights for the CN stretch at 2100 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
Figure 5 A) SERS and B) NR spectra of VX Conditions A) 1 vv (10 mgml) in MeOH B) pure sample both 100 mW of 785 nm 1-min acquisition time
A
B
two have been assigned to a C-S stretch and CH2bend respectively while the latter has been attributed to either a P-C stretch or an O-C-C stretch Although the 745 cm-1 band may alternatively be assigned to a C-S stretch based on the previous measurements of CEES and HD The relatively intense bands at 890 1106 1218 1445 and 1465 cm-1 also match predicted bands at 880 1108 1216 1440 and 1464 cm-1 that are assigned to a C-C stretch CH3 rock N-C3 stretch various C-H3 bends and C-H bends respectively Both the computer generated and the measured spectra contain numerous other less intense bands One is worth mentioning A unique band appears at 370 cm-1 that is predicted at 368 cm-1 and corresponds to an O-P=O bend The surface-enhanced Raman spectrum of VX is also rich with spectral features It has the unique low frequency band at 370 cm-1 as well as a second band at 380 cm-1 that is assigned tothe S-P-O bend predicted in the normal Raman spectrum at 388 cm-1 Based on the measured and predicted normal Raman spectra the following SERS assignments are given 460 cm-1 to the CH3-P=O bend 544 cm-1 to the PO2CS wag 738 cm-1 to a C-S stretch (based on arguments above) 890 cm-1
to a C-C stretch 1101 cm-1 to a CH3 rock and 1456
cm-1 to a C-H bend The ability of SERS to measure chemical agents in water containing real-world chemical interferents was tested by using the generic tap water described in Table 1 The ability to reproduce measurements was accomplished by preparing three separate water stock solutions which were used to prepare three sample solutions each of 1 mgmL KCN in the generic tap water The samples defined as vials 1-9 were then measured at 4 points per vial in a semi-random fashion such that errors associated with stock solution preparation and errors associated with instrument drift could be identified No trends were apparent that signified such systematic errors Each spectrum collected consisted of 20 averaged scans taking 16 seconds at 8 cm-1 resolution The laser power at the sample was measured periodically during the day and it ranged from 102 to 105 mW spacccc
spot Vial 1 Vial 2 Vial 41 13754 13004 128192 13519 12692 12909
1 3 13541 12721 126394 13462 12648 12651
ave 13569 12766 12755Vial 6 Vial 3 Vial 5
1 11586 14068 10757 2 11236 14402 115122 3 11376 14546 11559
4 10894 11763 11214ave 11273 13695 11261
Vial 7 Vial8 Vial 91 11198 15804 11023
2 11243 15716 88753 3 11562 15216 11218
4 11655 15084 11077ave 11415 15455 10548
CN AVG STDEV ERR1 mgmL pk ht 12526 1560 1245
stock solution
SPIE-2003-5269 21
The CN measurements consisting of 36 data points produced an average height of 125 for the 2100 cm-1 peak with a standard deviation of 156 or 125 (Table 2) The HD and VX measurements were performed precisely the same way (Tables 3 and 4) For HD the 624 cm-1 peak was used for analysis and it had an average height of 53 with a standard deviation of 068 or 129 while for VX the 544 cm-1 peak was used for analysis and it had an average height of 1051 with a standard deviation of 308 or 293 The greater error in the VX measurements can be somewhat attributed to Vial 4 which produced lower SER signal intensities But removing this vial from the data set changes the standard deviation to 233 only a modest improvement
spot Vial 1 Vial 2 Vial 41 525 609 4682 45 675 484
1 3 527 644 5094 559 807 669
ave 51525 68375 5325Vial 6 Vial 3 Vial 5
1 521 574 575 2 536 53 4492 3 509 508 451
4 565 594 379ave 53275 5515 4635
Vial 7 Vial 8 Vial 91 457 589 413
2 583 497 4273 3 544 505 4
4 528 5 509ave 528 52275 43725
HD AVG STDEV ERR1 mgmL pk ht 530 068 1291
stock solution
spot Vial 1 Vial 2 Vial 41 1464 1034 4672 1485 99 61
1 3 1491 989 5684 1041 777 553
ave 137025 9475 5495Vial 6 Vial 3 Vial 5
1 1058 942 1293 2 697 121 965
2 3 727 1206 11124 689 1254 746
ave 79275 1153 1029Vial 7 Vial 8 Vial 9
1 1127 783 163 2 1358 812 1614
3 3 1371 908 15124 1097 875 1397
ave 123825 8445 153825
VX AVG STDEV ERR1 mgmL pk ht 1051 308 2925
stock solution
4 CONCLUSIONS In this paper we examined the ability of surface-enhanced Raman spectroscopy to reproducibly measure CN HD and VX in tap water without chemical interference Both normal and surface-enhanced Raman spectra were examined to select unique bands suitable to identify and quantify these chemical agents For SER measurements the 2100 cm-1 C-N stretch was used for CN the 624 cm-1 C-Cl stretch was used for HD and the 544 cm-1 PO2CS wag was used for VX It was determined that 1 mgmL samples of each of these chemicals measured 36 times in glass vials coated with a silver-doped sol-gel reproduced measurements with standard deviations of 125 129 and 293 It was further found that the 10 chemicals added to simulate generic tap water did not interfere with or alter the SER spectra It should be noted that the concentrations used in this study were considerably greater than those required by the JSAWM program Current work involves improving SER sensitivity and designing sampling systems with better reproducibility This includes the development of fractal silver and gold structures within the sol-gel matrix and the development of chemically selective sol-gels
ACKNOWLEDGMENTS The authors are grateful for the support of the US Army (DAAD13-02-C-0015 Joint Service Agent Water Monitor program) The authors would also like to thank Janet Jensen Ronald Crosier and Kristina Gonser for helpful discussions
Table 4 Measured SER peak heights for the VX band at 544 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
Table 3 Measured SER peak heights for the HD band at 624 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
SPIE-2003-5269 22
REFERENCES 1 Jensen JL NC Wong W R Loerop SPIE 4775-03 (2001) 2 JSAWM Requirements at wwwsbccomapgeaarmymilRDAecbcrtPRODSERJSAWMjsawmhtml 3 Erickson B ldquoThe Chemical Weapons Convention Redefines Analytical Challengerdquo Anal Chem 70 397A (1998) 4 Johnston RL Hoefler CM Fargo JC and Moberley B ldquoThe Defense Nuclear Agencyrsquos
ChemicalBiochemical Weapons Agreements Verification Technology Research Development Test and Evaluation Program and its Requirements for On-Site Analysisrdquo AT-ONSITE 5-8 (1994)
5 Black RM Clarke RJ Read RW and Reid MT ldquoApplication of GC-MS and GC-tandem MS to the analysis of chemical warfare samples found to contain residues of the nerve agent sarin sulphur mustard and their degradation productsrdquo J Chrom 662 301-321 (1994)
6 USA Army Center for Health Promotion and Preventive Medicine (USACHPPM) Technical Guideline 230A 7 Hoffland LD Piffath RJ Bouck JB rdquoSpectral signatures of chemical agents and simulantsrdquo Optical
Engineering 24 982-984 (1985) 8 Braue EHJ Pannella MGrdquoCIRCLE CELL FT-IR Analysis of Chemical Warfare Agents in Aqueous Solutionsrdquo
App Spectrosc 44 1513-1520 (1990) 9 Tseng C-H Mann CK and Vickers TJ ldquoDetermination of Organics on Metal Surfaces by Raman
Spectroscopyrdquo App Spectrosc 47 1767-1771 (1993) 10 Staff Reporter ldquoUS Army Tests Lidar to Detect Biological Toxinsrdquo Photonic Spectra pg 50 December 1998 11 Christesen SD Raman cross sections of chemical agents and simulants App Spectrosc 42 318-321 (1988) 12 Weaver MJ Farquharson S Tadayyoni MA ldquoSurface-enhancement factors for Raman scattering at silver electrodesrdquo J Chem Phys 82 4867-4874 (1985) 13 Alak AM and Vo-Dinh T ldquoSERS of Organophosphorous Chemical Agentsrdquo Anal Chem 59 2149-2153 (1987) 14 Lee Y Farquharson S ldquoRapid chemical agent identification by SERSrdquo SPIE 4378 21-26 (2001) 15 Farquharson S Maksymiuk P Ong K Christesen S ldquoChemical agent identification by surface-enhanced Raman spectroscopyrdquo SPIE 4577 166-173 (2001) 16 Spencer KM Sylvia J Clauson S and Janni J ldquoSurface Enhanced Raman as a Water Monitor for Warfare
Agents in Waterrdquo SPIE 4577 158-165 (2001) 17 Tessier P Christesen S Ong K Clemente E Lenhoff A Kaler E Velev O ldquoOn-line spectroscopic characterization of sodium cyanide with nanostructured gold surface-enhanced Raman spectroscopy substratesrdquo
App Spectrosc 56 1524-1530 (2002) 18 Farquharson S W W Smith S Elliott and J F Sperry Rapid biological agent identification by surface- enhanced Raman spectroscopy SPIE 3855110-116 (1999) 19 Farquharson S WW Smith YH Lee S Elliott and J F Sperry Detection of biological signatures A
comparison of electrolytic and metal-doped sol-gel surface-enhanced Raman media SPIE 4575 62-72 (2002) 20 Guzelian AA Sylvia J Janni J Clauson S and Spencer KM ldquoSERS of whole cell bacteria and trace levels of biological moleculesrdquo SPIE 4577 182-192 (2001) 21 Shende C Inscore F Gift A Maksymiuk P Farquharson S ldquoRapid dipicolinic acid extraction from Bacillus spores detected by surface-enhanced Raman spectroscopyrdquo App Spectrosc 58 accepted 22 Lee Y and Farquharson S ldquoSERS Sample Vials Based on Sol-Gel Process for Trace Pesticide Analysisrdquo SPIE
4206 140-146 (2000) 23 Farquharson S and Lee Y ldquoTrace Drug Analysis by SERSrdquo SPIE 4200-16 (2000) 24 Lee Y Farquharson S and Rainey P M Surface-Enhanced Raman Sensor for Trace Chemical Detection in
Water SPIE 3857 76-84 (1999) 25 Lee Y Farquharson S Kwong H and Shahriari M ldquoSol-Gel Chemical Sensor for Surface-Enhanced Raman
Spectroscopyrdquo SPIE 3537 252-260 (1998) 26 Farquharson S A Gift P Maksymiuk F Inscore and W Smith ldquopH dependence of methyl phosphonic acid
dipicolinic acid and cyanide by surface-enhanced Raman spectroscopyrdquo SPIE 5269-19 (2003) 27 Lide DR Ed Handbook of Chemistry and Physics CRC Press 77th Ed (1996-7) 28 Sosa C RJ Bartlett K KuBulat and WB Person ldquoA theoretical study of harmonic vibrational frequencies and
infrared intensities of XCH2CH2SCH2CH2X and XCH2CH2SH (= H Cl)rdquo J Phys Chem 93 577-588 (1993) 29 Hameka HF and JO Jenson ldquoComputer-generated predictions of the structure and the IR and Raman spectra of
VXrdquo ERDEC-TR-065 May 1993
- Chemagents Appendicespdf
-
- SPIE2001-4575-Bioagent-SERSapdf
-
- Detection of bioagent signatures A comparison of electrolytic and metal-doped sol-gel surface-enhanced Raman media
- Stuart Farquharson Wayne Smith and Yuan Lee
-
- Real-Time Analyzers 87 Church Street East Hartford CT 06108
-
- ABSTRACT
- 4 CONCLUSIONS
- 5 ACKNOWLEDGEMENTS
- 6 REFERENCES
-
SPIE 2001-4575
67
Table 1 Comparison of E-SER and SG-SER Spectral Band Positions for the NA Bases and Adenosine Monophosphate
Adenine Cytosine Guanine Thymine Uracil AMP E-SERS SG-SERS E-SERS SG-SERS E-SERS SG-SERS E-SERS SG-SERS E-SERS SG-SERS E-SERS SG-SERS1647 1638 1634 1656 1655 1630 1587 1585 1510 1580 1551 1539 1456 1456 1465 1460 1462 1480 1453 1459 1394 1398 1425 1431 1435 1399 1404 1392 1374 1375 1373 1383 1370 1335 1332 1311 1307 1333 1331 1353 1348 1331 1329 1265 1273 1280 1292 1278 1276 1275 1279 1271 1183 1195 1222 1232 1221 1219 1204 1205 1180 1144 1097 1033 1029 1038 1040 1035 1051 1037 1041 1035 963 963 957 1001 1000 961 944 884 819 817 859 866 735 737 797 799 784 782 800 800 72738 742 630 630 653 664 667 684 603 602 590 611 466 561
Bands unique to SG-SERS The next chemical to be analyzed by both E-SERS and SG-SERS was adenosine monophosphate (AMP) The E-SER spectrum yields bands due to the adenine chemical functionality at 727 961 1233 1279 1331 1381 and 1486 cm-1 In addition phosphate bands are observed at 860 1097 1453 1587 and 1705 cm-1 (Figure 5) Other researchers have noted that the ribose component does not appear to contribute to the spectrum19 The AMP spectrum also changes as a function of potential As the electrode is swept more positive (here from -09 to -03VSCE) the phosphate bands at 860 1097 1453 and 1587 cm-1 increase in intensity compared to the adenine bands while a band at 1705 cm-1 appears The adenine bands at 1233 1381 and 1486 cm-1 virtually disappear These potential dependent spectral changes are consistent with earlier studies that show that phosphate is attracted to silver at potentials positive of the pzc but repelled at potentials negative of the pzc19
Wavenumbers (∆cm-1) Figure 3 E-SERS of A) 21x10-3M cytosine at -03VSCE 1000 scans B) ~10x10-5M guanine at -06VSCE 500 scans C) 23x10-3M thymine at -06VSCE 500 scans and D) 12x10-3M uracil at -093VSCE 500 scans All spectra at pH 92 750 mW 1064 nm at 8 cm-1
Wavenumbers (∆cm-1) Figure 4 SG-SERS of A) 21x10-3M cytosine 200 scans B) ~10x10-5M guanine 200 scans C) 23x10-3M thymine 200 scans and D) 12x10-3M uracil 500 scans All spectra at pH 92 75 mW 1064 nm at 8 cm-1
A
B
C
A
B
C
D D
SPIE 2001-4575
68
The SG-SER spectrum of AMP is considerably different The adenine bands virtually disappear except for the two primary bands which shift to742 and 1329 cm-1 While the phosphate band at 1459 cm-1 has gained considerable intensity In addition two new intense bands appear at 684 and 1539 cm-1 as well as a moderately intense band at 1180 cm-1 The SG-SER spectrum has greater similarity to the E-SER spectrum at -03VSCE and suggests that the silver particles embedded in the sol-gel behave as if at a potential positive of the pzc RNA samples extracted from E coli B subtilis and S aureus were next examined by both E-SERS and SG-SERS E-SER spectra of these samples yielded quality spectra in 10 minutes in which all of the major features can be identified (Figures 6 and 7) This includes guanine at 650 cm-1 adenine at 791 cm-1 cytosine and uracil combining at 790 cm-1 and phosphate at 1100 1335 (in combination with adenine and guanine) 1465 and 1570 cm-1 Surprisingly adenine which demonstrated the greatest surface-enhanced Raman effect does not dominate the ring-breathing mode portion of the spectrum The intensities of the other base-pairs bands are of the same order of magnitude This suggests that when the base-pairs are linked together as in RNA they are enhanced in concert In fact the relative intensities are very similar to a normal Raman spectrum of E coli RNA which shows the combined cytosine and uracil band at ca twice the intensity of the adenine band and ca four times the intensity of the guanine band Unfortunately this means that the independent enhancement factors for the NA bases can not be used to estimate relative concentrations For example the relative 791 and 734 cm-1 bands for B subtilis would indicate that the cytosine andor uracil concentration was at least 20 times the adenine concentration whereas each of the four RNA bases are known to contribute 15-35 Nevertheless it is worth noting that the three RNA samples yield different relative band intensities that were reproduced in numerous measurements Although the relative concentrations of the NA bases for these samples have not been determined these differences can be quantified If it is assumed that the 650 cm-1 band represents 25 guanine the 791 cm-1 band represents 25 adenine and the 790 cm-1 50 cytosine plus uracil in the E coli RNA spectrum then the relative concentrations can be estimated for the other RNA samples To aid this calculation the three spectra were normalized to the phosphate band at 1100 cm-1 which has been shown to correlate to the total phosphate concentration and can be used as an internal standard In addition a simple baseline correction was applied (Figure 7) This yields 15 adenine 30 guanine and 55 cytosine plus uracil for B subtilis RNA and 18 adenine 25 guanine and 57 cytosine plus uracil for S aureus RNA The average SN of these measurements was 26 with an average error of 8 of the value (SplusmnN) It is also worth noting that the three RNA spectra show a marked shift in a band near 825 cm-1 This band is assigned to the symmetric stretch of the O-P-O ester linkage9 The band appears at 815 cm-1 for S aureus shifting to 820 cm-1 for B subtilis and 830 cm-1 for E coli Others have used the normal Raman intensity of the band at 815 cm-1 as a direct indication of the amount of A-class helix present while the intensity of the band at 830 cm-1 has been used as a direct indication of the amount of B-class helix present However the latter is more associated with DNA than RNA
Wavenumbers (∆cm-1) Figure 5 E-SER spectra of 020 mgmL adenosine monophosphate at A) -03 and B) -09VSCE and C) SG-SER spectra Conditions sample in 01M KCl buffered to pH 92 A) and B) 750 mW C) 75 mW of 1064 64 scans (1-min) at 8 cm-1
Wavenumbers (∆cm-1) Figure 6 E-SERS of 01 mgmL RNA from E coli 02 mgmL RNA from B subtilis and 02 mgmL RNA from S aureus Conditions 01M KCl pH 92 -03VSCE 750 mW of 1064 nm 640 scans (10 min) at 8 cm-1
AMP RNA
E coli A
B
C
B subtilis
S aureus
SPIE 2001-4575
69
SG-SER spectra of reasonable quality were also obtained for E coli and B subtilis especially the latter (Figure 8) However the spectra differ substantially from the E-SERS of the same samples Both SG-SER spectra are dominated by adenine at 735 cm-1 and a band at 1030 cm-1 Although unassigned the latter does appear in the RNA E-SER spectra Bands at 1105 and 1565 cm-1 are likely due to phosphate while bands at 1320 and 1470 cm-1 are less confidently assigned to phosphate They are significantly less intense and somewhat shifted from their SG-SERS counterparts (1335 and 1455 cm-1) A band at 670 cm-1 may be due to guanine which was observed at 664 cm-1 for SG-SERS of the pure sample However the SG-SER spectrum of AMP also had an intense 667 cm-1 band A number of other bands occur at 890 1070 1165 1245 1290 1420 1505 cm-1 and remain unassigned The SG-SER spectra are somewhat disappointing in that only adenine and guanine contributions can be positively identified This limits the ability to determine relative NA base concentrations and distinguish bacterial RNA However several of the unassigned bands may be due to the bases (eg 1030 and 1420 cm-1 due to cytosine) Further experiments will be required to clarify this point
Ecoli
B subtilis
S aureas
P G
A C+U
OPO
A B
Wavenumbers (∆cm-1) Wavenumbers (∆cm-1) Figure 7 SER spectra of RNA from A) B subtilis with contributions indicated and B) E coli B subtilis and S aureus with baseline correction and peak positions used to calculate contributions indicated G = guanine A = adenine C+U = cytosine plus uracil P = phosphate (backbone) OPO = phosphate ester linkage (A- vs B-class helix)
Wavenumbers (∆cm-1) Wavenumbers (∆cm-1) Figure 9 E-SER (-03VSCE) and SG-SER spectra of RNA from A) E coli and B) B subtilis Sample conditions as in Figure 6 E-SER spectra at 750 mW SG-SERS at 75 mW
A B E-SERS
SG-SERS
E-SERS
SG-SERS
SPIE 2001-4575
70
A final comparison was made between the two SER methods by measuring dipicolinic acid (DPA) This chemical may be invaluable as a test for spore forming bacteria specifically B anthracis 50 to 90 of B anthracis sporilates During spore formation dipicolinic acid is synthesized and once completed 10-15 of the dry spore weight is composed of the Ca2+ complex located in the spore core42 Heating in water can be used to initiate germination at which point the exosporium breaks and releases the Ca dipicolinate which becomes dipicolinic acid in water The structure of this chemical strongly
4 CONCLUSIONS Towards the goal of developing a method and technology to rapidly detect and identify bioagents we have been investigating surface-enhanced Raman spectroscopy as a tool to measure relative concentrations of nucleic acid bases in RNA and determine bioagent taxonomy Initially we investigated E-SERS since this method has been extensively researched and the optimum sample conditions are well developed However this method requires a three-electrode sample cell and electrolyte solution Incorporation of an electrolytic cell into sample systems used to analyze for BWA as aerosols in water or on surfaces can be designed using flow injection analysis technologies but cross-contamination and plugging of sample lines seems inevitable For this reason we also investigated metal-doped sol-gels as a SER-active medium Previous studies have shown this material to be active in all solvents particularly water capable of continuous measurements in flowing systems and reproducible (quantitative) between coated sample vials Here we compared SG-SER spectra to traditional E-SER spectra of the nucleic acid base pairs adenosine monophosphate and RNA High quality spectra of adenine cytosine guanine thymine and uracil were obtained by both E-SERS and SG-SERS Both methods yielded very similar spectra for the NA bases including a pH dependent study of adenine Enhancement factors and detection limits for adenine were determined as 2x105 and 16x10-11g and 1x105 and 12x10-10g for E-SERS and SG-SERS respectively (normalized to 75 mW and 10-min acquisition time) Fifty percent of the silver particle surface area in the sol-gel matrix was assumed covered by adenine which may have been overestimated yielding a lower EF and higher detection limit It should also be realized that each E-SER spectrum required several attempts to optimize the measurement conditions (pH electrode potential etc) While each SG-SER spectrum involved no sample preparation and often represents the first and only attempt to make the measurement Quality spectra of RNA extracted from Escherichia coli Bacillus subtilis and Staphylococcus aureus were obtained by E-SERS that were easily interpreted Bands due to adenine guanine cytosine plus uracil and phosphate were identified The SER band intensity of the NA bases in the RNA samples were of the same order of magnitude suggesting that their interaction with the silver surface is concerted as is their Raman enhancement Interestingly the relative SER band
Wavenumbers (∆cm-1) Figure 10 A) Raman spectrum of solid dipicolinic acid B) Ax20 C) electrolytic SERS of 6x10-3 M dipicolinic acid in 01 M KCl at a potential of +07VSCE and pH of 4 and D) sol-gel SERS of 6x10-3 M dipicolinic acid Conditions for A and C as in Figure 1 C) 100 mW of 1064 nm 50 scans 8 cm-1
A B (Ax20)
C
D
suggested that it would be SER active However the E-SER spectrum was unstable and varied considerably as a function of potential A consistent spectrum was obtained at +06VSCE (Figure 10) This potential is not recommended for measurement because the surface is actively dissolving in solution The SG-SER spectrum was considerably more stable of higher quality and easily reproduced Bands at 660 825 1010 1390 1430 1570 1590 and 3075 cm-1 were observed Enhancement factors were determined for the two media using the symmetric ring stretching mode at 995 and 1010 cm-1 for the normal Raman and SER spectra respectively E-SERS yielded an EF of 5x103 while SG-SERS yielded an EF of 2x105 for DPA The SN of the latter suggests a detection limit of 20x10-10g (based on adenine coverage 75 mW and 10-min) The differences in SER activity for these two media may be attributed to the combined electrolytic potential of the solution chemical and metal15 Again the E-SERS suggests that the SG-SERS is at a potential positive of the pzc While the instability in the E-SERS may also be associated with surface interactions of two carboxylic acid groups of dipicolinic acid during the ORCs
DPA
SPIE 2001-4575
71
intensities for RNA extracted from E coli are very similar to those measured by normal Raman spectroscopy Although the relative percent that each of the NA bases contributed to each RNA sample was not determined reproducible band intensities allowed noting the following trends The percent adenine decreases while the combined percent cytosine and guanine increase for both B subtilis and S aureus compared to E coli Quality spectra were also obtained for the RNA samples by SG-SERS but only a few bands were readily identified Calculations of NA base concentrations by SG-SERS will require further research In light of recent events we summarize the capabilities of these SERS media in terms of rapid detection of B anthracis and dipicolinic acid However these capabilities must be qualified First and foremost the level to which SERS can distinguish bacteria or viruses has not yet been determined Development of a database of both DNA and RNA base concentrations for BWAs and common bacteria to establish the level of taxonomic identification is ongoing Second rapid collection of aerosol water or surface samples is being addressed by others who report trapping particles on filters from 100 liters of air per minute Third although not presented here we have developed methods to extract RNA or DNA from cells and spores for SER analysis within 10 minutes Finally we assume a detection limit of 3600 spores per 100 liters of air is required although a 50 lethal dosage of anthrax has not been established With these qualifications a mass detection limit for RNA using SERS is estimated as follows A single measurement is performed in ca 20 minutes (140 liters collected in 14 min RNA extracted in 8 min spectral acquisition and analysis in 10 min) The average human breaths 7 liters per minute therefore the analyzer must at the very minimum detect 5000 spores in 140 liters of air One spore is approximately 2x10-18m3 (1x1x2 microm) and if a density of 075 gcm3 is assumed this corresponds to a mass of 15x10-12g Each spore contains 4-12 RNA or 12x10-13g RNA for 8 If we assume 23 of the RNA can be isolated for analysis during lysis then the proposed instrument must be able to detect 4x10-10g RNA from 5000 spores per 70 liters of air within 10 minutes As noted above the mass detection limits for adenine were estimated at 16x10-11g and 12x10-10g for E-SERS and SG-SERS respectively Although these detection limits suggests that RNA from 5000 spores is detectable with the current instrumentation it is highly likely that only a portion of an RNA segment (eg 120-nucleotide 5S rRNA) is in contact with the metal surface and will contribute to the SER effect The SN for the RNA spectra were 110th of the average SN for the four individual RNA bases suggesting a 10 contribution Furthermore effective taxonomy will likely require knowing the NA base concentrations to 1 of the value (eg 25plusmn025) Again the average measurement error for the bases is 12 These values suggest that the E-SERS is within a factor of 4 of the required detection limit whereas the SG-SERS detection limit must be improved by 25 times The same arguments can be applied to the detection of dipicolinic acid If we assume a spore releases 10 by weight DPA during germination then the proposed instrument must be able to detect 75x10-10g DPA from 5000 spores per 70 liters of air within 10 minutes The detection limit for SG-SERS was estimated at 20x10-10g and suggest that the vials are suitable to perform a rapid screen for anthrax A series of concentration dependent measurements are currently being performed to verify this assertion Finally we note that the measurements performed here employed an FT-Raman spectrometer This instrumentation was chosen over dispersive Raman instrumentation for the high wavelength accuracy afforded by the HeNe reference laser (Connes Advantage43) which would allow reliable spectral subtraction matching of observed spectra to stored library spectra and confident use of chemometric approaches Such data analysis is likely to be required to enhance BWA identification However this instrumentation sacrifices sensitivity due to a less efficient detector (InGaAs vs Si) less efficient Raman scattering and less efficient generation of plasmon modes Substantial improvements can be obtained using visible excitation and Si detection and these measurements are underway
5 ACKNOWLEDGEMENTS The authors are grateful to Drs D Cookmeyer and S Tove of the US Army Research Office (Contract Number DAAH04-96-C-0078) for their interest and support of this research The authors would also like top acknowledge Dr R Yin and J Jensen for supporting development of the metal-doped sol-gels (Contract Number DAAD13-01-C-0019) They also thank Dr Wilfred H Nelson for assistance in spectral interpretations
SPIE 2001-4575
72
6 REFERENCES 1 Roberts WL and Rainey PM Clin Chem 39 1872-1877 (1993) 2 Pasechnik VA CC Shone and P Hambleton Bioseperations 3 267-283 (1993) 3 Jackson PJ ME Hugh-Jones DM Adair G Green KK Hill CR Kuske LM Grinberg FA Abramova and P
Keim Proc Natl Acad Sci 95 1224-1229 (1998) 4 Jensen JL NC Wong W R Loerop SPIE 4775-03 (2001) 5 Snyder A Pet al SPIE 3853-15 (1999) 6 Danosky T R and McFadden P N in press (1997) 7 Woodruff WH Farquharson S Science 201 831-833 (1978) 8 Lord RC and Thomas GJJr Spectrochemica Acta 23A 2551-2591 (1967) 9 Thomas GJJr Biochim Biophys Acta 213 417-423 (1970) 10 Placzek G Handbuch der Radiologie 2 EMarx ed Akademische Verlagagescellschatt Liepzig 1934 UCRL
Trans No 526 (1959) 11 Jeanmaire DL and RP Van Duyne J Electroanalytical Chem 84 1-20 (1977) 12 Chada S Manoharan R Moenne-Loccoz P Nelson WH Peticolas WL and Sperry JF Applied Spectroscopy
47 38-43 (1993) 13 Manoharan R Ghiamati E Chada S Nelson WH and Sperry JF Applied Spectroscopy 47 2145-2150 (1993) 14 Todd EA Morris MD Applied Spectroscopy 48 545-548 (1994) 15 Wentrup-Byrne E Sarinas S and Fredericks PM Applied Spectroscopy 47 1192-1197 (1993) 16 Weldon MK VR Zhelyaskov and MD Morris Applied Spectroscopy 52 265-269 (1998) 17 Kneipp K and J Fleming J Mol Structure 145 173-179 (1986) 18 Pothier NJ and Force RK Applied Spectroscopy 46 147-151 (1992) 19 Ervin KM E Koglin JM Sequaris P Valenta and HW Nurnberg J Electroanal Chem 114 179-194 (1980) 20 Kim SK TH Joo SW Suh and MS Kim J Raman Spectrosc 17 381-386 (1986) 21 Pothier NJ and Force RK Analytical Chemistry 62 678-680 (1990) 22 Kneipp K Y Wang RR Dasari and MS Feld Applied Spectroscopy 49 780-784 (1995) 23 Nie S and Emory SR Science 275 1102 (1997) 24 Graham D WE Smith AMT Linacre CH Munro ND Watson and PC White Analytical Chemistry 69 4703-
4707 (1997) 25 Fleischmann M PJ Hendra and AJ McQuillan Chem Phys Lett 26 163-166 (1974) 26 Pettinger B J Chemical Phys 85 7442-7451 (1986) 27 Surface-Enhanced Raman Scattering Section Four Theory SPIE MS 10 M Kerker and B Thompson Eds (1990) 28 Wang D-S and Kerker M SPIE (M Kerker and B Thompson Eds) MS 10 417-429 (1990) 29 Weaver MJ Farquharson S Tadayyoni MA J Chem Phys 82 4867-4874 (1985) 30 Norrod KL Sudnik LM Rousell D and Rowlen KL Applied Spectroscopy 51 994-1001 (1997) 31 Farquharson S Weaver WJ Lay PA Magnuson RH and Taube H J Am Chem Soc 105 3350-3351 (1983) 32 Lee PC and Meisel D ldquoAdsorption and Surface-Enhanced Raman of Dyes on Silver and gold Solsrdquo J Phys Chem
86 3391-3395 (1982) 33 Li Y-S and Wang Y Applied Spectroscopy 46 142-146 (1992) 34 Bello JM DL Stokes and T Vo-Dinh Analytical Chemistry 61 1779-1783 (1989) 35 van Duyne RP J C Hulteen D A Treichel M T Smith M L Duval and T R Jensen J Phys Chem B1033854-
3863 (1999) 36 Farquharson S and W W Smith W H Nelson and J F Sperry SPIE 3533-27 207-214 (1998) 37 Lee Y H W Smith S Farquharson H C Kwon M R Shahriari and P M Rainey SPIE 3537 252-260 (1998) 38 Lee Y-H S Farquharson and P M Rainey SPIE 3857 76-84 (1999) 39 Current Protocols in Molecular Biology Wiley Interscience1003-1006 (1987) 40 Farquharson S Smith W Carangelo R C and Brouillette C SPIE 3859 14-23 (1999) 41 Chase D B and JF Rabolt Fourier Transform Raman Spectroscopy Acad Press Ch1 p 131 (1994) 42 Brock TD MT Madigan JM Martinko and J Parker Biology of Microorganisms 7th Ed Prentice Hall p 76-80
(1994) 43 Connes J Rev Opt Theor Instrum 40 45 (1961)
166
Chemical agent identification by surface-enhanced Raman spectroscopy
Stuart Farquharson and Paul Maksymiuk
Real-Time Analyzers 87 Church Street East Hartford CT 06108
Kate Ong and Steven D Christesen US Army SBCCOM Aberdeen Proving Ground MD 21010
ABSTRACT The recent distribution of anthrax through the US postal system and the subsequent infection and death of several postal and national media employees amplifies the need for methods to rapidly detect identify and quantify this and other chemical and biological warfare agents The US military has also identified water supplies as a likely method of warfare agent deployment and is funding the development of a Joint Service Agent Water Monitor (JSAWM) In an effort to aid military personnel and the public at large we are developing a portable analyzer capable of identifying and quantifying chemical agents rapidly either on-demand or continuously The approach is based on the ability of Raman spectroscopy to identify molecular structure through the abundant vibration information provided in spectra and the ability of surface-enhanced Raman spectroscopy (SERS) to detect extremely low concentrations (eg part-per-billion) through the enhancement of Raman scattering by six orders of magnitude or more A key element to the analyzer design is a new SER active medium that is capable of quantitative reversible measurements The medium consists of silver or gold nanoparticles incorporated into a sol-gel matrix The porous silica network offers a unique environment for stabilizing SER active metals and the high surface area increases the interaction between the analyte and metal particles Here we present the use of new sol-gels that also selectively enhance chemicals based on polarity and charge Base-line measurements of chemical agents and their hydrolysis products are presented and compared to the JSAWM goal of 30 micrograms per liter detection Keywords Chemical warfare agent hydrolysis product SERS Raman spectroscopy sol-gel nanoparticle
1 INTRODUCTION Since September 11 2001 the threat of terrorist attacks and biological warfare within US borders has become a sobering reality The simplicity in manufacturing ease of deployment and the relatively low cost of chemical warfare agents (CWAs) raises public concern that they may also be used by terrorists Indeed terrorists released sarin (GB) in the Tokyo subway in 19951 Countering terrorism and terrorist attacks requires recognizing likely deployment scenarios and having the required technology to rapidly detect the deployment event One method of deployment has been long identified by the US military distribution through water supplies To counter this threat the Department of Defense is funding or monitoring the capabilities of 200 technologies with the goal of developing a Joint Service Agent Water Monitor (JSAWM) that is field portable2 The JSAWM must be capable of identifying the chemical agents with no false-positives and quantifying the agents at microgram per liter concentrations within 10 minutes (Table 1)3 This includes the analysis of drinking water supplies distribution and storage systems as well as potable water supplies The earliest technologies developed for CWA detection were based on electrochemical ionization or colorimetric analysis (eg phosgene tape) Although these analyzers were easy to use they were not generally agent specific and suffered from false-positives4 More traditional laboratory methods have also been investigated and in particular combined gas chromatography and mass spectrometry (GCMS) has been very successful at eliminating false-positives56 However GCMS requires extraction repeated calibration and long analysis times (typically 20 to 60 minutes)6 making it labor intensive and less than desirable for field use More rapid analysis of agents in the solid liquid and gas phase has been To whom correspondence should be addressed emailfarqureal-time-analyzerscom
Vibrational Spectroscopy-based Sensor Systems Steven D Christesen Arthur J Sedlacek III EditorsProceedings of SPIE Vol 4557 (2002) copy 2002 SPIE middot0277-786X02$1500
167
Table 1 Chemical Agent Structures Hydrolysis Half-lives and JSAWM Thresholds Agent Short-Hand Chemical Structure Hydrolysis
Half-Life JSAWM
Thresholds
Sarin (GB) F-[O=P-CH3]-O-CH(CH3)2 213 hours 32 microgL Soman (GD) F-[O=P-CH3]-O-CH(CH3)-(C-(CH3)3) 23 hours 32 microgL Tabun (GA) (CH3)2-N-[O=P-CN]-O-C2H5 41 hours 32 microgL VX C2H5O-[O=P-CH3]-S-(CH2)2-N-(CH(CH2)2)2 821 hours 32 microgL EA2192 HO-[O=P-CH3]-S-(CH2)2-N-(CH(CH2)2)2 gt9 years 32 microgL Mustard (H) ClCH2CH2-S-CH2CH2Cl encapsulates 47 microgL Lewisite (L) ClCH=CH-As-Cl2 rapid 27 microgL HCN HCN rapid 20 mgL BZ C7NH12-O-[C=O]-COH(C6H5)2 23 microgL T-2 Toxin 87 microgL at pH 7 to 75 and 20 to 25 oC
demonstrated by vibrational spectroscopy7-10 Hoffland et al7 reported infrared absorbance spectra and absolute Raman cross sections for several chemical agents while Christesen measured Raman cross sections for sarin tabun mustard gas and VX (ethyl S-2-diisopropylamino ethyl methylphosphonothioate)11 Again however these techniques also have limitations Raman spectroscopy is simply not a very sensitive technique and detection limits are typically 01 (1000 ppm) While infrared spectroscopy would have limited value in analyzing poisoned water since the very strong infrared absorption of water would obscure most other chemicals present Nevertheless efforts to overcome these limitations have been demonstrated Braue and Pannella8 quantified the G-series nerve agents (tabun sarin and soman) in terms of infrared attenuated total reflectance using a circle-cell And Alak and Vo-Dinh demonstrated the possibility of surface-enhanced Raman spectroscopy (SERS) to identify CWAs by measuring several organophosphonates that simulate the nerve agents12 However quantitative measurements have not been demonstrated for the SER-active material used (silver coated on alumina particles) or other SER-active media13 Recently we developed silver-doped sol-gels to promote the SER effect14-17 The porous silica network of the sol-gel matrix offers a unique environment for stabilizing SER-active metal particles and the sol-gel provides a high surface area that effectively increases the number of molecules observed within the Raman scattering volume The silver-doped sol-gels have been coated on the internal walls of standard glass vials to allow rapid SER analysis of small sample quantities (lt 01 mL) without preparation We have used p-aminobenzoic acid (PABA) as a test chemical to demonstrate surface enhancements greater than 106 reversible measurements in a flowing system reproducible measurements from vial-to-vial and batch-to-batch and measurements in multiple solvents including water14-17 Recently we used these vials to measure Tabun (GB) and Sarin and several hydrolysis products pinacolyl methyl phosphonate (PMP from Soman) and methyl phosphonic acid (MPA from all G-agents Figure 1) Although a number of unique vibrational bands are observed (eg C-N stretch doublet and P-C stretch) the G-agents were only observed for 5 concentrations and all spectra required baseline corrections Figure 1 Surface enhanced Raman spectra of ~5 vv A) Tabun and B) Sarin C) 1 vv PMP and D) 10 ppm MPA using sol-gel sample vials 785 nm excitation 1-min scan and CCD detection Performed at Aberdeen Proving Ground
C A
Wavenumber (∆cm-1) Wavenumber (∆cm-1)
D B
790
545 C-N
2135 2190 P-C 770
760 1290
Proc SPIE Vol 4577
168
Nevertheless MPA was readily observed for a 10 ppm sample with an estimated detection limit of 04 ppm (based on a signal-to-noise ratio of 3 for the 760 cm-1 band intensity) This measurement provides encouragement in that SERS may satisfy the needs of a JSAWM Furthermore MPA is also a hydrolysis product of VX and V-gas and EA2192 (Figure 2) and may prove a valuable indicator of agent usage Figure 2 Hydrolysis of Sarin to form hydrofluoric acid (HF) methylphosphonic acid 1-methylethyl ester (MPAMME) methyl phosphonic acid (MPA) and 2-propanol With this initial albeit modest success we began analyzing chemicals with various sol-gel compositions that we have been developing Here we describe four sol-gel compositions that select for 1) polar-positive 2) polar-negative 3) weakly polar-positive and 4) weakly polar-negative chemical species The ability of these sol-gels to select and enhance Raman scattering is described for several test chemicals and MPA
2 EXPERIMENTAL The chemicals analyzed as well as all chemicals used to prepare the metal-doped sol-gels were obtained at their purest commercially available grade from Aldrich (Milwaukee WI) The sol-gel designed to select for polar-negative species was prepared from a silver amine complex tetramethyl orthosilicate (TMOS) and methanol After mixing 02 mL of the sol-gel solution was transferred into a glass vial (2 mL) dried and heated The incorporated silver ions were then reduced using dilute sodium borohydride The vials were washed and dried prior to the addition of a sample solution In a similar manner the sol-gel designed to select for polar-positive species was prepared from a gold salt TMOS and methanol The sol-gel designed to select for weakly polar-negative species was prepared from a silver amine complex tetraethyl orthosilicate (TEOS) and methanol And the last sol-gel designed to select for weakly polar-positive species was prepared from a gold salt TEOS and methanol All samples were prepared in a chemical hood and transferred into plain or SER-active vials for analysis Normal Raman spectral measurements employed 1-mL pure samples that were placed in a 1-cm3 cuvette and weighed This yielded a powder density that allowed accurate calculation of molecules in the optical collection field SERS measurements employed 1-mg sample per mL water concentrations unless otherwise stated Once prepared a 01 mL sample was placed into one of the four selective sample vials which in turn was placed into the sample compartment of a Raman spectrometer for analysis A prototype Fourier transform Raman spectrometer (RTA) was used for these measurements18 The system consisted of a NdYAG laser (Brimrose) for excitation at 1064 nm an interferometer built by On-Line Technologies (OLT East Hartford CT) for frequency separation an uncooled InGaAs detector for signal detection (RTA) and an Intel 400 MHz Pentium II based laptop computer (Dell Round Rock TX) for interferometric control data acquisition (OLT) and analysis (LabVIEW by National Instruments Austin TX) Additional components included a Notch filter (Kaiser Ann Arbor MI) and interferometer entrance and exit optics (Edmund Scientific Barrington NJ) Fiber optics were used to deliver the excitation beam to the sample and the scattered radiation to the interferometer (1 meter lengths of 200 and 365 micron core diameter respectively Spectran Avon CT) A second Notch filter (Kaiser) was used as a beam splitter to direct the excitation beam along the same axis as the collected radiation A microscope object (20x04 Newport Irvine CA) was used to focus the beam into the sample and to collect the scattered radiation back along the same axis In this co-axial backscattering arrangement the excitation beam was passed through the outside of a glass vial and focused onto the silver-doped sol-gel film (01-03 mm thickness) containing the sample
3 RESULTS AND DISCUSSION p-aminobenzoic acid (PABA) and phenyl acetylene (PA) and were used to refine the selectivity and SER-activity of the four different metal-doped sol-gels PABA is a popular chemical used to evaluate the performance of SER-active media Here the polar end groups can be used to test selectivity of the polar-negative and polar-positive sol-gels PA is essentially non-
2O+ H HF + +OH
OH3H C
OP
OF
CH
CH3
3
3H C
OCP
OOH
CH
CH3
3
3H C
O
CPHO
CH
CH
3
3
C
Sarin MPAMME MPA 2-propanol
Proc SPIE Vol 4577
169
polar but a high electron density in the cylindrical π cloud around the carbon-carbon triple bond allows testing the selectivity of the weakly polar-negative and weakly polar-positive sol-gels As Figure 3 illustrates PABA passes through the polar sol-gel and is enhanced by either the silver or gold particles At 1 mgml the concentration of neutral PABA is ca 20 times that of the ionized form (pKa = 48) For electropositive silver the PABA anion is expected to interact through the carboxylate group and the associated vibrational modes are expected to dominate the spectrum Conversely for electronegative gold either form of PABA is expected to interact through the amine group The clear differences in our spectra support this expectation Furthermore bands at 840 and 1405 cm-1 assigned to a COO- bend and stretch respectively are significantly more intense for silver than gold Additional bands at 1140 and 1195 cm-1 are assigned to CH bending modes while bands at 1450 1500 and 1605 cm-1 are assigned to ring vibrational modes A very similar SER spectrum for PABA on a silver-coated alumina substrate has previously been reported with similar assignments19 For the gold-doped sol-gel new bands appear at 690 1355 and 1585 cm-1 The first band is assigned to a ring-H bending mode the second band to a ring-N- stretching mode and the third band to a possible NH2 scissors mode or ring mode The second band is not observed in the normal Raman spectrum but infrared bands occur at this frequency for aromatic ring-secondary amine stretching modes The scissors mode occurs at this frequency in Raman spectra for several chemicals but is absent in the PABA Raman spectrum Alternatively this mode may be the1600 cm-1 ring mode that has been shifted by the gold interaction Again a very similar SER spectrum of PABA has been reported but surprisingly using silver (colloids)2021 not gold as the enhancement medium These researchers also assumed the primary interaction of PABA with silver was through the carboxylate anion and made assignments accordingly For example they assigned the 1359 cm-1 to a COO- stretch not to the amine group as we have They also favor the ring stretching mode assignment for the 1582 cm-1 band Finally it should be said that other researchers have argued that the most dominant band in the SER spectra at 1450 cm-1 a ring vibration mode suggests that PABA lies flat on the surface and the π-orbitals dominate the surface interaction22
Figure 3 SER spectra of A) PABA using polar-negative and B) polar-positive sol-gels and C) PA using weakly polar-negative and D) weakly polar-positive sol-gels PABA is 1 mgmL PA is 1 vv Spectral conditions 75 mw 1064 nm 100 scans (15 min) 8 cm-1 resolution Non-polar PA passes through the non-polar sol-gels and is also enhanced by both metals The spectra are easily understood For electropositive silver PA interacts through the cylindrical triple bond π electron cloud and a -CequivC- doublet occurs near 2000 cm-1 The interaction is reasonably strong since this band appears at 2112 cm-1 in the normal Raman spectrum For electronegative gold this interaction is unlikely and only very weak bands occur near 2000 cm-1 The remaining bands are at 1000 cm-1 1200 cm-1 doublet and 1595 cm-1 all appear in the normal Raman spectra at virtually the same frequencies and are assigned to the symmetric ring-breathing mode CH bending modes and the trigonal ring-breathing mode respectively The polarnon-polar selectivity of the polar-negative and weakly polar-negative sol-gels was tested by adding a 11 molar mixture of PABA and PA The selective enhancement is quite good (Figure 4) The spectrum obtained using the polar sol-gel represents 78 PABA and 22 PA while the spectrum obtained using the weakly polar sol-gel represents 9 PABA and 91 PA The band peak intensities at 2000 cm-1 for PA and 1450 cm-1 for PABA were used for these calculations and are expanded in Figure 4 for clarity
C A
Wavenumber (∆cm-1) Wavenumber (∆cm-1)
2NH COOH
D B
C CH
Proc SPIE Vol 4577
170
Figure 4 SERS of 11 MM of PABA and PA in A) polar-negative and C) weakly polar-negative sol-gels The lower traces compare the pure chemicals B) 1 mgml PABA in polar-negative sol-gel and D) 1 PA in weakly polar-negative sol-gel while the insets magnify the minority species for clarity (x5 in A and x10 in B) Spectral conditions as in Figure 3 Following this development of selective sol-gels that maintained SER activity we measured cyanide and MPA (Figure 5) Not surprisingly the best sensitivity for both hydrolysis products was obtained using the polar-negative sol-gel The interaction of the cyanide anion with the silver surface is sufficient to shift the CequivN stretch observed at 2080 cm-1 in the normal Raman spectrum to 2145 cm-1 in the surface-enhanced Raman spectrum Furthermore the band is substantially broadened This anion has been extensively studied by electrolytic SERS and this shift and broadening have been attributed to the formation of a tetrahedral Ag(CN)3
2- surface structure23 Figure 5 Surface-enhanced (upper traces) and normal Raman spectra (bottom traces) of A) CN- and B) MPA in silver-doped TMOS SERS conditions as in Figure 3 and 1 mgmL Note MPA yields two distinct spectra for neutral (top) and acidic pH (middle) The normal Raman spectra employed pure powders 500 scans and 900 mW of 1064 nm SER measurements of MPA with the polar-negative sol-gel yielded two unique spectral signatures that depended on solution pH (Figure 5) For more neutral solutions the P-C stretch of MPA at 762 cm-1 dominates and the CH2 stretch at 2922 cm-1 appears The SN is sufficiently high that the anti-Stokes Raman shift at -762 cm-1 is observed For deprotonated MPA an oxygen-surface mode appears at 325 cm-1 (as well as its anti-Stokes complement) suggesting a strong interaction This results in substantial enhancement of the P-O-C mode at 1051 cm-1 (upper trace) Others report that this mode dominates the infrared spectra of nerve agents measured in water8 Comparison of the two spectra suggests the following molecule-to-surface orientations The appearance of the oxygen-surface and P-O-C modes in the upper spectrum of Figure 5B indicates that the tetrahedral molecule interacts with the silver surface through the deprotonated oxygen and is oriented end-on The
A B
Wavenumber (∆cm-1)
C A
Wavenumber (∆cm-1) Wavenumber (∆cm-1)
D B
Wavenumber (∆cm-1)
Proc SPIE Vol 4577
171
dominance of the P-C and the CH2 stretches and the disappearance of the P-O-C mode in the upper spectrum suggest the molecule is oriented with the methyl-hydroxyl-hydroxyl face towards the surface However considerably more research must be performed to verify these points Both chemical forms were measured as a function of concentration to determine the onset of monolayer coverage and estimate expected detection limits (Figure 6) Below monolayer coverage the signal to concentration dependence should be linear and the SN of any spectral measurement in this range can be used to predict the detection limit In the spectra presented here the peak height was used as the signal while the noise as root-mean-squared (RMS) was measured between 4400-4600 cm-1 Since noise is distributed evenly throughout the spectrum when transformed this region was used since it does not have any contributions from signals or baseline offsets Figure 6 shows a series of spectra for MPA along with a plot of the 1050 cm-1 band intensity (the noise was identical for this series) as a function of sample concentration A clear discontinuity appears in the vicinity of 01 mgmL (19 ppm) indicating the onset of monolayer coverage A detection limit defined as a SN of 3 was calculated for the 01 and 005gmL samples at 24x10-4 and 25x10-4 gL respectively A more modest detection limit of 101x10-4 gL was obtained using the 760 cm-1 band in the second series of concentration measurements These detection limits correspond to 46 and 190 ppb for a 3-min scan and 75 mW of incident laser power
Figure 6 A) Concentration dependence of MPA SERS measured in silver-doped TMOS) B) Concentrations are 001 005 01 05 1 gL (188 94 188 94 188 ppm) I760 series (bull) and I1050 series (∆)
Enhancement factors for cyanide and methyl phosphonic acid can be determined by estimating the number of molecules contributing to the surface-enhanced and normal Raman spectra The enhancement factor EF can be defined by the following equation
EF = (ISERSINR)bull(MNRMSERS) bull(PNRPSERS) bull(TNRTSERS)12
where I is the spectral band intensity M is the sample mass P is the incident laser power and T is the measurement time (or number of scans) for the two measurements For the normal Raman spectra a cylindrical scattering volume is assumed based on the laser area (28x10-7m2 6x10-4m diameter spot) and the penetration depth (1x10-3 m)24 The density of KCN and MPA as powders were measured at 0572 and 0516 gcm3 indicating that 16x10-4 and 144x10-4 g produced the normal Raman signals in Figure 5 respectively The number of molecules contributing to the SER spectra are those on the silver particles embedded in the sol-gel The total silver surface area can be determined from the average particle size concentration and the scattering volume Previous scanning electron micrographs showed the average silver particle size to be 40 nm in diameter (335x10-23m3)17 The silver concentration is 012M based on the reactant molar concentrations and dilution factors And the scattering volume is 76x10-
11m3 again based on a cylindrical scattering volume defined by a laser area of 28x10-7m2 and a sol-gel thickness of 27x10-
4m This volume contains 123x10-6g of silver equivalent to 35x109 silver particles with a collective surface area of 18x10-
5m2 However it may reasonably be assumed that at least half of the silver surface is in contact with the sol-gel matrix and unavailable for analyte interaction If we assume monolayer coverage and that each CN molecule occupies 15x10-20m2 then approximately 62x1014 molecules or 27x10-8g of CN contribute to the SER spectrum (20x10-19m2 46x1013 molecules 74x10-9g for MPA) Accordingly the EF for cyanide equals 48x104 ((180599) bull(16x10-427x10-8) bull(90075) bull(500100)12) The EF for MPA is considerably higher at 87x106 ((60326) bull(144x10-474x10-9) bull(90075) bull(500200)12)
0
20
40
60
80
100
120
140
0 02 04 06 08 1 12[MPA] (mgmL)
I (76
0)
0
100
200
300
400
500
600
I (10
50)
Wavenumber (∆cm-1)
A B
Proc SPIE Vol 4577
172
4 CONCLUSIONS Here we present for the first time surface-enhanced Raman spectra of Tabun and Sarin obtained using silver-doped sol-gels However the inferior enhancement suggested employing SER-active sol-gels varying in composition to improve both detection limits and selectivity of the target analytes To this end we successfully demonstrated the capabilities of four sol-gels that select for 1) polar-positive 2) polar-negative 3) weakly polar-positive and 4) weakly polar-negative chemical species p-aminobenzoic acid was used to show that silver could be used to attract polar-negative chemicals or functional groups (carboxylate anion) and that gold could be used to attract polar-positive chemicals (amine) while a mixture of p-aminobenzoic acid and phenylacetylene was used to show that tetramethyl orthosilicate preferentially solvates polar chemicals and that tetraethyl orthosilicate preferentially solvates weakly polar chemicals This increased sample control was applied to cyanide and methyl phosphonic acid two hydrolysis products of chemical warfare agents Exceptional results were obtained for methyl phosphonic acid allowing measurement of 1x10-2 gL for a 3-min scan and 75 mW of incident laser power with an estimated detection limit of 245x10-4 gL and an enhancement factor of 87x106 However this detection limit is 76 times less sensitive than required for the JSAWM (32x10-6gL for the G-agents) Finally we note that the measurements performed here employed an FT-Raman spectrometer This instrumentation was chosen over dispersive Raman instrumentation for the high wavelength accuracy afforded by the HeNe reference laser (Connes Advantage25) which would allow reliable spectral subtraction matching of observed spectra to stored library spectra and confident use of chemometric approaches Such data analysis is likely to be required to identify the chemical agents as well as distinguish hydrolysis products However this instrumentation which employs 1064 nm excitation and InGaAs detection sacrifices sensitivity We believe that the measurement sensitivity can be improved by at least two orders of magnitude by using shorter laser excitation wavelengths (eg 532 nm) This would provide more efficient Raman scattering (fourth power dependence on laser excitation wavelength) more efficient generation of plasmon modes and allow using more efficient detector material (Si vs InGaAs) These modifications are underway
5 ACKNOWLEDGEMENTS The authors would like to thank Dr R Yin and J Jensen of the US Army for supporting this work (Contract Number DAAD13-01-C-0019) They would also like to thank Advanced Fuel Research for making their laboratory facilities available
6 REFERENCES 1 Tu Anthony ldquoOverview of Sarin Terrorist Incidents in Japan in 1994 and 1995rdquo 6th CBW Protection Symposium
Stockholm Sweden 10-15 May 1998 2 Jensen JL NC Wong W R Loerop SPIE 4775-03 (2001) 3 JSAWM Requirements at wwwsbccomapgeaarmymilRDAecbcrtPRODSERJSAWMjsawmhtml 4 ldquoThe Chemical Weapons Convention Redefines Analytical Challengerdquo Analytical Chemistry News amp Features June 1
397A (1998) 5 Johnston RL Hoefler CM Fargo JC and Moberley B ldquoThe Defense Nuclear Agencyrsquos ChemicalBiochemical
Weapons Agreements Verification Technology Research Development Test and Evaluation Program and its Requirements for On-Site Analysisrdquo AT-ONSITE 5-8 (1994)
6 Black RM Clarke RJ Read RW and Reid MT ldquoApplication of gas chromatography-mass spectrometry and gas chromatography-tandem mass spectrometry to the analysis of chemical warfare samples found to contain residues of the nerve agent sarin sulphur mustard and their degradation productsrdquo J Chromatography 662 301-321 (1994)
7 Hoffland LD Piffath RJ Bouck JBrdquoSpectral signatures of chemical agents and simulantsrdquo Optical Engineering 24 982-984 (1985)
8 Braue EHJ Pannella MGrdquoCIRCLE CELL FT-IR Analysis of Chemical Warfare Agents in Aqueous Solutionsrdquo Applied Spectroscopy 44 1513-1520 (1990)
9 Tseng C-H Mann CK and Vickers TJ ldquoDetermination of Organics on Metal Surfaces by Raman Spectroscopyrdquo Applied Spectroscopy 47 1767-1771 (1993)
10 Staff Reporter ldquoUS Army Tests Lidar to Detect Biological Toxinsrdquo Photonic Spectra pg 50 December 1998
Proc SPIE Vol 4577
Proc SPIE Vol 4577
173
11 Christesen SD Raman cross sections of chemical agents and simulants Applied Spectroscopy 42 318-321 (1988) 12 Alak AM and Vo-Dinh T ldquoSERS of Organophosphorous Chemical Agentsrdquo Analytical Chemistry 59 2149-2153
(1987) 13 Norrod KL Sudnik LM Rousell D and Rowlen KL ldquoQuantitative Comparison of Five SERS Substrates
Sensitivity and Detection Limitrdquo Applied Spectroscopy 51 994-1001 (1997) 14 Lee Y and Farquharson S ldquoSERS Sample Vials Based on Sol-Gel Process for Trace Pesticide Analysisrdquo SPIE 4206
140-146 (2000) 15 Farquharson S and Lee Y ldquoTrace Drug Analysis by Surface-Enhanced Raman Spectroscopyrdquo SPIE 4200-16 (2000) 16 Lee Y Farquharson S and Rainey P M Surface-Enhanced Raman Sensor for Trace Chemical Detection in Water
SPIE 3857 76-84 (1999) 17 Lee Y Farquharson S Kwong H and Shahriari M ldquoSurface-Enhanced Raman Sensor for Surface-Enhanced Raman
Spectroscopyrdquo SPIE 3537 252-260 (1998) 18 Farquharson S Smith W Carangelo R C and Brouillette C ldquoIndustrial Raman Providing Easy Immediate Cost
Effective Chemical Analysis Anywhererdquo SPIE 3859 14-23 (1999) 19 Narayanan VA JM Bello JD Stokes and T Vo-Dinh Analusis 19 307-310 (1991) 20 Laserna JJ E L Torres and JD Winefordner Analytica Chemica Acta 469-480 (1987) 21 Torres EL and JD Winefordner Analytical Chemistry 59 1626-1632 (1987) 22 Suh JS DP DiLella M Moskovits J Phys Chem 87 1540-1544 (1983) 23 Benner RE R Dornhaus R Chang and BL Laube Correlations in the Raman spectra of cyanide complexes adsorbed
at silver electrodes with voltammograms Surface Science 101 341 (1980) 24 Chase D B and JF Rabolt Fourier Transform Raman Spectroscopy Acad Press Ch1 p 131 (1994) 25 Connes J Rev Opt Theor Instrum 40 45 (1961)
APPLIED SPECTROSCOPY 351
focusing the 488 nm laser beam 2 mm deep into thebulk of the crystals to avoid contributions from potentialdepletion layers As shown in Fig 1b the dependence ofthe Raman shift on the degree of deuteration is almostperfectly linear and ts very well with D 5 22684cmR1 24526 where D is the degree of deuteration (in )and R is the spectral mean of the PO4 vibration in cm21A linear correlation coef cient of 0998 indicates an ex-cellent linear dependence of the Raman peak shift withdegree of deuteration This result shows that the shift ofthe PO 4 peak is simply caused by the linear increase inatomic mass due to isotope substitution which decreasesthe length of hydrogen-like bonds
This excellent linear dependence allows us to map thepro le of the DH exchange layer at the surface of DKDPcrystals by acquiring Raman spectra and determining theposition of the PO 4 peak for various depths This methodis preferable over other methods such as determining thestrength of the OD vibration directly (eg at 715 cm21)because the position of the most intense peak in the Ra-man spectrum can be measured more precisely than theintensity of some of the weakest peaks in the spectrumThis is demonstrated in Fig 2 where depth-dependentRaman spectra (Fig 2a) and the resulting exchange layerpro les for two DKDP crystals are shown (Fig 2b) Thespectra in Fig 2a were obtained from a depth scan of aDKDP crystal with 75 degree of deuteration in thebulk grown at 45 8C The spectra start out as DKDP with30 deuteration close to the surface and approach thebulk DKDP spectrum within a few micrometers of depthThe fact that the relative degree of deuteration does notextend to 0 D is due to the limited depth resolution ofthe Raman microprobe which averages over 4 mm indepth Figure 2b depicts the resulting DH exchange layerpro les for this and a second crystal grown at 63 8Crespectively Both crystals had the same exposure to am-bient conditions and their main difference is the temper-ature at which they were grown The different exchangelayer pro les indicate that crystals grown at differenttemperatures have differen t proton conductiv ities 13
which leads to a difference in their rate of deuteriumdepletion The parameters controlling this behavior arecurrently the objective of a detailed study the results ofwhich will be reported elsewhere
CONCLUSION
In conclusion we have shown that the shift of the to-tally symmetric PO 4 stretch mode in the Raman spectrumof DKDP crystals scales linearly with degree of deuter-ation This allows us to correlate Raman peak positionsto deuteration levels in these crystals We have presenteda new technique to determine DH diffusion pro les inDKDP frequency conversion crystals based on micro-Ra-man spectroscopy This technique is fast inexpensiveand works under various environmental conditionswhich will allow us to better understand and control deu-terium depletion in DKDP crystals
ACKNOWLEDGMENTS
We would like to thank M Runkel for rst discovering DKDP crack-ing R Floyd for providing DKDP crystals and L Chase and A Burn-ham for their support and helpful discussions This work was performedunder the auspices of the US Department of Energy by the University
of California Lawrence Livermore National Laboratory under contract W-7405-Eng-48
1 J J De Yoreo A K Burnham and P K Whitman Int Mat Rev47 113 (2002)
2 C E Barker R A Sacks B M Van Wonterghern J A Caird JR Murray J H Campbell K Kyle R B Ehrlich and N DNielsen Proc SPIE-Int Soc Opt Eng 2633 501 (1995)
3 T Suratwala paper to be published4 Cleveland Crystals Inc httpwwwclevelandcrystalscom
KDPshtmltable5 E A Popova I T Savatinova and I A Velichko Sov Phys Solid
State 12 1543 (1971)6 I P Kaminow R C C Leite and S P S Porto J Phys Chem
Solids 26 2085 (1965)7 J A Subramony B J Marquardt J W Macklin and B Kahr
Chem Mat 11 1312 (1999)8 H Tanaka M Tokunaga and I Tatsuzaki Solid State Commun
49 153 (1984)9 R J Nelmes G M Meyer and J E Tibballs J Phys C 15 59
(1982)10 M A Yakshin D W Kim Y S Kim Y Y Broslavets O E
Sidoryuk and S Goldstein Laser Physics 7 941 (1997)11 I Takenaga Y Tominaga S Endo and M Kobayashi Solid State
Commun 84 931 (1992)12 C Krenn personal communication13 M Sharon and A K Kalia J Solid State Chem 21 171 (1977)
Rapid Dipicolinic Acid Extractionfrom Bacillus Spores Detectedby Surface-EnhancedRaman Spectroscopy
STUART FARQUHARSON ALAND GIFT PAUL MAKSYMIUK andFRANK E INSCOREReal-Time Analyzers Inc East Hartford Connecticut06108
Index Headings Dipicolinic acid Bacillus spores Anthrax Surface-enhanced Raman spectroscopy
INTRODUCTION
The anxiety caused by the distribution of anthrax en-dospores through the US postal system in October 2001was exacerbated by the long time required for positiveidenti cation of the Bacillus anthracis spores and the un-known extent of their distribution Since that time manymethods capable of rapid eld analysis have been inves-tigated to augment or replace the laboratory method ofgrowing microorganisms in culture media which takesdays to perform12 Prominent among these approachesare polymerase chain reactions (PCR)3 immunoassaysand detection of calcium dipicolinate as a biochemicalsignature PCR employs primers to separate organism-
Received 10 October 2003 accepted 14 November 2003 Author to whom correspondence should be sent
352 Volume 58 Number 3 2004
speci c nucleic acid sequences (eg capsular protein en-coding gene for Bacillus anthracis)4 and polymerases toamplify the segment until it is detectable Recently am-pli cation times have been substantially reduced andcomplete analysis can now be performed in an hour orless Immunoassay methods are also being developed thatuse competitive binding of the bioagent (as an antigen)and its labeled conjugate for a limited number of anti-bodies Although analyses can be performed in under 30minutes a well-de ned anthrax antigen has not yet beenidenti ed5ndash7 and consequently the false-positive rate isunacceptably high8
A number of other methods are being developed witha focus on the detection of calcium dipicolinate (CaDPA)and its derivatives as a B anthracis signature This is sobecause only spore-forming bacteria contain CaDPA andthe most common potentially interfering spores such aspollen and mold spores do not Relatively fast methodshave been developed to chemically extract CaDPA andthen detect it directly by uorescence9 or indirectly byluminescence1011 In the latter case hot dodecylamine(DDA) has been used to extract dipicolinic acid (DPA)and terbium has been utilized to form a highly lumines-cent DPA complex11 Although measurements have beenperformed in as little as ve minutes it was found thatas many as three concentration-dependent complexes canform each with different lifetimes This coupled withthe fact that the Tb31 cation produces the same lumines-cence spectrum makes determinations of low spore con-centrations problematic
It has been long known that Raman spectra of Bacillispores are dominated by bands associated with CaDPA12
and that these spectra may provide a suitable anthrax sig-nature at the genus level13 Since that time considerableimprovements in Raman instrumentation have led to lab-oratory measurements of single Bacilli spores14 and to eld measurements of spores captured from a mail-sort-ing system15 However the single spore measurementsrequired complex instrumentation that is not ruggedwhile the eld measurements required milligrams of sam-ple Furthermore the Raman spectra of both measure-ments contained uorescence contributions that would in-crease uncertainty in quanti cation
In related research we demonstrated that nanogramquantities of DPA could be detected by uorescence-freesurface-enhanced Raman spectroscopy (SERS)16 We alsodemonstrated that microliter volumes of chemicals canbe detected by SERS using metal-doped sol-gel-packedglass capillaries17 Towards the goal of developing a rap-id eld SERS-based anthrax spore detector we havecombined our previous research and we now report thatDPA can be extracted from a 10 mg B cereus spore sam-ple using DDA in 1 minute and can be detected by SERSin an additional 1 minute
EXPERIMENTAL
Dipicolinic acid (26-pyridinedicarboxylic acid DPA)and dodecylamine (DDA) were used as received fromSigma-Aldrich (Milwaukee WI) Lyophilized B cereusspores prepared according to the literature13 were sup-plied by the University of Rhode Island and used as re-ceived Multiple particles approximately 01 mm3 each
were separated and weighed at 5 to 15 mg representing05 to 15 million spores The sample masses were con-sistent with a previous determination of spore density at0081 gmL that indicated a high degree of entrained air
All chemicals used to prepare the silver-doped sol-gelcoated capillaries were also obtained and used as receivedfrom Sigma-Aldrich According to previously publishedprocedures17 two precursor solutions were preparedmixed and then drawn into 1-mm-diameter glass capil-laries The silver amine precursor consisted of a 51 vvratio of 1 N AgNO 3 to 28 NH3OH while the alkoxideprecursor consisted of a 21 vv ratio of methanol to te-tramethyl orthosilicate The alkoxide precursors weremixed with silver amine precursor in an 81 vv ratioApproximately 015 mL was drawn into the capillarycoating a 15-mm length After sol-gel formation the in-corporated silver ions were reduced with dilute sodiumborohydride which was followed by a water wash to re-move residual reducing agent
A 100 mL drop of a 50 mM DDA solution in ethanolpre-heated to 78 8C was added to each of the B cereusparticles to digest the spore coat After 1 minute the re-sultant solution was drawn into a SER-active capillarythat was immediately xed horizontally to an XY posi-tioning stage (Conix Research Spring eld OR) just in-side the focal point of an f 07 aspheric lens The lensfocused the beam into the sample and collected the scat-tered radiation back along the same axis A dichroic lter(Omega Optical Brattleborough VT) was used to re ectthe excitation laser to the lens and pass the Raman scat-tered radiation collected by the lens An f 2 achromatwas used to collimate the laser beam exiting a 200-mm-core-diameter source ber optic while a second f 2 ach-romat was used to focus the scattered radiation into a 365mm ber optic (Spectran Avon CT) A short-pass lterwas placed in the excitation beam path to block the sil-icon Raman scattering generated in the source ber fromre ecting off sampling optics and reaching the detectorA long-pass lter was placed in the collection beam pathto block the sample Rayleigh scattering from reachingthe detector A 785 nm diode laser (Process InstrumentsInc model 785-600 Salt Lake City UT) was used todeliver 100 to 150 mW of power to the sample A Fouriertransform Raman spectrometer (Real-Time Analyzersmodel IRA-785 East Hartford CT) and a silicon photo-avalanche detector (Perkin Elmer model C30902S Stam-ford CT) were used to acquire the SER spectra
RESULTS AND DISCUSSION
As an initial experiment the SER spectrum of 1 gLof DPA in water was measured using the newly devel-oped silver-doped sol-gel-coated capillaries (Fig 1A) Atthis concentration a high signal-to-noise ratio (SN) isobtained in 1 min In fact a reasonable spectrum is ob-tained in the same time frame for 1 mgL (Fig 1B) TheSER spectra are reasonably similar to the normal Raman(NR) spectrum obtained for a saturated solution of DPAin 1 N KOH (Fig 1C) and the following band shifts areobserved (NR to SER) 647 to 657 cm21 817 to 815cm21 998 to 1008 cm21 1384 to 1382 cm21 1434 to1428 cm21 and 1569 to 1567 cm21 Many of these bandshave been previously assigned1213 such as 998 cm21 to
APPLIED SPECTROSCOPY 353
FIG 1 SERS of DPA in water using silver-doped sol-gel-coated glasscapillary for (A) 1 gL and (B) 1 mgL (C ) NR of saturated DPA in 1N KOH in a glass capillary Spectral conditions (A) and (B) 150 mWof 785 nm 1-min acquisition time (C) 450 mW of 785 nm 5-minacquisition time both 8 cm 21 resolution
FIG 2 SERS of DPA extracted from 10 mg B cereus particle using100 mL of 50 mM hot DDA acquired in (A) 1 minute and (B) 2 seconds(C ) Attempted SERS of 50 mM hot DDA in ethanol using silver-dopedsol-gel-coated glass capillary acquired in 1 min Spectral conditions150 mW of 785 nm 8 cm21 resolution
the symmetric ring stretch 1384 cm21 to the OndashCndashOsymmetric stretch 1428 cm21 to the symmetric ring CndashH bend and 1569 cm21 to the asymmetric OndashCndashOstretch
The rst B cereus samples consisted of 2 mg of sporesin 2 mL of 5 mM hot DDA The samples were main-tained at 78 8C for 40 min and while hot approximately10 mL was drawn into a SER-active capillary Since spec-tra of DPA were obtained for these initial samples small-er spore masses higher DDA concentrations and shorterheating periods were examined In due course it wasfound that 10 mg of spores could be digested by 100 mLof 50 mM hot DDA in one minute and detected (Fig2A) In fact the signal was suf ciently intense that it canbe observed in as little as two seconds (Fig 2B) Theamount of DPA that was extracted was estimated to bebetween 5 and 10 mgL by comparing the signal intensityof the 1008 cm21 band to that measured for DPA in waterThis is consistent with previous research that found thatthe majority of the DPA is extracted from spores usingDDA11 and that B cereus spores contain approximately10 DPA by weight18 The SN of 127 for the 1008 cm21
band in the 1-minute SER spectrum suggests a limit ofdetection of approximately 250 ng of B cereus sporesbased on a SN of 3 Finally it should be noted that DDAdid not produce a detectable SER spectrum as shown inFig 2C
CONCLUSION
We have demonstrated that by combining rapid extrac-tion of dipicolinic acid from Bacillus cereus spores withchemical identi cation by surface-enhanced Raman spec-troscopy as little as 10 mg of spores can be detected Infact the entire measurement from the time of adding hotdodecylamine to the spores to the time when the dipi-colinic acid SER spectrum is acquired and analyzedcould be performed in less than two minutes The abilityof this method to distinguish between spore-forming bac-teria such as Bacillus anthracis and non-DPA containingpowders could help prevent costly shutdowns associated
with the appearance of suspicious material or intentionalmailing of common substances as an anthrax hoax Thismethod could also prove useful in detecting the locationof anthrax endospores in mail distribution facilities if an-other veri ed attack should occur
Research continues to fully characterize the surface-enhanced Raman spectroscopy signal intensities as afunction of sample concentration and to explore otherextractants that do not require the use of elevated tem-perature
ACKNOWLEDGMENTS
The authors are grateful for the support of the National Science Foun-dation (DMI-0296116 and DMI-0215819) and the US Army(DAAD13-02-C-0015 Joint Service Agent Water Monitor program)The authors are indebted to Chetan Shende for preparing the sol-gelcapillaries The authors also thank James Gillespie Nicholas Fell andAugustus Fountain for providing important background informationMark Farquharson for laboratory support and Professor Jay Sperry ofthe University of Rhode Island for supplying B cereus spores
1 V A Pasechnik C C Shone and P Hambleton Bioseparations 3267 (1993)
2 P J Jackson M E Hugh-Jones D M Adair G Green K K HillC R Kuske L M Grinberg F A Abramova and P Keim ProcNatl Acad Sci USA 95 1224 (1998)
3 B R Glick and J J Pasternak Molecular Biology Principles andApplications of Recombinant DNA (ASM Press Washington DC1994)
4 C A Bell J R Uhl T L Had eld J C David R F Meyer TF Smith and F R Cockerill III J Clin Microbiol 40 2897(2002)
5 D L Gatto-Menking H Yu J G Bruno M T Goode M Millerand A W Zulich Biosens Bioelectron 10 501 (1995)
6 J J Quinlan and P M Foegeding J Rapid Methods AutomationMicrobiol 6 1 (1998)
7 A A Hindle and E A H Hall Analyst (Cambridge UK) 1241599 (1999)
8 M S Ascher US Department of Health amp Human Services(httpwwwhhsgovophppresentationsAscherdoc)
9 R Nudelman B V Bronk and S Efrima Appl Spectrosc 54 445(2000)
10 D L Rosen C Sharpless and L B McBrown Anal Chem 691082 (1997)
354 Volume 58 Number 3 2004
11 P M Pellegrino N F Fell Jr and J B Gillespie Anal ChimActa 455 167 (2002)
12 W H Woodruff T G Spiro and C Gilvarg Biochem BiophysRes Commun 58 197 (1974)
13 E Ghiamati R S Manoharan W H Nelson and J F SperryAppl Spectrosc 46 357 (1992)
14 A P Esposito C E Talley T Huser C W Hollars C M Schal-dach and S M Lane Appl Spectrosc 57 868 (2003)
15 S Farquharson L Grigely V Khitrov W W Smith J F Sperryand G Fenerty J Raman Spectrosc paper accep ted (2003)
16 S Farquharson W W Smith S Elliott and J F Sperry SPIE-IntSoc Opt Eng 3855 110 (1999)
17 S Farquharson and P Maksymiuk Appl Spectrosc 57 479(2003)
18 F W Janssen A J Lund and L E Anderson Science (Washing-ton DC) 127 26 (1958)
SPIE -2003-5269 117
pH dependence of methyl phosphonic acid dipicolinic acid and cyanide by surface-enhanced Raman spectroscopy
Stuart Farquharson Alan Gift Paul Maksymiuk Frank Inscore and Wayne Smith
Real-Time Analyzers 87 Church Street East Hartford CT 06108
ABSTRACT US and Coalition forces fighting terrorism in Afghanistan and Iraq must consider a wide range of attack scenarios in addition to car bombings Among these is the intentional poisoning of water supplies to obstruct military operations To counter such attacks the military is developing portable analyzers that can identify and quantify potential chemical agents in water supplies at microgram per liter concentrations within 10 minutes To aid this effort we have been investigating the value of a surface-enhanced Raman spectroscopy based portable analyzer In particular we have been developing silver-doped sol-gels to generate SER spectra of chemical agents and their hydrolysis products Here we present SER spectra of methyl phosphonic acid and cyanide as a function of pH an important factor affecting quantitation measurements which to our knowledge has not been examined In addition dipicolinic acid a chemical signature associated with anthrax-causing spores is also presented Keywords Chemical warfare agents agent detection agent hydrolysis SERS Raman spectroscopy homeland security
1 INTRODUCTION In the past decade the Unites States and its allies have been challenged by a different kind of warfare exemplified by the terrorist attacks of September 11 2001 Suicide bombings and the use of chemical agents are the norm and military personnel must consider a wide range of attack scenarios Among these is the intentional poisoning of water supplies to obstruct military operations in Afghanistan and Iraq Currently colorimetric paper is used to detect agents on-site while gas chromatography combined with mass spectrometry (GCMS) is used in mobile support laboratories However both methods have severe drawbacks The paper changes color in response to contact with many chemicals besides CWAs causing a high incidence of false positives1 while GCMS although very chemically specific requires hours to perform and constant re-calibration234 Military operations would be greatly aided by a portable analyzer that can identify and quantify potential chemical agents at concentrations that impact safety This includes the analysis of drinking water supplies distribution and storage systems To meet this goal the Department of Defense has been investigating numerous approaches under the auspices of the Joint Service Agent Water Monitor (JSAWM) program5 The JSAWM must be capable of identifying the chemical agents with no false-positives and quantifying the agents at microgram per liter concentrations within 10 minutes678 Recently we and others have been investigating the ability of surface-enhanced Raman spectroscopy (SERS) to measure chemical agents9-12 bioagents13-17 and their hydrolysis products in water SERS employs the interaction of surface plasmon modes of metal particles with target analytes to increase scattering efficiency by as much as 1 million times18 In 1987 the potential of this approach to measure CWAs was demonstrated by measuring a series of pesticides19 Several of these organophosphonates have chemical structures similar to CWAs in particular P=O functional groups In our studies we have been employing silver-doped sol-gels to promote the SER effect The porous silica network of the sol-gel matrix offers a unique environment for immobilizing and stabilizing SER-active metal particles20-23 The sol-gels have been coated on the internal walls of standard glass vials to allow rapid SER analysis of small sample quantities without preparation We have measured over 100 chemicals with enhancements of 104 to 106 demonstrated reversible measurements in a flowing system reproducible measurements from vial-to-vial and measurements in multiple solvents including water20-23 Previously we used these vials to perform preliminary measurements of cyanide (CN) methylphosphonic acid (MPA) and dipicolinic acid (DPA) MPA is a hydrolysis product of the nerve agents (eg sarin Reaction 1) and may be a valuable indicator of nerve agent usage particularly since the alkyl methylphosphonic acids are relatively more stable than their corresponding parent complexes24 DPA is
SPIE -2003-5269 118
a chemical signature of spore forming bacteria such as Bacillus anthracis And in light of the inability to rapidly detect the anthrax spores distributed through the US mail in October 2001 a number of methods are being developed to extract and analyze this signature Reaction 1 Stepwise hydrolysis of Sarin to form hydrofluoric acid (HF) isopropyl methylphosphonic acid (IMPA) then methyl phosphonic acid (MPA) and 2-propanol In our previous SERS investigations MPA and DPA were measured at 50 and 100 mgL respectively In both cases limits of detection (LOD) were estimated at 100 microgL providing encouragement in that SERS may satisfy the needs of the JSAWM Since it has been shown that pH can substantially influence the intensity of SER bands25 which would clearly influence quantitative analysis we undertook the present study to determine the severity of these effects for cyanide methyl phosphonic acid and dipicolinic acid Furthermore we previously observed a band at 1050 cm-1 for MPA6 possibly due to an anion formed at basic pH Here we investigate the source of this spectral anomaly
2 EXPERIMENTAL All chemicals including potassium cyanide methyl phosphonic acid dipicolinic acid and those used to prepare the silver-doped and gold-doped sol-gel coated vials were obtained and used as received from Sigma-Aldrich All samples were prepared in HPLC grade water (Fischer Scientific Fair Lawn NJ) for SERS measurements The pH of these samples was adjusted using dilute nitric acid or potassium hydroxide and verified using a pH electrode (Corning Inc Corning NY) that had been calibrated with pH 400 700 and 1000 buffered standards from Fischer Scientific Once prepared the samples were transferred into the silver-doped sol-gel vials (Real-Time Analyzers Inc East Hartford CT) The vials were coated in a manner similar to that previously reported by adding ammonium hydroxide to a solution of silver nitrate tetramethyl orthosilicate (TMOS) and methanol20 Gold-vials were coated by adding nitric acid to a solution of gold tetrachloride TMOS and methanol The two precursor solutions were prepared mixed and transferred to 2-mL glass vials dried and heated After sol-gel formation the incorporated metal ions were reduced with dilute sodium borohydride (1mgmL) which was followed by a water wash to remove residual reducing agent After the resultant analyte solution was introduced the SER-active vial was immediately fixed horizontally to an XY positioning stage (Conix Research Springfield OR) just inside the focal point of an f07 aspheric lens The lens focused the beam into the sample and collected the scattered radiation back along the same axis A dichroic filter (Omega Optical Brattleborough VT) was used to reflect the excitation laser to the lens and pass the Raman scattered radiation collected by the lens An f2 achromat was used to collimate the laser beam exiting a 200 microm core diameter source fiber optic while a second f2 achromat was used to focus the scattered radiation into a 365 microm fiber optic (Spectran Avon CT) A short pass filter was placed in the excitation beam path to block the silicon Raman scattering generated in the source fiber from reflecting off sampling optics and reaching the detector A long pass filter was placed in the collection beam path to block the sample Rayleigh scattering from reaching the detector A 785 nm diode laser (Process Instruments Inc model 785-600 Salt Lake City UT) was used to deliver 100 to 150 mW of power to the sample A Fourier transform Raman spectrometer (Real-Time Analyzers model IRA-785 East Hartford CT) and a silicon photo-avalanche detector (Perkin Elmer model C30902S Stamford CT) were used to acquire the SER spectra
3 RESULTS AND DISCUSSION In a previous study of MPA6 aimed at developing a concentration calibration curve and determining limits of detection (LOD) we observed an anomaly at 1050 cm-1 Since it was found that the band intensity changed as a function of concentration the band must be associated with a sample parameter Two possible parameters photon flux and pH are examined here The first parameter was investigated by irradiating a 1mgmL MPA sample in a SER-active vial with laser powers of 200 mW and above and monitoring spectral changes It was immediately found that the 1050 cm-1 band
2O+ H HF + +OH
OH3H C
OP
OF
CH
CH3
3
3H C
OCP
OOH
CH
CH3
3
3H C
O
CPHO
CH
CH
3
3
C
Sarin IMPA MPA 2-propanol
SPIE -2003-5269 119
appeared and grew as a function of time and that the higher the laser power the faster the growth Figure 1 shows the growth of the 1050 cm-1 band over the course of 30 minutes when using 150 mW of 785 nm excitation while Figure 2A shows that the growth can be fit with a first order exponential equation namely I1050 = 03+05e-013t Furthermore the 760 cm-1 band could be fit with a first order decay equation with an identical exponential rate constant ie I760 = 08-
08e-013t The rates represent classical first order kinetics and their correspondence allows one to conclude that MPA is being transformed one-for-one into a photo-generated product At this time the photoproduct has not been positively identified but phosphonic (phosphorous) acid and phosphonate are likely candidates since the symmetric P(OH)3 stretch occurs at ~1050cm-1 Our studies show that a reduction in laser power to 100 mW at the sample for MPA essentially eliminates this degradation process This laser power or lower was used for further measurements
Methyl phosphonic acid is a diprotic acid that stepwise dissociates into two anions MPA- and MPA= according to the following reactions26
MPA MPA- + H+ pKa1 = 212 Reaction 2
MPA- MPA= + H+ pKa2 = 729 Reaction 3 The relative concentrations of MPA MPA- and MPA= can be determined at any pH by expressing [MPA] and [MPA=] in terms of [MPA-] using Reactions 2 and 3 and summing all three to equal the total starting concentration here 2 mgmL (0021M MW = 9602) viz
[MPA] + [MPA-] + [MPA=] = 0021M Equation 1 substituting from Reactions 1 and 2
([H+][MPA-])K1a + [MPA-] + (K2a[MPA-])[H+] = 0021M Equation 2 rearranging [MPA-] = 0021M(1+[H+]K1a + K2a[H+]) Equation 3 The relative concentrations of MPA MPA- and MPA= as a function of pH are shown in Figure 3 It is worth noting that near neutral pH both MPA- and MPA= will be present To confirm that the SER signal followed this pH dependence a starting solution consisting of 20 mg of MPA in 10 mL HPLC grade water was prepared and brought to pH of 20 using dilute nitric acid From this solution 2 mL were added to a SER-active vial and the SER spectrum recorded At this pH a peak at 760 cm-1 was barely discernable The 2 mL solution was returned to the starting solution and the pH was re-measured to correct for any changes that the silver-doped sol-gel vials might cause In most cases the change was less than 02 pH units and the pH is reported as the before and after average Next the pH of the
Figure 2 A) Exponential growth of 1050 cm-1 band and B) exponential decay of 760 cm-1 band for spectral series in Figure 1
Figure 1 Growth of 1050 cm-1 band as a function of time due to exposure to 150 mW of 785 nm Spectra are 5 sec each collected every 100-sec from 0 to 30-min
B
570 770 970 1170 1370Raman Shift (cm-1)
Arbit
rary
Unit
s
MPA Photodegradation
600 800 1000 1200 1400 Wavenumber (cm-1)
A
0 10 20 30 0 10 20 30 time (min) time (min)
Ram
an In
tens
ity (r
elat
ive)
30
min
0
SPIE -2003-5269 120
starting solution was adjusted to 325 using dilute KOH Again 2 mL were added to a vial and the SER spectrum recorded At this pH a reasonably strong 760 cm-1 band was observed This process was repeated as spectra were recorded at pHs of 70 74 75 79 85 and 100 A total of 1 mL of KOH was added diluting the total concentration by 10 Next the pH of the starting solution was made acidic by adding dilute nitric acid dropwise This time spectra were recorded at pHs of 72 69 64 and 37 Figure 4 shows the SER spectra for representative pHs (spectra were left out to simplify the figure) while Figure 3 shows the 760 cm-1 peak intensities as a function of pH (The band intensities were adjusted to compensate for dilution effects caused by the addition of HNO3 and KOH then normalized to 0021 M for the most intense band observed at pH 37) It is clear from Figure 3 that the 760 cm-1 band follows the MPA- concentration as a function of pH and must be assigned to this anion No bands were observed that corresponded to MPA or MPA= The lack of an MPA SER spectrum may be due to the absence of an attraction between the neutral analyte and the electropositive silver surface The same reasoning suggests that a strong SER spectrum should be observed for MPA= but it is not and a satisfactory explanation has not been found
0000
0005
0010
0015
0020
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14pH
Con
cent
ratio
n [M
]
pK1 = 212 pK2 = 729
MPA- MPA=
MPA
Figure 4 SER spectra of 002M MPA as a function of pH Conditions 100 mW of 785 nm 36 scans (1 min) 8 cm-1 recorded 2 min after sample introduction pH 19 and 20 not apparent on this scale pH 69 and 74 near identical to 70 and 75 and not shown for clarity
00
02
04
06
08
10
12
14
16
18
0 5 10 15 20 25 30Measurement Number
Ram
an In
tens
ity (7
60 c
m-1
)
Figure 3 Concentration dependence of MPA MPA- and MPA= as a function of pH for a 002M sample Intensity of 760 cm-1
band from Figure 6 as a function of pH ( for increasing basic adjustment for increasing acidic adjustment error was measured at ~10 for pH 64)
Figure 5 SER spectra of 002M MPA at pH 64 measured around a vial at three heights (9 points per height) Conditions as in Figure 4 but 10-sec scans
Since these measurements involved the removal and replacement of the SER-active vial in the sample holder to remove and add sample variation in the intensity as a function of vial position was minimized by illuminating the exact same height along the vial wall But this does not account for variability of the SERS response of the sol-gel coating around the vial To analyze this effect a vial containing MPA at pH 64 was rotated at ~ 40o
intervals at the original height and 18rdquo above and below this value Figure 5 shows the intensity of the 760 cm-1
band for the 27 positions It was found that the average value was 137plusmn014 an RSD of 10 overall and 5 for each height An error bar is included in Figure 4 for the pH 64 measurement
SPIE -2003-5269 121
With the above analyses in mind a preliminary investigation of the SER spectral response for MPA (as MPA-) as a function of concentration was performed A single vial was used for these measurements beginning with 1 mgL followed by measurements of 10 100 and 1000 mgL In all cases the pH was ~7 and 3 positions around the vial were measured per concentration Since the 760 cm-1 band was not observed for concentrations of 1 or 10 mgL using 100 mW of 785 nm the laser power at the sample was raised to 200 mW beginning with the 10 mgL concentration Photo-degradation was largely avoided (and not observed) by exposing the sample for only 33 seconds per spectral acquisition Representative spectra for 10 100 and 1000 mgL are shown in Figure 6 while a plot of the 760 cm-1 band intensity as a function of concentration is shown in Figure 7 These values were also used to estimate limits of detection based on the signal-to-noise ratio (SN) of the 760 cm-1 band using 100 mW of 785 nm laser excitation and a 10-min acquisition time As summarized in the Figure 7 inset the lower the measured concentration the lower the predicted LOD Even if as estimated 210 microgL could be measured using the silver-doped sol-gel vials an improvement of a factor of 70 is still required to meet JSAWM goals of measuring 3 microgL in 10-minutes Similar to MPA DPA is a diprotic acid (pKa1 = 216 and pKa2 = 692) and variations in pH will effect the relative concentrations of DPA DPA- and DPA= and possibly the SER spectra and band intensities This could prove significant if an acid or base is used to denature anthrax spores with the goal of extracting and analyzing DPA The MPA pH study described above was mimicked for DPA except that the starting solution consisted of 20 mg of DPA in 20 mL HPLC grade water (60x10-3M MW = 1671) The initial solution had a pH of 245 which was made basic by dilute KOH to pHs of 355 433 487 559 1069 and 1166 SER spectra were recorded at each pH using 100 mW of 785 nm and a 44-sec acquisition time Next one drop of concentrated nitric acid was used to remake the solution acidic at a pH of 200 Again sequential pH measurements were performed at 383 510 735 and 822 The solution pH was made acidic a third time but to pH 219 171 then 135 Throughout this process no more than 20 drops of acid or base were added and therefore the concentration was diluted by no more than 10 Most of the spectral bands showed a minor decrease in intensity as a function of increasing pH values However the bands at 525 and 795 cm-1 showed the most dramatic changes which occurred at acid pH Figure 8 shows the SER spectra of DPA for the spectral region and pH range of interest The identity of the DPA species was determined by plotting the normalized peak intensities with the lowest value set to 0 and the highest to 0006 M as a function of pH and overlaying these values on a plot of the relative concentrations for DPA DPA- and DPA= as previously done for MPA (Figure 9) As can be seen the 525 cm-1 band clearly corresponds to DPA The correspondence of the 795 cm-1 band to this species is less clear as the band retains intensity until pH of 55 This can be attributed to contribution to the overlapping band at 810 cm-1 which does not change as a function of pH The fact that most bands are observed at all pHs suggest that the primary interaction with silver is through the ring nitrogen This is supported by the fact that the most intense band occurs at 1008 cm-1 attributed to a symmetric ring breathing mode and that this interaction has been characterized for pyridine in numerous papers27
0
1
2
3
4
5
6
7
0 200 400 600 800 1000 1200
MPA Concentration (mgL)
760
Ban
d In
tens
ity
Figure 6 SER spectra of MPA in water at A) 1000 B) 100 and C) 10 mgL Conditions pH of 7 silver-doped sol-gel coated vial 200 mW of 785 nm 33-sec 8 cm-1 resolution
Figure 7 Plot of SER intensity of 760 cm-1 band of MPA as a function of concentration using 200 mW of 785 nm Inset table includes average intensity LOD standard deviation and percent deviation for each concentration but for 100 mW and 10 min
conc (mgL) LOD ave stddev dev10 021 002 1021100 072 010 14301000 312 040 1280
A
B C
SPIE -2003-5269 122
The ability to detect dipicolinic acid (DPA) by SERS is demonstrated in Figure 10 Even at 1 mgL the primary bands are visible The intensity of the dominant ring breathing mode observed at 1008 cm-1 for DPA in water (pH = 25 to 55) is plotted as a function of concentration in Figure 11 These values were also used to estimate limits of detection based on the SN of the 1008 cm-1 band using 100 mW of 785 nm laser excitation and a 10-min acquisition time Again the lower the measured concentration the lower the predicted LOD (see Figure 11 inset) and detection of 160 microgL is possible Gastrointestinal anthrax requires significant more spores than inhalation anthrax28 and a limit of detection might be placed at 1 million spores in 1 liter of water or 10 microgL Since each spore contains ~10 CaDPA by weight29 a goal for DPA might be 1 microgL indicating that the present measurements must be improved by nearly two orders of magnitude Due to the increased hazards of handling HCN gas KCN salt was used for these experiments Nevertheless all sample preparations were performed in a chemical hood KCN completely dissolves in water but its conjugate acid HCN is formed and has a Ka of 615x10-1030 viz
HCN CN- + H+ pKa = 921 Reaction 4
Figure 8 SER spectra of 1 mgmL DPA as a function of pH Conditions 100 mW 785 nm 100 scans (44-sec)
Figure 9 Concentration dependence of DPA DPA- and DPA= as a function of pH for a 0006M sample Intensity of 525 () and 795 (diams) cm-1 bands from Fig 8 as a function of pH
pH
135 171 219 383
0
0001
0002
0003
0004
0005
0006
0007
0 2 4 6 8 10 12 14pH
Con
cent
ratio
n [M
]
DPADPA-DPA=795525
DPA DPA=DPA-
pK2 = 692pK1 = 216
795 525
0
05
1
15
2
25
0 200 400 600 800 1000 1200
DPA Concentration (mgL)
1008
Ban
d In
tens
ity
conc (mgL) lod-10min-100mw1 017
10 016100 103
1000 355
Figure 10 SER spectra of DPA in water at A) 1000 B) 100 C) 10 and D) 1 mgL Conditions pH of 25-55 silver-doped sol-gel coated vial 175 mW of 785 nm 1-min 8 cm-1 D) has been multiplied by x10 to make bands visible
A
B C
D
Figure 11 Plot of SER intensity of 1008 cm-1 band of DPA as a function of concentration using 175 mW of 785 nm Inset table includes LOD in mgL for each concentration but for 100 mW and 10 min
SPIE -2003-5269 123
Consequently the cyanide concentration must be determined for each initial KCN concentration Specifically the samples prepared with concentrations of 01 1 10 100 and 1000 mgL of KCN produced CN- concentrations of 63x10-3 033 69 89 and 964 mgL at pHs of 816 90 967 102 and 107 respectively The pH dependence for the HCN and CN- concentrations are shown in Figure 12 Thus as the amount of KCN added to the solution decreases so does the pH of the solution (becomes less basic) and according to Reaction 4 the relative amount of CN- to HCN also decreases For example in the preparation of a 01 mgL solution of KCN the pH is shifted from 7 for pure water to only 816 and only 63 of the starting material becomes CN- or 63x10-3 mgL In comparison for a solution of 1000 mgL the pH is shifted from 7 to 107 and 96 of the starting material becomes CN- This is significant since the cyanide ion is better able to adsorb onto the silver particles and become SERS active SER spectra of 10 100 and 1000 mgL of KCN samples are shown in Figure 13 and a corresponding plot of the 2100 cm-1 band intensity as a function of concentration is shown in Figure 14
000102030405060708091011
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14pH
Con
cent
ratio
n [m
gm
L]
pKa = 921
CN -HCN
The SER spectra of cyanide are dominated by the single CN stretch at 2100 cm-1 which occurs in normal Raman spectra of solutions at 2080 cm-1 However a low frequency mode occurs at 135 cm-1 due to a Ag-CN stretch (not shown) indicative of a strong surface interaction It is also observed that as the concentration decreases the CN stretch shifts to 2140 cm-1 This shift has been attribute to the formation of a tetrahedral Ag(CN)3
2- surface structure31 as well
Figure 14 Concentration dependence of KCN SERS measured under conditions in Fig 11 Concentrations are 1 01 and 001 mgml Intensities are measured for the CN stretch at 2100 cm-1 Inset table includes LOD in mgL for each concentration in Figs 13 and 15 but for 100 mW and 10 min
Figure 13 SER spectra of KCN in water at A) 1000 B) 100 and C) 10 mgL Conditions pHs of 107 102 and 97 silver-doped sol-gel coated vial 100 mW of 785 nm 1-min 8 cm-1 C) has been multiplied by x10 to make band visible
Figure 12 Concentration dependence of HCN and CN- as a function of pH for a 1 mgmL sample Calculated intensity of 2100 () cm-1 band for a 1 mgml sample at pHs of 816 90 967 102 and 107
A
B
C
0
20
40
60
80
100
120
140
0 200 400 600 800 1000 1200CN Concentration (mgL)
2100
Ban
d In
tens
ity
Figure 15 SER spectra of KCN in water at A) 10 B) 1 and C) 01 mgL Conditions pHs of 97 12 and 12 gold-doped sol-gel coated vial 100 mW of 785 nm 1-min 8 cm-1 C) has been multiplied by x10 to make band visible
A
B
C
conc (mgL) lod-10min-100mw Condition01 001 Au-pH 121 007 Au-pH 12
69 003 Au-pH 9769 016 Ag-pH 9789 022 Ag-pH 102946 113 Ag-pH 107
SPIE -2003-5269 124
as to CN adsorbed to two different surface sites32 Alternatively the 2140 cm-1 band could be attributed to HCN since this species dominates at lower concentrations However it is unlikely that this species would be attracted to the electropositive silver surface Further both peaks should be present at pHs between 85 and 105 but this is not observed It has also been suggested that at concentrations near and above monolayer coverage the CN- species is forced to adsorb end-on due to crowding and at lower concentrations the molecule can reorient to lie flat33 This suggests that the 2100 and 2140 cm-1 bands correspond to the end-on and flat orientations respectively As Figures 13 and 14 show the intensity of the CN stretch for the 89 mgL sample is nearly as intense as the 964 mgmL band This suggests that the Raman signal for the flat orientation is more enhanced However more extensive measurements are required to verify this point Since resent research has suggested that cyanide may be more effectively detected on gold measurements of KCN solutions were also performed using gold-doped sol-gel vials Preliminary measurements are shown in Figure 15 for samples prepared from 01 1 and 10 mgL KCN Since the pHs are 816 90 967 the resultant CN- concentrations are 63x10-3 033 and 69 mgL Initially only the highest concentration was observed and the signal intensity was significantly better than the equivalent concentration measured using silver In an effort to shift Reaction 4 to the left transforming HCN to CN- (Le Chatelierrsquos principle) KOH was added to the lower concentration samples producing solutions with pH 12 This effectively forces all of the cyanide in solution to be CN- or 01 and 10 mgL respectively More importantly the CN stretch is now observed in the SER spectra The band appears at 2125 cm-1 as has been previously reported for gold12 As calculated for MPA and DPA LODs can be estimated from this data For the three concentrations of cyanide on silver the LODs are 016 to 11 mgL for 100 mW of 785 nm laser excitation and a 10-min acquisition time For gold pH adjusted the LODs are10 to 70 microgL an improvement of more than 10 times silver Nevertheless either substrate is sufficient to meet the JSAWM goals of measuring 3 mgL in 10-minutes as the requirements form cyanide are much less stringent than the nerve agents
4 CONCLUSIONS Here we examined the surface-enhanced Raman spectral response for methyl phosphonic acid dipicolinic acid and cyanide as a function of pH It was determined that the most prominent peak at 760 cm-1 reaches a maximum intensity between pH 3 and 7 and corresponds to the MPA- species Neither the MPA nor MPA= species appear to generate a SER spectrum and consequently no spectra were observed below pH 2 or above pH 8 In this study we also found that higher laser powers could cause photodegradation of MPA signified by the exponential growth of a band at 1050 cm-1 which is tentatively assigned to phosphorous acid Unlike MPA DPA was observed at all pHs This is attributed to the dominant interaction of the pyridine functional group with silver Minor spectral changes were observed at acid pHs and were assigned to neutral DPA Like MPA SER spectra of cyanide were pH dependent No spectra were observed for the HCN species while CN- was best observed at pHs more basic than 8 Preliminary concentration studies for the three analytes allowed estimating limits of detection for MPA DPA and CN using 100 mW of 785 nm and a 10-min acquisition time of 210 165 and 70 microgL respectively Although the latter value suggests that cyanide can be measured at sufficiently low concentrations to meet JSAWM goal improvements by 100 to 200 times are required for MPA and DPA It is clear from this study that pH of the sample is important and must be taken into account when developing concentration calibration curves as well as fieldable analyzers Future work will include tailoring the sol-gel with specific functional groups to dictate sample pH and thereby optimize sensitivity
ACKNOWLEDGMENTS The authors are grateful for the support of the US Army (DAAD13-02-C-0015 Joint Service Agent Water Monitor program) The authors would also like to thank Dr Steve Christensen of the US Army SBCCOM for helpful discussions and Mr Chetan Shende of Real-Time Analyzers for assistance in development of the gold-doped sol-gels
REFERENCES 1 Erickson B Analytical Chemistry News amp Features June 1 397A (1998)
SPIE -2003-5269 125
2 Johnston RL Hoefler CM Fargo JC and Moberley B AT-ONSITE 5-8 (1994) 3 Black RM RJ Clarke RW Read and MT Reid J Chromatography 662 301-321 (1994) 4 USA Army Center for Health Promotion and Preventive Medicine (USACHPPM) Technical Guideline 230A 5 Jensen JL NC Wong W R Loerop SPIE 4775-03 (2001) 6 Hoenig SL Handbook of Chemical Warfare and Terrorism Greenwood Press Wesport CT (2002) 7 Munro NB SS Talmage GD Griffin LC Waters AP Watson JF King and V Hauschild Env Health
Persp 107 933-974 (1999) 8 Holstege CP Kirk M Sidell FR Crit Care Clin13 923-42 (1997) 9 Farquharson S P Maksymiuk K Ong and S Christesen SPIE 4577 166-173 (2001) 10 Lee Y and S Farquharson SPIE 4378 21-26 (2001) 11 Spencer KM J Sylvia S Clauson and J Janni SPIE 4577 158-165 (2001) 12 Tessier P S Christesen K Ong E Clemente A Lenhoff E Kaler and O Velev Applied Spectroscopy 56
1524-1530 (2002) 13 Farquharson S WW Smith S Elliott and JF Sperry SPIE 3533 207-214 (1998) 14 Farquharson S WW Smith S Elliott and JF Sperry SPIE 3855110-116 (1999) 15 Farquharson S WW Smith YH Lee S Elliott and JF Sperry SPIE 4575 62-72 (2002) 16 Guzelian AA J Sylvia J Janni S Clauson and KM Spencer SPIE 4577 182-192 (2001) 17 Shende C F Inscore A Gift P Maksymiuk and S Farquharson in press 18 Weaver MJ S Farquharson and MA Tadayyoni J Chem Phys 82 4867-4874 (1985) 19 Alak AM and T Vo-Dinh Analytical Chemistry 59 2149-2153 (1987) 20 Lee Y and S Farquharson SPIE 4206 140-146 (2000) 21 Farquharson S and Y Lee SPIE 4200-16 (2000) 22 Lee Y S Farquharson and P M Rainey SPIE 3857 76-84 (1999) 23 Lee Y S Farquharson H Kwong and M Shahriari SPIE 3537 252-260 (1998) 24 Wang J M Pumera G Collins and A Mulchandani Analytical Chemistry 74 6121-6125 (2002) 25 Dou X YM Jung Z-Q Cao and Y Ozaki Applied Spectroscopy 53 1440-1447 (1999) 26 Data supplied by S Christesen and K Ewing 27 Kerker M and B Thompson Eds SPIE MS 10 (1990) 28 Inglesby TV DA Henderson JG Bartlett JAMA 287 2236 (2002) 29 FW Janssen AJ Lund and LE Anderson Science 127 26 (1958) 30 Lide DR Ed Handbook of Chemistry and Physics CRC Press 77th Ed (1996-7) 31 Billmann J G Kovacs and A Otto Surf Sci 92 153 (1980) 32 Murray CA and S Bodoff Phys Rev B 32 671 (1985) 33 Kellogg D and J Pemberton J Phys Chem 91 1120 (1987)
SPIE-2003-5269 16
Chemical agent detection by surface-enhanced Raman spectroscopy
Stuart Farquharson Alan Gift Paul Maksymiuk Frank Inscore and Wayne Smith
Real-Time Analyzers 87 Church Street East Hartford CT 06108
Kevin Morrisey and Steven D Christesen US Army SBCCOM Aberdeen Proving Ground MD 21010
ABSTRACT
In the past decade the Unites States and its allies have been challenged by a different kind of warfare exemplified by the terrorist attacks of September 11 2001 Although suicide bombings are the most often used form of terror military personnel must consider a wide range of attack scenarios Among these is the intentional poisoning of water supplies to obstruct military operations in Afghanistan and Iraq To counter such attacks the military is developing portable analyzers that can identify and quantify potential chemical agents in water supplies at microgram per liter concentrations within 10 minutes To aid this effort we have been investigating the value of a surface-enhanced Raman spectroscopy based portable analyzer In particular we have been developing silver-doped sol-gels to generate SER spectra of chemical agents and their hydrolysis products Here we present SER spectra of several chemical agents measured in a generic tap water Repeat measurements were performed to establish statistical error associated with SERS obtained using the sol-gel coated vials Keywords Chemical agents chemical agent detection SERS Raman spectroscopy
1 INTRODUCTION
In the autumn of 2001 terrorism within US borders became a sobering reality While extensive efforts are being implemented to secure the homeland US and Coalition forces in Afghanistan and Iraq are constantly faced with terrorist attacks In addition to car-bombings the military has identified several non-traditional attack scenarios including poisoning of water supplies by chemical warfare agents (CWAs) To counter this threat the Department of Defense has been investigating numerous approaches to detect such attacks under the auspices of the Joint Service Agent Water Monitor (JSAWM) program1 The JSAWM must be capable of identifying the chemical agents with no false-positives and quantifying the agents at microgram per liter concentrations within 10 minutes2 This includes the analysis of drinking water supplies distribution and storage systems Currently colorimetric paper is used to detect agents on-site while gas chromatography combined with mass spectrometry (GCMS) is used in mobile support laboratories However both methods have severe drawbacks The paper changes color in response to contact with many chemicals besides CWAs causing a high incidence of false positives3 while GCMS although very chemically specific requires up to an hour to perform and regular re-calibration456 Vibrational spectroscopy has also been investigated as a potential method of rapidly detecting CWAs7-
11 as early as 1985 in the case of infrared spectroscopy8 and 1988 in the case of Raman spectroscopy11 Again however these techniques also have limitations when it comes to measuring trace poisons in water Infrared spectra would be dominated by the very strong absorption of water which would obscure absorptions by most other chemicals present Whereas Raman spectroscopy is simply not a very sensitive technique and detection limits are typically grams per liter Surface-enhanced Raman spectroscopy (SERS) offers several advantages over conventional vibrational methods and that may provide the necessary sensitivity required for detecting trace quantities of chemical agents in water SERS employs the interaction of surface plasmon modes of metal particles with target analytes to increase scattering efficiency by as much as 1 million times12 In 1987 the potential of this approach to measure CWAs was demonstrated by measuring a series of pesticides13 Several of these organophosphonates have chemical structures similar to CWAs in particular P=O functional groups In the past few years we and others have further explored the ability of SERS to detect CWAs14-17 and even bioagents 18-21 We have been employing silver-doped sol-gels to promote the SER effect
SPIE-2003-5269 17
in these studies The porous silica network of the sol-gel matrix offers a unique environment for immobilizing and stabilizing SER-active metal particles22-25 The silver-doped sol-gels have been coated on the internal walls of standard glass vials to allow rapid SER analysis of small sample quantities without preparation We have measured over 100 chemicals with enhancements of 104 to 106 demonstrated reversible measurements in a flowing system reproducible measurements from vial-to-vial and measurements in multiple solvents including water21-25 Previously we used these vials to perform preliminary measurements of cyanide (CN) and methylphosphonic acid (MPA) Most of the nerve agents form MPA during hydrolysis while Tabun forms CN a chemical agent in its own right In another paper including in these proceedings we examined the limits of detection (LOD) for MPA by measuring a series of concentrations down to 50 mgL and estimated a limit of detection of 100 microgL26 These measurements provide encouragement in that SERS may satisfy the needs of the JSAWM To further establish the viability of SERS in particular silver-doped sol-gels here we present analysis of cyanide mustard and VX in tap water The measurements performed at the US Armyrsquos Edgewood Chemical Biological Center Aberdeen MD also included numerous repeat measurements to establish reproducibility
2 EXPERIMENTAL 2a General All chemicals including potassium cyanide 2-chloroethylethyl sulfide and those used to prepare the silver-doped and gold-doped sol-gel coated vials were obtained and used as received from Sigma-Aldrich All samples were prepared in a chemical hood using HPLC grade water unless otherwise noted (Fischer Scientific Fair Lawn NJ) for SERS measurements Once prepared the samples were transferred into the silver-doped sol-gel vials (Real-Time Analyzers Inc East Hartford CT) The vials were coated in a manner similar to that previously reported by adding ammonium hydroxide to a solution of silver nitrate tetramethyl orthosilicate (TMOS) and methanol22 The two precursor solutions were prepared mixed and transferred to 2-mL glass vials dried and heated After sol-gel formation the incorporated metal ions were reduced with dilute sodium borohydride (1mgmL) which is followed by a water wash to remove residual reducing agent After the resultant analyte solution was introduced the SER-active vial was fixed horizontally to an XY positioning stage (Conix Research Springfield OR) just inside the focal point of an f07 aspheric lens The lens focused the beam into the sample and collected the scattered radiation back along the same axis A dichroic filter (Omega Optical Brattleborough VT) was used to reflect the excitation laser to the lens and pass the Raman scattered radiation collected by the lens An f2 achromat was used to collimate the laser beam exiting a 200 microm core diameter source fiber optic while a second f2 achromat was used to focus the scattered radiation into a 365 microm fiber optic (Spectran Avon CT) A short pass filter was placed in the excitation beam path to block the silicon Raman scattering generated in the source fiber from reflecting off sampling optics and reaching the detector A long pass filter was placed in the collection beam path to block the sample Rayleigh scattering from reaching the detector A 785 nm diode laser (Process Instruments Inc model 785-600 Salt Lake City UT) was used to deliver 100 to 150 mW of power to the sample A Fourier transform Raman spectrometer (Real-Time Analyzers model IRA-785 East Hartford CT) and a silicon photo-avalanche detector (Perkin Elmer model C30902S Stamford CT) were used to acquire the SER spectra
2b Edgewood Chemical Biological Center The surface-enhanced Raman spectral measurements at the US Armyrsquos Chemical Biological center presented here were all performed on September 12 2003 To expedite measurements a plate was machined to hold up to 12 SER-active sample vials (Figure 1) The plate fit a standard XY plate reader that could be programmed Pure KCN bis-(2-chloroethyl)sulfide (distilled mustard gas HD) and ethyl S-2-diisopropylamino ethyl methylphosphonothioate (VX) were obtained on-site and used to prepare 1 gL tap water solutions in a chemical hood with appropriate safety equipment Simulated tap water was prepared by adding 10 chemicals most often found in tap water at appropriate concentrations to distilled water (Table 1) SER measurements were also performed in a chemical hood For added safety the FT-Raman instrument was placed outside the laboratory and 30 foot fiber optic and electrical cables were used to allow remote SERS measurements and plate manipulation For each experiment 1gL samples were prepared and added to 9 individual vials which were then loaded on the plate In some cases a tenth vial was included as a blank
Table 1 Generic Tap Water Composition
Compound mgL NaHCO3 100 CaSO4 27 MgSO4bull7H2O 67 NaNO3 1 Fulvic Acid 1 K2HPO4 07 KH2PO4 03 (NH4)2HSO4 001 NaCl 001 FeSO4 0001 pH 76-78
SPIE-2003-5269 18
A software program was written that allowed selecting the sequence that the vials were measured the number of positions along the length of the vials to measure (1 to 5) and the number of scans to co-add During sample analysis the program displayed the vial being analyzed the point being analyzed and the spectrum as it was being acquired Once all the data was collected a second software program was written to rapidly analyze the data The spectra collected for all the vials on a plate could be loaded at one time and then the spectra for each point could be displayed simultaneously or separately The user could then select the Raman peak to analyze in terms of peak height or area This was accomplished by selecting points on either side of the peak to define a baseline of zero The peak height or area could then be computed for all of the spectra loaded and then exported to a spreadsheet for statistical analysis
Figure 1 A) Vial Holder 6 slots to hold 2 vials each end-to-end B) Measurement Configuration Program user selects vials to measure sequence number of points per vial (1 to 5) and number of scans per point C) Spectral Acquisition Program shows spectrum being collected which vial and position D) Spectra Analysis Program user selects spectra to analyze by plate vial and point (s) as well as two wavenumbers defining the peak and the baseline to subtract The image is of 5 repeat measurements of 10 mgL KCN in generic tap water 16 sec each 100 mW of 785 nm
3 RESULTS AND DISCUSSION Raman and surface-enhanced Raman spectra were obtained for potassium cyanide bis-(2-chloroethyl)sulfide and ethyl S-2-diisopropylamino ethyl methylphosphonothioate representing three classes of chemical agents cyanides mustards and nerve agents respectively Spectra were also obtained for 2-chloroethyl ethyl sulfide (CEES) a structural analogue to HD which was included in the study to aid in assigning spectral bands KCN salt was used for cyanide experiments to avoid the increased hazards of handling HCN gas KCN completely dissolves in water forming its conjugate acid HCN according to its Ka of 615x10-1027 and at a concentration of 1 mgmL results in a pH 107 solution This is significant in that only CN- appears to interact sufficiently with silver to produce a SER spectrum and no spectral signal is observed below pH 726 Figure 2 shows the SER and normal Raman spectra for KCN The SERS spectrum of 1mgml KCN in water shows a single intense somewhat broad feature at
A B
C D
SPIE-2003-5269 19
2100 cm-1 assigned to the single CequivN stretch The band is much sharper in the normal Raman spectra of the solid KCN salt at 2074 cm-1 This band does broaden and shift to 2080 cm-1 in solution (not shown) However the observed SERS frequency is attributed to interaction with silver and low frequency mode at 135 cm-1 attributed to a Ag-CN stretch (not shown) supports this conclusion
cm-1 band to a C-S stretch but the authors concede that it is in fact more likely a C-Cl stretch28 It appears that the most intense Raman bands at 648 692 and 747 cm-1 shift to 620 660 and 720 cm-1 in the SER spectra and are tentatively assigned as above The width of these bands suggests that they overlap underlying spectral features Additional bands in the Raman spectra occur at 972 1034 1049 1263 1286 1423 1442 2865 2935 and 2960 cm-1 Corresponding bands occur in the SER spectra at 964 1015 1054 1286 1410 1447 2865 and 2935 cm-1 Most of these bands are associated with alkane modes specifically the bands at approximately 1040 cm-1 to a C-C stretch 1290 cm-1 to a CH2 in-phase twist 1440 cm-1 to a CH2 wag 2865 cm-1 to a symmetric CH2 stretch and 2965 cm-1 to an asymmetric CH2 stretch The Raman and SER spectra of sulfur mustard were measured at the Edgewood center (Figure 4) Both spectra are largely similar to CEES The C-Cl and C-S bands in the Raman spectrum of HD now occur at 640 655 700 739 and 760 cm-1 and are more resolved possibly due to the increased molecular symmetry Theoretical calculations indicate that the first three bands are due to C-Cl stretching modes and the latter two to C-S stretching modes28 Only the C-Cl bands maintain significant intensity in the SER spectra occurring at 624 and 643 cm-1 which is attributed to the
Prior to measurements of HD CEES was examined by Raman and SER spectroscopy (Figure 3) CEES also known as half-mustard is essentially identical to HD except one of the chlorine end atoms is replaced by a hydrogen atom Again although not as toxic as HD CEES is a blister agent and dilute aqueous samples were prepared using appropriate safety equipment Both the Raman and SER spectra of CEES are similar and dominated by bands between 600 and 800 cm-1 These are associated with C-Cl and C-S stretching modes which are tentatively assigned to 648 and 747 cm-1 in the Raman spectra respectively The shoulder at 630 cm-1 the overlapped band at 660 cm-1 and the strong band at 692 cm-1 could also be due to these modes or their asymmetric counterparts It is worth noting that theoretical calculations assign the 692
Figure 2 A) SER and B) NR spectra of KCN Conditions A) 1 mgml in tap water 100 mW of 785 nm at sample 1-min acquisition time B) solid 300 mW of 785 nm 5-min All spectra are 8 cm-1 resolution
A
B
Figure 3 A) SER and B) NR spectra of CEES Conditions A) 1 vv (10 mgml) in MeOH 100 mW of 785 nm 1-min acquisition time B) neat 300 mW of 785 nm 5-min
A
B
Cl-CH2-CH2-S-CH2-CH3
A
B
Figure 4 A) SER and B) NR of HD Conditions A) 1mgml in tap water B) pure both 100 mW of 785 nm 1-min
Cl-CH2-CH2-S-CH2-CH2-Cl
SPIE-2003-5269 20
expected strong interaction between chlorine and silver and adds support to the assignment of this band to a C-Cl stretch Weaker overlapping bands occur at 670 692 and 724 cm-1 the latter possibly due to C-S stretching modes Again the alkane modes are apparent in the normal Raman spectra of HD but only a broad feature at 1300 to 1450 cm-1 suggests CH2 contributions in the SER spectrum Although the observed bands in the VX spectrum have not been assigned (Figure 4) a computer generated Raman spectrum29 predicts many of the same features with surprising accuracy and are used here Two intense bands at 460 and 530 cm-1 closely match predicted bands at 463 and 546 cm-1 assigned to a CH3-P=O bend and a PO2CS wag Three highly overlapped bands occur at 694 745 and 771 cm-1 matching predicted bands at 713 730 and 760 cm-1 The first
Table 2 Measured SER peak heights for the CN stretch at 2100 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
Figure 5 A) SERS and B) NR spectra of VX Conditions A) 1 vv (10 mgml) in MeOH B) pure sample both 100 mW of 785 nm 1-min acquisition time
A
B
two have been assigned to a C-S stretch and CH2bend respectively while the latter has been attributed to either a P-C stretch or an O-C-C stretch Although the 745 cm-1 band may alternatively be assigned to a C-S stretch based on the previous measurements of CEES and HD The relatively intense bands at 890 1106 1218 1445 and 1465 cm-1 also match predicted bands at 880 1108 1216 1440 and 1464 cm-1 that are assigned to a C-C stretch CH3 rock N-C3 stretch various C-H3 bends and C-H bends respectively Both the computer generated and the measured spectra contain numerous other less intense bands One is worth mentioning A unique band appears at 370 cm-1 that is predicted at 368 cm-1 and corresponds to an O-P=O bend The surface-enhanced Raman spectrum of VX is also rich with spectral features It has the unique low frequency band at 370 cm-1 as well as a second band at 380 cm-1 that is assigned tothe S-P-O bend predicted in the normal Raman spectrum at 388 cm-1 Based on the measured and predicted normal Raman spectra the following SERS assignments are given 460 cm-1 to the CH3-P=O bend 544 cm-1 to the PO2CS wag 738 cm-1 to a C-S stretch (based on arguments above) 890 cm-1
to a C-C stretch 1101 cm-1 to a CH3 rock and 1456
cm-1 to a C-H bend The ability of SERS to measure chemical agents in water containing real-world chemical interferents was tested by using the generic tap water described in Table 1 The ability to reproduce measurements was accomplished by preparing three separate water stock solutions which were used to prepare three sample solutions each of 1 mgmL KCN in the generic tap water The samples defined as vials 1-9 were then measured at 4 points per vial in a semi-random fashion such that errors associated with stock solution preparation and errors associated with instrument drift could be identified No trends were apparent that signified such systematic errors Each spectrum collected consisted of 20 averaged scans taking 16 seconds at 8 cm-1 resolution The laser power at the sample was measured periodically during the day and it ranged from 102 to 105 mW spacccc
spot Vial 1 Vial 2 Vial 41 13754 13004 128192 13519 12692 12909
1 3 13541 12721 126394 13462 12648 12651
ave 13569 12766 12755Vial 6 Vial 3 Vial 5
1 11586 14068 10757 2 11236 14402 115122 3 11376 14546 11559
4 10894 11763 11214ave 11273 13695 11261
Vial 7 Vial8 Vial 91 11198 15804 11023
2 11243 15716 88753 3 11562 15216 11218
4 11655 15084 11077ave 11415 15455 10548
CN AVG STDEV ERR1 mgmL pk ht 12526 1560 1245
stock solution
SPIE-2003-5269 21
The CN measurements consisting of 36 data points produced an average height of 125 for the 2100 cm-1 peak with a standard deviation of 156 or 125 (Table 2) The HD and VX measurements were performed precisely the same way (Tables 3 and 4) For HD the 624 cm-1 peak was used for analysis and it had an average height of 53 with a standard deviation of 068 or 129 while for VX the 544 cm-1 peak was used for analysis and it had an average height of 1051 with a standard deviation of 308 or 293 The greater error in the VX measurements can be somewhat attributed to Vial 4 which produced lower SER signal intensities But removing this vial from the data set changes the standard deviation to 233 only a modest improvement
spot Vial 1 Vial 2 Vial 41 525 609 4682 45 675 484
1 3 527 644 5094 559 807 669
ave 51525 68375 5325Vial 6 Vial 3 Vial 5
1 521 574 575 2 536 53 4492 3 509 508 451
4 565 594 379ave 53275 5515 4635
Vial 7 Vial 8 Vial 91 457 589 413
2 583 497 4273 3 544 505 4
4 528 5 509ave 528 52275 43725
HD AVG STDEV ERR1 mgmL pk ht 530 068 1291
stock solution
spot Vial 1 Vial 2 Vial 41 1464 1034 4672 1485 99 61
1 3 1491 989 5684 1041 777 553
ave 137025 9475 5495Vial 6 Vial 3 Vial 5
1 1058 942 1293 2 697 121 965
2 3 727 1206 11124 689 1254 746
ave 79275 1153 1029Vial 7 Vial 8 Vial 9
1 1127 783 163 2 1358 812 1614
3 3 1371 908 15124 1097 875 1397
ave 123825 8445 153825
VX AVG STDEV ERR1 mgmL pk ht 1051 308 2925
stock solution
4 CONCLUSIONS In this paper we examined the ability of surface-enhanced Raman spectroscopy to reproducibly measure CN HD and VX in tap water without chemical interference Both normal and surface-enhanced Raman spectra were examined to select unique bands suitable to identify and quantify these chemical agents For SER measurements the 2100 cm-1 C-N stretch was used for CN the 624 cm-1 C-Cl stretch was used for HD and the 544 cm-1 PO2CS wag was used for VX It was determined that 1 mgmL samples of each of these chemicals measured 36 times in glass vials coated with a silver-doped sol-gel reproduced measurements with standard deviations of 125 129 and 293 It was further found that the 10 chemicals added to simulate generic tap water did not interfere with or alter the SER spectra It should be noted that the concentrations used in this study were considerably greater than those required by the JSAWM program Current work involves improving SER sensitivity and designing sampling systems with better reproducibility This includes the development of fractal silver and gold structures within the sol-gel matrix and the development of chemically selective sol-gels
ACKNOWLEDGMENTS The authors are grateful for the support of the US Army (DAAD13-02-C-0015 Joint Service Agent Water Monitor program) The authors would also like to thank Janet Jensen Ronald Crosier and Kristina Gonser for helpful discussions
Table 4 Measured SER peak heights for the VX band at 544 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
Table 3 Measured SER peak heights for the HD band at 624 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
SPIE-2003-5269 22
REFERENCES 1 Jensen JL NC Wong W R Loerop SPIE 4775-03 (2001) 2 JSAWM Requirements at wwwsbccomapgeaarmymilRDAecbcrtPRODSERJSAWMjsawmhtml 3 Erickson B ldquoThe Chemical Weapons Convention Redefines Analytical Challengerdquo Anal Chem 70 397A (1998) 4 Johnston RL Hoefler CM Fargo JC and Moberley B ldquoThe Defense Nuclear Agencyrsquos
ChemicalBiochemical Weapons Agreements Verification Technology Research Development Test and Evaluation Program and its Requirements for On-Site Analysisrdquo AT-ONSITE 5-8 (1994)
5 Black RM Clarke RJ Read RW and Reid MT ldquoApplication of GC-MS and GC-tandem MS to the analysis of chemical warfare samples found to contain residues of the nerve agent sarin sulphur mustard and their degradation productsrdquo J Chrom 662 301-321 (1994)
6 USA Army Center for Health Promotion and Preventive Medicine (USACHPPM) Technical Guideline 230A 7 Hoffland LD Piffath RJ Bouck JB rdquoSpectral signatures of chemical agents and simulantsrdquo Optical
Engineering 24 982-984 (1985) 8 Braue EHJ Pannella MGrdquoCIRCLE CELL FT-IR Analysis of Chemical Warfare Agents in Aqueous Solutionsrdquo
App Spectrosc 44 1513-1520 (1990) 9 Tseng C-H Mann CK and Vickers TJ ldquoDetermination of Organics on Metal Surfaces by Raman
Spectroscopyrdquo App Spectrosc 47 1767-1771 (1993) 10 Staff Reporter ldquoUS Army Tests Lidar to Detect Biological Toxinsrdquo Photonic Spectra pg 50 December 1998 11 Christesen SD Raman cross sections of chemical agents and simulants App Spectrosc 42 318-321 (1988) 12 Weaver MJ Farquharson S Tadayyoni MA ldquoSurface-enhancement factors for Raman scattering at silver electrodesrdquo J Chem Phys 82 4867-4874 (1985) 13 Alak AM and Vo-Dinh T ldquoSERS of Organophosphorous Chemical Agentsrdquo Anal Chem 59 2149-2153 (1987) 14 Lee Y Farquharson S ldquoRapid chemical agent identification by SERSrdquo SPIE 4378 21-26 (2001) 15 Farquharson S Maksymiuk P Ong K Christesen S ldquoChemical agent identification by surface-enhanced Raman spectroscopyrdquo SPIE 4577 166-173 (2001) 16 Spencer KM Sylvia J Clauson S and Janni J ldquoSurface Enhanced Raman as a Water Monitor for Warfare
Agents in Waterrdquo SPIE 4577 158-165 (2001) 17 Tessier P Christesen S Ong K Clemente E Lenhoff A Kaler E Velev O ldquoOn-line spectroscopic characterization of sodium cyanide with nanostructured gold surface-enhanced Raman spectroscopy substratesrdquo
App Spectrosc 56 1524-1530 (2002) 18 Farquharson S W W Smith S Elliott and J F Sperry Rapid biological agent identification by surface- enhanced Raman spectroscopy SPIE 3855110-116 (1999) 19 Farquharson S WW Smith YH Lee S Elliott and J F Sperry Detection of biological signatures A
comparison of electrolytic and metal-doped sol-gel surface-enhanced Raman media SPIE 4575 62-72 (2002) 20 Guzelian AA Sylvia J Janni J Clauson S and Spencer KM ldquoSERS of whole cell bacteria and trace levels of biological moleculesrdquo SPIE 4577 182-192 (2001) 21 Shende C Inscore F Gift A Maksymiuk P Farquharson S ldquoRapid dipicolinic acid extraction from Bacillus spores detected by surface-enhanced Raman spectroscopyrdquo App Spectrosc 58 accepted 22 Lee Y and Farquharson S ldquoSERS Sample Vials Based on Sol-Gel Process for Trace Pesticide Analysisrdquo SPIE
4206 140-146 (2000) 23 Farquharson S and Lee Y ldquoTrace Drug Analysis by SERSrdquo SPIE 4200-16 (2000) 24 Lee Y Farquharson S and Rainey P M Surface-Enhanced Raman Sensor for Trace Chemical Detection in
Water SPIE 3857 76-84 (1999) 25 Lee Y Farquharson S Kwong H and Shahriari M ldquoSol-Gel Chemical Sensor for Surface-Enhanced Raman
Spectroscopyrdquo SPIE 3537 252-260 (1998) 26 Farquharson S A Gift P Maksymiuk F Inscore and W Smith ldquopH dependence of methyl phosphonic acid
dipicolinic acid and cyanide by surface-enhanced Raman spectroscopyrdquo SPIE 5269-19 (2003) 27 Lide DR Ed Handbook of Chemistry and Physics CRC Press 77th Ed (1996-7) 28 Sosa C RJ Bartlett K KuBulat and WB Person ldquoA theoretical study of harmonic vibrational frequencies and
infrared intensities of XCH2CH2SCH2CH2X and XCH2CH2SH (= H Cl)rdquo J Phys Chem 93 577-588 (1993) 29 Hameka HF and JO Jenson ldquoComputer-generated predictions of the structure and the IR and Raman spectra of
VXrdquo ERDEC-TR-065 May 1993
- Chemagents Appendicespdf
-
- SPIE2001-4575-Bioagent-SERSapdf
-
- Detection of bioagent signatures A comparison of electrolytic and metal-doped sol-gel surface-enhanced Raman media
- Stuart Farquharson Wayne Smith and Yuan Lee
-
- Real-Time Analyzers 87 Church Street East Hartford CT 06108
-
- ABSTRACT
- 4 CONCLUSIONS
- 5 ACKNOWLEDGEMENTS
- 6 REFERENCES
-
SPIE 2001-4575
68
The SG-SER spectrum of AMP is considerably different The adenine bands virtually disappear except for the two primary bands which shift to742 and 1329 cm-1 While the phosphate band at 1459 cm-1 has gained considerable intensity In addition two new intense bands appear at 684 and 1539 cm-1 as well as a moderately intense band at 1180 cm-1 The SG-SER spectrum has greater similarity to the E-SER spectrum at -03VSCE and suggests that the silver particles embedded in the sol-gel behave as if at a potential positive of the pzc RNA samples extracted from E coli B subtilis and S aureus were next examined by both E-SERS and SG-SERS E-SER spectra of these samples yielded quality spectra in 10 minutes in which all of the major features can be identified (Figures 6 and 7) This includes guanine at 650 cm-1 adenine at 791 cm-1 cytosine and uracil combining at 790 cm-1 and phosphate at 1100 1335 (in combination with adenine and guanine) 1465 and 1570 cm-1 Surprisingly adenine which demonstrated the greatest surface-enhanced Raman effect does not dominate the ring-breathing mode portion of the spectrum The intensities of the other base-pairs bands are of the same order of magnitude This suggests that when the base-pairs are linked together as in RNA they are enhanced in concert In fact the relative intensities are very similar to a normal Raman spectrum of E coli RNA which shows the combined cytosine and uracil band at ca twice the intensity of the adenine band and ca four times the intensity of the guanine band Unfortunately this means that the independent enhancement factors for the NA bases can not be used to estimate relative concentrations For example the relative 791 and 734 cm-1 bands for B subtilis would indicate that the cytosine andor uracil concentration was at least 20 times the adenine concentration whereas each of the four RNA bases are known to contribute 15-35 Nevertheless it is worth noting that the three RNA samples yield different relative band intensities that were reproduced in numerous measurements Although the relative concentrations of the NA bases for these samples have not been determined these differences can be quantified If it is assumed that the 650 cm-1 band represents 25 guanine the 791 cm-1 band represents 25 adenine and the 790 cm-1 50 cytosine plus uracil in the E coli RNA spectrum then the relative concentrations can be estimated for the other RNA samples To aid this calculation the three spectra were normalized to the phosphate band at 1100 cm-1 which has been shown to correlate to the total phosphate concentration and can be used as an internal standard In addition a simple baseline correction was applied (Figure 7) This yields 15 adenine 30 guanine and 55 cytosine plus uracil for B subtilis RNA and 18 adenine 25 guanine and 57 cytosine plus uracil for S aureus RNA The average SN of these measurements was 26 with an average error of 8 of the value (SplusmnN) It is also worth noting that the three RNA spectra show a marked shift in a band near 825 cm-1 This band is assigned to the symmetric stretch of the O-P-O ester linkage9 The band appears at 815 cm-1 for S aureus shifting to 820 cm-1 for B subtilis and 830 cm-1 for E coli Others have used the normal Raman intensity of the band at 815 cm-1 as a direct indication of the amount of A-class helix present while the intensity of the band at 830 cm-1 has been used as a direct indication of the amount of B-class helix present However the latter is more associated with DNA than RNA
Wavenumbers (∆cm-1) Figure 5 E-SER spectra of 020 mgmL adenosine monophosphate at A) -03 and B) -09VSCE and C) SG-SER spectra Conditions sample in 01M KCl buffered to pH 92 A) and B) 750 mW C) 75 mW of 1064 64 scans (1-min) at 8 cm-1
Wavenumbers (∆cm-1) Figure 6 E-SERS of 01 mgmL RNA from E coli 02 mgmL RNA from B subtilis and 02 mgmL RNA from S aureus Conditions 01M KCl pH 92 -03VSCE 750 mW of 1064 nm 640 scans (10 min) at 8 cm-1
AMP RNA
E coli A
B
C
B subtilis
S aureus
SPIE 2001-4575
69
SG-SER spectra of reasonable quality were also obtained for E coli and B subtilis especially the latter (Figure 8) However the spectra differ substantially from the E-SERS of the same samples Both SG-SER spectra are dominated by adenine at 735 cm-1 and a band at 1030 cm-1 Although unassigned the latter does appear in the RNA E-SER spectra Bands at 1105 and 1565 cm-1 are likely due to phosphate while bands at 1320 and 1470 cm-1 are less confidently assigned to phosphate They are significantly less intense and somewhat shifted from their SG-SERS counterparts (1335 and 1455 cm-1) A band at 670 cm-1 may be due to guanine which was observed at 664 cm-1 for SG-SERS of the pure sample However the SG-SER spectrum of AMP also had an intense 667 cm-1 band A number of other bands occur at 890 1070 1165 1245 1290 1420 1505 cm-1 and remain unassigned The SG-SER spectra are somewhat disappointing in that only adenine and guanine contributions can be positively identified This limits the ability to determine relative NA base concentrations and distinguish bacterial RNA However several of the unassigned bands may be due to the bases (eg 1030 and 1420 cm-1 due to cytosine) Further experiments will be required to clarify this point
Ecoli
B subtilis
S aureas
P G
A C+U
OPO
A B
Wavenumbers (∆cm-1) Wavenumbers (∆cm-1) Figure 7 SER spectra of RNA from A) B subtilis with contributions indicated and B) E coli B subtilis and S aureus with baseline correction and peak positions used to calculate contributions indicated G = guanine A = adenine C+U = cytosine plus uracil P = phosphate (backbone) OPO = phosphate ester linkage (A- vs B-class helix)
Wavenumbers (∆cm-1) Wavenumbers (∆cm-1) Figure 9 E-SER (-03VSCE) and SG-SER spectra of RNA from A) E coli and B) B subtilis Sample conditions as in Figure 6 E-SER spectra at 750 mW SG-SERS at 75 mW
A B E-SERS
SG-SERS
E-SERS
SG-SERS
SPIE 2001-4575
70
A final comparison was made between the two SER methods by measuring dipicolinic acid (DPA) This chemical may be invaluable as a test for spore forming bacteria specifically B anthracis 50 to 90 of B anthracis sporilates During spore formation dipicolinic acid is synthesized and once completed 10-15 of the dry spore weight is composed of the Ca2+ complex located in the spore core42 Heating in water can be used to initiate germination at which point the exosporium breaks and releases the Ca dipicolinate which becomes dipicolinic acid in water The structure of this chemical strongly
4 CONCLUSIONS Towards the goal of developing a method and technology to rapidly detect and identify bioagents we have been investigating surface-enhanced Raman spectroscopy as a tool to measure relative concentrations of nucleic acid bases in RNA and determine bioagent taxonomy Initially we investigated E-SERS since this method has been extensively researched and the optimum sample conditions are well developed However this method requires a three-electrode sample cell and electrolyte solution Incorporation of an electrolytic cell into sample systems used to analyze for BWA as aerosols in water or on surfaces can be designed using flow injection analysis technologies but cross-contamination and plugging of sample lines seems inevitable For this reason we also investigated metal-doped sol-gels as a SER-active medium Previous studies have shown this material to be active in all solvents particularly water capable of continuous measurements in flowing systems and reproducible (quantitative) between coated sample vials Here we compared SG-SER spectra to traditional E-SER spectra of the nucleic acid base pairs adenosine monophosphate and RNA High quality spectra of adenine cytosine guanine thymine and uracil were obtained by both E-SERS and SG-SERS Both methods yielded very similar spectra for the NA bases including a pH dependent study of adenine Enhancement factors and detection limits for adenine were determined as 2x105 and 16x10-11g and 1x105 and 12x10-10g for E-SERS and SG-SERS respectively (normalized to 75 mW and 10-min acquisition time) Fifty percent of the silver particle surface area in the sol-gel matrix was assumed covered by adenine which may have been overestimated yielding a lower EF and higher detection limit It should also be realized that each E-SER spectrum required several attempts to optimize the measurement conditions (pH electrode potential etc) While each SG-SER spectrum involved no sample preparation and often represents the first and only attempt to make the measurement Quality spectra of RNA extracted from Escherichia coli Bacillus subtilis and Staphylococcus aureus were obtained by E-SERS that were easily interpreted Bands due to adenine guanine cytosine plus uracil and phosphate were identified The SER band intensity of the NA bases in the RNA samples were of the same order of magnitude suggesting that their interaction with the silver surface is concerted as is their Raman enhancement Interestingly the relative SER band
Wavenumbers (∆cm-1) Figure 10 A) Raman spectrum of solid dipicolinic acid B) Ax20 C) electrolytic SERS of 6x10-3 M dipicolinic acid in 01 M KCl at a potential of +07VSCE and pH of 4 and D) sol-gel SERS of 6x10-3 M dipicolinic acid Conditions for A and C as in Figure 1 C) 100 mW of 1064 nm 50 scans 8 cm-1
A B (Ax20)
C
D
suggested that it would be SER active However the E-SER spectrum was unstable and varied considerably as a function of potential A consistent spectrum was obtained at +06VSCE (Figure 10) This potential is not recommended for measurement because the surface is actively dissolving in solution The SG-SER spectrum was considerably more stable of higher quality and easily reproduced Bands at 660 825 1010 1390 1430 1570 1590 and 3075 cm-1 were observed Enhancement factors were determined for the two media using the symmetric ring stretching mode at 995 and 1010 cm-1 for the normal Raman and SER spectra respectively E-SERS yielded an EF of 5x103 while SG-SERS yielded an EF of 2x105 for DPA The SN of the latter suggests a detection limit of 20x10-10g (based on adenine coverage 75 mW and 10-min) The differences in SER activity for these two media may be attributed to the combined electrolytic potential of the solution chemical and metal15 Again the E-SERS suggests that the SG-SERS is at a potential positive of the pzc While the instability in the E-SERS may also be associated with surface interactions of two carboxylic acid groups of dipicolinic acid during the ORCs
DPA
SPIE 2001-4575
71
intensities for RNA extracted from E coli are very similar to those measured by normal Raman spectroscopy Although the relative percent that each of the NA bases contributed to each RNA sample was not determined reproducible band intensities allowed noting the following trends The percent adenine decreases while the combined percent cytosine and guanine increase for both B subtilis and S aureus compared to E coli Quality spectra were also obtained for the RNA samples by SG-SERS but only a few bands were readily identified Calculations of NA base concentrations by SG-SERS will require further research In light of recent events we summarize the capabilities of these SERS media in terms of rapid detection of B anthracis and dipicolinic acid However these capabilities must be qualified First and foremost the level to which SERS can distinguish bacteria or viruses has not yet been determined Development of a database of both DNA and RNA base concentrations for BWAs and common bacteria to establish the level of taxonomic identification is ongoing Second rapid collection of aerosol water or surface samples is being addressed by others who report trapping particles on filters from 100 liters of air per minute Third although not presented here we have developed methods to extract RNA or DNA from cells and spores for SER analysis within 10 minutes Finally we assume a detection limit of 3600 spores per 100 liters of air is required although a 50 lethal dosage of anthrax has not been established With these qualifications a mass detection limit for RNA using SERS is estimated as follows A single measurement is performed in ca 20 minutes (140 liters collected in 14 min RNA extracted in 8 min spectral acquisition and analysis in 10 min) The average human breaths 7 liters per minute therefore the analyzer must at the very minimum detect 5000 spores in 140 liters of air One spore is approximately 2x10-18m3 (1x1x2 microm) and if a density of 075 gcm3 is assumed this corresponds to a mass of 15x10-12g Each spore contains 4-12 RNA or 12x10-13g RNA for 8 If we assume 23 of the RNA can be isolated for analysis during lysis then the proposed instrument must be able to detect 4x10-10g RNA from 5000 spores per 70 liters of air within 10 minutes As noted above the mass detection limits for adenine were estimated at 16x10-11g and 12x10-10g for E-SERS and SG-SERS respectively Although these detection limits suggests that RNA from 5000 spores is detectable with the current instrumentation it is highly likely that only a portion of an RNA segment (eg 120-nucleotide 5S rRNA) is in contact with the metal surface and will contribute to the SER effect The SN for the RNA spectra were 110th of the average SN for the four individual RNA bases suggesting a 10 contribution Furthermore effective taxonomy will likely require knowing the NA base concentrations to 1 of the value (eg 25plusmn025) Again the average measurement error for the bases is 12 These values suggest that the E-SERS is within a factor of 4 of the required detection limit whereas the SG-SERS detection limit must be improved by 25 times The same arguments can be applied to the detection of dipicolinic acid If we assume a spore releases 10 by weight DPA during germination then the proposed instrument must be able to detect 75x10-10g DPA from 5000 spores per 70 liters of air within 10 minutes The detection limit for SG-SERS was estimated at 20x10-10g and suggest that the vials are suitable to perform a rapid screen for anthrax A series of concentration dependent measurements are currently being performed to verify this assertion Finally we note that the measurements performed here employed an FT-Raman spectrometer This instrumentation was chosen over dispersive Raman instrumentation for the high wavelength accuracy afforded by the HeNe reference laser (Connes Advantage43) which would allow reliable spectral subtraction matching of observed spectra to stored library spectra and confident use of chemometric approaches Such data analysis is likely to be required to enhance BWA identification However this instrumentation sacrifices sensitivity due to a less efficient detector (InGaAs vs Si) less efficient Raman scattering and less efficient generation of plasmon modes Substantial improvements can be obtained using visible excitation and Si detection and these measurements are underway
5 ACKNOWLEDGEMENTS The authors are grateful to Drs D Cookmeyer and S Tove of the US Army Research Office (Contract Number DAAH04-96-C-0078) for their interest and support of this research The authors would also like top acknowledge Dr R Yin and J Jensen for supporting development of the metal-doped sol-gels (Contract Number DAAD13-01-C-0019) They also thank Dr Wilfred H Nelson for assistance in spectral interpretations
SPIE 2001-4575
72
6 REFERENCES 1 Roberts WL and Rainey PM Clin Chem 39 1872-1877 (1993) 2 Pasechnik VA CC Shone and P Hambleton Bioseperations 3 267-283 (1993) 3 Jackson PJ ME Hugh-Jones DM Adair G Green KK Hill CR Kuske LM Grinberg FA Abramova and P
Keim Proc Natl Acad Sci 95 1224-1229 (1998) 4 Jensen JL NC Wong W R Loerop SPIE 4775-03 (2001) 5 Snyder A Pet al SPIE 3853-15 (1999) 6 Danosky T R and McFadden P N in press (1997) 7 Woodruff WH Farquharson S Science 201 831-833 (1978) 8 Lord RC and Thomas GJJr Spectrochemica Acta 23A 2551-2591 (1967) 9 Thomas GJJr Biochim Biophys Acta 213 417-423 (1970) 10 Placzek G Handbuch der Radiologie 2 EMarx ed Akademische Verlagagescellschatt Liepzig 1934 UCRL
Trans No 526 (1959) 11 Jeanmaire DL and RP Van Duyne J Electroanalytical Chem 84 1-20 (1977) 12 Chada S Manoharan R Moenne-Loccoz P Nelson WH Peticolas WL and Sperry JF Applied Spectroscopy
47 38-43 (1993) 13 Manoharan R Ghiamati E Chada S Nelson WH and Sperry JF Applied Spectroscopy 47 2145-2150 (1993) 14 Todd EA Morris MD Applied Spectroscopy 48 545-548 (1994) 15 Wentrup-Byrne E Sarinas S and Fredericks PM Applied Spectroscopy 47 1192-1197 (1993) 16 Weldon MK VR Zhelyaskov and MD Morris Applied Spectroscopy 52 265-269 (1998) 17 Kneipp K and J Fleming J Mol Structure 145 173-179 (1986) 18 Pothier NJ and Force RK Applied Spectroscopy 46 147-151 (1992) 19 Ervin KM E Koglin JM Sequaris P Valenta and HW Nurnberg J Electroanal Chem 114 179-194 (1980) 20 Kim SK TH Joo SW Suh and MS Kim J Raman Spectrosc 17 381-386 (1986) 21 Pothier NJ and Force RK Analytical Chemistry 62 678-680 (1990) 22 Kneipp K Y Wang RR Dasari and MS Feld Applied Spectroscopy 49 780-784 (1995) 23 Nie S and Emory SR Science 275 1102 (1997) 24 Graham D WE Smith AMT Linacre CH Munro ND Watson and PC White Analytical Chemistry 69 4703-
4707 (1997) 25 Fleischmann M PJ Hendra and AJ McQuillan Chem Phys Lett 26 163-166 (1974) 26 Pettinger B J Chemical Phys 85 7442-7451 (1986) 27 Surface-Enhanced Raman Scattering Section Four Theory SPIE MS 10 M Kerker and B Thompson Eds (1990) 28 Wang D-S and Kerker M SPIE (M Kerker and B Thompson Eds) MS 10 417-429 (1990) 29 Weaver MJ Farquharson S Tadayyoni MA J Chem Phys 82 4867-4874 (1985) 30 Norrod KL Sudnik LM Rousell D and Rowlen KL Applied Spectroscopy 51 994-1001 (1997) 31 Farquharson S Weaver WJ Lay PA Magnuson RH and Taube H J Am Chem Soc 105 3350-3351 (1983) 32 Lee PC and Meisel D ldquoAdsorption and Surface-Enhanced Raman of Dyes on Silver and gold Solsrdquo J Phys Chem
86 3391-3395 (1982) 33 Li Y-S and Wang Y Applied Spectroscopy 46 142-146 (1992) 34 Bello JM DL Stokes and T Vo-Dinh Analytical Chemistry 61 1779-1783 (1989) 35 van Duyne RP J C Hulteen D A Treichel M T Smith M L Duval and T R Jensen J Phys Chem B1033854-
3863 (1999) 36 Farquharson S and W W Smith W H Nelson and J F Sperry SPIE 3533-27 207-214 (1998) 37 Lee Y H W Smith S Farquharson H C Kwon M R Shahriari and P M Rainey SPIE 3537 252-260 (1998) 38 Lee Y-H S Farquharson and P M Rainey SPIE 3857 76-84 (1999) 39 Current Protocols in Molecular Biology Wiley Interscience1003-1006 (1987) 40 Farquharson S Smith W Carangelo R C and Brouillette C SPIE 3859 14-23 (1999) 41 Chase D B and JF Rabolt Fourier Transform Raman Spectroscopy Acad Press Ch1 p 131 (1994) 42 Brock TD MT Madigan JM Martinko and J Parker Biology of Microorganisms 7th Ed Prentice Hall p 76-80
(1994) 43 Connes J Rev Opt Theor Instrum 40 45 (1961)
166
Chemical agent identification by surface-enhanced Raman spectroscopy
Stuart Farquharson and Paul Maksymiuk
Real-Time Analyzers 87 Church Street East Hartford CT 06108
Kate Ong and Steven D Christesen US Army SBCCOM Aberdeen Proving Ground MD 21010
ABSTRACT The recent distribution of anthrax through the US postal system and the subsequent infection and death of several postal and national media employees amplifies the need for methods to rapidly detect identify and quantify this and other chemical and biological warfare agents The US military has also identified water supplies as a likely method of warfare agent deployment and is funding the development of a Joint Service Agent Water Monitor (JSAWM) In an effort to aid military personnel and the public at large we are developing a portable analyzer capable of identifying and quantifying chemical agents rapidly either on-demand or continuously The approach is based on the ability of Raman spectroscopy to identify molecular structure through the abundant vibration information provided in spectra and the ability of surface-enhanced Raman spectroscopy (SERS) to detect extremely low concentrations (eg part-per-billion) through the enhancement of Raman scattering by six orders of magnitude or more A key element to the analyzer design is a new SER active medium that is capable of quantitative reversible measurements The medium consists of silver or gold nanoparticles incorporated into a sol-gel matrix The porous silica network offers a unique environment for stabilizing SER active metals and the high surface area increases the interaction between the analyte and metal particles Here we present the use of new sol-gels that also selectively enhance chemicals based on polarity and charge Base-line measurements of chemical agents and their hydrolysis products are presented and compared to the JSAWM goal of 30 micrograms per liter detection Keywords Chemical warfare agent hydrolysis product SERS Raman spectroscopy sol-gel nanoparticle
1 INTRODUCTION Since September 11 2001 the threat of terrorist attacks and biological warfare within US borders has become a sobering reality The simplicity in manufacturing ease of deployment and the relatively low cost of chemical warfare agents (CWAs) raises public concern that they may also be used by terrorists Indeed terrorists released sarin (GB) in the Tokyo subway in 19951 Countering terrorism and terrorist attacks requires recognizing likely deployment scenarios and having the required technology to rapidly detect the deployment event One method of deployment has been long identified by the US military distribution through water supplies To counter this threat the Department of Defense is funding or monitoring the capabilities of 200 technologies with the goal of developing a Joint Service Agent Water Monitor (JSAWM) that is field portable2 The JSAWM must be capable of identifying the chemical agents with no false-positives and quantifying the agents at microgram per liter concentrations within 10 minutes (Table 1)3 This includes the analysis of drinking water supplies distribution and storage systems as well as potable water supplies The earliest technologies developed for CWA detection were based on electrochemical ionization or colorimetric analysis (eg phosgene tape) Although these analyzers were easy to use they were not generally agent specific and suffered from false-positives4 More traditional laboratory methods have also been investigated and in particular combined gas chromatography and mass spectrometry (GCMS) has been very successful at eliminating false-positives56 However GCMS requires extraction repeated calibration and long analysis times (typically 20 to 60 minutes)6 making it labor intensive and less than desirable for field use More rapid analysis of agents in the solid liquid and gas phase has been To whom correspondence should be addressed emailfarqureal-time-analyzerscom
Vibrational Spectroscopy-based Sensor Systems Steven D Christesen Arthur J Sedlacek III EditorsProceedings of SPIE Vol 4557 (2002) copy 2002 SPIE middot0277-786X02$1500
167
Table 1 Chemical Agent Structures Hydrolysis Half-lives and JSAWM Thresholds Agent Short-Hand Chemical Structure Hydrolysis
Half-Life JSAWM
Thresholds
Sarin (GB) F-[O=P-CH3]-O-CH(CH3)2 213 hours 32 microgL Soman (GD) F-[O=P-CH3]-O-CH(CH3)-(C-(CH3)3) 23 hours 32 microgL Tabun (GA) (CH3)2-N-[O=P-CN]-O-C2H5 41 hours 32 microgL VX C2H5O-[O=P-CH3]-S-(CH2)2-N-(CH(CH2)2)2 821 hours 32 microgL EA2192 HO-[O=P-CH3]-S-(CH2)2-N-(CH(CH2)2)2 gt9 years 32 microgL Mustard (H) ClCH2CH2-S-CH2CH2Cl encapsulates 47 microgL Lewisite (L) ClCH=CH-As-Cl2 rapid 27 microgL HCN HCN rapid 20 mgL BZ C7NH12-O-[C=O]-COH(C6H5)2 23 microgL T-2 Toxin 87 microgL at pH 7 to 75 and 20 to 25 oC
demonstrated by vibrational spectroscopy7-10 Hoffland et al7 reported infrared absorbance spectra and absolute Raman cross sections for several chemical agents while Christesen measured Raman cross sections for sarin tabun mustard gas and VX (ethyl S-2-diisopropylamino ethyl methylphosphonothioate)11 Again however these techniques also have limitations Raman spectroscopy is simply not a very sensitive technique and detection limits are typically 01 (1000 ppm) While infrared spectroscopy would have limited value in analyzing poisoned water since the very strong infrared absorption of water would obscure most other chemicals present Nevertheless efforts to overcome these limitations have been demonstrated Braue and Pannella8 quantified the G-series nerve agents (tabun sarin and soman) in terms of infrared attenuated total reflectance using a circle-cell And Alak and Vo-Dinh demonstrated the possibility of surface-enhanced Raman spectroscopy (SERS) to identify CWAs by measuring several organophosphonates that simulate the nerve agents12 However quantitative measurements have not been demonstrated for the SER-active material used (silver coated on alumina particles) or other SER-active media13 Recently we developed silver-doped sol-gels to promote the SER effect14-17 The porous silica network of the sol-gel matrix offers a unique environment for stabilizing SER-active metal particles and the sol-gel provides a high surface area that effectively increases the number of molecules observed within the Raman scattering volume The silver-doped sol-gels have been coated on the internal walls of standard glass vials to allow rapid SER analysis of small sample quantities (lt 01 mL) without preparation We have used p-aminobenzoic acid (PABA) as a test chemical to demonstrate surface enhancements greater than 106 reversible measurements in a flowing system reproducible measurements from vial-to-vial and batch-to-batch and measurements in multiple solvents including water14-17 Recently we used these vials to measure Tabun (GB) and Sarin and several hydrolysis products pinacolyl methyl phosphonate (PMP from Soman) and methyl phosphonic acid (MPA from all G-agents Figure 1) Although a number of unique vibrational bands are observed (eg C-N stretch doublet and P-C stretch) the G-agents were only observed for 5 concentrations and all spectra required baseline corrections Figure 1 Surface enhanced Raman spectra of ~5 vv A) Tabun and B) Sarin C) 1 vv PMP and D) 10 ppm MPA using sol-gel sample vials 785 nm excitation 1-min scan and CCD detection Performed at Aberdeen Proving Ground
C A
Wavenumber (∆cm-1) Wavenumber (∆cm-1)
D B
790
545 C-N
2135 2190 P-C 770
760 1290
Proc SPIE Vol 4577
168
Nevertheless MPA was readily observed for a 10 ppm sample with an estimated detection limit of 04 ppm (based on a signal-to-noise ratio of 3 for the 760 cm-1 band intensity) This measurement provides encouragement in that SERS may satisfy the needs of a JSAWM Furthermore MPA is also a hydrolysis product of VX and V-gas and EA2192 (Figure 2) and may prove a valuable indicator of agent usage Figure 2 Hydrolysis of Sarin to form hydrofluoric acid (HF) methylphosphonic acid 1-methylethyl ester (MPAMME) methyl phosphonic acid (MPA) and 2-propanol With this initial albeit modest success we began analyzing chemicals with various sol-gel compositions that we have been developing Here we describe four sol-gel compositions that select for 1) polar-positive 2) polar-negative 3) weakly polar-positive and 4) weakly polar-negative chemical species The ability of these sol-gels to select and enhance Raman scattering is described for several test chemicals and MPA
2 EXPERIMENTAL The chemicals analyzed as well as all chemicals used to prepare the metal-doped sol-gels were obtained at their purest commercially available grade from Aldrich (Milwaukee WI) The sol-gel designed to select for polar-negative species was prepared from a silver amine complex tetramethyl orthosilicate (TMOS) and methanol After mixing 02 mL of the sol-gel solution was transferred into a glass vial (2 mL) dried and heated The incorporated silver ions were then reduced using dilute sodium borohydride The vials were washed and dried prior to the addition of a sample solution In a similar manner the sol-gel designed to select for polar-positive species was prepared from a gold salt TMOS and methanol The sol-gel designed to select for weakly polar-negative species was prepared from a silver amine complex tetraethyl orthosilicate (TEOS) and methanol And the last sol-gel designed to select for weakly polar-positive species was prepared from a gold salt TEOS and methanol All samples were prepared in a chemical hood and transferred into plain or SER-active vials for analysis Normal Raman spectral measurements employed 1-mL pure samples that were placed in a 1-cm3 cuvette and weighed This yielded a powder density that allowed accurate calculation of molecules in the optical collection field SERS measurements employed 1-mg sample per mL water concentrations unless otherwise stated Once prepared a 01 mL sample was placed into one of the four selective sample vials which in turn was placed into the sample compartment of a Raman spectrometer for analysis A prototype Fourier transform Raman spectrometer (RTA) was used for these measurements18 The system consisted of a NdYAG laser (Brimrose) for excitation at 1064 nm an interferometer built by On-Line Technologies (OLT East Hartford CT) for frequency separation an uncooled InGaAs detector for signal detection (RTA) and an Intel 400 MHz Pentium II based laptop computer (Dell Round Rock TX) for interferometric control data acquisition (OLT) and analysis (LabVIEW by National Instruments Austin TX) Additional components included a Notch filter (Kaiser Ann Arbor MI) and interferometer entrance and exit optics (Edmund Scientific Barrington NJ) Fiber optics were used to deliver the excitation beam to the sample and the scattered radiation to the interferometer (1 meter lengths of 200 and 365 micron core diameter respectively Spectran Avon CT) A second Notch filter (Kaiser) was used as a beam splitter to direct the excitation beam along the same axis as the collected radiation A microscope object (20x04 Newport Irvine CA) was used to focus the beam into the sample and to collect the scattered radiation back along the same axis In this co-axial backscattering arrangement the excitation beam was passed through the outside of a glass vial and focused onto the silver-doped sol-gel film (01-03 mm thickness) containing the sample
3 RESULTS AND DISCUSSION p-aminobenzoic acid (PABA) and phenyl acetylene (PA) and were used to refine the selectivity and SER-activity of the four different metal-doped sol-gels PABA is a popular chemical used to evaluate the performance of SER-active media Here the polar end groups can be used to test selectivity of the polar-negative and polar-positive sol-gels PA is essentially non-
2O+ H HF + +OH
OH3H C
OP
OF
CH
CH3
3
3H C
OCP
OOH
CH
CH3
3
3H C
O
CPHO
CH
CH
3
3
C
Sarin MPAMME MPA 2-propanol
Proc SPIE Vol 4577
169
polar but a high electron density in the cylindrical π cloud around the carbon-carbon triple bond allows testing the selectivity of the weakly polar-negative and weakly polar-positive sol-gels As Figure 3 illustrates PABA passes through the polar sol-gel and is enhanced by either the silver or gold particles At 1 mgml the concentration of neutral PABA is ca 20 times that of the ionized form (pKa = 48) For electropositive silver the PABA anion is expected to interact through the carboxylate group and the associated vibrational modes are expected to dominate the spectrum Conversely for electronegative gold either form of PABA is expected to interact through the amine group The clear differences in our spectra support this expectation Furthermore bands at 840 and 1405 cm-1 assigned to a COO- bend and stretch respectively are significantly more intense for silver than gold Additional bands at 1140 and 1195 cm-1 are assigned to CH bending modes while bands at 1450 1500 and 1605 cm-1 are assigned to ring vibrational modes A very similar SER spectrum for PABA on a silver-coated alumina substrate has previously been reported with similar assignments19 For the gold-doped sol-gel new bands appear at 690 1355 and 1585 cm-1 The first band is assigned to a ring-H bending mode the second band to a ring-N- stretching mode and the third band to a possible NH2 scissors mode or ring mode The second band is not observed in the normal Raman spectrum but infrared bands occur at this frequency for aromatic ring-secondary amine stretching modes The scissors mode occurs at this frequency in Raman spectra for several chemicals but is absent in the PABA Raman spectrum Alternatively this mode may be the1600 cm-1 ring mode that has been shifted by the gold interaction Again a very similar SER spectrum of PABA has been reported but surprisingly using silver (colloids)2021 not gold as the enhancement medium These researchers also assumed the primary interaction of PABA with silver was through the carboxylate anion and made assignments accordingly For example they assigned the 1359 cm-1 to a COO- stretch not to the amine group as we have They also favor the ring stretching mode assignment for the 1582 cm-1 band Finally it should be said that other researchers have argued that the most dominant band in the SER spectra at 1450 cm-1 a ring vibration mode suggests that PABA lies flat on the surface and the π-orbitals dominate the surface interaction22
Figure 3 SER spectra of A) PABA using polar-negative and B) polar-positive sol-gels and C) PA using weakly polar-negative and D) weakly polar-positive sol-gels PABA is 1 mgmL PA is 1 vv Spectral conditions 75 mw 1064 nm 100 scans (15 min) 8 cm-1 resolution Non-polar PA passes through the non-polar sol-gels and is also enhanced by both metals The spectra are easily understood For electropositive silver PA interacts through the cylindrical triple bond π electron cloud and a -CequivC- doublet occurs near 2000 cm-1 The interaction is reasonably strong since this band appears at 2112 cm-1 in the normal Raman spectrum For electronegative gold this interaction is unlikely and only very weak bands occur near 2000 cm-1 The remaining bands are at 1000 cm-1 1200 cm-1 doublet and 1595 cm-1 all appear in the normal Raman spectra at virtually the same frequencies and are assigned to the symmetric ring-breathing mode CH bending modes and the trigonal ring-breathing mode respectively The polarnon-polar selectivity of the polar-negative and weakly polar-negative sol-gels was tested by adding a 11 molar mixture of PABA and PA The selective enhancement is quite good (Figure 4) The spectrum obtained using the polar sol-gel represents 78 PABA and 22 PA while the spectrum obtained using the weakly polar sol-gel represents 9 PABA and 91 PA The band peak intensities at 2000 cm-1 for PA and 1450 cm-1 for PABA were used for these calculations and are expanded in Figure 4 for clarity
C A
Wavenumber (∆cm-1) Wavenumber (∆cm-1)
2NH COOH
D B
C CH
Proc SPIE Vol 4577
170
Figure 4 SERS of 11 MM of PABA and PA in A) polar-negative and C) weakly polar-negative sol-gels The lower traces compare the pure chemicals B) 1 mgml PABA in polar-negative sol-gel and D) 1 PA in weakly polar-negative sol-gel while the insets magnify the minority species for clarity (x5 in A and x10 in B) Spectral conditions as in Figure 3 Following this development of selective sol-gels that maintained SER activity we measured cyanide and MPA (Figure 5) Not surprisingly the best sensitivity for both hydrolysis products was obtained using the polar-negative sol-gel The interaction of the cyanide anion with the silver surface is sufficient to shift the CequivN stretch observed at 2080 cm-1 in the normal Raman spectrum to 2145 cm-1 in the surface-enhanced Raman spectrum Furthermore the band is substantially broadened This anion has been extensively studied by electrolytic SERS and this shift and broadening have been attributed to the formation of a tetrahedral Ag(CN)3
2- surface structure23 Figure 5 Surface-enhanced (upper traces) and normal Raman spectra (bottom traces) of A) CN- and B) MPA in silver-doped TMOS SERS conditions as in Figure 3 and 1 mgmL Note MPA yields two distinct spectra for neutral (top) and acidic pH (middle) The normal Raman spectra employed pure powders 500 scans and 900 mW of 1064 nm SER measurements of MPA with the polar-negative sol-gel yielded two unique spectral signatures that depended on solution pH (Figure 5) For more neutral solutions the P-C stretch of MPA at 762 cm-1 dominates and the CH2 stretch at 2922 cm-1 appears The SN is sufficiently high that the anti-Stokes Raman shift at -762 cm-1 is observed For deprotonated MPA an oxygen-surface mode appears at 325 cm-1 (as well as its anti-Stokes complement) suggesting a strong interaction This results in substantial enhancement of the P-O-C mode at 1051 cm-1 (upper trace) Others report that this mode dominates the infrared spectra of nerve agents measured in water8 Comparison of the two spectra suggests the following molecule-to-surface orientations The appearance of the oxygen-surface and P-O-C modes in the upper spectrum of Figure 5B indicates that the tetrahedral molecule interacts with the silver surface through the deprotonated oxygen and is oriented end-on The
A B
Wavenumber (∆cm-1)
C A
Wavenumber (∆cm-1) Wavenumber (∆cm-1)
D B
Wavenumber (∆cm-1)
Proc SPIE Vol 4577
171
dominance of the P-C and the CH2 stretches and the disappearance of the P-O-C mode in the upper spectrum suggest the molecule is oriented with the methyl-hydroxyl-hydroxyl face towards the surface However considerably more research must be performed to verify these points Both chemical forms were measured as a function of concentration to determine the onset of monolayer coverage and estimate expected detection limits (Figure 6) Below monolayer coverage the signal to concentration dependence should be linear and the SN of any spectral measurement in this range can be used to predict the detection limit In the spectra presented here the peak height was used as the signal while the noise as root-mean-squared (RMS) was measured between 4400-4600 cm-1 Since noise is distributed evenly throughout the spectrum when transformed this region was used since it does not have any contributions from signals or baseline offsets Figure 6 shows a series of spectra for MPA along with a plot of the 1050 cm-1 band intensity (the noise was identical for this series) as a function of sample concentration A clear discontinuity appears in the vicinity of 01 mgmL (19 ppm) indicating the onset of monolayer coverage A detection limit defined as a SN of 3 was calculated for the 01 and 005gmL samples at 24x10-4 and 25x10-4 gL respectively A more modest detection limit of 101x10-4 gL was obtained using the 760 cm-1 band in the second series of concentration measurements These detection limits correspond to 46 and 190 ppb for a 3-min scan and 75 mW of incident laser power
Figure 6 A) Concentration dependence of MPA SERS measured in silver-doped TMOS) B) Concentrations are 001 005 01 05 1 gL (188 94 188 94 188 ppm) I760 series (bull) and I1050 series (∆)
Enhancement factors for cyanide and methyl phosphonic acid can be determined by estimating the number of molecules contributing to the surface-enhanced and normal Raman spectra The enhancement factor EF can be defined by the following equation
EF = (ISERSINR)bull(MNRMSERS) bull(PNRPSERS) bull(TNRTSERS)12
where I is the spectral band intensity M is the sample mass P is the incident laser power and T is the measurement time (or number of scans) for the two measurements For the normal Raman spectra a cylindrical scattering volume is assumed based on the laser area (28x10-7m2 6x10-4m diameter spot) and the penetration depth (1x10-3 m)24 The density of KCN and MPA as powders were measured at 0572 and 0516 gcm3 indicating that 16x10-4 and 144x10-4 g produced the normal Raman signals in Figure 5 respectively The number of molecules contributing to the SER spectra are those on the silver particles embedded in the sol-gel The total silver surface area can be determined from the average particle size concentration and the scattering volume Previous scanning electron micrographs showed the average silver particle size to be 40 nm in diameter (335x10-23m3)17 The silver concentration is 012M based on the reactant molar concentrations and dilution factors And the scattering volume is 76x10-
11m3 again based on a cylindrical scattering volume defined by a laser area of 28x10-7m2 and a sol-gel thickness of 27x10-
4m This volume contains 123x10-6g of silver equivalent to 35x109 silver particles with a collective surface area of 18x10-
5m2 However it may reasonably be assumed that at least half of the silver surface is in contact with the sol-gel matrix and unavailable for analyte interaction If we assume monolayer coverage and that each CN molecule occupies 15x10-20m2 then approximately 62x1014 molecules or 27x10-8g of CN contribute to the SER spectrum (20x10-19m2 46x1013 molecules 74x10-9g for MPA) Accordingly the EF for cyanide equals 48x104 ((180599) bull(16x10-427x10-8) bull(90075) bull(500100)12) The EF for MPA is considerably higher at 87x106 ((60326) bull(144x10-474x10-9) bull(90075) bull(500200)12)
0
20
40
60
80
100
120
140
0 02 04 06 08 1 12[MPA] (mgmL)
I (76
0)
0
100
200
300
400
500
600
I (10
50)
Wavenumber (∆cm-1)
A B
Proc SPIE Vol 4577
172
4 CONCLUSIONS Here we present for the first time surface-enhanced Raman spectra of Tabun and Sarin obtained using silver-doped sol-gels However the inferior enhancement suggested employing SER-active sol-gels varying in composition to improve both detection limits and selectivity of the target analytes To this end we successfully demonstrated the capabilities of four sol-gels that select for 1) polar-positive 2) polar-negative 3) weakly polar-positive and 4) weakly polar-negative chemical species p-aminobenzoic acid was used to show that silver could be used to attract polar-negative chemicals or functional groups (carboxylate anion) and that gold could be used to attract polar-positive chemicals (amine) while a mixture of p-aminobenzoic acid and phenylacetylene was used to show that tetramethyl orthosilicate preferentially solvates polar chemicals and that tetraethyl orthosilicate preferentially solvates weakly polar chemicals This increased sample control was applied to cyanide and methyl phosphonic acid two hydrolysis products of chemical warfare agents Exceptional results were obtained for methyl phosphonic acid allowing measurement of 1x10-2 gL for a 3-min scan and 75 mW of incident laser power with an estimated detection limit of 245x10-4 gL and an enhancement factor of 87x106 However this detection limit is 76 times less sensitive than required for the JSAWM (32x10-6gL for the G-agents) Finally we note that the measurements performed here employed an FT-Raman spectrometer This instrumentation was chosen over dispersive Raman instrumentation for the high wavelength accuracy afforded by the HeNe reference laser (Connes Advantage25) which would allow reliable spectral subtraction matching of observed spectra to stored library spectra and confident use of chemometric approaches Such data analysis is likely to be required to identify the chemical agents as well as distinguish hydrolysis products However this instrumentation which employs 1064 nm excitation and InGaAs detection sacrifices sensitivity We believe that the measurement sensitivity can be improved by at least two orders of magnitude by using shorter laser excitation wavelengths (eg 532 nm) This would provide more efficient Raman scattering (fourth power dependence on laser excitation wavelength) more efficient generation of plasmon modes and allow using more efficient detector material (Si vs InGaAs) These modifications are underway
5 ACKNOWLEDGEMENTS The authors would like to thank Dr R Yin and J Jensen of the US Army for supporting this work (Contract Number DAAD13-01-C-0019) They would also like to thank Advanced Fuel Research for making their laboratory facilities available
6 REFERENCES 1 Tu Anthony ldquoOverview of Sarin Terrorist Incidents in Japan in 1994 and 1995rdquo 6th CBW Protection Symposium
Stockholm Sweden 10-15 May 1998 2 Jensen JL NC Wong W R Loerop SPIE 4775-03 (2001) 3 JSAWM Requirements at wwwsbccomapgeaarmymilRDAecbcrtPRODSERJSAWMjsawmhtml 4 ldquoThe Chemical Weapons Convention Redefines Analytical Challengerdquo Analytical Chemistry News amp Features June 1
397A (1998) 5 Johnston RL Hoefler CM Fargo JC and Moberley B ldquoThe Defense Nuclear Agencyrsquos ChemicalBiochemical
Weapons Agreements Verification Technology Research Development Test and Evaluation Program and its Requirements for On-Site Analysisrdquo AT-ONSITE 5-8 (1994)
6 Black RM Clarke RJ Read RW and Reid MT ldquoApplication of gas chromatography-mass spectrometry and gas chromatography-tandem mass spectrometry to the analysis of chemical warfare samples found to contain residues of the nerve agent sarin sulphur mustard and their degradation productsrdquo J Chromatography 662 301-321 (1994)
7 Hoffland LD Piffath RJ Bouck JBrdquoSpectral signatures of chemical agents and simulantsrdquo Optical Engineering 24 982-984 (1985)
8 Braue EHJ Pannella MGrdquoCIRCLE CELL FT-IR Analysis of Chemical Warfare Agents in Aqueous Solutionsrdquo Applied Spectroscopy 44 1513-1520 (1990)
9 Tseng C-H Mann CK and Vickers TJ ldquoDetermination of Organics on Metal Surfaces by Raman Spectroscopyrdquo Applied Spectroscopy 47 1767-1771 (1993)
10 Staff Reporter ldquoUS Army Tests Lidar to Detect Biological Toxinsrdquo Photonic Spectra pg 50 December 1998
Proc SPIE Vol 4577
Proc SPIE Vol 4577
173
11 Christesen SD Raman cross sections of chemical agents and simulants Applied Spectroscopy 42 318-321 (1988) 12 Alak AM and Vo-Dinh T ldquoSERS of Organophosphorous Chemical Agentsrdquo Analytical Chemistry 59 2149-2153
(1987) 13 Norrod KL Sudnik LM Rousell D and Rowlen KL ldquoQuantitative Comparison of Five SERS Substrates
Sensitivity and Detection Limitrdquo Applied Spectroscopy 51 994-1001 (1997) 14 Lee Y and Farquharson S ldquoSERS Sample Vials Based on Sol-Gel Process for Trace Pesticide Analysisrdquo SPIE 4206
140-146 (2000) 15 Farquharson S and Lee Y ldquoTrace Drug Analysis by Surface-Enhanced Raman Spectroscopyrdquo SPIE 4200-16 (2000) 16 Lee Y Farquharson S and Rainey P M Surface-Enhanced Raman Sensor for Trace Chemical Detection in Water
SPIE 3857 76-84 (1999) 17 Lee Y Farquharson S Kwong H and Shahriari M ldquoSurface-Enhanced Raman Sensor for Surface-Enhanced Raman
Spectroscopyrdquo SPIE 3537 252-260 (1998) 18 Farquharson S Smith W Carangelo R C and Brouillette C ldquoIndustrial Raman Providing Easy Immediate Cost
Effective Chemical Analysis Anywhererdquo SPIE 3859 14-23 (1999) 19 Narayanan VA JM Bello JD Stokes and T Vo-Dinh Analusis 19 307-310 (1991) 20 Laserna JJ E L Torres and JD Winefordner Analytica Chemica Acta 469-480 (1987) 21 Torres EL and JD Winefordner Analytical Chemistry 59 1626-1632 (1987) 22 Suh JS DP DiLella M Moskovits J Phys Chem 87 1540-1544 (1983) 23 Benner RE R Dornhaus R Chang and BL Laube Correlations in the Raman spectra of cyanide complexes adsorbed
at silver electrodes with voltammograms Surface Science 101 341 (1980) 24 Chase D B and JF Rabolt Fourier Transform Raman Spectroscopy Acad Press Ch1 p 131 (1994) 25 Connes J Rev Opt Theor Instrum 40 45 (1961)
APPLIED SPECTROSCOPY 351
focusing the 488 nm laser beam 2 mm deep into thebulk of the crystals to avoid contributions from potentialdepletion layers As shown in Fig 1b the dependence ofthe Raman shift on the degree of deuteration is almostperfectly linear and ts very well with D 5 22684cmR1 24526 where D is the degree of deuteration (in )and R is the spectral mean of the PO4 vibration in cm21A linear correlation coef cient of 0998 indicates an ex-cellent linear dependence of the Raman peak shift withdegree of deuteration This result shows that the shift ofthe PO 4 peak is simply caused by the linear increase inatomic mass due to isotope substitution which decreasesthe length of hydrogen-like bonds
This excellent linear dependence allows us to map thepro le of the DH exchange layer at the surface of DKDPcrystals by acquiring Raman spectra and determining theposition of the PO 4 peak for various depths This methodis preferable over other methods such as determining thestrength of the OD vibration directly (eg at 715 cm21)because the position of the most intense peak in the Ra-man spectrum can be measured more precisely than theintensity of some of the weakest peaks in the spectrumThis is demonstrated in Fig 2 where depth-dependentRaman spectra (Fig 2a) and the resulting exchange layerpro les for two DKDP crystals are shown (Fig 2b) Thespectra in Fig 2a were obtained from a depth scan of aDKDP crystal with 75 degree of deuteration in thebulk grown at 45 8C The spectra start out as DKDP with30 deuteration close to the surface and approach thebulk DKDP spectrum within a few micrometers of depthThe fact that the relative degree of deuteration does notextend to 0 D is due to the limited depth resolution ofthe Raman microprobe which averages over 4 mm indepth Figure 2b depicts the resulting DH exchange layerpro les for this and a second crystal grown at 63 8Crespectively Both crystals had the same exposure to am-bient conditions and their main difference is the temper-ature at which they were grown The different exchangelayer pro les indicate that crystals grown at differenttemperatures have differen t proton conductiv ities 13
which leads to a difference in their rate of deuteriumdepletion The parameters controlling this behavior arecurrently the objective of a detailed study the results ofwhich will be reported elsewhere
CONCLUSION
In conclusion we have shown that the shift of the to-tally symmetric PO 4 stretch mode in the Raman spectrumof DKDP crystals scales linearly with degree of deuter-ation This allows us to correlate Raman peak positionsto deuteration levels in these crystals We have presenteda new technique to determine DH diffusion pro les inDKDP frequency conversion crystals based on micro-Ra-man spectroscopy This technique is fast inexpensiveand works under various environmental conditionswhich will allow us to better understand and control deu-terium depletion in DKDP crystals
ACKNOWLEDGMENTS
We would like to thank M Runkel for rst discovering DKDP crack-ing R Floyd for providing DKDP crystals and L Chase and A Burn-ham for their support and helpful discussions This work was performedunder the auspices of the US Department of Energy by the University
of California Lawrence Livermore National Laboratory under contract W-7405-Eng-48
1 J J De Yoreo A K Burnham and P K Whitman Int Mat Rev47 113 (2002)
2 C E Barker R A Sacks B M Van Wonterghern J A Caird JR Murray J H Campbell K Kyle R B Ehrlich and N DNielsen Proc SPIE-Int Soc Opt Eng 2633 501 (1995)
3 T Suratwala paper to be published4 Cleveland Crystals Inc httpwwwclevelandcrystalscom
KDPshtmltable5 E A Popova I T Savatinova and I A Velichko Sov Phys Solid
State 12 1543 (1971)6 I P Kaminow R C C Leite and S P S Porto J Phys Chem
Solids 26 2085 (1965)7 J A Subramony B J Marquardt J W Macklin and B Kahr
Chem Mat 11 1312 (1999)8 H Tanaka M Tokunaga and I Tatsuzaki Solid State Commun
49 153 (1984)9 R J Nelmes G M Meyer and J E Tibballs J Phys C 15 59
(1982)10 M A Yakshin D W Kim Y S Kim Y Y Broslavets O E
Sidoryuk and S Goldstein Laser Physics 7 941 (1997)11 I Takenaga Y Tominaga S Endo and M Kobayashi Solid State
Commun 84 931 (1992)12 C Krenn personal communication13 M Sharon and A K Kalia J Solid State Chem 21 171 (1977)
Rapid Dipicolinic Acid Extractionfrom Bacillus Spores Detectedby Surface-EnhancedRaman Spectroscopy
STUART FARQUHARSON ALAND GIFT PAUL MAKSYMIUK andFRANK E INSCOREReal-Time Analyzers Inc East Hartford Connecticut06108
Index Headings Dipicolinic acid Bacillus spores Anthrax Surface-enhanced Raman spectroscopy
INTRODUCTION
The anxiety caused by the distribution of anthrax en-dospores through the US postal system in October 2001was exacerbated by the long time required for positiveidenti cation of the Bacillus anthracis spores and the un-known extent of their distribution Since that time manymethods capable of rapid eld analysis have been inves-tigated to augment or replace the laboratory method ofgrowing microorganisms in culture media which takesdays to perform12 Prominent among these approachesare polymerase chain reactions (PCR)3 immunoassaysand detection of calcium dipicolinate as a biochemicalsignature PCR employs primers to separate organism-
Received 10 October 2003 accepted 14 November 2003 Author to whom correspondence should be sent
352 Volume 58 Number 3 2004
speci c nucleic acid sequences (eg capsular protein en-coding gene for Bacillus anthracis)4 and polymerases toamplify the segment until it is detectable Recently am-pli cation times have been substantially reduced andcomplete analysis can now be performed in an hour orless Immunoassay methods are also being developed thatuse competitive binding of the bioagent (as an antigen)and its labeled conjugate for a limited number of anti-bodies Although analyses can be performed in under 30minutes a well-de ned anthrax antigen has not yet beenidenti ed5ndash7 and consequently the false-positive rate isunacceptably high8
A number of other methods are being developed witha focus on the detection of calcium dipicolinate (CaDPA)and its derivatives as a B anthracis signature This is sobecause only spore-forming bacteria contain CaDPA andthe most common potentially interfering spores such aspollen and mold spores do not Relatively fast methodshave been developed to chemically extract CaDPA andthen detect it directly by uorescence9 or indirectly byluminescence1011 In the latter case hot dodecylamine(DDA) has been used to extract dipicolinic acid (DPA)and terbium has been utilized to form a highly lumines-cent DPA complex11 Although measurements have beenperformed in as little as ve minutes it was found thatas many as three concentration-dependent complexes canform each with different lifetimes This coupled withthe fact that the Tb31 cation produces the same lumines-cence spectrum makes determinations of low spore con-centrations problematic
It has been long known that Raman spectra of Bacillispores are dominated by bands associated with CaDPA12
and that these spectra may provide a suitable anthrax sig-nature at the genus level13 Since that time considerableimprovements in Raman instrumentation have led to lab-oratory measurements of single Bacilli spores14 and to eld measurements of spores captured from a mail-sort-ing system15 However the single spore measurementsrequired complex instrumentation that is not ruggedwhile the eld measurements required milligrams of sam-ple Furthermore the Raman spectra of both measure-ments contained uorescence contributions that would in-crease uncertainty in quanti cation
In related research we demonstrated that nanogramquantities of DPA could be detected by uorescence-freesurface-enhanced Raman spectroscopy (SERS)16 We alsodemonstrated that microliter volumes of chemicals canbe detected by SERS using metal-doped sol-gel-packedglass capillaries17 Towards the goal of developing a rap-id eld SERS-based anthrax spore detector we havecombined our previous research and we now report thatDPA can be extracted from a 10 mg B cereus spore sam-ple using DDA in 1 minute and can be detected by SERSin an additional 1 minute
EXPERIMENTAL
Dipicolinic acid (26-pyridinedicarboxylic acid DPA)and dodecylamine (DDA) were used as received fromSigma-Aldrich (Milwaukee WI) Lyophilized B cereusspores prepared according to the literature13 were sup-plied by the University of Rhode Island and used as re-ceived Multiple particles approximately 01 mm3 each
were separated and weighed at 5 to 15 mg representing05 to 15 million spores The sample masses were con-sistent with a previous determination of spore density at0081 gmL that indicated a high degree of entrained air
All chemicals used to prepare the silver-doped sol-gelcoated capillaries were also obtained and used as receivedfrom Sigma-Aldrich According to previously publishedprocedures17 two precursor solutions were preparedmixed and then drawn into 1-mm-diameter glass capil-laries The silver amine precursor consisted of a 51 vvratio of 1 N AgNO 3 to 28 NH3OH while the alkoxideprecursor consisted of a 21 vv ratio of methanol to te-tramethyl orthosilicate The alkoxide precursors weremixed with silver amine precursor in an 81 vv ratioApproximately 015 mL was drawn into the capillarycoating a 15-mm length After sol-gel formation the in-corporated silver ions were reduced with dilute sodiumborohydride which was followed by a water wash to re-move residual reducing agent
A 100 mL drop of a 50 mM DDA solution in ethanolpre-heated to 78 8C was added to each of the B cereusparticles to digest the spore coat After 1 minute the re-sultant solution was drawn into a SER-active capillarythat was immediately xed horizontally to an XY posi-tioning stage (Conix Research Spring eld OR) just in-side the focal point of an f 07 aspheric lens The lensfocused the beam into the sample and collected the scat-tered radiation back along the same axis A dichroic lter(Omega Optical Brattleborough VT) was used to re ectthe excitation laser to the lens and pass the Raman scat-tered radiation collected by the lens An f 2 achromatwas used to collimate the laser beam exiting a 200-mm-core-diameter source ber optic while a second f 2 ach-romat was used to focus the scattered radiation into a 365mm ber optic (Spectran Avon CT) A short-pass lterwas placed in the excitation beam path to block the sil-icon Raman scattering generated in the source ber fromre ecting off sampling optics and reaching the detectorA long-pass lter was placed in the collection beam pathto block the sample Rayleigh scattering from reachingthe detector A 785 nm diode laser (Process InstrumentsInc model 785-600 Salt Lake City UT) was used todeliver 100 to 150 mW of power to the sample A Fouriertransform Raman spectrometer (Real-Time Analyzersmodel IRA-785 East Hartford CT) and a silicon photo-avalanche detector (Perkin Elmer model C30902S Stam-ford CT) were used to acquire the SER spectra
RESULTS AND DISCUSSION
As an initial experiment the SER spectrum of 1 gLof DPA in water was measured using the newly devel-oped silver-doped sol-gel-coated capillaries (Fig 1A) Atthis concentration a high signal-to-noise ratio (SN) isobtained in 1 min In fact a reasonable spectrum is ob-tained in the same time frame for 1 mgL (Fig 1B) TheSER spectra are reasonably similar to the normal Raman(NR) spectrum obtained for a saturated solution of DPAin 1 N KOH (Fig 1C) and the following band shifts areobserved (NR to SER) 647 to 657 cm21 817 to 815cm21 998 to 1008 cm21 1384 to 1382 cm21 1434 to1428 cm21 and 1569 to 1567 cm21 Many of these bandshave been previously assigned1213 such as 998 cm21 to
APPLIED SPECTROSCOPY 353
FIG 1 SERS of DPA in water using silver-doped sol-gel-coated glasscapillary for (A) 1 gL and (B) 1 mgL (C ) NR of saturated DPA in 1N KOH in a glass capillary Spectral conditions (A) and (B) 150 mWof 785 nm 1-min acquisition time (C) 450 mW of 785 nm 5-minacquisition time both 8 cm 21 resolution
FIG 2 SERS of DPA extracted from 10 mg B cereus particle using100 mL of 50 mM hot DDA acquired in (A) 1 minute and (B) 2 seconds(C ) Attempted SERS of 50 mM hot DDA in ethanol using silver-dopedsol-gel-coated glass capillary acquired in 1 min Spectral conditions150 mW of 785 nm 8 cm21 resolution
the symmetric ring stretch 1384 cm21 to the OndashCndashOsymmetric stretch 1428 cm21 to the symmetric ring CndashH bend and 1569 cm21 to the asymmetric OndashCndashOstretch
The rst B cereus samples consisted of 2 mg of sporesin 2 mL of 5 mM hot DDA The samples were main-tained at 78 8C for 40 min and while hot approximately10 mL was drawn into a SER-active capillary Since spec-tra of DPA were obtained for these initial samples small-er spore masses higher DDA concentrations and shorterheating periods were examined In due course it wasfound that 10 mg of spores could be digested by 100 mLof 50 mM hot DDA in one minute and detected (Fig2A) In fact the signal was suf ciently intense that it canbe observed in as little as two seconds (Fig 2B) Theamount of DPA that was extracted was estimated to bebetween 5 and 10 mgL by comparing the signal intensityof the 1008 cm21 band to that measured for DPA in waterThis is consistent with previous research that found thatthe majority of the DPA is extracted from spores usingDDA11 and that B cereus spores contain approximately10 DPA by weight18 The SN of 127 for the 1008 cm21
band in the 1-minute SER spectrum suggests a limit ofdetection of approximately 250 ng of B cereus sporesbased on a SN of 3 Finally it should be noted that DDAdid not produce a detectable SER spectrum as shown inFig 2C
CONCLUSION
We have demonstrated that by combining rapid extrac-tion of dipicolinic acid from Bacillus cereus spores withchemical identi cation by surface-enhanced Raman spec-troscopy as little as 10 mg of spores can be detected Infact the entire measurement from the time of adding hotdodecylamine to the spores to the time when the dipi-colinic acid SER spectrum is acquired and analyzedcould be performed in less than two minutes The abilityof this method to distinguish between spore-forming bac-teria such as Bacillus anthracis and non-DPA containingpowders could help prevent costly shutdowns associated
with the appearance of suspicious material or intentionalmailing of common substances as an anthrax hoax Thismethod could also prove useful in detecting the locationof anthrax endospores in mail distribution facilities if an-other veri ed attack should occur
Research continues to fully characterize the surface-enhanced Raman spectroscopy signal intensities as afunction of sample concentration and to explore otherextractants that do not require the use of elevated tem-perature
ACKNOWLEDGMENTS
The authors are grateful for the support of the National Science Foun-dation (DMI-0296116 and DMI-0215819) and the US Army(DAAD13-02-C-0015 Joint Service Agent Water Monitor program)The authors are indebted to Chetan Shende for preparing the sol-gelcapillaries The authors also thank James Gillespie Nicholas Fell andAugustus Fountain for providing important background informationMark Farquharson for laboratory support and Professor Jay Sperry ofthe University of Rhode Island for supplying B cereus spores
1 V A Pasechnik C C Shone and P Hambleton Bioseparations 3267 (1993)
2 P J Jackson M E Hugh-Jones D M Adair G Green K K HillC R Kuske L M Grinberg F A Abramova and P Keim ProcNatl Acad Sci USA 95 1224 (1998)
3 B R Glick and J J Pasternak Molecular Biology Principles andApplications of Recombinant DNA (ASM Press Washington DC1994)
4 C A Bell J R Uhl T L Had eld J C David R F Meyer TF Smith and F R Cockerill III J Clin Microbiol 40 2897(2002)
5 D L Gatto-Menking H Yu J G Bruno M T Goode M Millerand A W Zulich Biosens Bioelectron 10 501 (1995)
6 J J Quinlan and P M Foegeding J Rapid Methods AutomationMicrobiol 6 1 (1998)
7 A A Hindle and E A H Hall Analyst (Cambridge UK) 1241599 (1999)
8 M S Ascher US Department of Health amp Human Services(httpwwwhhsgovophppresentationsAscherdoc)
9 R Nudelman B V Bronk and S Efrima Appl Spectrosc 54 445(2000)
10 D L Rosen C Sharpless and L B McBrown Anal Chem 691082 (1997)
354 Volume 58 Number 3 2004
11 P M Pellegrino N F Fell Jr and J B Gillespie Anal ChimActa 455 167 (2002)
12 W H Woodruff T G Spiro and C Gilvarg Biochem BiophysRes Commun 58 197 (1974)
13 E Ghiamati R S Manoharan W H Nelson and J F SperryAppl Spectrosc 46 357 (1992)
14 A P Esposito C E Talley T Huser C W Hollars C M Schal-dach and S M Lane Appl Spectrosc 57 868 (2003)
15 S Farquharson L Grigely V Khitrov W W Smith J F Sperryand G Fenerty J Raman Spectrosc paper accep ted (2003)
16 S Farquharson W W Smith S Elliott and J F Sperry SPIE-IntSoc Opt Eng 3855 110 (1999)
17 S Farquharson and P Maksymiuk Appl Spectrosc 57 479(2003)
18 F W Janssen A J Lund and L E Anderson Science (Washing-ton DC) 127 26 (1958)
SPIE -2003-5269 117
pH dependence of methyl phosphonic acid dipicolinic acid and cyanide by surface-enhanced Raman spectroscopy
Stuart Farquharson Alan Gift Paul Maksymiuk Frank Inscore and Wayne Smith
Real-Time Analyzers 87 Church Street East Hartford CT 06108
ABSTRACT US and Coalition forces fighting terrorism in Afghanistan and Iraq must consider a wide range of attack scenarios in addition to car bombings Among these is the intentional poisoning of water supplies to obstruct military operations To counter such attacks the military is developing portable analyzers that can identify and quantify potential chemical agents in water supplies at microgram per liter concentrations within 10 minutes To aid this effort we have been investigating the value of a surface-enhanced Raman spectroscopy based portable analyzer In particular we have been developing silver-doped sol-gels to generate SER spectra of chemical agents and their hydrolysis products Here we present SER spectra of methyl phosphonic acid and cyanide as a function of pH an important factor affecting quantitation measurements which to our knowledge has not been examined In addition dipicolinic acid a chemical signature associated with anthrax-causing spores is also presented Keywords Chemical warfare agents agent detection agent hydrolysis SERS Raman spectroscopy homeland security
1 INTRODUCTION In the past decade the Unites States and its allies have been challenged by a different kind of warfare exemplified by the terrorist attacks of September 11 2001 Suicide bombings and the use of chemical agents are the norm and military personnel must consider a wide range of attack scenarios Among these is the intentional poisoning of water supplies to obstruct military operations in Afghanistan and Iraq Currently colorimetric paper is used to detect agents on-site while gas chromatography combined with mass spectrometry (GCMS) is used in mobile support laboratories However both methods have severe drawbacks The paper changes color in response to contact with many chemicals besides CWAs causing a high incidence of false positives1 while GCMS although very chemically specific requires hours to perform and constant re-calibration234 Military operations would be greatly aided by a portable analyzer that can identify and quantify potential chemical agents at concentrations that impact safety This includes the analysis of drinking water supplies distribution and storage systems To meet this goal the Department of Defense has been investigating numerous approaches under the auspices of the Joint Service Agent Water Monitor (JSAWM) program5 The JSAWM must be capable of identifying the chemical agents with no false-positives and quantifying the agents at microgram per liter concentrations within 10 minutes678 Recently we and others have been investigating the ability of surface-enhanced Raman spectroscopy (SERS) to measure chemical agents9-12 bioagents13-17 and their hydrolysis products in water SERS employs the interaction of surface plasmon modes of metal particles with target analytes to increase scattering efficiency by as much as 1 million times18 In 1987 the potential of this approach to measure CWAs was demonstrated by measuring a series of pesticides19 Several of these organophosphonates have chemical structures similar to CWAs in particular P=O functional groups In our studies we have been employing silver-doped sol-gels to promote the SER effect The porous silica network of the sol-gel matrix offers a unique environment for immobilizing and stabilizing SER-active metal particles20-23 The sol-gels have been coated on the internal walls of standard glass vials to allow rapid SER analysis of small sample quantities without preparation We have measured over 100 chemicals with enhancements of 104 to 106 demonstrated reversible measurements in a flowing system reproducible measurements from vial-to-vial and measurements in multiple solvents including water20-23 Previously we used these vials to perform preliminary measurements of cyanide (CN) methylphosphonic acid (MPA) and dipicolinic acid (DPA) MPA is a hydrolysis product of the nerve agents (eg sarin Reaction 1) and may be a valuable indicator of nerve agent usage particularly since the alkyl methylphosphonic acids are relatively more stable than their corresponding parent complexes24 DPA is
SPIE -2003-5269 118
a chemical signature of spore forming bacteria such as Bacillus anthracis And in light of the inability to rapidly detect the anthrax spores distributed through the US mail in October 2001 a number of methods are being developed to extract and analyze this signature Reaction 1 Stepwise hydrolysis of Sarin to form hydrofluoric acid (HF) isopropyl methylphosphonic acid (IMPA) then methyl phosphonic acid (MPA) and 2-propanol In our previous SERS investigations MPA and DPA were measured at 50 and 100 mgL respectively In both cases limits of detection (LOD) were estimated at 100 microgL providing encouragement in that SERS may satisfy the needs of the JSAWM Since it has been shown that pH can substantially influence the intensity of SER bands25 which would clearly influence quantitative analysis we undertook the present study to determine the severity of these effects for cyanide methyl phosphonic acid and dipicolinic acid Furthermore we previously observed a band at 1050 cm-1 for MPA6 possibly due to an anion formed at basic pH Here we investigate the source of this spectral anomaly
2 EXPERIMENTAL All chemicals including potassium cyanide methyl phosphonic acid dipicolinic acid and those used to prepare the silver-doped and gold-doped sol-gel coated vials were obtained and used as received from Sigma-Aldrich All samples were prepared in HPLC grade water (Fischer Scientific Fair Lawn NJ) for SERS measurements The pH of these samples was adjusted using dilute nitric acid or potassium hydroxide and verified using a pH electrode (Corning Inc Corning NY) that had been calibrated with pH 400 700 and 1000 buffered standards from Fischer Scientific Once prepared the samples were transferred into the silver-doped sol-gel vials (Real-Time Analyzers Inc East Hartford CT) The vials were coated in a manner similar to that previously reported by adding ammonium hydroxide to a solution of silver nitrate tetramethyl orthosilicate (TMOS) and methanol20 Gold-vials were coated by adding nitric acid to a solution of gold tetrachloride TMOS and methanol The two precursor solutions were prepared mixed and transferred to 2-mL glass vials dried and heated After sol-gel formation the incorporated metal ions were reduced with dilute sodium borohydride (1mgmL) which was followed by a water wash to remove residual reducing agent After the resultant analyte solution was introduced the SER-active vial was immediately fixed horizontally to an XY positioning stage (Conix Research Springfield OR) just inside the focal point of an f07 aspheric lens The lens focused the beam into the sample and collected the scattered radiation back along the same axis A dichroic filter (Omega Optical Brattleborough VT) was used to reflect the excitation laser to the lens and pass the Raman scattered radiation collected by the lens An f2 achromat was used to collimate the laser beam exiting a 200 microm core diameter source fiber optic while a second f2 achromat was used to focus the scattered radiation into a 365 microm fiber optic (Spectran Avon CT) A short pass filter was placed in the excitation beam path to block the silicon Raman scattering generated in the source fiber from reflecting off sampling optics and reaching the detector A long pass filter was placed in the collection beam path to block the sample Rayleigh scattering from reaching the detector A 785 nm diode laser (Process Instruments Inc model 785-600 Salt Lake City UT) was used to deliver 100 to 150 mW of power to the sample A Fourier transform Raman spectrometer (Real-Time Analyzers model IRA-785 East Hartford CT) and a silicon photo-avalanche detector (Perkin Elmer model C30902S Stamford CT) were used to acquire the SER spectra
3 RESULTS AND DISCUSSION In a previous study of MPA6 aimed at developing a concentration calibration curve and determining limits of detection (LOD) we observed an anomaly at 1050 cm-1 Since it was found that the band intensity changed as a function of concentration the band must be associated with a sample parameter Two possible parameters photon flux and pH are examined here The first parameter was investigated by irradiating a 1mgmL MPA sample in a SER-active vial with laser powers of 200 mW and above and monitoring spectral changes It was immediately found that the 1050 cm-1 band
2O+ H HF + +OH
OH3H C
OP
OF
CH
CH3
3
3H C
OCP
OOH
CH
CH3
3
3H C
O
CPHO
CH
CH
3
3
C
Sarin IMPA MPA 2-propanol
SPIE -2003-5269 119
appeared and grew as a function of time and that the higher the laser power the faster the growth Figure 1 shows the growth of the 1050 cm-1 band over the course of 30 minutes when using 150 mW of 785 nm excitation while Figure 2A shows that the growth can be fit with a first order exponential equation namely I1050 = 03+05e-013t Furthermore the 760 cm-1 band could be fit with a first order decay equation with an identical exponential rate constant ie I760 = 08-
08e-013t The rates represent classical first order kinetics and their correspondence allows one to conclude that MPA is being transformed one-for-one into a photo-generated product At this time the photoproduct has not been positively identified but phosphonic (phosphorous) acid and phosphonate are likely candidates since the symmetric P(OH)3 stretch occurs at ~1050cm-1 Our studies show that a reduction in laser power to 100 mW at the sample for MPA essentially eliminates this degradation process This laser power or lower was used for further measurements
Methyl phosphonic acid is a diprotic acid that stepwise dissociates into two anions MPA- and MPA= according to the following reactions26
MPA MPA- + H+ pKa1 = 212 Reaction 2
MPA- MPA= + H+ pKa2 = 729 Reaction 3 The relative concentrations of MPA MPA- and MPA= can be determined at any pH by expressing [MPA] and [MPA=] in terms of [MPA-] using Reactions 2 and 3 and summing all three to equal the total starting concentration here 2 mgmL (0021M MW = 9602) viz
[MPA] + [MPA-] + [MPA=] = 0021M Equation 1 substituting from Reactions 1 and 2
([H+][MPA-])K1a + [MPA-] + (K2a[MPA-])[H+] = 0021M Equation 2 rearranging [MPA-] = 0021M(1+[H+]K1a + K2a[H+]) Equation 3 The relative concentrations of MPA MPA- and MPA= as a function of pH are shown in Figure 3 It is worth noting that near neutral pH both MPA- and MPA= will be present To confirm that the SER signal followed this pH dependence a starting solution consisting of 20 mg of MPA in 10 mL HPLC grade water was prepared and brought to pH of 20 using dilute nitric acid From this solution 2 mL were added to a SER-active vial and the SER spectrum recorded At this pH a peak at 760 cm-1 was barely discernable The 2 mL solution was returned to the starting solution and the pH was re-measured to correct for any changes that the silver-doped sol-gel vials might cause In most cases the change was less than 02 pH units and the pH is reported as the before and after average Next the pH of the
Figure 2 A) Exponential growth of 1050 cm-1 band and B) exponential decay of 760 cm-1 band for spectral series in Figure 1
Figure 1 Growth of 1050 cm-1 band as a function of time due to exposure to 150 mW of 785 nm Spectra are 5 sec each collected every 100-sec from 0 to 30-min
B
570 770 970 1170 1370Raman Shift (cm-1)
Arbit
rary
Unit
s
MPA Photodegradation
600 800 1000 1200 1400 Wavenumber (cm-1)
A
0 10 20 30 0 10 20 30 time (min) time (min)
Ram
an In
tens
ity (r
elat
ive)
30
min
0
SPIE -2003-5269 120
starting solution was adjusted to 325 using dilute KOH Again 2 mL were added to a vial and the SER spectrum recorded At this pH a reasonably strong 760 cm-1 band was observed This process was repeated as spectra were recorded at pHs of 70 74 75 79 85 and 100 A total of 1 mL of KOH was added diluting the total concentration by 10 Next the pH of the starting solution was made acidic by adding dilute nitric acid dropwise This time spectra were recorded at pHs of 72 69 64 and 37 Figure 4 shows the SER spectra for representative pHs (spectra were left out to simplify the figure) while Figure 3 shows the 760 cm-1 peak intensities as a function of pH (The band intensities were adjusted to compensate for dilution effects caused by the addition of HNO3 and KOH then normalized to 0021 M for the most intense band observed at pH 37) It is clear from Figure 3 that the 760 cm-1 band follows the MPA- concentration as a function of pH and must be assigned to this anion No bands were observed that corresponded to MPA or MPA= The lack of an MPA SER spectrum may be due to the absence of an attraction between the neutral analyte and the electropositive silver surface The same reasoning suggests that a strong SER spectrum should be observed for MPA= but it is not and a satisfactory explanation has not been found
0000
0005
0010
0015
0020
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14pH
Con
cent
ratio
n [M
]
pK1 = 212 pK2 = 729
MPA- MPA=
MPA
Figure 4 SER spectra of 002M MPA as a function of pH Conditions 100 mW of 785 nm 36 scans (1 min) 8 cm-1 recorded 2 min after sample introduction pH 19 and 20 not apparent on this scale pH 69 and 74 near identical to 70 and 75 and not shown for clarity
00
02
04
06
08
10
12
14
16
18
0 5 10 15 20 25 30Measurement Number
Ram
an In
tens
ity (7
60 c
m-1
)
Figure 3 Concentration dependence of MPA MPA- and MPA= as a function of pH for a 002M sample Intensity of 760 cm-1
band from Figure 6 as a function of pH ( for increasing basic adjustment for increasing acidic adjustment error was measured at ~10 for pH 64)
Figure 5 SER spectra of 002M MPA at pH 64 measured around a vial at three heights (9 points per height) Conditions as in Figure 4 but 10-sec scans
Since these measurements involved the removal and replacement of the SER-active vial in the sample holder to remove and add sample variation in the intensity as a function of vial position was minimized by illuminating the exact same height along the vial wall But this does not account for variability of the SERS response of the sol-gel coating around the vial To analyze this effect a vial containing MPA at pH 64 was rotated at ~ 40o
intervals at the original height and 18rdquo above and below this value Figure 5 shows the intensity of the 760 cm-1
band for the 27 positions It was found that the average value was 137plusmn014 an RSD of 10 overall and 5 for each height An error bar is included in Figure 4 for the pH 64 measurement
SPIE -2003-5269 121
With the above analyses in mind a preliminary investigation of the SER spectral response for MPA (as MPA-) as a function of concentration was performed A single vial was used for these measurements beginning with 1 mgL followed by measurements of 10 100 and 1000 mgL In all cases the pH was ~7 and 3 positions around the vial were measured per concentration Since the 760 cm-1 band was not observed for concentrations of 1 or 10 mgL using 100 mW of 785 nm the laser power at the sample was raised to 200 mW beginning with the 10 mgL concentration Photo-degradation was largely avoided (and not observed) by exposing the sample for only 33 seconds per spectral acquisition Representative spectra for 10 100 and 1000 mgL are shown in Figure 6 while a plot of the 760 cm-1 band intensity as a function of concentration is shown in Figure 7 These values were also used to estimate limits of detection based on the signal-to-noise ratio (SN) of the 760 cm-1 band using 100 mW of 785 nm laser excitation and a 10-min acquisition time As summarized in the Figure 7 inset the lower the measured concentration the lower the predicted LOD Even if as estimated 210 microgL could be measured using the silver-doped sol-gel vials an improvement of a factor of 70 is still required to meet JSAWM goals of measuring 3 microgL in 10-minutes Similar to MPA DPA is a diprotic acid (pKa1 = 216 and pKa2 = 692) and variations in pH will effect the relative concentrations of DPA DPA- and DPA= and possibly the SER spectra and band intensities This could prove significant if an acid or base is used to denature anthrax spores with the goal of extracting and analyzing DPA The MPA pH study described above was mimicked for DPA except that the starting solution consisted of 20 mg of DPA in 20 mL HPLC grade water (60x10-3M MW = 1671) The initial solution had a pH of 245 which was made basic by dilute KOH to pHs of 355 433 487 559 1069 and 1166 SER spectra were recorded at each pH using 100 mW of 785 nm and a 44-sec acquisition time Next one drop of concentrated nitric acid was used to remake the solution acidic at a pH of 200 Again sequential pH measurements were performed at 383 510 735 and 822 The solution pH was made acidic a third time but to pH 219 171 then 135 Throughout this process no more than 20 drops of acid or base were added and therefore the concentration was diluted by no more than 10 Most of the spectral bands showed a minor decrease in intensity as a function of increasing pH values However the bands at 525 and 795 cm-1 showed the most dramatic changes which occurred at acid pH Figure 8 shows the SER spectra of DPA for the spectral region and pH range of interest The identity of the DPA species was determined by plotting the normalized peak intensities with the lowest value set to 0 and the highest to 0006 M as a function of pH and overlaying these values on a plot of the relative concentrations for DPA DPA- and DPA= as previously done for MPA (Figure 9) As can be seen the 525 cm-1 band clearly corresponds to DPA The correspondence of the 795 cm-1 band to this species is less clear as the band retains intensity until pH of 55 This can be attributed to contribution to the overlapping band at 810 cm-1 which does not change as a function of pH The fact that most bands are observed at all pHs suggest that the primary interaction with silver is through the ring nitrogen This is supported by the fact that the most intense band occurs at 1008 cm-1 attributed to a symmetric ring breathing mode and that this interaction has been characterized for pyridine in numerous papers27
0
1
2
3
4
5
6
7
0 200 400 600 800 1000 1200
MPA Concentration (mgL)
760
Ban
d In
tens
ity
Figure 6 SER spectra of MPA in water at A) 1000 B) 100 and C) 10 mgL Conditions pH of 7 silver-doped sol-gel coated vial 200 mW of 785 nm 33-sec 8 cm-1 resolution
Figure 7 Plot of SER intensity of 760 cm-1 band of MPA as a function of concentration using 200 mW of 785 nm Inset table includes average intensity LOD standard deviation and percent deviation for each concentration but for 100 mW and 10 min
conc (mgL) LOD ave stddev dev10 021 002 1021100 072 010 14301000 312 040 1280
A
B C
SPIE -2003-5269 122
The ability to detect dipicolinic acid (DPA) by SERS is demonstrated in Figure 10 Even at 1 mgL the primary bands are visible The intensity of the dominant ring breathing mode observed at 1008 cm-1 for DPA in water (pH = 25 to 55) is plotted as a function of concentration in Figure 11 These values were also used to estimate limits of detection based on the SN of the 1008 cm-1 band using 100 mW of 785 nm laser excitation and a 10-min acquisition time Again the lower the measured concentration the lower the predicted LOD (see Figure 11 inset) and detection of 160 microgL is possible Gastrointestinal anthrax requires significant more spores than inhalation anthrax28 and a limit of detection might be placed at 1 million spores in 1 liter of water or 10 microgL Since each spore contains ~10 CaDPA by weight29 a goal for DPA might be 1 microgL indicating that the present measurements must be improved by nearly two orders of magnitude Due to the increased hazards of handling HCN gas KCN salt was used for these experiments Nevertheless all sample preparations were performed in a chemical hood KCN completely dissolves in water but its conjugate acid HCN is formed and has a Ka of 615x10-1030 viz
HCN CN- + H+ pKa = 921 Reaction 4
Figure 8 SER spectra of 1 mgmL DPA as a function of pH Conditions 100 mW 785 nm 100 scans (44-sec)
Figure 9 Concentration dependence of DPA DPA- and DPA= as a function of pH for a 0006M sample Intensity of 525 () and 795 (diams) cm-1 bands from Fig 8 as a function of pH
pH
135 171 219 383
0
0001
0002
0003
0004
0005
0006
0007
0 2 4 6 8 10 12 14pH
Con
cent
ratio
n [M
]
DPADPA-DPA=795525
DPA DPA=DPA-
pK2 = 692pK1 = 216
795 525
0
05
1
15
2
25
0 200 400 600 800 1000 1200
DPA Concentration (mgL)
1008
Ban
d In
tens
ity
conc (mgL) lod-10min-100mw1 017
10 016100 103
1000 355
Figure 10 SER spectra of DPA in water at A) 1000 B) 100 C) 10 and D) 1 mgL Conditions pH of 25-55 silver-doped sol-gel coated vial 175 mW of 785 nm 1-min 8 cm-1 D) has been multiplied by x10 to make bands visible
A
B C
D
Figure 11 Plot of SER intensity of 1008 cm-1 band of DPA as a function of concentration using 175 mW of 785 nm Inset table includes LOD in mgL for each concentration but for 100 mW and 10 min
SPIE -2003-5269 123
Consequently the cyanide concentration must be determined for each initial KCN concentration Specifically the samples prepared with concentrations of 01 1 10 100 and 1000 mgL of KCN produced CN- concentrations of 63x10-3 033 69 89 and 964 mgL at pHs of 816 90 967 102 and 107 respectively The pH dependence for the HCN and CN- concentrations are shown in Figure 12 Thus as the amount of KCN added to the solution decreases so does the pH of the solution (becomes less basic) and according to Reaction 4 the relative amount of CN- to HCN also decreases For example in the preparation of a 01 mgL solution of KCN the pH is shifted from 7 for pure water to only 816 and only 63 of the starting material becomes CN- or 63x10-3 mgL In comparison for a solution of 1000 mgL the pH is shifted from 7 to 107 and 96 of the starting material becomes CN- This is significant since the cyanide ion is better able to adsorb onto the silver particles and become SERS active SER spectra of 10 100 and 1000 mgL of KCN samples are shown in Figure 13 and a corresponding plot of the 2100 cm-1 band intensity as a function of concentration is shown in Figure 14
000102030405060708091011
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14pH
Con
cent
ratio
n [m
gm
L]
pKa = 921
CN -HCN
The SER spectra of cyanide are dominated by the single CN stretch at 2100 cm-1 which occurs in normal Raman spectra of solutions at 2080 cm-1 However a low frequency mode occurs at 135 cm-1 due to a Ag-CN stretch (not shown) indicative of a strong surface interaction It is also observed that as the concentration decreases the CN stretch shifts to 2140 cm-1 This shift has been attribute to the formation of a tetrahedral Ag(CN)3
2- surface structure31 as well
Figure 14 Concentration dependence of KCN SERS measured under conditions in Fig 11 Concentrations are 1 01 and 001 mgml Intensities are measured for the CN stretch at 2100 cm-1 Inset table includes LOD in mgL for each concentration in Figs 13 and 15 but for 100 mW and 10 min
Figure 13 SER spectra of KCN in water at A) 1000 B) 100 and C) 10 mgL Conditions pHs of 107 102 and 97 silver-doped sol-gel coated vial 100 mW of 785 nm 1-min 8 cm-1 C) has been multiplied by x10 to make band visible
Figure 12 Concentration dependence of HCN and CN- as a function of pH for a 1 mgmL sample Calculated intensity of 2100 () cm-1 band for a 1 mgml sample at pHs of 816 90 967 102 and 107
A
B
C
0
20
40
60
80
100
120
140
0 200 400 600 800 1000 1200CN Concentration (mgL)
2100
Ban
d In
tens
ity
Figure 15 SER spectra of KCN in water at A) 10 B) 1 and C) 01 mgL Conditions pHs of 97 12 and 12 gold-doped sol-gel coated vial 100 mW of 785 nm 1-min 8 cm-1 C) has been multiplied by x10 to make band visible
A
B
C
conc (mgL) lod-10min-100mw Condition01 001 Au-pH 121 007 Au-pH 12
69 003 Au-pH 9769 016 Ag-pH 9789 022 Ag-pH 102946 113 Ag-pH 107
SPIE -2003-5269 124
as to CN adsorbed to two different surface sites32 Alternatively the 2140 cm-1 band could be attributed to HCN since this species dominates at lower concentrations However it is unlikely that this species would be attracted to the electropositive silver surface Further both peaks should be present at pHs between 85 and 105 but this is not observed It has also been suggested that at concentrations near and above monolayer coverage the CN- species is forced to adsorb end-on due to crowding and at lower concentrations the molecule can reorient to lie flat33 This suggests that the 2100 and 2140 cm-1 bands correspond to the end-on and flat orientations respectively As Figures 13 and 14 show the intensity of the CN stretch for the 89 mgL sample is nearly as intense as the 964 mgmL band This suggests that the Raman signal for the flat orientation is more enhanced However more extensive measurements are required to verify this point Since resent research has suggested that cyanide may be more effectively detected on gold measurements of KCN solutions were also performed using gold-doped sol-gel vials Preliminary measurements are shown in Figure 15 for samples prepared from 01 1 and 10 mgL KCN Since the pHs are 816 90 967 the resultant CN- concentrations are 63x10-3 033 and 69 mgL Initially only the highest concentration was observed and the signal intensity was significantly better than the equivalent concentration measured using silver In an effort to shift Reaction 4 to the left transforming HCN to CN- (Le Chatelierrsquos principle) KOH was added to the lower concentration samples producing solutions with pH 12 This effectively forces all of the cyanide in solution to be CN- or 01 and 10 mgL respectively More importantly the CN stretch is now observed in the SER spectra The band appears at 2125 cm-1 as has been previously reported for gold12 As calculated for MPA and DPA LODs can be estimated from this data For the three concentrations of cyanide on silver the LODs are 016 to 11 mgL for 100 mW of 785 nm laser excitation and a 10-min acquisition time For gold pH adjusted the LODs are10 to 70 microgL an improvement of more than 10 times silver Nevertheless either substrate is sufficient to meet the JSAWM goals of measuring 3 mgL in 10-minutes as the requirements form cyanide are much less stringent than the nerve agents
4 CONCLUSIONS Here we examined the surface-enhanced Raman spectral response for methyl phosphonic acid dipicolinic acid and cyanide as a function of pH It was determined that the most prominent peak at 760 cm-1 reaches a maximum intensity between pH 3 and 7 and corresponds to the MPA- species Neither the MPA nor MPA= species appear to generate a SER spectrum and consequently no spectra were observed below pH 2 or above pH 8 In this study we also found that higher laser powers could cause photodegradation of MPA signified by the exponential growth of a band at 1050 cm-1 which is tentatively assigned to phosphorous acid Unlike MPA DPA was observed at all pHs This is attributed to the dominant interaction of the pyridine functional group with silver Minor spectral changes were observed at acid pHs and were assigned to neutral DPA Like MPA SER spectra of cyanide were pH dependent No spectra were observed for the HCN species while CN- was best observed at pHs more basic than 8 Preliminary concentration studies for the three analytes allowed estimating limits of detection for MPA DPA and CN using 100 mW of 785 nm and a 10-min acquisition time of 210 165 and 70 microgL respectively Although the latter value suggests that cyanide can be measured at sufficiently low concentrations to meet JSAWM goal improvements by 100 to 200 times are required for MPA and DPA It is clear from this study that pH of the sample is important and must be taken into account when developing concentration calibration curves as well as fieldable analyzers Future work will include tailoring the sol-gel with specific functional groups to dictate sample pH and thereby optimize sensitivity
ACKNOWLEDGMENTS The authors are grateful for the support of the US Army (DAAD13-02-C-0015 Joint Service Agent Water Monitor program) The authors would also like to thank Dr Steve Christensen of the US Army SBCCOM for helpful discussions and Mr Chetan Shende of Real-Time Analyzers for assistance in development of the gold-doped sol-gels
REFERENCES 1 Erickson B Analytical Chemistry News amp Features June 1 397A (1998)
SPIE -2003-5269 125
2 Johnston RL Hoefler CM Fargo JC and Moberley B AT-ONSITE 5-8 (1994) 3 Black RM RJ Clarke RW Read and MT Reid J Chromatography 662 301-321 (1994) 4 USA Army Center for Health Promotion and Preventive Medicine (USACHPPM) Technical Guideline 230A 5 Jensen JL NC Wong W R Loerop SPIE 4775-03 (2001) 6 Hoenig SL Handbook of Chemical Warfare and Terrorism Greenwood Press Wesport CT (2002) 7 Munro NB SS Talmage GD Griffin LC Waters AP Watson JF King and V Hauschild Env Health
Persp 107 933-974 (1999) 8 Holstege CP Kirk M Sidell FR Crit Care Clin13 923-42 (1997) 9 Farquharson S P Maksymiuk K Ong and S Christesen SPIE 4577 166-173 (2001) 10 Lee Y and S Farquharson SPIE 4378 21-26 (2001) 11 Spencer KM J Sylvia S Clauson and J Janni SPIE 4577 158-165 (2001) 12 Tessier P S Christesen K Ong E Clemente A Lenhoff E Kaler and O Velev Applied Spectroscopy 56
1524-1530 (2002) 13 Farquharson S WW Smith S Elliott and JF Sperry SPIE 3533 207-214 (1998) 14 Farquharson S WW Smith S Elliott and JF Sperry SPIE 3855110-116 (1999) 15 Farquharson S WW Smith YH Lee S Elliott and JF Sperry SPIE 4575 62-72 (2002) 16 Guzelian AA J Sylvia J Janni S Clauson and KM Spencer SPIE 4577 182-192 (2001) 17 Shende C F Inscore A Gift P Maksymiuk and S Farquharson in press 18 Weaver MJ S Farquharson and MA Tadayyoni J Chem Phys 82 4867-4874 (1985) 19 Alak AM and T Vo-Dinh Analytical Chemistry 59 2149-2153 (1987) 20 Lee Y and S Farquharson SPIE 4206 140-146 (2000) 21 Farquharson S and Y Lee SPIE 4200-16 (2000) 22 Lee Y S Farquharson and P M Rainey SPIE 3857 76-84 (1999) 23 Lee Y S Farquharson H Kwong and M Shahriari SPIE 3537 252-260 (1998) 24 Wang J M Pumera G Collins and A Mulchandani Analytical Chemistry 74 6121-6125 (2002) 25 Dou X YM Jung Z-Q Cao and Y Ozaki Applied Spectroscopy 53 1440-1447 (1999) 26 Data supplied by S Christesen and K Ewing 27 Kerker M and B Thompson Eds SPIE MS 10 (1990) 28 Inglesby TV DA Henderson JG Bartlett JAMA 287 2236 (2002) 29 FW Janssen AJ Lund and LE Anderson Science 127 26 (1958) 30 Lide DR Ed Handbook of Chemistry and Physics CRC Press 77th Ed (1996-7) 31 Billmann J G Kovacs and A Otto Surf Sci 92 153 (1980) 32 Murray CA and S Bodoff Phys Rev B 32 671 (1985) 33 Kellogg D and J Pemberton J Phys Chem 91 1120 (1987)
SPIE-2003-5269 16
Chemical agent detection by surface-enhanced Raman spectroscopy
Stuart Farquharson Alan Gift Paul Maksymiuk Frank Inscore and Wayne Smith
Real-Time Analyzers 87 Church Street East Hartford CT 06108
Kevin Morrisey and Steven D Christesen US Army SBCCOM Aberdeen Proving Ground MD 21010
ABSTRACT
In the past decade the Unites States and its allies have been challenged by a different kind of warfare exemplified by the terrorist attacks of September 11 2001 Although suicide bombings are the most often used form of terror military personnel must consider a wide range of attack scenarios Among these is the intentional poisoning of water supplies to obstruct military operations in Afghanistan and Iraq To counter such attacks the military is developing portable analyzers that can identify and quantify potential chemical agents in water supplies at microgram per liter concentrations within 10 minutes To aid this effort we have been investigating the value of a surface-enhanced Raman spectroscopy based portable analyzer In particular we have been developing silver-doped sol-gels to generate SER spectra of chemical agents and their hydrolysis products Here we present SER spectra of several chemical agents measured in a generic tap water Repeat measurements were performed to establish statistical error associated with SERS obtained using the sol-gel coated vials Keywords Chemical agents chemical agent detection SERS Raman spectroscopy
1 INTRODUCTION
In the autumn of 2001 terrorism within US borders became a sobering reality While extensive efforts are being implemented to secure the homeland US and Coalition forces in Afghanistan and Iraq are constantly faced with terrorist attacks In addition to car-bombings the military has identified several non-traditional attack scenarios including poisoning of water supplies by chemical warfare agents (CWAs) To counter this threat the Department of Defense has been investigating numerous approaches to detect such attacks under the auspices of the Joint Service Agent Water Monitor (JSAWM) program1 The JSAWM must be capable of identifying the chemical agents with no false-positives and quantifying the agents at microgram per liter concentrations within 10 minutes2 This includes the analysis of drinking water supplies distribution and storage systems Currently colorimetric paper is used to detect agents on-site while gas chromatography combined with mass spectrometry (GCMS) is used in mobile support laboratories However both methods have severe drawbacks The paper changes color in response to contact with many chemicals besides CWAs causing a high incidence of false positives3 while GCMS although very chemically specific requires up to an hour to perform and regular re-calibration456 Vibrational spectroscopy has also been investigated as a potential method of rapidly detecting CWAs7-
11 as early as 1985 in the case of infrared spectroscopy8 and 1988 in the case of Raman spectroscopy11 Again however these techniques also have limitations when it comes to measuring trace poisons in water Infrared spectra would be dominated by the very strong absorption of water which would obscure absorptions by most other chemicals present Whereas Raman spectroscopy is simply not a very sensitive technique and detection limits are typically grams per liter Surface-enhanced Raman spectroscopy (SERS) offers several advantages over conventional vibrational methods and that may provide the necessary sensitivity required for detecting trace quantities of chemical agents in water SERS employs the interaction of surface plasmon modes of metal particles with target analytes to increase scattering efficiency by as much as 1 million times12 In 1987 the potential of this approach to measure CWAs was demonstrated by measuring a series of pesticides13 Several of these organophosphonates have chemical structures similar to CWAs in particular P=O functional groups In the past few years we and others have further explored the ability of SERS to detect CWAs14-17 and even bioagents 18-21 We have been employing silver-doped sol-gels to promote the SER effect
SPIE-2003-5269 17
in these studies The porous silica network of the sol-gel matrix offers a unique environment for immobilizing and stabilizing SER-active metal particles22-25 The silver-doped sol-gels have been coated on the internal walls of standard glass vials to allow rapid SER analysis of small sample quantities without preparation We have measured over 100 chemicals with enhancements of 104 to 106 demonstrated reversible measurements in a flowing system reproducible measurements from vial-to-vial and measurements in multiple solvents including water21-25 Previously we used these vials to perform preliminary measurements of cyanide (CN) and methylphosphonic acid (MPA) Most of the nerve agents form MPA during hydrolysis while Tabun forms CN a chemical agent in its own right In another paper including in these proceedings we examined the limits of detection (LOD) for MPA by measuring a series of concentrations down to 50 mgL and estimated a limit of detection of 100 microgL26 These measurements provide encouragement in that SERS may satisfy the needs of the JSAWM To further establish the viability of SERS in particular silver-doped sol-gels here we present analysis of cyanide mustard and VX in tap water The measurements performed at the US Armyrsquos Edgewood Chemical Biological Center Aberdeen MD also included numerous repeat measurements to establish reproducibility
2 EXPERIMENTAL 2a General All chemicals including potassium cyanide 2-chloroethylethyl sulfide and those used to prepare the silver-doped and gold-doped sol-gel coated vials were obtained and used as received from Sigma-Aldrich All samples were prepared in a chemical hood using HPLC grade water unless otherwise noted (Fischer Scientific Fair Lawn NJ) for SERS measurements Once prepared the samples were transferred into the silver-doped sol-gel vials (Real-Time Analyzers Inc East Hartford CT) The vials were coated in a manner similar to that previously reported by adding ammonium hydroxide to a solution of silver nitrate tetramethyl orthosilicate (TMOS) and methanol22 The two precursor solutions were prepared mixed and transferred to 2-mL glass vials dried and heated After sol-gel formation the incorporated metal ions were reduced with dilute sodium borohydride (1mgmL) which is followed by a water wash to remove residual reducing agent After the resultant analyte solution was introduced the SER-active vial was fixed horizontally to an XY positioning stage (Conix Research Springfield OR) just inside the focal point of an f07 aspheric lens The lens focused the beam into the sample and collected the scattered radiation back along the same axis A dichroic filter (Omega Optical Brattleborough VT) was used to reflect the excitation laser to the lens and pass the Raman scattered radiation collected by the lens An f2 achromat was used to collimate the laser beam exiting a 200 microm core diameter source fiber optic while a second f2 achromat was used to focus the scattered radiation into a 365 microm fiber optic (Spectran Avon CT) A short pass filter was placed in the excitation beam path to block the silicon Raman scattering generated in the source fiber from reflecting off sampling optics and reaching the detector A long pass filter was placed in the collection beam path to block the sample Rayleigh scattering from reaching the detector A 785 nm diode laser (Process Instruments Inc model 785-600 Salt Lake City UT) was used to deliver 100 to 150 mW of power to the sample A Fourier transform Raman spectrometer (Real-Time Analyzers model IRA-785 East Hartford CT) and a silicon photo-avalanche detector (Perkin Elmer model C30902S Stamford CT) were used to acquire the SER spectra
2b Edgewood Chemical Biological Center The surface-enhanced Raman spectral measurements at the US Armyrsquos Chemical Biological center presented here were all performed on September 12 2003 To expedite measurements a plate was machined to hold up to 12 SER-active sample vials (Figure 1) The plate fit a standard XY plate reader that could be programmed Pure KCN bis-(2-chloroethyl)sulfide (distilled mustard gas HD) and ethyl S-2-diisopropylamino ethyl methylphosphonothioate (VX) were obtained on-site and used to prepare 1 gL tap water solutions in a chemical hood with appropriate safety equipment Simulated tap water was prepared by adding 10 chemicals most often found in tap water at appropriate concentrations to distilled water (Table 1) SER measurements were also performed in a chemical hood For added safety the FT-Raman instrument was placed outside the laboratory and 30 foot fiber optic and electrical cables were used to allow remote SERS measurements and plate manipulation For each experiment 1gL samples were prepared and added to 9 individual vials which were then loaded on the plate In some cases a tenth vial was included as a blank
Table 1 Generic Tap Water Composition
Compound mgL NaHCO3 100 CaSO4 27 MgSO4bull7H2O 67 NaNO3 1 Fulvic Acid 1 K2HPO4 07 KH2PO4 03 (NH4)2HSO4 001 NaCl 001 FeSO4 0001 pH 76-78
SPIE-2003-5269 18
A software program was written that allowed selecting the sequence that the vials were measured the number of positions along the length of the vials to measure (1 to 5) and the number of scans to co-add During sample analysis the program displayed the vial being analyzed the point being analyzed and the spectrum as it was being acquired Once all the data was collected a second software program was written to rapidly analyze the data The spectra collected for all the vials on a plate could be loaded at one time and then the spectra for each point could be displayed simultaneously or separately The user could then select the Raman peak to analyze in terms of peak height or area This was accomplished by selecting points on either side of the peak to define a baseline of zero The peak height or area could then be computed for all of the spectra loaded and then exported to a spreadsheet for statistical analysis
Figure 1 A) Vial Holder 6 slots to hold 2 vials each end-to-end B) Measurement Configuration Program user selects vials to measure sequence number of points per vial (1 to 5) and number of scans per point C) Spectral Acquisition Program shows spectrum being collected which vial and position D) Spectra Analysis Program user selects spectra to analyze by plate vial and point (s) as well as two wavenumbers defining the peak and the baseline to subtract The image is of 5 repeat measurements of 10 mgL KCN in generic tap water 16 sec each 100 mW of 785 nm
3 RESULTS AND DISCUSSION Raman and surface-enhanced Raman spectra were obtained for potassium cyanide bis-(2-chloroethyl)sulfide and ethyl S-2-diisopropylamino ethyl methylphosphonothioate representing three classes of chemical agents cyanides mustards and nerve agents respectively Spectra were also obtained for 2-chloroethyl ethyl sulfide (CEES) a structural analogue to HD which was included in the study to aid in assigning spectral bands KCN salt was used for cyanide experiments to avoid the increased hazards of handling HCN gas KCN completely dissolves in water forming its conjugate acid HCN according to its Ka of 615x10-1027 and at a concentration of 1 mgmL results in a pH 107 solution This is significant in that only CN- appears to interact sufficiently with silver to produce a SER spectrum and no spectral signal is observed below pH 726 Figure 2 shows the SER and normal Raman spectra for KCN The SERS spectrum of 1mgml KCN in water shows a single intense somewhat broad feature at
A B
C D
SPIE-2003-5269 19
2100 cm-1 assigned to the single CequivN stretch The band is much sharper in the normal Raman spectra of the solid KCN salt at 2074 cm-1 This band does broaden and shift to 2080 cm-1 in solution (not shown) However the observed SERS frequency is attributed to interaction with silver and low frequency mode at 135 cm-1 attributed to a Ag-CN stretch (not shown) supports this conclusion
cm-1 band to a C-S stretch but the authors concede that it is in fact more likely a C-Cl stretch28 It appears that the most intense Raman bands at 648 692 and 747 cm-1 shift to 620 660 and 720 cm-1 in the SER spectra and are tentatively assigned as above The width of these bands suggests that they overlap underlying spectral features Additional bands in the Raman spectra occur at 972 1034 1049 1263 1286 1423 1442 2865 2935 and 2960 cm-1 Corresponding bands occur in the SER spectra at 964 1015 1054 1286 1410 1447 2865 and 2935 cm-1 Most of these bands are associated with alkane modes specifically the bands at approximately 1040 cm-1 to a C-C stretch 1290 cm-1 to a CH2 in-phase twist 1440 cm-1 to a CH2 wag 2865 cm-1 to a symmetric CH2 stretch and 2965 cm-1 to an asymmetric CH2 stretch The Raman and SER spectra of sulfur mustard were measured at the Edgewood center (Figure 4) Both spectra are largely similar to CEES The C-Cl and C-S bands in the Raman spectrum of HD now occur at 640 655 700 739 and 760 cm-1 and are more resolved possibly due to the increased molecular symmetry Theoretical calculations indicate that the first three bands are due to C-Cl stretching modes and the latter two to C-S stretching modes28 Only the C-Cl bands maintain significant intensity in the SER spectra occurring at 624 and 643 cm-1 which is attributed to the
Prior to measurements of HD CEES was examined by Raman and SER spectroscopy (Figure 3) CEES also known as half-mustard is essentially identical to HD except one of the chlorine end atoms is replaced by a hydrogen atom Again although not as toxic as HD CEES is a blister agent and dilute aqueous samples were prepared using appropriate safety equipment Both the Raman and SER spectra of CEES are similar and dominated by bands between 600 and 800 cm-1 These are associated with C-Cl and C-S stretching modes which are tentatively assigned to 648 and 747 cm-1 in the Raman spectra respectively The shoulder at 630 cm-1 the overlapped band at 660 cm-1 and the strong band at 692 cm-1 could also be due to these modes or their asymmetric counterparts It is worth noting that theoretical calculations assign the 692
Figure 2 A) SER and B) NR spectra of KCN Conditions A) 1 mgml in tap water 100 mW of 785 nm at sample 1-min acquisition time B) solid 300 mW of 785 nm 5-min All spectra are 8 cm-1 resolution
A
B
Figure 3 A) SER and B) NR spectra of CEES Conditions A) 1 vv (10 mgml) in MeOH 100 mW of 785 nm 1-min acquisition time B) neat 300 mW of 785 nm 5-min
A
B
Cl-CH2-CH2-S-CH2-CH3
A
B
Figure 4 A) SER and B) NR of HD Conditions A) 1mgml in tap water B) pure both 100 mW of 785 nm 1-min
Cl-CH2-CH2-S-CH2-CH2-Cl
SPIE-2003-5269 20
expected strong interaction between chlorine and silver and adds support to the assignment of this band to a C-Cl stretch Weaker overlapping bands occur at 670 692 and 724 cm-1 the latter possibly due to C-S stretching modes Again the alkane modes are apparent in the normal Raman spectra of HD but only a broad feature at 1300 to 1450 cm-1 suggests CH2 contributions in the SER spectrum Although the observed bands in the VX spectrum have not been assigned (Figure 4) a computer generated Raman spectrum29 predicts many of the same features with surprising accuracy and are used here Two intense bands at 460 and 530 cm-1 closely match predicted bands at 463 and 546 cm-1 assigned to a CH3-P=O bend and a PO2CS wag Three highly overlapped bands occur at 694 745 and 771 cm-1 matching predicted bands at 713 730 and 760 cm-1 The first
Table 2 Measured SER peak heights for the CN stretch at 2100 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
Figure 5 A) SERS and B) NR spectra of VX Conditions A) 1 vv (10 mgml) in MeOH B) pure sample both 100 mW of 785 nm 1-min acquisition time
A
B
two have been assigned to a C-S stretch and CH2bend respectively while the latter has been attributed to either a P-C stretch or an O-C-C stretch Although the 745 cm-1 band may alternatively be assigned to a C-S stretch based on the previous measurements of CEES and HD The relatively intense bands at 890 1106 1218 1445 and 1465 cm-1 also match predicted bands at 880 1108 1216 1440 and 1464 cm-1 that are assigned to a C-C stretch CH3 rock N-C3 stretch various C-H3 bends and C-H bends respectively Both the computer generated and the measured spectra contain numerous other less intense bands One is worth mentioning A unique band appears at 370 cm-1 that is predicted at 368 cm-1 and corresponds to an O-P=O bend The surface-enhanced Raman spectrum of VX is also rich with spectral features It has the unique low frequency band at 370 cm-1 as well as a second band at 380 cm-1 that is assigned tothe S-P-O bend predicted in the normal Raman spectrum at 388 cm-1 Based on the measured and predicted normal Raman spectra the following SERS assignments are given 460 cm-1 to the CH3-P=O bend 544 cm-1 to the PO2CS wag 738 cm-1 to a C-S stretch (based on arguments above) 890 cm-1
to a C-C stretch 1101 cm-1 to a CH3 rock and 1456
cm-1 to a C-H bend The ability of SERS to measure chemical agents in water containing real-world chemical interferents was tested by using the generic tap water described in Table 1 The ability to reproduce measurements was accomplished by preparing three separate water stock solutions which were used to prepare three sample solutions each of 1 mgmL KCN in the generic tap water The samples defined as vials 1-9 were then measured at 4 points per vial in a semi-random fashion such that errors associated with stock solution preparation and errors associated with instrument drift could be identified No trends were apparent that signified such systematic errors Each spectrum collected consisted of 20 averaged scans taking 16 seconds at 8 cm-1 resolution The laser power at the sample was measured periodically during the day and it ranged from 102 to 105 mW spacccc
spot Vial 1 Vial 2 Vial 41 13754 13004 128192 13519 12692 12909
1 3 13541 12721 126394 13462 12648 12651
ave 13569 12766 12755Vial 6 Vial 3 Vial 5
1 11586 14068 10757 2 11236 14402 115122 3 11376 14546 11559
4 10894 11763 11214ave 11273 13695 11261
Vial 7 Vial8 Vial 91 11198 15804 11023
2 11243 15716 88753 3 11562 15216 11218
4 11655 15084 11077ave 11415 15455 10548
CN AVG STDEV ERR1 mgmL pk ht 12526 1560 1245
stock solution
SPIE-2003-5269 21
The CN measurements consisting of 36 data points produced an average height of 125 for the 2100 cm-1 peak with a standard deviation of 156 or 125 (Table 2) The HD and VX measurements were performed precisely the same way (Tables 3 and 4) For HD the 624 cm-1 peak was used for analysis and it had an average height of 53 with a standard deviation of 068 or 129 while for VX the 544 cm-1 peak was used for analysis and it had an average height of 1051 with a standard deviation of 308 or 293 The greater error in the VX measurements can be somewhat attributed to Vial 4 which produced lower SER signal intensities But removing this vial from the data set changes the standard deviation to 233 only a modest improvement
spot Vial 1 Vial 2 Vial 41 525 609 4682 45 675 484
1 3 527 644 5094 559 807 669
ave 51525 68375 5325Vial 6 Vial 3 Vial 5
1 521 574 575 2 536 53 4492 3 509 508 451
4 565 594 379ave 53275 5515 4635
Vial 7 Vial 8 Vial 91 457 589 413
2 583 497 4273 3 544 505 4
4 528 5 509ave 528 52275 43725
HD AVG STDEV ERR1 mgmL pk ht 530 068 1291
stock solution
spot Vial 1 Vial 2 Vial 41 1464 1034 4672 1485 99 61
1 3 1491 989 5684 1041 777 553
ave 137025 9475 5495Vial 6 Vial 3 Vial 5
1 1058 942 1293 2 697 121 965
2 3 727 1206 11124 689 1254 746
ave 79275 1153 1029Vial 7 Vial 8 Vial 9
1 1127 783 163 2 1358 812 1614
3 3 1371 908 15124 1097 875 1397
ave 123825 8445 153825
VX AVG STDEV ERR1 mgmL pk ht 1051 308 2925
stock solution
4 CONCLUSIONS In this paper we examined the ability of surface-enhanced Raman spectroscopy to reproducibly measure CN HD and VX in tap water without chemical interference Both normal and surface-enhanced Raman spectra were examined to select unique bands suitable to identify and quantify these chemical agents For SER measurements the 2100 cm-1 C-N stretch was used for CN the 624 cm-1 C-Cl stretch was used for HD and the 544 cm-1 PO2CS wag was used for VX It was determined that 1 mgmL samples of each of these chemicals measured 36 times in glass vials coated with a silver-doped sol-gel reproduced measurements with standard deviations of 125 129 and 293 It was further found that the 10 chemicals added to simulate generic tap water did not interfere with or alter the SER spectra It should be noted that the concentrations used in this study were considerably greater than those required by the JSAWM program Current work involves improving SER sensitivity and designing sampling systems with better reproducibility This includes the development of fractal silver and gold structures within the sol-gel matrix and the development of chemically selective sol-gels
ACKNOWLEDGMENTS The authors are grateful for the support of the US Army (DAAD13-02-C-0015 Joint Service Agent Water Monitor program) The authors would also like to thank Janet Jensen Ronald Crosier and Kristina Gonser for helpful discussions
Table 4 Measured SER peak heights for the VX band at 544 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
Table 3 Measured SER peak heights for the HD band at 624 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
SPIE-2003-5269 22
REFERENCES 1 Jensen JL NC Wong W R Loerop SPIE 4775-03 (2001) 2 JSAWM Requirements at wwwsbccomapgeaarmymilRDAecbcrtPRODSERJSAWMjsawmhtml 3 Erickson B ldquoThe Chemical Weapons Convention Redefines Analytical Challengerdquo Anal Chem 70 397A (1998) 4 Johnston RL Hoefler CM Fargo JC and Moberley B ldquoThe Defense Nuclear Agencyrsquos
ChemicalBiochemical Weapons Agreements Verification Technology Research Development Test and Evaluation Program and its Requirements for On-Site Analysisrdquo AT-ONSITE 5-8 (1994)
5 Black RM Clarke RJ Read RW and Reid MT ldquoApplication of GC-MS and GC-tandem MS to the analysis of chemical warfare samples found to contain residues of the nerve agent sarin sulphur mustard and their degradation productsrdquo J Chrom 662 301-321 (1994)
6 USA Army Center for Health Promotion and Preventive Medicine (USACHPPM) Technical Guideline 230A 7 Hoffland LD Piffath RJ Bouck JB rdquoSpectral signatures of chemical agents and simulantsrdquo Optical
Engineering 24 982-984 (1985) 8 Braue EHJ Pannella MGrdquoCIRCLE CELL FT-IR Analysis of Chemical Warfare Agents in Aqueous Solutionsrdquo
App Spectrosc 44 1513-1520 (1990) 9 Tseng C-H Mann CK and Vickers TJ ldquoDetermination of Organics on Metal Surfaces by Raman
Spectroscopyrdquo App Spectrosc 47 1767-1771 (1993) 10 Staff Reporter ldquoUS Army Tests Lidar to Detect Biological Toxinsrdquo Photonic Spectra pg 50 December 1998 11 Christesen SD Raman cross sections of chemical agents and simulants App Spectrosc 42 318-321 (1988) 12 Weaver MJ Farquharson S Tadayyoni MA ldquoSurface-enhancement factors for Raman scattering at silver electrodesrdquo J Chem Phys 82 4867-4874 (1985) 13 Alak AM and Vo-Dinh T ldquoSERS of Organophosphorous Chemical Agentsrdquo Anal Chem 59 2149-2153 (1987) 14 Lee Y Farquharson S ldquoRapid chemical agent identification by SERSrdquo SPIE 4378 21-26 (2001) 15 Farquharson S Maksymiuk P Ong K Christesen S ldquoChemical agent identification by surface-enhanced Raman spectroscopyrdquo SPIE 4577 166-173 (2001) 16 Spencer KM Sylvia J Clauson S and Janni J ldquoSurface Enhanced Raman as a Water Monitor for Warfare
Agents in Waterrdquo SPIE 4577 158-165 (2001) 17 Tessier P Christesen S Ong K Clemente E Lenhoff A Kaler E Velev O ldquoOn-line spectroscopic characterization of sodium cyanide with nanostructured gold surface-enhanced Raman spectroscopy substratesrdquo
App Spectrosc 56 1524-1530 (2002) 18 Farquharson S W W Smith S Elliott and J F Sperry Rapid biological agent identification by surface- enhanced Raman spectroscopy SPIE 3855110-116 (1999) 19 Farquharson S WW Smith YH Lee S Elliott and J F Sperry Detection of biological signatures A
comparison of electrolytic and metal-doped sol-gel surface-enhanced Raman media SPIE 4575 62-72 (2002) 20 Guzelian AA Sylvia J Janni J Clauson S and Spencer KM ldquoSERS of whole cell bacteria and trace levels of biological moleculesrdquo SPIE 4577 182-192 (2001) 21 Shende C Inscore F Gift A Maksymiuk P Farquharson S ldquoRapid dipicolinic acid extraction from Bacillus spores detected by surface-enhanced Raman spectroscopyrdquo App Spectrosc 58 accepted 22 Lee Y and Farquharson S ldquoSERS Sample Vials Based on Sol-Gel Process for Trace Pesticide Analysisrdquo SPIE
4206 140-146 (2000) 23 Farquharson S and Lee Y ldquoTrace Drug Analysis by SERSrdquo SPIE 4200-16 (2000) 24 Lee Y Farquharson S and Rainey P M Surface-Enhanced Raman Sensor for Trace Chemical Detection in
Water SPIE 3857 76-84 (1999) 25 Lee Y Farquharson S Kwong H and Shahriari M ldquoSol-Gel Chemical Sensor for Surface-Enhanced Raman
Spectroscopyrdquo SPIE 3537 252-260 (1998) 26 Farquharson S A Gift P Maksymiuk F Inscore and W Smith ldquopH dependence of methyl phosphonic acid
dipicolinic acid and cyanide by surface-enhanced Raman spectroscopyrdquo SPIE 5269-19 (2003) 27 Lide DR Ed Handbook of Chemistry and Physics CRC Press 77th Ed (1996-7) 28 Sosa C RJ Bartlett K KuBulat and WB Person ldquoA theoretical study of harmonic vibrational frequencies and
infrared intensities of XCH2CH2SCH2CH2X and XCH2CH2SH (= H Cl)rdquo J Phys Chem 93 577-588 (1993) 29 Hameka HF and JO Jenson ldquoComputer-generated predictions of the structure and the IR and Raman spectra of
VXrdquo ERDEC-TR-065 May 1993
- Chemagents Appendicespdf
-
- SPIE2001-4575-Bioagent-SERSapdf
-
- Detection of bioagent signatures A comparison of electrolytic and metal-doped sol-gel surface-enhanced Raman media
- Stuart Farquharson Wayne Smith and Yuan Lee
-
- Real-Time Analyzers 87 Church Street East Hartford CT 06108
-
- ABSTRACT
- 4 CONCLUSIONS
- 5 ACKNOWLEDGEMENTS
- 6 REFERENCES
-
SPIE 2001-4575
69
SG-SER spectra of reasonable quality were also obtained for E coli and B subtilis especially the latter (Figure 8) However the spectra differ substantially from the E-SERS of the same samples Both SG-SER spectra are dominated by adenine at 735 cm-1 and a band at 1030 cm-1 Although unassigned the latter does appear in the RNA E-SER spectra Bands at 1105 and 1565 cm-1 are likely due to phosphate while bands at 1320 and 1470 cm-1 are less confidently assigned to phosphate They are significantly less intense and somewhat shifted from their SG-SERS counterparts (1335 and 1455 cm-1) A band at 670 cm-1 may be due to guanine which was observed at 664 cm-1 for SG-SERS of the pure sample However the SG-SER spectrum of AMP also had an intense 667 cm-1 band A number of other bands occur at 890 1070 1165 1245 1290 1420 1505 cm-1 and remain unassigned The SG-SER spectra are somewhat disappointing in that only adenine and guanine contributions can be positively identified This limits the ability to determine relative NA base concentrations and distinguish bacterial RNA However several of the unassigned bands may be due to the bases (eg 1030 and 1420 cm-1 due to cytosine) Further experiments will be required to clarify this point
Ecoli
B subtilis
S aureas
P G
A C+U
OPO
A B
Wavenumbers (∆cm-1) Wavenumbers (∆cm-1) Figure 7 SER spectra of RNA from A) B subtilis with contributions indicated and B) E coli B subtilis and S aureus with baseline correction and peak positions used to calculate contributions indicated G = guanine A = adenine C+U = cytosine plus uracil P = phosphate (backbone) OPO = phosphate ester linkage (A- vs B-class helix)
Wavenumbers (∆cm-1) Wavenumbers (∆cm-1) Figure 9 E-SER (-03VSCE) and SG-SER spectra of RNA from A) E coli and B) B subtilis Sample conditions as in Figure 6 E-SER spectra at 750 mW SG-SERS at 75 mW
A B E-SERS
SG-SERS
E-SERS
SG-SERS
SPIE 2001-4575
70
A final comparison was made between the two SER methods by measuring dipicolinic acid (DPA) This chemical may be invaluable as a test for spore forming bacteria specifically B anthracis 50 to 90 of B anthracis sporilates During spore formation dipicolinic acid is synthesized and once completed 10-15 of the dry spore weight is composed of the Ca2+ complex located in the spore core42 Heating in water can be used to initiate germination at which point the exosporium breaks and releases the Ca dipicolinate which becomes dipicolinic acid in water The structure of this chemical strongly
4 CONCLUSIONS Towards the goal of developing a method and technology to rapidly detect and identify bioagents we have been investigating surface-enhanced Raman spectroscopy as a tool to measure relative concentrations of nucleic acid bases in RNA and determine bioagent taxonomy Initially we investigated E-SERS since this method has been extensively researched and the optimum sample conditions are well developed However this method requires a three-electrode sample cell and electrolyte solution Incorporation of an electrolytic cell into sample systems used to analyze for BWA as aerosols in water or on surfaces can be designed using flow injection analysis technologies but cross-contamination and plugging of sample lines seems inevitable For this reason we also investigated metal-doped sol-gels as a SER-active medium Previous studies have shown this material to be active in all solvents particularly water capable of continuous measurements in flowing systems and reproducible (quantitative) between coated sample vials Here we compared SG-SER spectra to traditional E-SER spectra of the nucleic acid base pairs adenosine monophosphate and RNA High quality spectra of adenine cytosine guanine thymine and uracil were obtained by both E-SERS and SG-SERS Both methods yielded very similar spectra for the NA bases including a pH dependent study of adenine Enhancement factors and detection limits for adenine were determined as 2x105 and 16x10-11g and 1x105 and 12x10-10g for E-SERS and SG-SERS respectively (normalized to 75 mW and 10-min acquisition time) Fifty percent of the silver particle surface area in the sol-gel matrix was assumed covered by adenine which may have been overestimated yielding a lower EF and higher detection limit It should also be realized that each E-SER spectrum required several attempts to optimize the measurement conditions (pH electrode potential etc) While each SG-SER spectrum involved no sample preparation and often represents the first and only attempt to make the measurement Quality spectra of RNA extracted from Escherichia coli Bacillus subtilis and Staphylococcus aureus were obtained by E-SERS that were easily interpreted Bands due to adenine guanine cytosine plus uracil and phosphate were identified The SER band intensity of the NA bases in the RNA samples were of the same order of magnitude suggesting that their interaction with the silver surface is concerted as is their Raman enhancement Interestingly the relative SER band
Wavenumbers (∆cm-1) Figure 10 A) Raman spectrum of solid dipicolinic acid B) Ax20 C) electrolytic SERS of 6x10-3 M dipicolinic acid in 01 M KCl at a potential of +07VSCE and pH of 4 and D) sol-gel SERS of 6x10-3 M dipicolinic acid Conditions for A and C as in Figure 1 C) 100 mW of 1064 nm 50 scans 8 cm-1
A B (Ax20)
C
D
suggested that it would be SER active However the E-SER spectrum was unstable and varied considerably as a function of potential A consistent spectrum was obtained at +06VSCE (Figure 10) This potential is not recommended for measurement because the surface is actively dissolving in solution The SG-SER spectrum was considerably more stable of higher quality and easily reproduced Bands at 660 825 1010 1390 1430 1570 1590 and 3075 cm-1 were observed Enhancement factors were determined for the two media using the symmetric ring stretching mode at 995 and 1010 cm-1 for the normal Raman and SER spectra respectively E-SERS yielded an EF of 5x103 while SG-SERS yielded an EF of 2x105 for DPA The SN of the latter suggests a detection limit of 20x10-10g (based on adenine coverage 75 mW and 10-min) The differences in SER activity for these two media may be attributed to the combined electrolytic potential of the solution chemical and metal15 Again the E-SERS suggests that the SG-SERS is at a potential positive of the pzc While the instability in the E-SERS may also be associated with surface interactions of two carboxylic acid groups of dipicolinic acid during the ORCs
DPA
SPIE 2001-4575
71
intensities for RNA extracted from E coli are very similar to those measured by normal Raman spectroscopy Although the relative percent that each of the NA bases contributed to each RNA sample was not determined reproducible band intensities allowed noting the following trends The percent adenine decreases while the combined percent cytosine and guanine increase for both B subtilis and S aureus compared to E coli Quality spectra were also obtained for the RNA samples by SG-SERS but only a few bands were readily identified Calculations of NA base concentrations by SG-SERS will require further research In light of recent events we summarize the capabilities of these SERS media in terms of rapid detection of B anthracis and dipicolinic acid However these capabilities must be qualified First and foremost the level to which SERS can distinguish bacteria or viruses has not yet been determined Development of a database of both DNA and RNA base concentrations for BWAs and common bacteria to establish the level of taxonomic identification is ongoing Second rapid collection of aerosol water or surface samples is being addressed by others who report trapping particles on filters from 100 liters of air per minute Third although not presented here we have developed methods to extract RNA or DNA from cells and spores for SER analysis within 10 minutes Finally we assume a detection limit of 3600 spores per 100 liters of air is required although a 50 lethal dosage of anthrax has not been established With these qualifications a mass detection limit for RNA using SERS is estimated as follows A single measurement is performed in ca 20 minutes (140 liters collected in 14 min RNA extracted in 8 min spectral acquisition and analysis in 10 min) The average human breaths 7 liters per minute therefore the analyzer must at the very minimum detect 5000 spores in 140 liters of air One spore is approximately 2x10-18m3 (1x1x2 microm) and if a density of 075 gcm3 is assumed this corresponds to a mass of 15x10-12g Each spore contains 4-12 RNA or 12x10-13g RNA for 8 If we assume 23 of the RNA can be isolated for analysis during lysis then the proposed instrument must be able to detect 4x10-10g RNA from 5000 spores per 70 liters of air within 10 minutes As noted above the mass detection limits for adenine were estimated at 16x10-11g and 12x10-10g for E-SERS and SG-SERS respectively Although these detection limits suggests that RNA from 5000 spores is detectable with the current instrumentation it is highly likely that only a portion of an RNA segment (eg 120-nucleotide 5S rRNA) is in contact with the metal surface and will contribute to the SER effect The SN for the RNA spectra were 110th of the average SN for the four individual RNA bases suggesting a 10 contribution Furthermore effective taxonomy will likely require knowing the NA base concentrations to 1 of the value (eg 25plusmn025) Again the average measurement error for the bases is 12 These values suggest that the E-SERS is within a factor of 4 of the required detection limit whereas the SG-SERS detection limit must be improved by 25 times The same arguments can be applied to the detection of dipicolinic acid If we assume a spore releases 10 by weight DPA during germination then the proposed instrument must be able to detect 75x10-10g DPA from 5000 spores per 70 liters of air within 10 minutes The detection limit for SG-SERS was estimated at 20x10-10g and suggest that the vials are suitable to perform a rapid screen for anthrax A series of concentration dependent measurements are currently being performed to verify this assertion Finally we note that the measurements performed here employed an FT-Raman spectrometer This instrumentation was chosen over dispersive Raman instrumentation for the high wavelength accuracy afforded by the HeNe reference laser (Connes Advantage43) which would allow reliable spectral subtraction matching of observed spectra to stored library spectra and confident use of chemometric approaches Such data analysis is likely to be required to enhance BWA identification However this instrumentation sacrifices sensitivity due to a less efficient detector (InGaAs vs Si) less efficient Raman scattering and less efficient generation of plasmon modes Substantial improvements can be obtained using visible excitation and Si detection and these measurements are underway
5 ACKNOWLEDGEMENTS The authors are grateful to Drs D Cookmeyer and S Tove of the US Army Research Office (Contract Number DAAH04-96-C-0078) for their interest and support of this research The authors would also like top acknowledge Dr R Yin and J Jensen for supporting development of the metal-doped sol-gels (Contract Number DAAD13-01-C-0019) They also thank Dr Wilfred H Nelson for assistance in spectral interpretations
SPIE 2001-4575
72
6 REFERENCES 1 Roberts WL and Rainey PM Clin Chem 39 1872-1877 (1993) 2 Pasechnik VA CC Shone and P Hambleton Bioseperations 3 267-283 (1993) 3 Jackson PJ ME Hugh-Jones DM Adair G Green KK Hill CR Kuske LM Grinberg FA Abramova and P
Keim Proc Natl Acad Sci 95 1224-1229 (1998) 4 Jensen JL NC Wong W R Loerop SPIE 4775-03 (2001) 5 Snyder A Pet al SPIE 3853-15 (1999) 6 Danosky T R and McFadden P N in press (1997) 7 Woodruff WH Farquharson S Science 201 831-833 (1978) 8 Lord RC and Thomas GJJr Spectrochemica Acta 23A 2551-2591 (1967) 9 Thomas GJJr Biochim Biophys Acta 213 417-423 (1970) 10 Placzek G Handbuch der Radiologie 2 EMarx ed Akademische Verlagagescellschatt Liepzig 1934 UCRL
Trans No 526 (1959) 11 Jeanmaire DL and RP Van Duyne J Electroanalytical Chem 84 1-20 (1977) 12 Chada S Manoharan R Moenne-Loccoz P Nelson WH Peticolas WL and Sperry JF Applied Spectroscopy
47 38-43 (1993) 13 Manoharan R Ghiamati E Chada S Nelson WH and Sperry JF Applied Spectroscopy 47 2145-2150 (1993) 14 Todd EA Morris MD Applied Spectroscopy 48 545-548 (1994) 15 Wentrup-Byrne E Sarinas S and Fredericks PM Applied Spectroscopy 47 1192-1197 (1993) 16 Weldon MK VR Zhelyaskov and MD Morris Applied Spectroscopy 52 265-269 (1998) 17 Kneipp K and J Fleming J Mol Structure 145 173-179 (1986) 18 Pothier NJ and Force RK Applied Spectroscopy 46 147-151 (1992) 19 Ervin KM E Koglin JM Sequaris P Valenta and HW Nurnberg J Electroanal Chem 114 179-194 (1980) 20 Kim SK TH Joo SW Suh and MS Kim J Raman Spectrosc 17 381-386 (1986) 21 Pothier NJ and Force RK Analytical Chemistry 62 678-680 (1990) 22 Kneipp K Y Wang RR Dasari and MS Feld Applied Spectroscopy 49 780-784 (1995) 23 Nie S and Emory SR Science 275 1102 (1997) 24 Graham D WE Smith AMT Linacre CH Munro ND Watson and PC White Analytical Chemistry 69 4703-
4707 (1997) 25 Fleischmann M PJ Hendra and AJ McQuillan Chem Phys Lett 26 163-166 (1974) 26 Pettinger B J Chemical Phys 85 7442-7451 (1986) 27 Surface-Enhanced Raman Scattering Section Four Theory SPIE MS 10 M Kerker and B Thompson Eds (1990) 28 Wang D-S and Kerker M SPIE (M Kerker and B Thompson Eds) MS 10 417-429 (1990) 29 Weaver MJ Farquharson S Tadayyoni MA J Chem Phys 82 4867-4874 (1985) 30 Norrod KL Sudnik LM Rousell D and Rowlen KL Applied Spectroscopy 51 994-1001 (1997) 31 Farquharson S Weaver WJ Lay PA Magnuson RH and Taube H J Am Chem Soc 105 3350-3351 (1983) 32 Lee PC and Meisel D ldquoAdsorption and Surface-Enhanced Raman of Dyes on Silver and gold Solsrdquo J Phys Chem
86 3391-3395 (1982) 33 Li Y-S and Wang Y Applied Spectroscopy 46 142-146 (1992) 34 Bello JM DL Stokes and T Vo-Dinh Analytical Chemistry 61 1779-1783 (1989) 35 van Duyne RP J C Hulteen D A Treichel M T Smith M L Duval and T R Jensen J Phys Chem B1033854-
3863 (1999) 36 Farquharson S and W W Smith W H Nelson and J F Sperry SPIE 3533-27 207-214 (1998) 37 Lee Y H W Smith S Farquharson H C Kwon M R Shahriari and P M Rainey SPIE 3537 252-260 (1998) 38 Lee Y-H S Farquharson and P M Rainey SPIE 3857 76-84 (1999) 39 Current Protocols in Molecular Biology Wiley Interscience1003-1006 (1987) 40 Farquharson S Smith W Carangelo R C and Brouillette C SPIE 3859 14-23 (1999) 41 Chase D B and JF Rabolt Fourier Transform Raman Spectroscopy Acad Press Ch1 p 131 (1994) 42 Brock TD MT Madigan JM Martinko and J Parker Biology of Microorganisms 7th Ed Prentice Hall p 76-80
(1994) 43 Connes J Rev Opt Theor Instrum 40 45 (1961)
166
Chemical agent identification by surface-enhanced Raman spectroscopy
Stuart Farquharson and Paul Maksymiuk
Real-Time Analyzers 87 Church Street East Hartford CT 06108
Kate Ong and Steven D Christesen US Army SBCCOM Aberdeen Proving Ground MD 21010
ABSTRACT The recent distribution of anthrax through the US postal system and the subsequent infection and death of several postal and national media employees amplifies the need for methods to rapidly detect identify and quantify this and other chemical and biological warfare agents The US military has also identified water supplies as a likely method of warfare agent deployment and is funding the development of a Joint Service Agent Water Monitor (JSAWM) In an effort to aid military personnel and the public at large we are developing a portable analyzer capable of identifying and quantifying chemical agents rapidly either on-demand or continuously The approach is based on the ability of Raman spectroscopy to identify molecular structure through the abundant vibration information provided in spectra and the ability of surface-enhanced Raman spectroscopy (SERS) to detect extremely low concentrations (eg part-per-billion) through the enhancement of Raman scattering by six orders of magnitude or more A key element to the analyzer design is a new SER active medium that is capable of quantitative reversible measurements The medium consists of silver or gold nanoparticles incorporated into a sol-gel matrix The porous silica network offers a unique environment for stabilizing SER active metals and the high surface area increases the interaction between the analyte and metal particles Here we present the use of new sol-gels that also selectively enhance chemicals based on polarity and charge Base-line measurements of chemical agents and their hydrolysis products are presented and compared to the JSAWM goal of 30 micrograms per liter detection Keywords Chemical warfare agent hydrolysis product SERS Raman spectroscopy sol-gel nanoparticle
1 INTRODUCTION Since September 11 2001 the threat of terrorist attacks and biological warfare within US borders has become a sobering reality The simplicity in manufacturing ease of deployment and the relatively low cost of chemical warfare agents (CWAs) raises public concern that they may also be used by terrorists Indeed terrorists released sarin (GB) in the Tokyo subway in 19951 Countering terrorism and terrorist attacks requires recognizing likely deployment scenarios and having the required technology to rapidly detect the deployment event One method of deployment has been long identified by the US military distribution through water supplies To counter this threat the Department of Defense is funding or monitoring the capabilities of 200 technologies with the goal of developing a Joint Service Agent Water Monitor (JSAWM) that is field portable2 The JSAWM must be capable of identifying the chemical agents with no false-positives and quantifying the agents at microgram per liter concentrations within 10 minutes (Table 1)3 This includes the analysis of drinking water supplies distribution and storage systems as well as potable water supplies The earliest technologies developed for CWA detection were based on electrochemical ionization or colorimetric analysis (eg phosgene tape) Although these analyzers were easy to use they were not generally agent specific and suffered from false-positives4 More traditional laboratory methods have also been investigated and in particular combined gas chromatography and mass spectrometry (GCMS) has been very successful at eliminating false-positives56 However GCMS requires extraction repeated calibration and long analysis times (typically 20 to 60 minutes)6 making it labor intensive and less than desirable for field use More rapid analysis of agents in the solid liquid and gas phase has been To whom correspondence should be addressed emailfarqureal-time-analyzerscom
Vibrational Spectroscopy-based Sensor Systems Steven D Christesen Arthur J Sedlacek III EditorsProceedings of SPIE Vol 4557 (2002) copy 2002 SPIE middot0277-786X02$1500
167
Table 1 Chemical Agent Structures Hydrolysis Half-lives and JSAWM Thresholds Agent Short-Hand Chemical Structure Hydrolysis
Half-Life JSAWM
Thresholds
Sarin (GB) F-[O=P-CH3]-O-CH(CH3)2 213 hours 32 microgL Soman (GD) F-[O=P-CH3]-O-CH(CH3)-(C-(CH3)3) 23 hours 32 microgL Tabun (GA) (CH3)2-N-[O=P-CN]-O-C2H5 41 hours 32 microgL VX C2H5O-[O=P-CH3]-S-(CH2)2-N-(CH(CH2)2)2 821 hours 32 microgL EA2192 HO-[O=P-CH3]-S-(CH2)2-N-(CH(CH2)2)2 gt9 years 32 microgL Mustard (H) ClCH2CH2-S-CH2CH2Cl encapsulates 47 microgL Lewisite (L) ClCH=CH-As-Cl2 rapid 27 microgL HCN HCN rapid 20 mgL BZ C7NH12-O-[C=O]-COH(C6H5)2 23 microgL T-2 Toxin 87 microgL at pH 7 to 75 and 20 to 25 oC
demonstrated by vibrational spectroscopy7-10 Hoffland et al7 reported infrared absorbance spectra and absolute Raman cross sections for several chemical agents while Christesen measured Raman cross sections for sarin tabun mustard gas and VX (ethyl S-2-diisopropylamino ethyl methylphosphonothioate)11 Again however these techniques also have limitations Raman spectroscopy is simply not a very sensitive technique and detection limits are typically 01 (1000 ppm) While infrared spectroscopy would have limited value in analyzing poisoned water since the very strong infrared absorption of water would obscure most other chemicals present Nevertheless efforts to overcome these limitations have been demonstrated Braue and Pannella8 quantified the G-series nerve agents (tabun sarin and soman) in terms of infrared attenuated total reflectance using a circle-cell And Alak and Vo-Dinh demonstrated the possibility of surface-enhanced Raman spectroscopy (SERS) to identify CWAs by measuring several organophosphonates that simulate the nerve agents12 However quantitative measurements have not been demonstrated for the SER-active material used (silver coated on alumina particles) or other SER-active media13 Recently we developed silver-doped sol-gels to promote the SER effect14-17 The porous silica network of the sol-gel matrix offers a unique environment for stabilizing SER-active metal particles and the sol-gel provides a high surface area that effectively increases the number of molecules observed within the Raman scattering volume The silver-doped sol-gels have been coated on the internal walls of standard glass vials to allow rapid SER analysis of small sample quantities (lt 01 mL) without preparation We have used p-aminobenzoic acid (PABA) as a test chemical to demonstrate surface enhancements greater than 106 reversible measurements in a flowing system reproducible measurements from vial-to-vial and batch-to-batch and measurements in multiple solvents including water14-17 Recently we used these vials to measure Tabun (GB) and Sarin and several hydrolysis products pinacolyl methyl phosphonate (PMP from Soman) and methyl phosphonic acid (MPA from all G-agents Figure 1) Although a number of unique vibrational bands are observed (eg C-N stretch doublet and P-C stretch) the G-agents were only observed for 5 concentrations and all spectra required baseline corrections Figure 1 Surface enhanced Raman spectra of ~5 vv A) Tabun and B) Sarin C) 1 vv PMP and D) 10 ppm MPA using sol-gel sample vials 785 nm excitation 1-min scan and CCD detection Performed at Aberdeen Proving Ground
C A
Wavenumber (∆cm-1) Wavenumber (∆cm-1)
D B
790
545 C-N
2135 2190 P-C 770
760 1290
Proc SPIE Vol 4577
168
Nevertheless MPA was readily observed for a 10 ppm sample with an estimated detection limit of 04 ppm (based on a signal-to-noise ratio of 3 for the 760 cm-1 band intensity) This measurement provides encouragement in that SERS may satisfy the needs of a JSAWM Furthermore MPA is also a hydrolysis product of VX and V-gas and EA2192 (Figure 2) and may prove a valuable indicator of agent usage Figure 2 Hydrolysis of Sarin to form hydrofluoric acid (HF) methylphosphonic acid 1-methylethyl ester (MPAMME) methyl phosphonic acid (MPA) and 2-propanol With this initial albeit modest success we began analyzing chemicals with various sol-gel compositions that we have been developing Here we describe four sol-gel compositions that select for 1) polar-positive 2) polar-negative 3) weakly polar-positive and 4) weakly polar-negative chemical species The ability of these sol-gels to select and enhance Raman scattering is described for several test chemicals and MPA
2 EXPERIMENTAL The chemicals analyzed as well as all chemicals used to prepare the metal-doped sol-gels were obtained at their purest commercially available grade from Aldrich (Milwaukee WI) The sol-gel designed to select for polar-negative species was prepared from a silver amine complex tetramethyl orthosilicate (TMOS) and methanol After mixing 02 mL of the sol-gel solution was transferred into a glass vial (2 mL) dried and heated The incorporated silver ions were then reduced using dilute sodium borohydride The vials were washed and dried prior to the addition of a sample solution In a similar manner the sol-gel designed to select for polar-positive species was prepared from a gold salt TMOS and methanol The sol-gel designed to select for weakly polar-negative species was prepared from a silver amine complex tetraethyl orthosilicate (TEOS) and methanol And the last sol-gel designed to select for weakly polar-positive species was prepared from a gold salt TEOS and methanol All samples were prepared in a chemical hood and transferred into plain or SER-active vials for analysis Normal Raman spectral measurements employed 1-mL pure samples that were placed in a 1-cm3 cuvette and weighed This yielded a powder density that allowed accurate calculation of molecules in the optical collection field SERS measurements employed 1-mg sample per mL water concentrations unless otherwise stated Once prepared a 01 mL sample was placed into one of the four selective sample vials which in turn was placed into the sample compartment of a Raman spectrometer for analysis A prototype Fourier transform Raman spectrometer (RTA) was used for these measurements18 The system consisted of a NdYAG laser (Brimrose) for excitation at 1064 nm an interferometer built by On-Line Technologies (OLT East Hartford CT) for frequency separation an uncooled InGaAs detector for signal detection (RTA) and an Intel 400 MHz Pentium II based laptop computer (Dell Round Rock TX) for interferometric control data acquisition (OLT) and analysis (LabVIEW by National Instruments Austin TX) Additional components included a Notch filter (Kaiser Ann Arbor MI) and interferometer entrance and exit optics (Edmund Scientific Barrington NJ) Fiber optics were used to deliver the excitation beam to the sample and the scattered radiation to the interferometer (1 meter lengths of 200 and 365 micron core diameter respectively Spectran Avon CT) A second Notch filter (Kaiser) was used as a beam splitter to direct the excitation beam along the same axis as the collected radiation A microscope object (20x04 Newport Irvine CA) was used to focus the beam into the sample and to collect the scattered radiation back along the same axis In this co-axial backscattering arrangement the excitation beam was passed through the outside of a glass vial and focused onto the silver-doped sol-gel film (01-03 mm thickness) containing the sample
3 RESULTS AND DISCUSSION p-aminobenzoic acid (PABA) and phenyl acetylene (PA) and were used to refine the selectivity and SER-activity of the four different metal-doped sol-gels PABA is a popular chemical used to evaluate the performance of SER-active media Here the polar end groups can be used to test selectivity of the polar-negative and polar-positive sol-gels PA is essentially non-
2O+ H HF + +OH
OH3H C
OP
OF
CH
CH3
3
3H C
OCP
OOH
CH
CH3
3
3H C
O
CPHO
CH
CH
3
3
C
Sarin MPAMME MPA 2-propanol
Proc SPIE Vol 4577
169
polar but a high electron density in the cylindrical π cloud around the carbon-carbon triple bond allows testing the selectivity of the weakly polar-negative and weakly polar-positive sol-gels As Figure 3 illustrates PABA passes through the polar sol-gel and is enhanced by either the silver or gold particles At 1 mgml the concentration of neutral PABA is ca 20 times that of the ionized form (pKa = 48) For electropositive silver the PABA anion is expected to interact through the carboxylate group and the associated vibrational modes are expected to dominate the spectrum Conversely for electronegative gold either form of PABA is expected to interact through the amine group The clear differences in our spectra support this expectation Furthermore bands at 840 and 1405 cm-1 assigned to a COO- bend and stretch respectively are significantly more intense for silver than gold Additional bands at 1140 and 1195 cm-1 are assigned to CH bending modes while bands at 1450 1500 and 1605 cm-1 are assigned to ring vibrational modes A very similar SER spectrum for PABA on a silver-coated alumina substrate has previously been reported with similar assignments19 For the gold-doped sol-gel new bands appear at 690 1355 and 1585 cm-1 The first band is assigned to a ring-H bending mode the second band to a ring-N- stretching mode and the third band to a possible NH2 scissors mode or ring mode The second band is not observed in the normal Raman spectrum but infrared bands occur at this frequency for aromatic ring-secondary amine stretching modes The scissors mode occurs at this frequency in Raman spectra for several chemicals but is absent in the PABA Raman spectrum Alternatively this mode may be the1600 cm-1 ring mode that has been shifted by the gold interaction Again a very similar SER spectrum of PABA has been reported but surprisingly using silver (colloids)2021 not gold as the enhancement medium These researchers also assumed the primary interaction of PABA with silver was through the carboxylate anion and made assignments accordingly For example they assigned the 1359 cm-1 to a COO- stretch not to the amine group as we have They also favor the ring stretching mode assignment for the 1582 cm-1 band Finally it should be said that other researchers have argued that the most dominant band in the SER spectra at 1450 cm-1 a ring vibration mode suggests that PABA lies flat on the surface and the π-orbitals dominate the surface interaction22
Figure 3 SER spectra of A) PABA using polar-negative and B) polar-positive sol-gels and C) PA using weakly polar-negative and D) weakly polar-positive sol-gels PABA is 1 mgmL PA is 1 vv Spectral conditions 75 mw 1064 nm 100 scans (15 min) 8 cm-1 resolution Non-polar PA passes through the non-polar sol-gels and is also enhanced by both metals The spectra are easily understood For electropositive silver PA interacts through the cylindrical triple bond π electron cloud and a -CequivC- doublet occurs near 2000 cm-1 The interaction is reasonably strong since this band appears at 2112 cm-1 in the normal Raman spectrum For electronegative gold this interaction is unlikely and only very weak bands occur near 2000 cm-1 The remaining bands are at 1000 cm-1 1200 cm-1 doublet and 1595 cm-1 all appear in the normal Raman spectra at virtually the same frequencies and are assigned to the symmetric ring-breathing mode CH bending modes and the trigonal ring-breathing mode respectively The polarnon-polar selectivity of the polar-negative and weakly polar-negative sol-gels was tested by adding a 11 molar mixture of PABA and PA The selective enhancement is quite good (Figure 4) The spectrum obtained using the polar sol-gel represents 78 PABA and 22 PA while the spectrum obtained using the weakly polar sol-gel represents 9 PABA and 91 PA The band peak intensities at 2000 cm-1 for PA and 1450 cm-1 for PABA were used for these calculations and are expanded in Figure 4 for clarity
C A
Wavenumber (∆cm-1) Wavenumber (∆cm-1)
2NH COOH
D B
C CH
Proc SPIE Vol 4577
170
Figure 4 SERS of 11 MM of PABA and PA in A) polar-negative and C) weakly polar-negative sol-gels The lower traces compare the pure chemicals B) 1 mgml PABA in polar-negative sol-gel and D) 1 PA in weakly polar-negative sol-gel while the insets magnify the minority species for clarity (x5 in A and x10 in B) Spectral conditions as in Figure 3 Following this development of selective sol-gels that maintained SER activity we measured cyanide and MPA (Figure 5) Not surprisingly the best sensitivity for both hydrolysis products was obtained using the polar-negative sol-gel The interaction of the cyanide anion with the silver surface is sufficient to shift the CequivN stretch observed at 2080 cm-1 in the normal Raman spectrum to 2145 cm-1 in the surface-enhanced Raman spectrum Furthermore the band is substantially broadened This anion has been extensively studied by electrolytic SERS and this shift and broadening have been attributed to the formation of a tetrahedral Ag(CN)3
2- surface structure23 Figure 5 Surface-enhanced (upper traces) and normal Raman spectra (bottom traces) of A) CN- and B) MPA in silver-doped TMOS SERS conditions as in Figure 3 and 1 mgmL Note MPA yields two distinct spectra for neutral (top) and acidic pH (middle) The normal Raman spectra employed pure powders 500 scans and 900 mW of 1064 nm SER measurements of MPA with the polar-negative sol-gel yielded two unique spectral signatures that depended on solution pH (Figure 5) For more neutral solutions the P-C stretch of MPA at 762 cm-1 dominates and the CH2 stretch at 2922 cm-1 appears The SN is sufficiently high that the anti-Stokes Raman shift at -762 cm-1 is observed For deprotonated MPA an oxygen-surface mode appears at 325 cm-1 (as well as its anti-Stokes complement) suggesting a strong interaction This results in substantial enhancement of the P-O-C mode at 1051 cm-1 (upper trace) Others report that this mode dominates the infrared spectra of nerve agents measured in water8 Comparison of the two spectra suggests the following molecule-to-surface orientations The appearance of the oxygen-surface and P-O-C modes in the upper spectrum of Figure 5B indicates that the tetrahedral molecule interacts with the silver surface through the deprotonated oxygen and is oriented end-on The
A B
Wavenumber (∆cm-1)
C A
Wavenumber (∆cm-1) Wavenumber (∆cm-1)
D B
Wavenumber (∆cm-1)
Proc SPIE Vol 4577
171
dominance of the P-C and the CH2 stretches and the disappearance of the P-O-C mode in the upper spectrum suggest the molecule is oriented with the methyl-hydroxyl-hydroxyl face towards the surface However considerably more research must be performed to verify these points Both chemical forms were measured as a function of concentration to determine the onset of monolayer coverage and estimate expected detection limits (Figure 6) Below monolayer coverage the signal to concentration dependence should be linear and the SN of any spectral measurement in this range can be used to predict the detection limit In the spectra presented here the peak height was used as the signal while the noise as root-mean-squared (RMS) was measured between 4400-4600 cm-1 Since noise is distributed evenly throughout the spectrum when transformed this region was used since it does not have any contributions from signals or baseline offsets Figure 6 shows a series of spectra for MPA along with a plot of the 1050 cm-1 band intensity (the noise was identical for this series) as a function of sample concentration A clear discontinuity appears in the vicinity of 01 mgmL (19 ppm) indicating the onset of monolayer coverage A detection limit defined as a SN of 3 was calculated for the 01 and 005gmL samples at 24x10-4 and 25x10-4 gL respectively A more modest detection limit of 101x10-4 gL was obtained using the 760 cm-1 band in the second series of concentration measurements These detection limits correspond to 46 and 190 ppb for a 3-min scan and 75 mW of incident laser power
Figure 6 A) Concentration dependence of MPA SERS measured in silver-doped TMOS) B) Concentrations are 001 005 01 05 1 gL (188 94 188 94 188 ppm) I760 series (bull) and I1050 series (∆)
Enhancement factors for cyanide and methyl phosphonic acid can be determined by estimating the number of molecules contributing to the surface-enhanced and normal Raman spectra The enhancement factor EF can be defined by the following equation
EF = (ISERSINR)bull(MNRMSERS) bull(PNRPSERS) bull(TNRTSERS)12
where I is the spectral band intensity M is the sample mass P is the incident laser power and T is the measurement time (or number of scans) for the two measurements For the normal Raman spectra a cylindrical scattering volume is assumed based on the laser area (28x10-7m2 6x10-4m diameter spot) and the penetration depth (1x10-3 m)24 The density of KCN and MPA as powders were measured at 0572 and 0516 gcm3 indicating that 16x10-4 and 144x10-4 g produced the normal Raman signals in Figure 5 respectively The number of molecules contributing to the SER spectra are those on the silver particles embedded in the sol-gel The total silver surface area can be determined from the average particle size concentration and the scattering volume Previous scanning electron micrographs showed the average silver particle size to be 40 nm in diameter (335x10-23m3)17 The silver concentration is 012M based on the reactant molar concentrations and dilution factors And the scattering volume is 76x10-
11m3 again based on a cylindrical scattering volume defined by a laser area of 28x10-7m2 and a sol-gel thickness of 27x10-
4m This volume contains 123x10-6g of silver equivalent to 35x109 silver particles with a collective surface area of 18x10-
5m2 However it may reasonably be assumed that at least half of the silver surface is in contact with the sol-gel matrix and unavailable for analyte interaction If we assume monolayer coverage and that each CN molecule occupies 15x10-20m2 then approximately 62x1014 molecules or 27x10-8g of CN contribute to the SER spectrum (20x10-19m2 46x1013 molecules 74x10-9g for MPA) Accordingly the EF for cyanide equals 48x104 ((180599) bull(16x10-427x10-8) bull(90075) bull(500100)12) The EF for MPA is considerably higher at 87x106 ((60326) bull(144x10-474x10-9) bull(90075) bull(500200)12)
0
20
40
60
80
100
120
140
0 02 04 06 08 1 12[MPA] (mgmL)
I (76
0)
0
100
200
300
400
500
600
I (10
50)
Wavenumber (∆cm-1)
A B
Proc SPIE Vol 4577
172
4 CONCLUSIONS Here we present for the first time surface-enhanced Raman spectra of Tabun and Sarin obtained using silver-doped sol-gels However the inferior enhancement suggested employing SER-active sol-gels varying in composition to improve both detection limits and selectivity of the target analytes To this end we successfully demonstrated the capabilities of four sol-gels that select for 1) polar-positive 2) polar-negative 3) weakly polar-positive and 4) weakly polar-negative chemical species p-aminobenzoic acid was used to show that silver could be used to attract polar-negative chemicals or functional groups (carboxylate anion) and that gold could be used to attract polar-positive chemicals (amine) while a mixture of p-aminobenzoic acid and phenylacetylene was used to show that tetramethyl orthosilicate preferentially solvates polar chemicals and that tetraethyl orthosilicate preferentially solvates weakly polar chemicals This increased sample control was applied to cyanide and methyl phosphonic acid two hydrolysis products of chemical warfare agents Exceptional results were obtained for methyl phosphonic acid allowing measurement of 1x10-2 gL for a 3-min scan and 75 mW of incident laser power with an estimated detection limit of 245x10-4 gL and an enhancement factor of 87x106 However this detection limit is 76 times less sensitive than required for the JSAWM (32x10-6gL for the G-agents) Finally we note that the measurements performed here employed an FT-Raman spectrometer This instrumentation was chosen over dispersive Raman instrumentation for the high wavelength accuracy afforded by the HeNe reference laser (Connes Advantage25) which would allow reliable spectral subtraction matching of observed spectra to stored library spectra and confident use of chemometric approaches Such data analysis is likely to be required to identify the chemical agents as well as distinguish hydrolysis products However this instrumentation which employs 1064 nm excitation and InGaAs detection sacrifices sensitivity We believe that the measurement sensitivity can be improved by at least two orders of magnitude by using shorter laser excitation wavelengths (eg 532 nm) This would provide more efficient Raman scattering (fourth power dependence on laser excitation wavelength) more efficient generation of plasmon modes and allow using more efficient detector material (Si vs InGaAs) These modifications are underway
5 ACKNOWLEDGEMENTS The authors would like to thank Dr R Yin and J Jensen of the US Army for supporting this work (Contract Number DAAD13-01-C-0019) They would also like to thank Advanced Fuel Research for making their laboratory facilities available
6 REFERENCES 1 Tu Anthony ldquoOverview of Sarin Terrorist Incidents in Japan in 1994 and 1995rdquo 6th CBW Protection Symposium
Stockholm Sweden 10-15 May 1998 2 Jensen JL NC Wong W R Loerop SPIE 4775-03 (2001) 3 JSAWM Requirements at wwwsbccomapgeaarmymilRDAecbcrtPRODSERJSAWMjsawmhtml 4 ldquoThe Chemical Weapons Convention Redefines Analytical Challengerdquo Analytical Chemistry News amp Features June 1
397A (1998) 5 Johnston RL Hoefler CM Fargo JC and Moberley B ldquoThe Defense Nuclear Agencyrsquos ChemicalBiochemical
Weapons Agreements Verification Technology Research Development Test and Evaluation Program and its Requirements for On-Site Analysisrdquo AT-ONSITE 5-8 (1994)
6 Black RM Clarke RJ Read RW and Reid MT ldquoApplication of gas chromatography-mass spectrometry and gas chromatography-tandem mass spectrometry to the analysis of chemical warfare samples found to contain residues of the nerve agent sarin sulphur mustard and their degradation productsrdquo J Chromatography 662 301-321 (1994)
7 Hoffland LD Piffath RJ Bouck JBrdquoSpectral signatures of chemical agents and simulantsrdquo Optical Engineering 24 982-984 (1985)
8 Braue EHJ Pannella MGrdquoCIRCLE CELL FT-IR Analysis of Chemical Warfare Agents in Aqueous Solutionsrdquo Applied Spectroscopy 44 1513-1520 (1990)
9 Tseng C-H Mann CK and Vickers TJ ldquoDetermination of Organics on Metal Surfaces by Raman Spectroscopyrdquo Applied Spectroscopy 47 1767-1771 (1993)
10 Staff Reporter ldquoUS Army Tests Lidar to Detect Biological Toxinsrdquo Photonic Spectra pg 50 December 1998
Proc SPIE Vol 4577
Proc SPIE Vol 4577
173
11 Christesen SD Raman cross sections of chemical agents and simulants Applied Spectroscopy 42 318-321 (1988) 12 Alak AM and Vo-Dinh T ldquoSERS of Organophosphorous Chemical Agentsrdquo Analytical Chemistry 59 2149-2153
(1987) 13 Norrod KL Sudnik LM Rousell D and Rowlen KL ldquoQuantitative Comparison of Five SERS Substrates
Sensitivity and Detection Limitrdquo Applied Spectroscopy 51 994-1001 (1997) 14 Lee Y and Farquharson S ldquoSERS Sample Vials Based on Sol-Gel Process for Trace Pesticide Analysisrdquo SPIE 4206
140-146 (2000) 15 Farquharson S and Lee Y ldquoTrace Drug Analysis by Surface-Enhanced Raman Spectroscopyrdquo SPIE 4200-16 (2000) 16 Lee Y Farquharson S and Rainey P M Surface-Enhanced Raman Sensor for Trace Chemical Detection in Water
SPIE 3857 76-84 (1999) 17 Lee Y Farquharson S Kwong H and Shahriari M ldquoSurface-Enhanced Raman Sensor for Surface-Enhanced Raman
Spectroscopyrdquo SPIE 3537 252-260 (1998) 18 Farquharson S Smith W Carangelo R C and Brouillette C ldquoIndustrial Raman Providing Easy Immediate Cost
Effective Chemical Analysis Anywhererdquo SPIE 3859 14-23 (1999) 19 Narayanan VA JM Bello JD Stokes and T Vo-Dinh Analusis 19 307-310 (1991) 20 Laserna JJ E L Torres and JD Winefordner Analytica Chemica Acta 469-480 (1987) 21 Torres EL and JD Winefordner Analytical Chemistry 59 1626-1632 (1987) 22 Suh JS DP DiLella M Moskovits J Phys Chem 87 1540-1544 (1983) 23 Benner RE R Dornhaus R Chang and BL Laube Correlations in the Raman spectra of cyanide complexes adsorbed
at silver electrodes with voltammograms Surface Science 101 341 (1980) 24 Chase D B and JF Rabolt Fourier Transform Raman Spectroscopy Acad Press Ch1 p 131 (1994) 25 Connes J Rev Opt Theor Instrum 40 45 (1961)
APPLIED SPECTROSCOPY 351
focusing the 488 nm laser beam 2 mm deep into thebulk of the crystals to avoid contributions from potentialdepletion layers As shown in Fig 1b the dependence ofthe Raman shift on the degree of deuteration is almostperfectly linear and ts very well with D 5 22684cmR1 24526 where D is the degree of deuteration (in )and R is the spectral mean of the PO4 vibration in cm21A linear correlation coef cient of 0998 indicates an ex-cellent linear dependence of the Raman peak shift withdegree of deuteration This result shows that the shift ofthe PO 4 peak is simply caused by the linear increase inatomic mass due to isotope substitution which decreasesthe length of hydrogen-like bonds
This excellent linear dependence allows us to map thepro le of the DH exchange layer at the surface of DKDPcrystals by acquiring Raman spectra and determining theposition of the PO 4 peak for various depths This methodis preferable over other methods such as determining thestrength of the OD vibration directly (eg at 715 cm21)because the position of the most intense peak in the Ra-man spectrum can be measured more precisely than theintensity of some of the weakest peaks in the spectrumThis is demonstrated in Fig 2 where depth-dependentRaman spectra (Fig 2a) and the resulting exchange layerpro les for two DKDP crystals are shown (Fig 2b) Thespectra in Fig 2a were obtained from a depth scan of aDKDP crystal with 75 degree of deuteration in thebulk grown at 45 8C The spectra start out as DKDP with30 deuteration close to the surface and approach thebulk DKDP spectrum within a few micrometers of depthThe fact that the relative degree of deuteration does notextend to 0 D is due to the limited depth resolution ofthe Raman microprobe which averages over 4 mm indepth Figure 2b depicts the resulting DH exchange layerpro les for this and a second crystal grown at 63 8Crespectively Both crystals had the same exposure to am-bient conditions and their main difference is the temper-ature at which they were grown The different exchangelayer pro les indicate that crystals grown at differenttemperatures have differen t proton conductiv ities 13
which leads to a difference in their rate of deuteriumdepletion The parameters controlling this behavior arecurrently the objective of a detailed study the results ofwhich will be reported elsewhere
CONCLUSION
In conclusion we have shown that the shift of the to-tally symmetric PO 4 stretch mode in the Raman spectrumof DKDP crystals scales linearly with degree of deuter-ation This allows us to correlate Raman peak positionsto deuteration levels in these crystals We have presenteda new technique to determine DH diffusion pro les inDKDP frequency conversion crystals based on micro-Ra-man spectroscopy This technique is fast inexpensiveand works under various environmental conditionswhich will allow us to better understand and control deu-terium depletion in DKDP crystals
ACKNOWLEDGMENTS
We would like to thank M Runkel for rst discovering DKDP crack-ing R Floyd for providing DKDP crystals and L Chase and A Burn-ham for their support and helpful discussions This work was performedunder the auspices of the US Department of Energy by the University
of California Lawrence Livermore National Laboratory under contract W-7405-Eng-48
1 J J De Yoreo A K Burnham and P K Whitman Int Mat Rev47 113 (2002)
2 C E Barker R A Sacks B M Van Wonterghern J A Caird JR Murray J H Campbell K Kyle R B Ehrlich and N DNielsen Proc SPIE-Int Soc Opt Eng 2633 501 (1995)
3 T Suratwala paper to be published4 Cleveland Crystals Inc httpwwwclevelandcrystalscom
KDPshtmltable5 E A Popova I T Savatinova and I A Velichko Sov Phys Solid
State 12 1543 (1971)6 I P Kaminow R C C Leite and S P S Porto J Phys Chem
Solids 26 2085 (1965)7 J A Subramony B J Marquardt J W Macklin and B Kahr
Chem Mat 11 1312 (1999)8 H Tanaka M Tokunaga and I Tatsuzaki Solid State Commun
49 153 (1984)9 R J Nelmes G M Meyer and J E Tibballs J Phys C 15 59
(1982)10 M A Yakshin D W Kim Y S Kim Y Y Broslavets O E
Sidoryuk and S Goldstein Laser Physics 7 941 (1997)11 I Takenaga Y Tominaga S Endo and M Kobayashi Solid State
Commun 84 931 (1992)12 C Krenn personal communication13 M Sharon and A K Kalia J Solid State Chem 21 171 (1977)
Rapid Dipicolinic Acid Extractionfrom Bacillus Spores Detectedby Surface-EnhancedRaman Spectroscopy
STUART FARQUHARSON ALAND GIFT PAUL MAKSYMIUK andFRANK E INSCOREReal-Time Analyzers Inc East Hartford Connecticut06108
Index Headings Dipicolinic acid Bacillus spores Anthrax Surface-enhanced Raman spectroscopy
INTRODUCTION
The anxiety caused by the distribution of anthrax en-dospores through the US postal system in October 2001was exacerbated by the long time required for positiveidenti cation of the Bacillus anthracis spores and the un-known extent of their distribution Since that time manymethods capable of rapid eld analysis have been inves-tigated to augment or replace the laboratory method ofgrowing microorganisms in culture media which takesdays to perform12 Prominent among these approachesare polymerase chain reactions (PCR)3 immunoassaysand detection of calcium dipicolinate as a biochemicalsignature PCR employs primers to separate organism-
Received 10 October 2003 accepted 14 November 2003 Author to whom correspondence should be sent
352 Volume 58 Number 3 2004
speci c nucleic acid sequences (eg capsular protein en-coding gene for Bacillus anthracis)4 and polymerases toamplify the segment until it is detectable Recently am-pli cation times have been substantially reduced andcomplete analysis can now be performed in an hour orless Immunoassay methods are also being developed thatuse competitive binding of the bioagent (as an antigen)and its labeled conjugate for a limited number of anti-bodies Although analyses can be performed in under 30minutes a well-de ned anthrax antigen has not yet beenidenti ed5ndash7 and consequently the false-positive rate isunacceptably high8
A number of other methods are being developed witha focus on the detection of calcium dipicolinate (CaDPA)and its derivatives as a B anthracis signature This is sobecause only spore-forming bacteria contain CaDPA andthe most common potentially interfering spores such aspollen and mold spores do not Relatively fast methodshave been developed to chemically extract CaDPA andthen detect it directly by uorescence9 or indirectly byluminescence1011 In the latter case hot dodecylamine(DDA) has been used to extract dipicolinic acid (DPA)and terbium has been utilized to form a highly lumines-cent DPA complex11 Although measurements have beenperformed in as little as ve minutes it was found thatas many as three concentration-dependent complexes canform each with different lifetimes This coupled withthe fact that the Tb31 cation produces the same lumines-cence spectrum makes determinations of low spore con-centrations problematic
It has been long known that Raman spectra of Bacillispores are dominated by bands associated with CaDPA12
and that these spectra may provide a suitable anthrax sig-nature at the genus level13 Since that time considerableimprovements in Raman instrumentation have led to lab-oratory measurements of single Bacilli spores14 and to eld measurements of spores captured from a mail-sort-ing system15 However the single spore measurementsrequired complex instrumentation that is not ruggedwhile the eld measurements required milligrams of sam-ple Furthermore the Raman spectra of both measure-ments contained uorescence contributions that would in-crease uncertainty in quanti cation
In related research we demonstrated that nanogramquantities of DPA could be detected by uorescence-freesurface-enhanced Raman spectroscopy (SERS)16 We alsodemonstrated that microliter volumes of chemicals canbe detected by SERS using metal-doped sol-gel-packedglass capillaries17 Towards the goal of developing a rap-id eld SERS-based anthrax spore detector we havecombined our previous research and we now report thatDPA can be extracted from a 10 mg B cereus spore sam-ple using DDA in 1 minute and can be detected by SERSin an additional 1 minute
EXPERIMENTAL
Dipicolinic acid (26-pyridinedicarboxylic acid DPA)and dodecylamine (DDA) were used as received fromSigma-Aldrich (Milwaukee WI) Lyophilized B cereusspores prepared according to the literature13 were sup-plied by the University of Rhode Island and used as re-ceived Multiple particles approximately 01 mm3 each
were separated and weighed at 5 to 15 mg representing05 to 15 million spores The sample masses were con-sistent with a previous determination of spore density at0081 gmL that indicated a high degree of entrained air
All chemicals used to prepare the silver-doped sol-gelcoated capillaries were also obtained and used as receivedfrom Sigma-Aldrich According to previously publishedprocedures17 two precursor solutions were preparedmixed and then drawn into 1-mm-diameter glass capil-laries The silver amine precursor consisted of a 51 vvratio of 1 N AgNO 3 to 28 NH3OH while the alkoxideprecursor consisted of a 21 vv ratio of methanol to te-tramethyl orthosilicate The alkoxide precursors weremixed with silver amine precursor in an 81 vv ratioApproximately 015 mL was drawn into the capillarycoating a 15-mm length After sol-gel formation the in-corporated silver ions were reduced with dilute sodiumborohydride which was followed by a water wash to re-move residual reducing agent
A 100 mL drop of a 50 mM DDA solution in ethanolpre-heated to 78 8C was added to each of the B cereusparticles to digest the spore coat After 1 minute the re-sultant solution was drawn into a SER-active capillarythat was immediately xed horizontally to an XY posi-tioning stage (Conix Research Spring eld OR) just in-side the focal point of an f 07 aspheric lens The lensfocused the beam into the sample and collected the scat-tered radiation back along the same axis A dichroic lter(Omega Optical Brattleborough VT) was used to re ectthe excitation laser to the lens and pass the Raman scat-tered radiation collected by the lens An f 2 achromatwas used to collimate the laser beam exiting a 200-mm-core-diameter source ber optic while a second f 2 ach-romat was used to focus the scattered radiation into a 365mm ber optic (Spectran Avon CT) A short-pass lterwas placed in the excitation beam path to block the sil-icon Raman scattering generated in the source ber fromre ecting off sampling optics and reaching the detectorA long-pass lter was placed in the collection beam pathto block the sample Rayleigh scattering from reachingthe detector A 785 nm diode laser (Process InstrumentsInc model 785-600 Salt Lake City UT) was used todeliver 100 to 150 mW of power to the sample A Fouriertransform Raman spectrometer (Real-Time Analyzersmodel IRA-785 East Hartford CT) and a silicon photo-avalanche detector (Perkin Elmer model C30902S Stam-ford CT) were used to acquire the SER spectra
RESULTS AND DISCUSSION
As an initial experiment the SER spectrum of 1 gLof DPA in water was measured using the newly devel-oped silver-doped sol-gel-coated capillaries (Fig 1A) Atthis concentration a high signal-to-noise ratio (SN) isobtained in 1 min In fact a reasonable spectrum is ob-tained in the same time frame for 1 mgL (Fig 1B) TheSER spectra are reasonably similar to the normal Raman(NR) spectrum obtained for a saturated solution of DPAin 1 N KOH (Fig 1C) and the following band shifts areobserved (NR to SER) 647 to 657 cm21 817 to 815cm21 998 to 1008 cm21 1384 to 1382 cm21 1434 to1428 cm21 and 1569 to 1567 cm21 Many of these bandshave been previously assigned1213 such as 998 cm21 to
APPLIED SPECTROSCOPY 353
FIG 1 SERS of DPA in water using silver-doped sol-gel-coated glasscapillary for (A) 1 gL and (B) 1 mgL (C ) NR of saturated DPA in 1N KOH in a glass capillary Spectral conditions (A) and (B) 150 mWof 785 nm 1-min acquisition time (C) 450 mW of 785 nm 5-minacquisition time both 8 cm 21 resolution
FIG 2 SERS of DPA extracted from 10 mg B cereus particle using100 mL of 50 mM hot DDA acquired in (A) 1 minute and (B) 2 seconds(C ) Attempted SERS of 50 mM hot DDA in ethanol using silver-dopedsol-gel-coated glass capillary acquired in 1 min Spectral conditions150 mW of 785 nm 8 cm21 resolution
the symmetric ring stretch 1384 cm21 to the OndashCndashOsymmetric stretch 1428 cm21 to the symmetric ring CndashH bend and 1569 cm21 to the asymmetric OndashCndashOstretch
The rst B cereus samples consisted of 2 mg of sporesin 2 mL of 5 mM hot DDA The samples were main-tained at 78 8C for 40 min and while hot approximately10 mL was drawn into a SER-active capillary Since spec-tra of DPA were obtained for these initial samples small-er spore masses higher DDA concentrations and shorterheating periods were examined In due course it wasfound that 10 mg of spores could be digested by 100 mLof 50 mM hot DDA in one minute and detected (Fig2A) In fact the signal was suf ciently intense that it canbe observed in as little as two seconds (Fig 2B) Theamount of DPA that was extracted was estimated to bebetween 5 and 10 mgL by comparing the signal intensityof the 1008 cm21 band to that measured for DPA in waterThis is consistent with previous research that found thatthe majority of the DPA is extracted from spores usingDDA11 and that B cereus spores contain approximately10 DPA by weight18 The SN of 127 for the 1008 cm21
band in the 1-minute SER spectrum suggests a limit ofdetection of approximately 250 ng of B cereus sporesbased on a SN of 3 Finally it should be noted that DDAdid not produce a detectable SER spectrum as shown inFig 2C
CONCLUSION
We have demonstrated that by combining rapid extrac-tion of dipicolinic acid from Bacillus cereus spores withchemical identi cation by surface-enhanced Raman spec-troscopy as little as 10 mg of spores can be detected Infact the entire measurement from the time of adding hotdodecylamine to the spores to the time when the dipi-colinic acid SER spectrum is acquired and analyzedcould be performed in less than two minutes The abilityof this method to distinguish between spore-forming bac-teria such as Bacillus anthracis and non-DPA containingpowders could help prevent costly shutdowns associated
with the appearance of suspicious material or intentionalmailing of common substances as an anthrax hoax Thismethod could also prove useful in detecting the locationof anthrax endospores in mail distribution facilities if an-other veri ed attack should occur
Research continues to fully characterize the surface-enhanced Raman spectroscopy signal intensities as afunction of sample concentration and to explore otherextractants that do not require the use of elevated tem-perature
ACKNOWLEDGMENTS
The authors are grateful for the support of the National Science Foun-dation (DMI-0296116 and DMI-0215819) and the US Army(DAAD13-02-C-0015 Joint Service Agent Water Monitor program)The authors are indebted to Chetan Shende for preparing the sol-gelcapillaries The authors also thank James Gillespie Nicholas Fell andAugustus Fountain for providing important background informationMark Farquharson for laboratory support and Professor Jay Sperry ofthe University of Rhode Island for supplying B cereus spores
1 V A Pasechnik C C Shone and P Hambleton Bioseparations 3267 (1993)
2 P J Jackson M E Hugh-Jones D M Adair G Green K K HillC R Kuske L M Grinberg F A Abramova and P Keim ProcNatl Acad Sci USA 95 1224 (1998)
3 B R Glick and J J Pasternak Molecular Biology Principles andApplications of Recombinant DNA (ASM Press Washington DC1994)
4 C A Bell J R Uhl T L Had eld J C David R F Meyer TF Smith and F R Cockerill III J Clin Microbiol 40 2897(2002)
5 D L Gatto-Menking H Yu J G Bruno M T Goode M Millerand A W Zulich Biosens Bioelectron 10 501 (1995)
6 J J Quinlan and P M Foegeding J Rapid Methods AutomationMicrobiol 6 1 (1998)
7 A A Hindle and E A H Hall Analyst (Cambridge UK) 1241599 (1999)
8 M S Ascher US Department of Health amp Human Services(httpwwwhhsgovophppresentationsAscherdoc)
9 R Nudelman B V Bronk and S Efrima Appl Spectrosc 54 445(2000)
10 D L Rosen C Sharpless and L B McBrown Anal Chem 691082 (1997)
354 Volume 58 Number 3 2004
11 P M Pellegrino N F Fell Jr and J B Gillespie Anal ChimActa 455 167 (2002)
12 W H Woodruff T G Spiro and C Gilvarg Biochem BiophysRes Commun 58 197 (1974)
13 E Ghiamati R S Manoharan W H Nelson and J F SperryAppl Spectrosc 46 357 (1992)
14 A P Esposito C E Talley T Huser C W Hollars C M Schal-dach and S M Lane Appl Spectrosc 57 868 (2003)
15 S Farquharson L Grigely V Khitrov W W Smith J F Sperryand G Fenerty J Raman Spectrosc paper accep ted (2003)
16 S Farquharson W W Smith S Elliott and J F Sperry SPIE-IntSoc Opt Eng 3855 110 (1999)
17 S Farquharson and P Maksymiuk Appl Spectrosc 57 479(2003)
18 F W Janssen A J Lund and L E Anderson Science (Washing-ton DC) 127 26 (1958)
SPIE -2003-5269 117
pH dependence of methyl phosphonic acid dipicolinic acid and cyanide by surface-enhanced Raman spectroscopy
Stuart Farquharson Alan Gift Paul Maksymiuk Frank Inscore and Wayne Smith
Real-Time Analyzers 87 Church Street East Hartford CT 06108
ABSTRACT US and Coalition forces fighting terrorism in Afghanistan and Iraq must consider a wide range of attack scenarios in addition to car bombings Among these is the intentional poisoning of water supplies to obstruct military operations To counter such attacks the military is developing portable analyzers that can identify and quantify potential chemical agents in water supplies at microgram per liter concentrations within 10 minutes To aid this effort we have been investigating the value of a surface-enhanced Raman spectroscopy based portable analyzer In particular we have been developing silver-doped sol-gels to generate SER spectra of chemical agents and their hydrolysis products Here we present SER spectra of methyl phosphonic acid and cyanide as a function of pH an important factor affecting quantitation measurements which to our knowledge has not been examined In addition dipicolinic acid a chemical signature associated with anthrax-causing spores is also presented Keywords Chemical warfare agents agent detection agent hydrolysis SERS Raman spectroscopy homeland security
1 INTRODUCTION In the past decade the Unites States and its allies have been challenged by a different kind of warfare exemplified by the terrorist attacks of September 11 2001 Suicide bombings and the use of chemical agents are the norm and military personnel must consider a wide range of attack scenarios Among these is the intentional poisoning of water supplies to obstruct military operations in Afghanistan and Iraq Currently colorimetric paper is used to detect agents on-site while gas chromatography combined with mass spectrometry (GCMS) is used in mobile support laboratories However both methods have severe drawbacks The paper changes color in response to contact with many chemicals besides CWAs causing a high incidence of false positives1 while GCMS although very chemically specific requires hours to perform and constant re-calibration234 Military operations would be greatly aided by a portable analyzer that can identify and quantify potential chemical agents at concentrations that impact safety This includes the analysis of drinking water supplies distribution and storage systems To meet this goal the Department of Defense has been investigating numerous approaches under the auspices of the Joint Service Agent Water Monitor (JSAWM) program5 The JSAWM must be capable of identifying the chemical agents with no false-positives and quantifying the agents at microgram per liter concentrations within 10 minutes678 Recently we and others have been investigating the ability of surface-enhanced Raman spectroscopy (SERS) to measure chemical agents9-12 bioagents13-17 and their hydrolysis products in water SERS employs the interaction of surface plasmon modes of metal particles with target analytes to increase scattering efficiency by as much as 1 million times18 In 1987 the potential of this approach to measure CWAs was demonstrated by measuring a series of pesticides19 Several of these organophosphonates have chemical structures similar to CWAs in particular P=O functional groups In our studies we have been employing silver-doped sol-gels to promote the SER effect The porous silica network of the sol-gel matrix offers a unique environment for immobilizing and stabilizing SER-active metal particles20-23 The sol-gels have been coated on the internal walls of standard glass vials to allow rapid SER analysis of small sample quantities without preparation We have measured over 100 chemicals with enhancements of 104 to 106 demonstrated reversible measurements in a flowing system reproducible measurements from vial-to-vial and measurements in multiple solvents including water20-23 Previously we used these vials to perform preliminary measurements of cyanide (CN) methylphosphonic acid (MPA) and dipicolinic acid (DPA) MPA is a hydrolysis product of the nerve agents (eg sarin Reaction 1) and may be a valuable indicator of nerve agent usage particularly since the alkyl methylphosphonic acids are relatively more stable than their corresponding parent complexes24 DPA is
SPIE -2003-5269 118
a chemical signature of spore forming bacteria such as Bacillus anthracis And in light of the inability to rapidly detect the anthrax spores distributed through the US mail in October 2001 a number of methods are being developed to extract and analyze this signature Reaction 1 Stepwise hydrolysis of Sarin to form hydrofluoric acid (HF) isopropyl methylphosphonic acid (IMPA) then methyl phosphonic acid (MPA) and 2-propanol In our previous SERS investigations MPA and DPA were measured at 50 and 100 mgL respectively In both cases limits of detection (LOD) were estimated at 100 microgL providing encouragement in that SERS may satisfy the needs of the JSAWM Since it has been shown that pH can substantially influence the intensity of SER bands25 which would clearly influence quantitative analysis we undertook the present study to determine the severity of these effects for cyanide methyl phosphonic acid and dipicolinic acid Furthermore we previously observed a band at 1050 cm-1 for MPA6 possibly due to an anion formed at basic pH Here we investigate the source of this spectral anomaly
2 EXPERIMENTAL All chemicals including potassium cyanide methyl phosphonic acid dipicolinic acid and those used to prepare the silver-doped and gold-doped sol-gel coated vials were obtained and used as received from Sigma-Aldrich All samples were prepared in HPLC grade water (Fischer Scientific Fair Lawn NJ) for SERS measurements The pH of these samples was adjusted using dilute nitric acid or potassium hydroxide and verified using a pH electrode (Corning Inc Corning NY) that had been calibrated with pH 400 700 and 1000 buffered standards from Fischer Scientific Once prepared the samples were transferred into the silver-doped sol-gel vials (Real-Time Analyzers Inc East Hartford CT) The vials were coated in a manner similar to that previously reported by adding ammonium hydroxide to a solution of silver nitrate tetramethyl orthosilicate (TMOS) and methanol20 Gold-vials were coated by adding nitric acid to a solution of gold tetrachloride TMOS and methanol The two precursor solutions were prepared mixed and transferred to 2-mL glass vials dried and heated After sol-gel formation the incorporated metal ions were reduced with dilute sodium borohydride (1mgmL) which was followed by a water wash to remove residual reducing agent After the resultant analyte solution was introduced the SER-active vial was immediately fixed horizontally to an XY positioning stage (Conix Research Springfield OR) just inside the focal point of an f07 aspheric lens The lens focused the beam into the sample and collected the scattered radiation back along the same axis A dichroic filter (Omega Optical Brattleborough VT) was used to reflect the excitation laser to the lens and pass the Raman scattered radiation collected by the lens An f2 achromat was used to collimate the laser beam exiting a 200 microm core diameter source fiber optic while a second f2 achromat was used to focus the scattered radiation into a 365 microm fiber optic (Spectran Avon CT) A short pass filter was placed in the excitation beam path to block the silicon Raman scattering generated in the source fiber from reflecting off sampling optics and reaching the detector A long pass filter was placed in the collection beam path to block the sample Rayleigh scattering from reaching the detector A 785 nm diode laser (Process Instruments Inc model 785-600 Salt Lake City UT) was used to deliver 100 to 150 mW of power to the sample A Fourier transform Raman spectrometer (Real-Time Analyzers model IRA-785 East Hartford CT) and a silicon photo-avalanche detector (Perkin Elmer model C30902S Stamford CT) were used to acquire the SER spectra
3 RESULTS AND DISCUSSION In a previous study of MPA6 aimed at developing a concentration calibration curve and determining limits of detection (LOD) we observed an anomaly at 1050 cm-1 Since it was found that the band intensity changed as a function of concentration the band must be associated with a sample parameter Two possible parameters photon flux and pH are examined here The first parameter was investigated by irradiating a 1mgmL MPA sample in a SER-active vial with laser powers of 200 mW and above and monitoring spectral changes It was immediately found that the 1050 cm-1 band
2O+ H HF + +OH
OH3H C
OP
OF
CH
CH3
3
3H C
OCP
OOH
CH
CH3
3
3H C
O
CPHO
CH
CH
3
3
C
Sarin IMPA MPA 2-propanol
SPIE -2003-5269 119
appeared and grew as a function of time and that the higher the laser power the faster the growth Figure 1 shows the growth of the 1050 cm-1 band over the course of 30 minutes when using 150 mW of 785 nm excitation while Figure 2A shows that the growth can be fit with a first order exponential equation namely I1050 = 03+05e-013t Furthermore the 760 cm-1 band could be fit with a first order decay equation with an identical exponential rate constant ie I760 = 08-
08e-013t The rates represent classical first order kinetics and their correspondence allows one to conclude that MPA is being transformed one-for-one into a photo-generated product At this time the photoproduct has not been positively identified but phosphonic (phosphorous) acid and phosphonate are likely candidates since the symmetric P(OH)3 stretch occurs at ~1050cm-1 Our studies show that a reduction in laser power to 100 mW at the sample for MPA essentially eliminates this degradation process This laser power or lower was used for further measurements
Methyl phosphonic acid is a diprotic acid that stepwise dissociates into two anions MPA- and MPA= according to the following reactions26
MPA MPA- + H+ pKa1 = 212 Reaction 2
MPA- MPA= + H+ pKa2 = 729 Reaction 3 The relative concentrations of MPA MPA- and MPA= can be determined at any pH by expressing [MPA] and [MPA=] in terms of [MPA-] using Reactions 2 and 3 and summing all three to equal the total starting concentration here 2 mgmL (0021M MW = 9602) viz
[MPA] + [MPA-] + [MPA=] = 0021M Equation 1 substituting from Reactions 1 and 2
([H+][MPA-])K1a + [MPA-] + (K2a[MPA-])[H+] = 0021M Equation 2 rearranging [MPA-] = 0021M(1+[H+]K1a + K2a[H+]) Equation 3 The relative concentrations of MPA MPA- and MPA= as a function of pH are shown in Figure 3 It is worth noting that near neutral pH both MPA- and MPA= will be present To confirm that the SER signal followed this pH dependence a starting solution consisting of 20 mg of MPA in 10 mL HPLC grade water was prepared and brought to pH of 20 using dilute nitric acid From this solution 2 mL were added to a SER-active vial and the SER spectrum recorded At this pH a peak at 760 cm-1 was barely discernable The 2 mL solution was returned to the starting solution and the pH was re-measured to correct for any changes that the silver-doped sol-gel vials might cause In most cases the change was less than 02 pH units and the pH is reported as the before and after average Next the pH of the
Figure 2 A) Exponential growth of 1050 cm-1 band and B) exponential decay of 760 cm-1 band for spectral series in Figure 1
Figure 1 Growth of 1050 cm-1 band as a function of time due to exposure to 150 mW of 785 nm Spectra are 5 sec each collected every 100-sec from 0 to 30-min
B
570 770 970 1170 1370Raman Shift (cm-1)
Arbit
rary
Unit
s
MPA Photodegradation
600 800 1000 1200 1400 Wavenumber (cm-1)
A
0 10 20 30 0 10 20 30 time (min) time (min)
Ram
an In
tens
ity (r
elat
ive)
30
min
0
SPIE -2003-5269 120
starting solution was adjusted to 325 using dilute KOH Again 2 mL were added to a vial and the SER spectrum recorded At this pH a reasonably strong 760 cm-1 band was observed This process was repeated as spectra were recorded at pHs of 70 74 75 79 85 and 100 A total of 1 mL of KOH was added diluting the total concentration by 10 Next the pH of the starting solution was made acidic by adding dilute nitric acid dropwise This time spectra were recorded at pHs of 72 69 64 and 37 Figure 4 shows the SER spectra for representative pHs (spectra were left out to simplify the figure) while Figure 3 shows the 760 cm-1 peak intensities as a function of pH (The band intensities were adjusted to compensate for dilution effects caused by the addition of HNO3 and KOH then normalized to 0021 M for the most intense band observed at pH 37) It is clear from Figure 3 that the 760 cm-1 band follows the MPA- concentration as a function of pH and must be assigned to this anion No bands were observed that corresponded to MPA or MPA= The lack of an MPA SER spectrum may be due to the absence of an attraction between the neutral analyte and the electropositive silver surface The same reasoning suggests that a strong SER spectrum should be observed for MPA= but it is not and a satisfactory explanation has not been found
0000
0005
0010
0015
0020
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14pH
Con
cent
ratio
n [M
]
pK1 = 212 pK2 = 729
MPA- MPA=
MPA
Figure 4 SER spectra of 002M MPA as a function of pH Conditions 100 mW of 785 nm 36 scans (1 min) 8 cm-1 recorded 2 min after sample introduction pH 19 and 20 not apparent on this scale pH 69 and 74 near identical to 70 and 75 and not shown for clarity
00
02
04
06
08
10
12
14
16
18
0 5 10 15 20 25 30Measurement Number
Ram
an In
tens
ity (7
60 c
m-1
)
Figure 3 Concentration dependence of MPA MPA- and MPA= as a function of pH for a 002M sample Intensity of 760 cm-1
band from Figure 6 as a function of pH ( for increasing basic adjustment for increasing acidic adjustment error was measured at ~10 for pH 64)
Figure 5 SER spectra of 002M MPA at pH 64 measured around a vial at three heights (9 points per height) Conditions as in Figure 4 but 10-sec scans
Since these measurements involved the removal and replacement of the SER-active vial in the sample holder to remove and add sample variation in the intensity as a function of vial position was minimized by illuminating the exact same height along the vial wall But this does not account for variability of the SERS response of the sol-gel coating around the vial To analyze this effect a vial containing MPA at pH 64 was rotated at ~ 40o
intervals at the original height and 18rdquo above and below this value Figure 5 shows the intensity of the 760 cm-1
band for the 27 positions It was found that the average value was 137plusmn014 an RSD of 10 overall and 5 for each height An error bar is included in Figure 4 for the pH 64 measurement
SPIE -2003-5269 121
With the above analyses in mind a preliminary investigation of the SER spectral response for MPA (as MPA-) as a function of concentration was performed A single vial was used for these measurements beginning with 1 mgL followed by measurements of 10 100 and 1000 mgL In all cases the pH was ~7 and 3 positions around the vial were measured per concentration Since the 760 cm-1 band was not observed for concentrations of 1 or 10 mgL using 100 mW of 785 nm the laser power at the sample was raised to 200 mW beginning with the 10 mgL concentration Photo-degradation was largely avoided (and not observed) by exposing the sample for only 33 seconds per spectral acquisition Representative spectra for 10 100 and 1000 mgL are shown in Figure 6 while a plot of the 760 cm-1 band intensity as a function of concentration is shown in Figure 7 These values were also used to estimate limits of detection based on the signal-to-noise ratio (SN) of the 760 cm-1 band using 100 mW of 785 nm laser excitation and a 10-min acquisition time As summarized in the Figure 7 inset the lower the measured concentration the lower the predicted LOD Even if as estimated 210 microgL could be measured using the silver-doped sol-gel vials an improvement of a factor of 70 is still required to meet JSAWM goals of measuring 3 microgL in 10-minutes Similar to MPA DPA is a diprotic acid (pKa1 = 216 and pKa2 = 692) and variations in pH will effect the relative concentrations of DPA DPA- and DPA= and possibly the SER spectra and band intensities This could prove significant if an acid or base is used to denature anthrax spores with the goal of extracting and analyzing DPA The MPA pH study described above was mimicked for DPA except that the starting solution consisted of 20 mg of DPA in 20 mL HPLC grade water (60x10-3M MW = 1671) The initial solution had a pH of 245 which was made basic by dilute KOH to pHs of 355 433 487 559 1069 and 1166 SER spectra were recorded at each pH using 100 mW of 785 nm and a 44-sec acquisition time Next one drop of concentrated nitric acid was used to remake the solution acidic at a pH of 200 Again sequential pH measurements were performed at 383 510 735 and 822 The solution pH was made acidic a third time but to pH 219 171 then 135 Throughout this process no more than 20 drops of acid or base were added and therefore the concentration was diluted by no more than 10 Most of the spectral bands showed a minor decrease in intensity as a function of increasing pH values However the bands at 525 and 795 cm-1 showed the most dramatic changes which occurred at acid pH Figure 8 shows the SER spectra of DPA for the spectral region and pH range of interest The identity of the DPA species was determined by plotting the normalized peak intensities with the lowest value set to 0 and the highest to 0006 M as a function of pH and overlaying these values on a plot of the relative concentrations for DPA DPA- and DPA= as previously done for MPA (Figure 9) As can be seen the 525 cm-1 band clearly corresponds to DPA The correspondence of the 795 cm-1 band to this species is less clear as the band retains intensity until pH of 55 This can be attributed to contribution to the overlapping band at 810 cm-1 which does not change as a function of pH The fact that most bands are observed at all pHs suggest that the primary interaction with silver is through the ring nitrogen This is supported by the fact that the most intense band occurs at 1008 cm-1 attributed to a symmetric ring breathing mode and that this interaction has been characterized for pyridine in numerous papers27
0
1
2
3
4
5
6
7
0 200 400 600 800 1000 1200
MPA Concentration (mgL)
760
Ban
d In
tens
ity
Figure 6 SER spectra of MPA in water at A) 1000 B) 100 and C) 10 mgL Conditions pH of 7 silver-doped sol-gel coated vial 200 mW of 785 nm 33-sec 8 cm-1 resolution
Figure 7 Plot of SER intensity of 760 cm-1 band of MPA as a function of concentration using 200 mW of 785 nm Inset table includes average intensity LOD standard deviation and percent deviation for each concentration but for 100 mW and 10 min
conc (mgL) LOD ave stddev dev10 021 002 1021100 072 010 14301000 312 040 1280
A
B C
SPIE -2003-5269 122
The ability to detect dipicolinic acid (DPA) by SERS is demonstrated in Figure 10 Even at 1 mgL the primary bands are visible The intensity of the dominant ring breathing mode observed at 1008 cm-1 for DPA in water (pH = 25 to 55) is plotted as a function of concentration in Figure 11 These values were also used to estimate limits of detection based on the SN of the 1008 cm-1 band using 100 mW of 785 nm laser excitation and a 10-min acquisition time Again the lower the measured concentration the lower the predicted LOD (see Figure 11 inset) and detection of 160 microgL is possible Gastrointestinal anthrax requires significant more spores than inhalation anthrax28 and a limit of detection might be placed at 1 million spores in 1 liter of water or 10 microgL Since each spore contains ~10 CaDPA by weight29 a goal for DPA might be 1 microgL indicating that the present measurements must be improved by nearly two orders of magnitude Due to the increased hazards of handling HCN gas KCN salt was used for these experiments Nevertheless all sample preparations were performed in a chemical hood KCN completely dissolves in water but its conjugate acid HCN is formed and has a Ka of 615x10-1030 viz
HCN CN- + H+ pKa = 921 Reaction 4
Figure 8 SER spectra of 1 mgmL DPA as a function of pH Conditions 100 mW 785 nm 100 scans (44-sec)
Figure 9 Concentration dependence of DPA DPA- and DPA= as a function of pH for a 0006M sample Intensity of 525 () and 795 (diams) cm-1 bands from Fig 8 as a function of pH
pH
135 171 219 383
0
0001
0002
0003
0004
0005
0006
0007
0 2 4 6 8 10 12 14pH
Con
cent
ratio
n [M
]
DPADPA-DPA=795525
DPA DPA=DPA-
pK2 = 692pK1 = 216
795 525
0
05
1
15
2
25
0 200 400 600 800 1000 1200
DPA Concentration (mgL)
1008
Ban
d In
tens
ity
conc (mgL) lod-10min-100mw1 017
10 016100 103
1000 355
Figure 10 SER spectra of DPA in water at A) 1000 B) 100 C) 10 and D) 1 mgL Conditions pH of 25-55 silver-doped sol-gel coated vial 175 mW of 785 nm 1-min 8 cm-1 D) has been multiplied by x10 to make bands visible
A
B C
D
Figure 11 Plot of SER intensity of 1008 cm-1 band of DPA as a function of concentration using 175 mW of 785 nm Inset table includes LOD in mgL for each concentration but for 100 mW and 10 min
SPIE -2003-5269 123
Consequently the cyanide concentration must be determined for each initial KCN concentration Specifically the samples prepared with concentrations of 01 1 10 100 and 1000 mgL of KCN produced CN- concentrations of 63x10-3 033 69 89 and 964 mgL at pHs of 816 90 967 102 and 107 respectively The pH dependence for the HCN and CN- concentrations are shown in Figure 12 Thus as the amount of KCN added to the solution decreases so does the pH of the solution (becomes less basic) and according to Reaction 4 the relative amount of CN- to HCN also decreases For example in the preparation of a 01 mgL solution of KCN the pH is shifted from 7 for pure water to only 816 and only 63 of the starting material becomes CN- or 63x10-3 mgL In comparison for a solution of 1000 mgL the pH is shifted from 7 to 107 and 96 of the starting material becomes CN- This is significant since the cyanide ion is better able to adsorb onto the silver particles and become SERS active SER spectra of 10 100 and 1000 mgL of KCN samples are shown in Figure 13 and a corresponding plot of the 2100 cm-1 band intensity as a function of concentration is shown in Figure 14
000102030405060708091011
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14pH
Con
cent
ratio
n [m
gm
L]
pKa = 921
CN -HCN
The SER spectra of cyanide are dominated by the single CN stretch at 2100 cm-1 which occurs in normal Raman spectra of solutions at 2080 cm-1 However a low frequency mode occurs at 135 cm-1 due to a Ag-CN stretch (not shown) indicative of a strong surface interaction It is also observed that as the concentration decreases the CN stretch shifts to 2140 cm-1 This shift has been attribute to the formation of a tetrahedral Ag(CN)3
2- surface structure31 as well
Figure 14 Concentration dependence of KCN SERS measured under conditions in Fig 11 Concentrations are 1 01 and 001 mgml Intensities are measured for the CN stretch at 2100 cm-1 Inset table includes LOD in mgL for each concentration in Figs 13 and 15 but for 100 mW and 10 min
Figure 13 SER spectra of KCN in water at A) 1000 B) 100 and C) 10 mgL Conditions pHs of 107 102 and 97 silver-doped sol-gel coated vial 100 mW of 785 nm 1-min 8 cm-1 C) has been multiplied by x10 to make band visible
Figure 12 Concentration dependence of HCN and CN- as a function of pH for a 1 mgmL sample Calculated intensity of 2100 () cm-1 band for a 1 mgml sample at pHs of 816 90 967 102 and 107
A
B
C
0
20
40
60
80
100
120
140
0 200 400 600 800 1000 1200CN Concentration (mgL)
2100
Ban
d In
tens
ity
Figure 15 SER spectra of KCN in water at A) 10 B) 1 and C) 01 mgL Conditions pHs of 97 12 and 12 gold-doped sol-gel coated vial 100 mW of 785 nm 1-min 8 cm-1 C) has been multiplied by x10 to make band visible
A
B
C
conc (mgL) lod-10min-100mw Condition01 001 Au-pH 121 007 Au-pH 12
69 003 Au-pH 9769 016 Ag-pH 9789 022 Ag-pH 102946 113 Ag-pH 107
SPIE -2003-5269 124
as to CN adsorbed to two different surface sites32 Alternatively the 2140 cm-1 band could be attributed to HCN since this species dominates at lower concentrations However it is unlikely that this species would be attracted to the electropositive silver surface Further both peaks should be present at pHs between 85 and 105 but this is not observed It has also been suggested that at concentrations near and above monolayer coverage the CN- species is forced to adsorb end-on due to crowding and at lower concentrations the molecule can reorient to lie flat33 This suggests that the 2100 and 2140 cm-1 bands correspond to the end-on and flat orientations respectively As Figures 13 and 14 show the intensity of the CN stretch for the 89 mgL sample is nearly as intense as the 964 mgmL band This suggests that the Raman signal for the flat orientation is more enhanced However more extensive measurements are required to verify this point Since resent research has suggested that cyanide may be more effectively detected on gold measurements of KCN solutions were also performed using gold-doped sol-gel vials Preliminary measurements are shown in Figure 15 for samples prepared from 01 1 and 10 mgL KCN Since the pHs are 816 90 967 the resultant CN- concentrations are 63x10-3 033 and 69 mgL Initially only the highest concentration was observed and the signal intensity was significantly better than the equivalent concentration measured using silver In an effort to shift Reaction 4 to the left transforming HCN to CN- (Le Chatelierrsquos principle) KOH was added to the lower concentration samples producing solutions with pH 12 This effectively forces all of the cyanide in solution to be CN- or 01 and 10 mgL respectively More importantly the CN stretch is now observed in the SER spectra The band appears at 2125 cm-1 as has been previously reported for gold12 As calculated for MPA and DPA LODs can be estimated from this data For the three concentrations of cyanide on silver the LODs are 016 to 11 mgL for 100 mW of 785 nm laser excitation and a 10-min acquisition time For gold pH adjusted the LODs are10 to 70 microgL an improvement of more than 10 times silver Nevertheless either substrate is sufficient to meet the JSAWM goals of measuring 3 mgL in 10-minutes as the requirements form cyanide are much less stringent than the nerve agents
4 CONCLUSIONS Here we examined the surface-enhanced Raman spectral response for methyl phosphonic acid dipicolinic acid and cyanide as a function of pH It was determined that the most prominent peak at 760 cm-1 reaches a maximum intensity between pH 3 and 7 and corresponds to the MPA- species Neither the MPA nor MPA= species appear to generate a SER spectrum and consequently no spectra were observed below pH 2 or above pH 8 In this study we also found that higher laser powers could cause photodegradation of MPA signified by the exponential growth of a band at 1050 cm-1 which is tentatively assigned to phosphorous acid Unlike MPA DPA was observed at all pHs This is attributed to the dominant interaction of the pyridine functional group with silver Minor spectral changes were observed at acid pHs and were assigned to neutral DPA Like MPA SER spectra of cyanide were pH dependent No spectra were observed for the HCN species while CN- was best observed at pHs more basic than 8 Preliminary concentration studies for the three analytes allowed estimating limits of detection for MPA DPA and CN using 100 mW of 785 nm and a 10-min acquisition time of 210 165 and 70 microgL respectively Although the latter value suggests that cyanide can be measured at sufficiently low concentrations to meet JSAWM goal improvements by 100 to 200 times are required for MPA and DPA It is clear from this study that pH of the sample is important and must be taken into account when developing concentration calibration curves as well as fieldable analyzers Future work will include tailoring the sol-gel with specific functional groups to dictate sample pH and thereby optimize sensitivity
ACKNOWLEDGMENTS The authors are grateful for the support of the US Army (DAAD13-02-C-0015 Joint Service Agent Water Monitor program) The authors would also like to thank Dr Steve Christensen of the US Army SBCCOM for helpful discussions and Mr Chetan Shende of Real-Time Analyzers for assistance in development of the gold-doped sol-gels
REFERENCES 1 Erickson B Analytical Chemistry News amp Features June 1 397A (1998)
SPIE -2003-5269 125
2 Johnston RL Hoefler CM Fargo JC and Moberley B AT-ONSITE 5-8 (1994) 3 Black RM RJ Clarke RW Read and MT Reid J Chromatography 662 301-321 (1994) 4 USA Army Center for Health Promotion and Preventive Medicine (USACHPPM) Technical Guideline 230A 5 Jensen JL NC Wong W R Loerop SPIE 4775-03 (2001) 6 Hoenig SL Handbook of Chemical Warfare and Terrorism Greenwood Press Wesport CT (2002) 7 Munro NB SS Talmage GD Griffin LC Waters AP Watson JF King and V Hauschild Env Health
Persp 107 933-974 (1999) 8 Holstege CP Kirk M Sidell FR Crit Care Clin13 923-42 (1997) 9 Farquharson S P Maksymiuk K Ong and S Christesen SPIE 4577 166-173 (2001) 10 Lee Y and S Farquharson SPIE 4378 21-26 (2001) 11 Spencer KM J Sylvia S Clauson and J Janni SPIE 4577 158-165 (2001) 12 Tessier P S Christesen K Ong E Clemente A Lenhoff E Kaler and O Velev Applied Spectroscopy 56
1524-1530 (2002) 13 Farquharson S WW Smith S Elliott and JF Sperry SPIE 3533 207-214 (1998) 14 Farquharson S WW Smith S Elliott and JF Sperry SPIE 3855110-116 (1999) 15 Farquharson S WW Smith YH Lee S Elliott and JF Sperry SPIE 4575 62-72 (2002) 16 Guzelian AA J Sylvia J Janni S Clauson and KM Spencer SPIE 4577 182-192 (2001) 17 Shende C F Inscore A Gift P Maksymiuk and S Farquharson in press 18 Weaver MJ S Farquharson and MA Tadayyoni J Chem Phys 82 4867-4874 (1985) 19 Alak AM and T Vo-Dinh Analytical Chemistry 59 2149-2153 (1987) 20 Lee Y and S Farquharson SPIE 4206 140-146 (2000) 21 Farquharson S and Y Lee SPIE 4200-16 (2000) 22 Lee Y S Farquharson and P M Rainey SPIE 3857 76-84 (1999) 23 Lee Y S Farquharson H Kwong and M Shahriari SPIE 3537 252-260 (1998) 24 Wang J M Pumera G Collins and A Mulchandani Analytical Chemistry 74 6121-6125 (2002) 25 Dou X YM Jung Z-Q Cao and Y Ozaki Applied Spectroscopy 53 1440-1447 (1999) 26 Data supplied by S Christesen and K Ewing 27 Kerker M and B Thompson Eds SPIE MS 10 (1990) 28 Inglesby TV DA Henderson JG Bartlett JAMA 287 2236 (2002) 29 FW Janssen AJ Lund and LE Anderson Science 127 26 (1958) 30 Lide DR Ed Handbook of Chemistry and Physics CRC Press 77th Ed (1996-7) 31 Billmann J G Kovacs and A Otto Surf Sci 92 153 (1980) 32 Murray CA and S Bodoff Phys Rev B 32 671 (1985) 33 Kellogg D and J Pemberton J Phys Chem 91 1120 (1987)
SPIE-2003-5269 16
Chemical agent detection by surface-enhanced Raman spectroscopy
Stuart Farquharson Alan Gift Paul Maksymiuk Frank Inscore and Wayne Smith
Real-Time Analyzers 87 Church Street East Hartford CT 06108
Kevin Morrisey and Steven D Christesen US Army SBCCOM Aberdeen Proving Ground MD 21010
ABSTRACT
In the past decade the Unites States and its allies have been challenged by a different kind of warfare exemplified by the terrorist attacks of September 11 2001 Although suicide bombings are the most often used form of terror military personnel must consider a wide range of attack scenarios Among these is the intentional poisoning of water supplies to obstruct military operations in Afghanistan and Iraq To counter such attacks the military is developing portable analyzers that can identify and quantify potential chemical agents in water supplies at microgram per liter concentrations within 10 minutes To aid this effort we have been investigating the value of a surface-enhanced Raman spectroscopy based portable analyzer In particular we have been developing silver-doped sol-gels to generate SER spectra of chemical agents and their hydrolysis products Here we present SER spectra of several chemical agents measured in a generic tap water Repeat measurements were performed to establish statistical error associated with SERS obtained using the sol-gel coated vials Keywords Chemical agents chemical agent detection SERS Raman spectroscopy
1 INTRODUCTION
In the autumn of 2001 terrorism within US borders became a sobering reality While extensive efforts are being implemented to secure the homeland US and Coalition forces in Afghanistan and Iraq are constantly faced with terrorist attacks In addition to car-bombings the military has identified several non-traditional attack scenarios including poisoning of water supplies by chemical warfare agents (CWAs) To counter this threat the Department of Defense has been investigating numerous approaches to detect such attacks under the auspices of the Joint Service Agent Water Monitor (JSAWM) program1 The JSAWM must be capable of identifying the chemical agents with no false-positives and quantifying the agents at microgram per liter concentrations within 10 minutes2 This includes the analysis of drinking water supplies distribution and storage systems Currently colorimetric paper is used to detect agents on-site while gas chromatography combined with mass spectrometry (GCMS) is used in mobile support laboratories However both methods have severe drawbacks The paper changes color in response to contact with many chemicals besides CWAs causing a high incidence of false positives3 while GCMS although very chemically specific requires up to an hour to perform and regular re-calibration456 Vibrational spectroscopy has also been investigated as a potential method of rapidly detecting CWAs7-
11 as early as 1985 in the case of infrared spectroscopy8 and 1988 in the case of Raman spectroscopy11 Again however these techniques also have limitations when it comes to measuring trace poisons in water Infrared spectra would be dominated by the very strong absorption of water which would obscure absorptions by most other chemicals present Whereas Raman spectroscopy is simply not a very sensitive technique and detection limits are typically grams per liter Surface-enhanced Raman spectroscopy (SERS) offers several advantages over conventional vibrational methods and that may provide the necessary sensitivity required for detecting trace quantities of chemical agents in water SERS employs the interaction of surface plasmon modes of metal particles with target analytes to increase scattering efficiency by as much as 1 million times12 In 1987 the potential of this approach to measure CWAs was demonstrated by measuring a series of pesticides13 Several of these organophosphonates have chemical structures similar to CWAs in particular P=O functional groups In the past few years we and others have further explored the ability of SERS to detect CWAs14-17 and even bioagents 18-21 We have been employing silver-doped sol-gels to promote the SER effect
SPIE-2003-5269 17
in these studies The porous silica network of the sol-gel matrix offers a unique environment for immobilizing and stabilizing SER-active metal particles22-25 The silver-doped sol-gels have been coated on the internal walls of standard glass vials to allow rapid SER analysis of small sample quantities without preparation We have measured over 100 chemicals with enhancements of 104 to 106 demonstrated reversible measurements in a flowing system reproducible measurements from vial-to-vial and measurements in multiple solvents including water21-25 Previously we used these vials to perform preliminary measurements of cyanide (CN) and methylphosphonic acid (MPA) Most of the nerve agents form MPA during hydrolysis while Tabun forms CN a chemical agent in its own right In another paper including in these proceedings we examined the limits of detection (LOD) for MPA by measuring a series of concentrations down to 50 mgL and estimated a limit of detection of 100 microgL26 These measurements provide encouragement in that SERS may satisfy the needs of the JSAWM To further establish the viability of SERS in particular silver-doped sol-gels here we present analysis of cyanide mustard and VX in tap water The measurements performed at the US Armyrsquos Edgewood Chemical Biological Center Aberdeen MD also included numerous repeat measurements to establish reproducibility
2 EXPERIMENTAL 2a General All chemicals including potassium cyanide 2-chloroethylethyl sulfide and those used to prepare the silver-doped and gold-doped sol-gel coated vials were obtained and used as received from Sigma-Aldrich All samples were prepared in a chemical hood using HPLC grade water unless otherwise noted (Fischer Scientific Fair Lawn NJ) for SERS measurements Once prepared the samples were transferred into the silver-doped sol-gel vials (Real-Time Analyzers Inc East Hartford CT) The vials were coated in a manner similar to that previously reported by adding ammonium hydroxide to a solution of silver nitrate tetramethyl orthosilicate (TMOS) and methanol22 The two precursor solutions were prepared mixed and transferred to 2-mL glass vials dried and heated After sol-gel formation the incorporated metal ions were reduced with dilute sodium borohydride (1mgmL) which is followed by a water wash to remove residual reducing agent After the resultant analyte solution was introduced the SER-active vial was fixed horizontally to an XY positioning stage (Conix Research Springfield OR) just inside the focal point of an f07 aspheric lens The lens focused the beam into the sample and collected the scattered radiation back along the same axis A dichroic filter (Omega Optical Brattleborough VT) was used to reflect the excitation laser to the lens and pass the Raman scattered radiation collected by the lens An f2 achromat was used to collimate the laser beam exiting a 200 microm core diameter source fiber optic while a second f2 achromat was used to focus the scattered radiation into a 365 microm fiber optic (Spectran Avon CT) A short pass filter was placed in the excitation beam path to block the silicon Raman scattering generated in the source fiber from reflecting off sampling optics and reaching the detector A long pass filter was placed in the collection beam path to block the sample Rayleigh scattering from reaching the detector A 785 nm diode laser (Process Instruments Inc model 785-600 Salt Lake City UT) was used to deliver 100 to 150 mW of power to the sample A Fourier transform Raman spectrometer (Real-Time Analyzers model IRA-785 East Hartford CT) and a silicon photo-avalanche detector (Perkin Elmer model C30902S Stamford CT) were used to acquire the SER spectra
2b Edgewood Chemical Biological Center The surface-enhanced Raman spectral measurements at the US Armyrsquos Chemical Biological center presented here were all performed on September 12 2003 To expedite measurements a plate was machined to hold up to 12 SER-active sample vials (Figure 1) The plate fit a standard XY plate reader that could be programmed Pure KCN bis-(2-chloroethyl)sulfide (distilled mustard gas HD) and ethyl S-2-diisopropylamino ethyl methylphosphonothioate (VX) were obtained on-site and used to prepare 1 gL tap water solutions in a chemical hood with appropriate safety equipment Simulated tap water was prepared by adding 10 chemicals most often found in tap water at appropriate concentrations to distilled water (Table 1) SER measurements were also performed in a chemical hood For added safety the FT-Raman instrument was placed outside the laboratory and 30 foot fiber optic and electrical cables were used to allow remote SERS measurements and plate manipulation For each experiment 1gL samples were prepared and added to 9 individual vials which were then loaded on the plate In some cases a tenth vial was included as a blank
Table 1 Generic Tap Water Composition
Compound mgL NaHCO3 100 CaSO4 27 MgSO4bull7H2O 67 NaNO3 1 Fulvic Acid 1 K2HPO4 07 KH2PO4 03 (NH4)2HSO4 001 NaCl 001 FeSO4 0001 pH 76-78
SPIE-2003-5269 18
A software program was written that allowed selecting the sequence that the vials were measured the number of positions along the length of the vials to measure (1 to 5) and the number of scans to co-add During sample analysis the program displayed the vial being analyzed the point being analyzed and the spectrum as it was being acquired Once all the data was collected a second software program was written to rapidly analyze the data The spectra collected for all the vials on a plate could be loaded at one time and then the spectra for each point could be displayed simultaneously or separately The user could then select the Raman peak to analyze in terms of peak height or area This was accomplished by selecting points on either side of the peak to define a baseline of zero The peak height or area could then be computed for all of the spectra loaded and then exported to a spreadsheet for statistical analysis
Figure 1 A) Vial Holder 6 slots to hold 2 vials each end-to-end B) Measurement Configuration Program user selects vials to measure sequence number of points per vial (1 to 5) and number of scans per point C) Spectral Acquisition Program shows spectrum being collected which vial and position D) Spectra Analysis Program user selects spectra to analyze by plate vial and point (s) as well as two wavenumbers defining the peak and the baseline to subtract The image is of 5 repeat measurements of 10 mgL KCN in generic tap water 16 sec each 100 mW of 785 nm
3 RESULTS AND DISCUSSION Raman and surface-enhanced Raman spectra were obtained for potassium cyanide bis-(2-chloroethyl)sulfide and ethyl S-2-diisopropylamino ethyl methylphosphonothioate representing three classes of chemical agents cyanides mustards and nerve agents respectively Spectra were also obtained for 2-chloroethyl ethyl sulfide (CEES) a structural analogue to HD which was included in the study to aid in assigning spectral bands KCN salt was used for cyanide experiments to avoid the increased hazards of handling HCN gas KCN completely dissolves in water forming its conjugate acid HCN according to its Ka of 615x10-1027 and at a concentration of 1 mgmL results in a pH 107 solution This is significant in that only CN- appears to interact sufficiently with silver to produce a SER spectrum and no spectral signal is observed below pH 726 Figure 2 shows the SER and normal Raman spectra for KCN The SERS spectrum of 1mgml KCN in water shows a single intense somewhat broad feature at
A B
C D
SPIE-2003-5269 19
2100 cm-1 assigned to the single CequivN stretch The band is much sharper in the normal Raman spectra of the solid KCN salt at 2074 cm-1 This band does broaden and shift to 2080 cm-1 in solution (not shown) However the observed SERS frequency is attributed to interaction with silver and low frequency mode at 135 cm-1 attributed to a Ag-CN stretch (not shown) supports this conclusion
cm-1 band to a C-S stretch but the authors concede that it is in fact more likely a C-Cl stretch28 It appears that the most intense Raman bands at 648 692 and 747 cm-1 shift to 620 660 and 720 cm-1 in the SER spectra and are tentatively assigned as above The width of these bands suggests that they overlap underlying spectral features Additional bands in the Raman spectra occur at 972 1034 1049 1263 1286 1423 1442 2865 2935 and 2960 cm-1 Corresponding bands occur in the SER spectra at 964 1015 1054 1286 1410 1447 2865 and 2935 cm-1 Most of these bands are associated with alkane modes specifically the bands at approximately 1040 cm-1 to a C-C stretch 1290 cm-1 to a CH2 in-phase twist 1440 cm-1 to a CH2 wag 2865 cm-1 to a symmetric CH2 stretch and 2965 cm-1 to an asymmetric CH2 stretch The Raman and SER spectra of sulfur mustard were measured at the Edgewood center (Figure 4) Both spectra are largely similar to CEES The C-Cl and C-S bands in the Raman spectrum of HD now occur at 640 655 700 739 and 760 cm-1 and are more resolved possibly due to the increased molecular symmetry Theoretical calculations indicate that the first three bands are due to C-Cl stretching modes and the latter two to C-S stretching modes28 Only the C-Cl bands maintain significant intensity in the SER spectra occurring at 624 and 643 cm-1 which is attributed to the
Prior to measurements of HD CEES was examined by Raman and SER spectroscopy (Figure 3) CEES also known as half-mustard is essentially identical to HD except one of the chlorine end atoms is replaced by a hydrogen atom Again although not as toxic as HD CEES is a blister agent and dilute aqueous samples were prepared using appropriate safety equipment Both the Raman and SER spectra of CEES are similar and dominated by bands between 600 and 800 cm-1 These are associated with C-Cl and C-S stretching modes which are tentatively assigned to 648 and 747 cm-1 in the Raman spectra respectively The shoulder at 630 cm-1 the overlapped band at 660 cm-1 and the strong band at 692 cm-1 could also be due to these modes or their asymmetric counterparts It is worth noting that theoretical calculations assign the 692
Figure 2 A) SER and B) NR spectra of KCN Conditions A) 1 mgml in tap water 100 mW of 785 nm at sample 1-min acquisition time B) solid 300 mW of 785 nm 5-min All spectra are 8 cm-1 resolution
A
B
Figure 3 A) SER and B) NR spectra of CEES Conditions A) 1 vv (10 mgml) in MeOH 100 mW of 785 nm 1-min acquisition time B) neat 300 mW of 785 nm 5-min
A
B
Cl-CH2-CH2-S-CH2-CH3
A
B
Figure 4 A) SER and B) NR of HD Conditions A) 1mgml in tap water B) pure both 100 mW of 785 nm 1-min
Cl-CH2-CH2-S-CH2-CH2-Cl
SPIE-2003-5269 20
expected strong interaction between chlorine and silver and adds support to the assignment of this band to a C-Cl stretch Weaker overlapping bands occur at 670 692 and 724 cm-1 the latter possibly due to C-S stretching modes Again the alkane modes are apparent in the normal Raman spectra of HD but only a broad feature at 1300 to 1450 cm-1 suggests CH2 contributions in the SER spectrum Although the observed bands in the VX spectrum have not been assigned (Figure 4) a computer generated Raman spectrum29 predicts many of the same features with surprising accuracy and are used here Two intense bands at 460 and 530 cm-1 closely match predicted bands at 463 and 546 cm-1 assigned to a CH3-P=O bend and a PO2CS wag Three highly overlapped bands occur at 694 745 and 771 cm-1 matching predicted bands at 713 730 and 760 cm-1 The first
Table 2 Measured SER peak heights for the CN stretch at 2100 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
Figure 5 A) SERS and B) NR spectra of VX Conditions A) 1 vv (10 mgml) in MeOH B) pure sample both 100 mW of 785 nm 1-min acquisition time
A
B
two have been assigned to a C-S stretch and CH2bend respectively while the latter has been attributed to either a P-C stretch or an O-C-C stretch Although the 745 cm-1 band may alternatively be assigned to a C-S stretch based on the previous measurements of CEES and HD The relatively intense bands at 890 1106 1218 1445 and 1465 cm-1 also match predicted bands at 880 1108 1216 1440 and 1464 cm-1 that are assigned to a C-C stretch CH3 rock N-C3 stretch various C-H3 bends and C-H bends respectively Both the computer generated and the measured spectra contain numerous other less intense bands One is worth mentioning A unique band appears at 370 cm-1 that is predicted at 368 cm-1 and corresponds to an O-P=O bend The surface-enhanced Raman spectrum of VX is also rich with spectral features It has the unique low frequency band at 370 cm-1 as well as a second band at 380 cm-1 that is assigned tothe S-P-O bend predicted in the normal Raman spectrum at 388 cm-1 Based on the measured and predicted normal Raman spectra the following SERS assignments are given 460 cm-1 to the CH3-P=O bend 544 cm-1 to the PO2CS wag 738 cm-1 to a C-S stretch (based on arguments above) 890 cm-1
to a C-C stretch 1101 cm-1 to a CH3 rock and 1456
cm-1 to a C-H bend The ability of SERS to measure chemical agents in water containing real-world chemical interferents was tested by using the generic tap water described in Table 1 The ability to reproduce measurements was accomplished by preparing three separate water stock solutions which were used to prepare three sample solutions each of 1 mgmL KCN in the generic tap water The samples defined as vials 1-9 were then measured at 4 points per vial in a semi-random fashion such that errors associated with stock solution preparation and errors associated with instrument drift could be identified No trends were apparent that signified such systematic errors Each spectrum collected consisted of 20 averaged scans taking 16 seconds at 8 cm-1 resolution The laser power at the sample was measured periodically during the day and it ranged from 102 to 105 mW spacccc
spot Vial 1 Vial 2 Vial 41 13754 13004 128192 13519 12692 12909
1 3 13541 12721 126394 13462 12648 12651
ave 13569 12766 12755Vial 6 Vial 3 Vial 5
1 11586 14068 10757 2 11236 14402 115122 3 11376 14546 11559
4 10894 11763 11214ave 11273 13695 11261
Vial 7 Vial8 Vial 91 11198 15804 11023
2 11243 15716 88753 3 11562 15216 11218
4 11655 15084 11077ave 11415 15455 10548
CN AVG STDEV ERR1 mgmL pk ht 12526 1560 1245
stock solution
SPIE-2003-5269 21
The CN measurements consisting of 36 data points produced an average height of 125 for the 2100 cm-1 peak with a standard deviation of 156 or 125 (Table 2) The HD and VX measurements were performed precisely the same way (Tables 3 and 4) For HD the 624 cm-1 peak was used for analysis and it had an average height of 53 with a standard deviation of 068 or 129 while for VX the 544 cm-1 peak was used for analysis and it had an average height of 1051 with a standard deviation of 308 or 293 The greater error in the VX measurements can be somewhat attributed to Vial 4 which produced lower SER signal intensities But removing this vial from the data set changes the standard deviation to 233 only a modest improvement
spot Vial 1 Vial 2 Vial 41 525 609 4682 45 675 484
1 3 527 644 5094 559 807 669
ave 51525 68375 5325Vial 6 Vial 3 Vial 5
1 521 574 575 2 536 53 4492 3 509 508 451
4 565 594 379ave 53275 5515 4635
Vial 7 Vial 8 Vial 91 457 589 413
2 583 497 4273 3 544 505 4
4 528 5 509ave 528 52275 43725
HD AVG STDEV ERR1 mgmL pk ht 530 068 1291
stock solution
spot Vial 1 Vial 2 Vial 41 1464 1034 4672 1485 99 61
1 3 1491 989 5684 1041 777 553
ave 137025 9475 5495Vial 6 Vial 3 Vial 5
1 1058 942 1293 2 697 121 965
2 3 727 1206 11124 689 1254 746
ave 79275 1153 1029Vial 7 Vial 8 Vial 9
1 1127 783 163 2 1358 812 1614
3 3 1371 908 15124 1097 875 1397
ave 123825 8445 153825
VX AVG STDEV ERR1 mgmL pk ht 1051 308 2925
stock solution
4 CONCLUSIONS In this paper we examined the ability of surface-enhanced Raman spectroscopy to reproducibly measure CN HD and VX in tap water without chemical interference Both normal and surface-enhanced Raman spectra were examined to select unique bands suitable to identify and quantify these chemical agents For SER measurements the 2100 cm-1 C-N stretch was used for CN the 624 cm-1 C-Cl stretch was used for HD and the 544 cm-1 PO2CS wag was used for VX It was determined that 1 mgmL samples of each of these chemicals measured 36 times in glass vials coated with a silver-doped sol-gel reproduced measurements with standard deviations of 125 129 and 293 It was further found that the 10 chemicals added to simulate generic tap water did not interfere with or alter the SER spectra It should be noted that the concentrations used in this study were considerably greater than those required by the JSAWM program Current work involves improving SER sensitivity and designing sampling systems with better reproducibility This includes the development of fractal silver and gold structures within the sol-gel matrix and the development of chemically selective sol-gels
ACKNOWLEDGMENTS The authors are grateful for the support of the US Army (DAAD13-02-C-0015 Joint Service Agent Water Monitor program) The authors would also like to thank Janet Jensen Ronald Crosier and Kristina Gonser for helpful discussions
Table 4 Measured SER peak heights for the VX band at 544 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
Table 3 Measured SER peak heights for the HD band at 624 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
SPIE-2003-5269 22
REFERENCES 1 Jensen JL NC Wong W R Loerop SPIE 4775-03 (2001) 2 JSAWM Requirements at wwwsbccomapgeaarmymilRDAecbcrtPRODSERJSAWMjsawmhtml 3 Erickson B ldquoThe Chemical Weapons Convention Redefines Analytical Challengerdquo Anal Chem 70 397A (1998) 4 Johnston RL Hoefler CM Fargo JC and Moberley B ldquoThe Defense Nuclear Agencyrsquos
ChemicalBiochemical Weapons Agreements Verification Technology Research Development Test and Evaluation Program and its Requirements for On-Site Analysisrdquo AT-ONSITE 5-8 (1994)
5 Black RM Clarke RJ Read RW and Reid MT ldquoApplication of GC-MS and GC-tandem MS to the analysis of chemical warfare samples found to contain residues of the nerve agent sarin sulphur mustard and their degradation productsrdquo J Chrom 662 301-321 (1994)
6 USA Army Center for Health Promotion and Preventive Medicine (USACHPPM) Technical Guideline 230A 7 Hoffland LD Piffath RJ Bouck JB rdquoSpectral signatures of chemical agents and simulantsrdquo Optical
Engineering 24 982-984 (1985) 8 Braue EHJ Pannella MGrdquoCIRCLE CELL FT-IR Analysis of Chemical Warfare Agents in Aqueous Solutionsrdquo
App Spectrosc 44 1513-1520 (1990) 9 Tseng C-H Mann CK and Vickers TJ ldquoDetermination of Organics on Metal Surfaces by Raman
Spectroscopyrdquo App Spectrosc 47 1767-1771 (1993) 10 Staff Reporter ldquoUS Army Tests Lidar to Detect Biological Toxinsrdquo Photonic Spectra pg 50 December 1998 11 Christesen SD Raman cross sections of chemical agents and simulants App Spectrosc 42 318-321 (1988) 12 Weaver MJ Farquharson S Tadayyoni MA ldquoSurface-enhancement factors for Raman scattering at silver electrodesrdquo J Chem Phys 82 4867-4874 (1985) 13 Alak AM and Vo-Dinh T ldquoSERS of Organophosphorous Chemical Agentsrdquo Anal Chem 59 2149-2153 (1987) 14 Lee Y Farquharson S ldquoRapid chemical agent identification by SERSrdquo SPIE 4378 21-26 (2001) 15 Farquharson S Maksymiuk P Ong K Christesen S ldquoChemical agent identification by surface-enhanced Raman spectroscopyrdquo SPIE 4577 166-173 (2001) 16 Spencer KM Sylvia J Clauson S and Janni J ldquoSurface Enhanced Raman as a Water Monitor for Warfare
Agents in Waterrdquo SPIE 4577 158-165 (2001) 17 Tessier P Christesen S Ong K Clemente E Lenhoff A Kaler E Velev O ldquoOn-line spectroscopic characterization of sodium cyanide with nanostructured gold surface-enhanced Raman spectroscopy substratesrdquo
App Spectrosc 56 1524-1530 (2002) 18 Farquharson S W W Smith S Elliott and J F Sperry Rapid biological agent identification by surface- enhanced Raman spectroscopy SPIE 3855110-116 (1999) 19 Farquharson S WW Smith YH Lee S Elliott and J F Sperry Detection of biological signatures A
comparison of electrolytic and metal-doped sol-gel surface-enhanced Raman media SPIE 4575 62-72 (2002) 20 Guzelian AA Sylvia J Janni J Clauson S and Spencer KM ldquoSERS of whole cell bacteria and trace levels of biological moleculesrdquo SPIE 4577 182-192 (2001) 21 Shende C Inscore F Gift A Maksymiuk P Farquharson S ldquoRapid dipicolinic acid extraction from Bacillus spores detected by surface-enhanced Raman spectroscopyrdquo App Spectrosc 58 accepted 22 Lee Y and Farquharson S ldquoSERS Sample Vials Based on Sol-Gel Process for Trace Pesticide Analysisrdquo SPIE
4206 140-146 (2000) 23 Farquharson S and Lee Y ldquoTrace Drug Analysis by SERSrdquo SPIE 4200-16 (2000) 24 Lee Y Farquharson S and Rainey P M Surface-Enhanced Raman Sensor for Trace Chemical Detection in
Water SPIE 3857 76-84 (1999) 25 Lee Y Farquharson S Kwong H and Shahriari M ldquoSol-Gel Chemical Sensor for Surface-Enhanced Raman
Spectroscopyrdquo SPIE 3537 252-260 (1998) 26 Farquharson S A Gift P Maksymiuk F Inscore and W Smith ldquopH dependence of methyl phosphonic acid
dipicolinic acid and cyanide by surface-enhanced Raman spectroscopyrdquo SPIE 5269-19 (2003) 27 Lide DR Ed Handbook of Chemistry and Physics CRC Press 77th Ed (1996-7) 28 Sosa C RJ Bartlett K KuBulat and WB Person ldquoA theoretical study of harmonic vibrational frequencies and
infrared intensities of XCH2CH2SCH2CH2X and XCH2CH2SH (= H Cl)rdquo J Phys Chem 93 577-588 (1993) 29 Hameka HF and JO Jenson ldquoComputer-generated predictions of the structure and the IR and Raman spectra of
VXrdquo ERDEC-TR-065 May 1993
- Chemagents Appendicespdf
-
- SPIE2001-4575-Bioagent-SERSapdf
-
- Detection of bioagent signatures A comparison of electrolytic and metal-doped sol-gel surface-enhanced Raman media
- Stuart Farquharson Wayne Smith and Yuan Lee
-
- Real-Time Analyzers 87 Church Street East Hartford CT 06108
-
- ABSTRACT
- 4 CONCLUSIONS
- 5 ACKNOWLEDGEMENTS
- 6 REFERENCES
-
SPIE 2001-4575
70
A final comparison was made between the two SER methods by measuring dipicolinic acid (DPA) This chemical may be invaluable as a test for spore forming bacteria specifically B anthracis 50 to 90 of B anthracis sporilates During spore formation dipicolinic acid is synthesized and once completed 10-15 of the dry spore weight is composed of the Ca2+ complex located in the spore core42 Heating in water can be used to initiate germination at which point the exosporium breaks and releases the Ca dipicolinate which becomes dipicolinic acid in water The structure of this chemical strongly
4 CONCLUSIONS Towards the goal of developing a method and technology to rapidly detect and identify bioagents we have been investigating surface-enhanced Raman spectroscopy as a tool to measure relative concentrations of nucleic acid bases in RNA and determine bioagent taxonomy Initially we investigated E-SERS since this method has been extensively researched and the optimum sample conditions are well developed However this method requires a three-electrode sample cell and electrolyte solution Incorporation of an electrolytic cell into sample systems used to analyze for BWA as aerosols in water or on surfaces can be designed using flow injection analysis technologies but cross-contamination and plugging of sample lines seems inevitable For this reason we also investigated metal-doped sol-gels as a SER-active medium Previous studies have shown this material to be active in all solvents particularly water capable of continuous measurements in flowing systems and reproducible (quantitative) between coated sample vials Here we compared SG-SER spectra to traditional E-SER spectra of the nucleic acid base pairs adenosine monophosphate and RNA High quality spectra of adenine cytosine guanine thymine and uracil were obtained by both E-SERS and SG-SERS Both methods yielded very similar spectra for the NA bases including a pH dependent study of adenine Enhancement factors and detection limits for adenine were determined as 2x105 and 16x10-11g and 1x105 and 12x10-10g for E-SERS and SG-SERS respectively (normalized to 75 mW and 10-min acquisition time) Fifty percent of the silver particle surface area in the sol-gel matrix was assumed covered by adenine which may have been overestimated yielding a lower EF and higher detection limit It should also be realized that each E-SER spectrum required several attempts to optimize the measurement conditions (pH electrode potential etc) While each SG-SER spectrum involved no sample preparation and often represents the first and only attempt to make the measurement Quality spectra of RNA extracted from Escherichia coli Bacillus subtilis and Staphylococcus aureus were obtained by E-SERS that were easily interpreted Bands due to adenine guanine cytosine plus uracil and phosphate were identified The SER band intensity of the NA bases in the RNA samples were of the same order of magnitude suggesting that their interaction with the silver surface is concerted as is their Raman enhancement Interestingly the relative SER band
Wavenumbers (∆cm-1) Figure 10 A) Raman spectrum of solid dipicolinic acid B) Ax20 C) electrolytic SERS of 6x10-3 M dipicolinic acid in 01 M KCl at a potential of +07VSCE and pH of 4 and D) sol-gel SERS of 6x10-3 M dipicolinic acid Conditions for A and C as in Figure 1 C) 100 mW of 1064 nm 50 scans 8 cm-1
A B (Ax20)
C
D
suggested that it would be SER active However the E-SER spectrum was unstable and varied considerably as a function of potential A consistent spectrum was obtained at +06VSCE (Figure 10) This potential is not recommended for measurement because the surface is actively dissolving in solution The SG-SER spectrum was considerably more stable of higher quality and easily reproduced Bands at 660 825 1010 1390 1430 1570 1590 and 3075 cm-1 were observed Enhancement factors were determined for the two media using the symmetric ring stretching mode at 995 and 1010 cm-1 for the normal Raman and SER spectra respectively E-SERS yielded an EF of 5x103 while SG-SERS yielded an EF of 2x105 for DPA The SN of the latter suggests a detection limit of 20x10-10g (based on adenine coverage 75 mW and 10-min) The differences in SER activity for these two media may be attributed to the combined electrolytic potential of the solution chemical and metal15 Again the E-SERS suggests that the SG-SERS is at a potential positive of the pzc While the instability in the E-SERS may also be associated with surface interactions of two carboxylic acid groups of dipicolinic acid during the ORCs
DPA
SPIE 2001-4575
71
intensities for RNA extracted from E coli are very similar to those measured by normal Raman spectroscopy Although the relative percent that each of the NA bases contributed to each RNA sample was not determined reproducible band intensities allowed noting the following trends The percent adenine decreases while the combined percent cytosine and guanine increase for both B subtilis and S aureus compared to E coli Quality spectra were also obtained for the RNA samples by SG-SERS but only a few bands were readily identified Calculations of NA base concentrations by SG-SERS will require further research In light of recent events we summarize the capabilities of these SERS media in terms of rapid detection of B anthracis and dipicolinic acid However these capabilities must be qualified First and foremost the level to which SERS can distinguish bacteria or viruses has not yet been determined Development of a database of both DNA and RNA base concentrations for BWAs and common bacteria to establish the level of taxonomic identification is ongoing Second rapid collection of aerosol water or surface samples is being addressed by others who report trapping particles on filters from 100 liters of air per minute Third although not presented here we have developed methods to extract RNA or DNA from cells and spores for SER analysis within 10 minutes Finally we assume a detection limit of 3600 spores per 100 liters of air is required although a 50 lethal dosage of anthrax has not been established With these qualifications a mass detection limit for RNA using SERS is estimated as follows A single measurement is performed in ca 20 minutes (140 liters collected in 14 min RNA extracted in 8 min spectral acquisition and analysis in 10 min) The average human breaths 7 liters per minute therefore the analyzer must at the very minimum detect 5000 spores in 140 liters of air One spore is approximately 2x10-18m3 (1x1x2 microm) and if a density of 075 gcm3 is assumed this corresponds to a mass of 15x10-12g Each spore contains 4-12 RNA or 12x10-13g RNA for 8 If we assume 23 of the RNA can be isolated for analysis during lysis then the proposed instrument must be able to detect 4x10-10g RNA from 5000 spores per 70 liters of air within 10 minutes As noted above the mass detection limits for adenine were estimated at 16x10-11g and 12x10-10g for E-SERS and SG-SERS respectively Although these detection limits suggests that RNA from 5000 spores is detectable with the current instrumentation it is highly likely that only a portion of an RNA segment (eg 120-nucleotide 5S rRNA) is in contact with the metal surface and will contribute to the SER effect The SN for the RNA spectra were 110th of the average SN for the four individual RNA bases suggesting a 10 contribution Furthermore effective taxonomy will likely require knowing the NA base concentrations to 1 of the value (eg 25plusmn025) Again the average measurement error for the bases is 12 These values suggest that the E-SERS is within a factor of 4 of the required detection limit whereas the SG-SERS detection limit must be improved by 25 times The same arguments can be applied to the detection of dipicolinic acid If we assume a spore releases 10 by weight DPA during germination then the proposed instrument must be able to detect 75x10-10g DPA from 5000 spores per 70 liters of air within 10 minutes The detection limit for SG-SERS was estimated at 20x10-10g and suggest that the vials are suitable to perform a rapid screen for anthrax A series of concentration dependent measurements are currently being performed to verify this assertion Finally we note that the measurements performed here employed an FT-Raman spectrometer This instrumentation was chosen over dispersive Raman instrumentation for the high wavelength accuracy afforded by the HeNe reference laser (Connes Advantage43) which would allow reliable spectral subtraction matching of observed spectra to stored library spectra and confident use of chemometric approaches Such data analysis is likely to be required to enhance BWA identification However this instrumentation sacrifices sensitivity due to a less efficient detector (InGaAs vs Si) less efficient Raman scattering and less efficient generation of plasmon modes Substantial improvements can be obtained using visible excitation and Si detection and these measurements are underway
5 ACKNOWLEDGEMENTS The authors are grateful to Drs D Cookmeyer and S Tove of the US Army Research Office (Contract Number DAAH04-96-C-0078) for their interest and support of this research The authors would also like top acknowledge Dr R Yin and J Jensen for supporting development of the metal-doped sol-gels (Contract Number DAAD13-01-C-0019) They also thank Dr Wilfred H Nelson for assistance in spectral interpretations
SPIE 2001-4575
72
6 REFERENCES 1 Roberts WL and Rainey PM Clin Chem 39 1872-1877 (1993) 2 Pasechnik VA CC Shone and P Hambleton Bioseperations 3 267-283 (1993) 3 Jackson PJ ME Hugh-Jones DM Adair G Green KK Hill CR Kuske LM Grinberg FA Abramova and P
Keim Proc Natl Acad Sci 95 1224-1229 (1998) 4 Jensen JL NC Wong W R Loerop SPIE 4775-03 (2001) 5 Snyder A Pet al SPIE 3853-15 (1999) 6 Danosky T R and McFadden P N in press (1997) 7 Woodruff WH Farquharson S Science 201 831-833 (1978) 8 Lord RC and Thomas GJJr Spectrochemica Acta 23A 2551-2591 (1967) 9 Thomas GJJr Biochim Biophys Acta 213 417-423 (1970) 10 Placzek G Handbuch der Radiologie 2 EMarx ed Akademische Verlagagescellschatt Liepzig 1934 UCRL
Trans No 526 (1959) 11 Jeanmaire DL and RP Van Duyne J Electroanalytical Chem 84 1-20 (1977) 12 Chada S Manoharan R Moenne-Loccoz P Nelson WH Peticolas WL and Sperry JF Applied Spectroscopy
47 38-43 (1993) 13 Manoharan R Ghiamati E Chada S Nelson WH and Sperry JF Applied Spectroscopy 47 2145-2150 (1993) 14 Todd EA Morris MD Applied Spectroscopy 48 545-548 (1994) 15 Wentrup-Byrne E Sarinas S and Fredericks PM Applied Spectroscopy 47 1192-1197 (1993) 16 Weldon MK VR Zhelyaskov and MD Morris Applied Spectroscopy 52 265-269 (1998) 17 Kneipp K and J Fleming J Mol Structure 145 173-179 (1986) 18 Pothier NJ and Force RK Applied Spectroscopy 46 147-151 (1992) 19 Ervin KM E Koglin JM Sequaris P Valenta and HW Nurnberg J Electroanal Chem 114 179-194 (1980) 20 Kim SK TH Joo SW Suh and MS Kim J Raman Spectrosc 17 381-386 (1986) 21 Pothier NJ and Force RK Analytical Chemistry 62 678-680 (1990) 22 Kneipp K Y Wang RR Dasari and MS Feld Applied Spectroscopy 49 780-784 (1995) 23 Nie S and Emory SR Science 275 1102 (1997) 24 Graham D WE Smith AMT Linacre CH Munro ND Watson and PC White Analytical Chemistry 69 4703-
4707 (1997) 25 Fleischmann M PJ Hendra and AJ McQuillan Chem Phys Lett 26 163-166 (1974) 26 Pettinger B J Chemical Phys 85 7442-7451 (1986) 27 Surface-Enhanced Raman Scattering Section Four Theory SPIE MS 10 M Kerker and B Thompson Eds (1990) 28 Wang D-S and Kerker M SPIE (M Kerker and B Thompson Eds) MS 10 417-429 (1990) 29 Weaver MJ Farquharson S Tadayyoni MA J Chem Phys 82 4867-4874 (1985) 30 Norrod KL Sudnik LM Rousell D and Rowlen KL Applied Spectroscopy 51 994-1001 (1997) 31 Farquharson S Weaver WJ Lay PA Magnuson RH and Taube H J Am Chem Soc 105 3350-3351 (1983) 32 Lee PC and Meisel D ldquoAdsorption and Surface-Enhanced Raman of Dyes on Silver and gold Solsrdquo J Phys Chem
86 3391-3395 (1982) 33 Li Y-S and Wang Y Applied Spectroscopy 46 142-146 (1992) 34 Bello JM DL Stokes and T Vo-Dinh Analytical Chemistry 61 1779-1783 (1989) 35 van Duyne RP J C Hulteen D A Treichel M T Smith M L Duval and T R Jensen J Phys Chem B1033854-
3863 (1999) 36 Farquharson S and W W Smith W H Nelson and J F Sperry SPIE 3533-27 207-214 (1998) 37 Lee Y H W Smith S Farquharson H C Kwon M R Shahriari and P M Rainey SPIE 3537 252-260 (1998) 38 Lee Y-H S Farquharson and P M Rainey SPIE 3857 76-84 (1999) 39 Current Protocols in Molecular Biology Wiley Interscience1003-1006 (1987) 40 Farquharson S Smith W Carangelo R C and Brouillette C SPIE 3859 14-23 (1999) 41 Chase D B and JF Rabolt Fourier Transform Raman Spectroscopy Acad Press Ch1 p 131 (1994) 42 Brock TD MT Madigan JM Martinko and J Parker Biology of Microorganisms 7th Ed Prentice Hall p 76-80
(1994) 43 Connes J Rev Opt Theor Instrum 40 45 (1961)
166
Chemical agent identification by surface-enhanced Raman spectroscopy
Stuart Farquharson and Paul Maksymiuk
Real-Time Analyzers 87 Church Street East Hartford CT 06108
Kate Ong and Steven D Christesen US Army SBCCOM Aberdeen Proving Ground MD 21010
ABSTRACT The recent distribution of anthrax through the US postal system and the subsequent infection and death of several postal and national media employees amplifies the need for methods to rapidly detect identify and quantify this and other chemical and biological warfare agents The US military has also identified water supplies as a likely method of warfare agent deployment and is funding the development of a Joint Service Agent Water Monitor (JSAWM) In an effort to aid military personnel and the public at large we are developing a portable analyzer capable of identifying and quantifying chemical agents rapidly either on-demand or continuously The approach is based on the ability of Raman spectroscopy to identify molecular structure through the abundant vibration information provided in spectra and the ability of surface-enhanced Raman spectroscopy (SERS) to detect extremely low concentrations (eg part-per-billion) through the enhancement of Raman scattering by six orders of magnitude or more A key element to the analyzer design is a new SER active medium that is capable of quantitative reversible measurements The medium consists of silver or gold nanoparticles incorporated into a sol-gel matrix The porous silica network offers a unique environment for stabilizing SER active metals and the high surface area increases the interaction between the analyte and metal particles Here we present the use of new sol-gels that also selectively enhance chemicals based on polarity and charge Base-line measurements of chemical agents and their hydrolysis products are presented and compared to the JSAWM goal of 30 micrograms per liter detection Keywords Chemical warfare agent hydrolysis product SERS Raman spectroscopy sol-gel nanoparticle
1 INTRODUCTION Since September 11 2001 the threat of terrorist attacks and biological warfare within US borders has become a sobering reality The simplicity in manufacturing ease of deployment and the relatively low cost of chemical warfare agents (CWAs) raises public concern that they may also be used by terrorists Indeed terrorists released sarin (GB) in the Tokyo subway in 19951 Countering terrorism and terrorist attacks requires recognizing likely deployment scenarios and having the required technology to rapidly detect the deployment event One method of deployment has been long identified by the US military distribution through water supplies To counter this threat the Department of Defense is funding or monitoring the capabilities of 200 technologies with the goal of developing a Joint Service Agent Water Monitor (JSAWM) that is field portable2 The JSAWM must be capable of identifying the chemical agents with no false-positives and quantifying the agents at microgram per liter concentrations within 10 minutes (Table 1)3 This includes the analysis of drinking water supplies distribution and storage systems as well as potable water supplies The earliest technologies developed for CWA detection were based on electrochemical ionization or colorimetric analysis (eg phosgene tape) Although these analyzers were easy to use they were not generally agent specific and suffered from false-positives4 More traditional laboratory methods have also been investigated and in particular combined gas chromatography and mass spectrometry (GCMS) has been very successful at eliminating false-positives56 However GCMS requires extraction repeated calibration and long analysis times (typically 20 to 60 minutes)6 making it labor intensive and less than desirable for field use More rapid analysis of agents in the solid liquid and gas phase has been To whom correspondence should be addressed emailfarqureal-time-analyzerscom
Vibrational Spectroscopy-based Sensor Systems Steven D Christesen Arthur J Sedlacek III EditorsProceedings of SPIE Vol 4557 (2002) copy 2002 SPIE middot0277-786X02$1500
167
Table 1 Chemical Agent Structures Hydrolysis Half-lives and JSAWM Thresholds Agent Short-Hand Chemical Structure Hydrolysis
Half-Life JSAWM
Thresholds
Sarin (GB) F-[O=P-CH3]-O-CH(CH3)2 213 hours 32 microgL Soman (GD) F-[O=P-CH3]-O-CH(CH3)-(C-(CH3)3) 23 hours 32 microgL Tabun (GA) (CH3)2-N-[O=P-CN]-O-C2H5 41 hours 32 microgL VX C2H5O-[O=P-CH3]-S-(CH2)2-N-(CH(CH2)2)2 821 hours 32 microgL EA2192 HO-[O=P-CH3]-S-(CH2)2-N-(CH(CH2)2)2 gt9 years 32 microgL Mustard (H) ClCH2CH2-S-CH2CH2Cl encapsulates 47 microgL Lewisite (L) ClCH=CH-As-Cl2 rapid 27 microgL HCN HCN rapid 20 mgL BZ C7NH12-O-[C=O]-COH(C6H5)2 23 microgL T-2 Toxin 87 microgL at pH 7 to 75 and 20 to 25 oC
demonstrated by vibrational spectroscopy7-10 Hoffland et al7 reported infrared absorbance spectra and absolute Raman cross sections for several chemical agents while Christesen measured Raman cross sections for sarin tabun mustard gas and VX (ethyl S-2-diisopropylamino ethyl methylphosphonothioate)11 Again however these techniques also have limitations Raman spectroscopy is simply not a very sensitive technique and detection limits are typically 01 (1000 ppm) While infrared spectroscopy would have limited value in analyzing poisoned water since the very strong infrared absorption of water would obscure most other chemicals present Nevertheless efforts to overcome these limitations have been demonstrated Braue and Pannella8 quantified the G-series nerve agents (tabun sarin and soman) in terms of infrared attenuated total reflectance using a circle-cell And Alak and Vo-Dinh demonstrated the possibility of surface-enhanced Raman spectroscopy (SERS) to identify CWAs by measuring several organophosphonates that simulate the nerve agents12 However quantitative measurements have not been demonstrated for the SER-active material used (silver coated on alumina particles) or other SER-active media13 Recently we developed silver-doped sol-gels to promote the SER effect14-17 The porous silica network of the sol-gel matrix offers a unique environment for stabilizing SER-active metal particles and the sol-gel provides a high surface area that effectively increases the number of molecules observed within the Raman scattering volume The silver-doped sol-gels have been coated on the internal walls of standard glass vials to allow rapid SER analysis of small sample quantities (lt 01 mL) without preparation We have used p-aminobenzoic acid (PABA) as a test chemical to demonstrate surface enhancements greater than 106 reversible measurements in a flowing system reproducible measurements from vial-to-vial and batch-to-batch and measurements in multiple solvents including water14-17 Recently we used these vials to measure Tabun (GB) and Sarin and several hydrolysis products pinacolyl methyl phosphonate (PMP from Soman) and methyl phosphonic acid (MPA from all G-agents Figure 1) Although a number of unique vibrational bands are observed (eg C-N stretch doublet and P-C stretch) the G-agents were only observed for 5 concentrations and all spectra required baseline corrections Figure 1 Surface enhanced Raman spectra of ~5 vv A) Tabun and B) Sarin C) 1 vv PMP and D) 10 ppm MPA using sol-gel sample vials 785 nm excitation 1-min scan and CCD detection Performed at Aberdeen Proving Ground
C A
Wavenumber (∆cm-1) Wavenumber (∆cm-1)
D B
790
545 C-N
2135 2190 P-C 770
760 1290
Proc SPIE Vol 4577
168
Nevertheless MPA was readily observed for a 10 ppm sample with an estimated detection limit of 04 ppm (based on a signal-to-noise ratio of 3 for the 760 cm-1 band intensity) This measurement provides encouragement in that SERS may satisfy the needs of a JSAWM Furthermore MPA is also a hydrolysis product of VX and V-gas and EA2192 (Figure 2) and may prove a valuable indicator of agent usage Figure 2 Hydrolysis of Sarin to form hydrofluoric acid (HF) methylphosphonic acid 1-methylethyl ester (MPAMME) methyl phosphonic acid (MPA) and 2-propanol With this initial albeit modest success we began analyzing chemicals with various sol-gel compositions that we have been developing Here we describe four sol-gel compositions that select for 1) polar-positive 2) polar-negative 3) weakly polar-positive and 4) weakly polar-negative chemical species The ability of these sol-gels to select and enhance Raman scattering is described for several test chemicals and MPA
2 EXPERIMENTAL The chemicals analyzed as well as all chemicals used to prepare the metal-doped sol-gels were obtained at their purest commercially available grade from Aldrich (Milwaukee WI) The sol-gel designed to select for polar-negative species was prepared from a silver amine complex tetramethyl orthosilicate (TMOS) and methanol After mixing 02 mL of the sol-gel solution was transferred into a glass vial (2 mL) dried and heated The incorporated silver ions were then reduced using dilute sodium borohydride The vials were washed and dried prior to the addition of a sample solution In a similar manner the sol-gel designed to select for polar-positive species was prepared from a gold salt TMOS and methanol The sol-gel designed to select for weakly polar-negative species was prepared from a silver amine complex tetraethyl orthosilicate (TEOS) and methanol And the last sol-gel designed to select for weakly polar-positive species was prepared from a gold salt TEOS and methanol All samples were prepared in a chemical hood and transferred into plain or SER-active vials for analysis Normal Raman spectral measurements employed 1-mL pure samples that were placed in a 1-cm3 cuvette and weighed This yielded a powder density that allowed accurate calculation of molecules in the optical collection field SERS measurements employed 1-mg sample per mL water concentrations unless otherwise stated Once prepared a 01 mL sample was placed into one of the four selective sample vials which in turn was placed into the sample compartment of a Raman spectrometer for analysis A prototype Fourier transform Raman spectrometer (RTA) was used for these measurements18 The system consisted of a NdYAG laser (Brimrose) for excitation at 1064 nm an interferometer built by On-Line Technologies (OLT East Hartford CT) for frequency separation an uncooled InGaAs detector for signal detection (RTA) and an Intel 400 MHz Pentium II based laptop computer (Dell Round Rock TX) for interferometric control data acquisition (OLT) and analysis (LabVIEW by National Instruments Austin TX) Additional components included a Notch filter (Kaiser Ann Arbor MI) and interferometer entrance and exit optics (Edmund Scientific Barrington NJ) Fiber optics were used to deliver the excitation beam to the sample and the scattered radiation to the interferometer (1 meter lengths of 200 and 365 micron core diameter respectively Spectran Avon CT) A second Notch filter (Kaiser) was used as a beam splitter to direct the excitation beam along the same axis as the collected radiation A microscope object (20x04 Newport Irvine CA) was used to focus the beam into the sample and to collect the scattered radiation back along the same axis In this co-axial backscattering arrangement the excitation beam was passed through the outside of a glass vial and focused onto the silver-doped sol-gel film (01-03 mm thickness) containing the sample
3 RESULTS AND DISCUSSION p-aminobenzoic acid (PABA) and phenyl acetylene (PA) and were used to refine the selectivity and SER-activity of the four different metal-doped sol-gels PABA is a popular chemical used to evaluate the performance of SER-active media Here the polar end groups can be used to test selectivity of the polar-negative and polar-positive sol-gels PA is essentially non-
2O+ H HF + +OH
OH3H C
OP
OF
CH
CH3
3
3H C
OCP
OOH
CH
CH3
3
3H C
O
CPHO
CH
CH
3
3
C
Sarin MPAMME MPA 2-propanol
Proc SPIE Vol 4577
169
polar but a high electron density in the cylindrical π cloud around the carbon-carbon triple bond allows testing the selectivity of the weakly polar-negative and weakly polar-positive sol-gels As Figure 3 illustrates PABA passes through the polar sol-gel and is enhanced by either the silver or gold particles At 1 mgml the concentration of neutral PABA is ca 20 times that of the ionized form (pKa = 48) For electropositive silver the PABA anion is expected to interact through the carboxylate group and the associated vibrational modes are expected to dominate the spectrum Conversely for electronegative gold either form of PABA is expected to interact through the amine group The clear differences in our spectra support this expectation Furthermore bands at 840 and 1405 cm-1 assigned to a COO- bend and stretch respectively are significantly more intense for silver than gold Additional bands at 1140 and 1195 cm-1 are assigned to CH bending modes while bands at 1450 1500 and 1605 cm-1 are assigned to ring vibrational modes A very similar SER spectrum for PABA on a silver-coated alumina substrate has previously been reported with similar assignments19 For the gold-doped sol-gel new bands appear at 690 1355 and 1585 cm-1 The first band is assigned to a ring-H bending mode the second band to a ring-N- stretching mode and the third band to a possible NH2 scissors mode or ring mode The second band is not observed in the normal Raman spectrum but infrared bands occur at this frequency for aromatic ring-secondary amine stretching modes The scissors mode occurs at this frequency in Raman spectra for several chemicals but is absent in the PABA Raman spectrum Alternatively this mode may be the1600 cm-1 ring mode that has been shifted by the gold interaction Again a very similar SER spectrum of PABA has been reported but surprisingly using silver (colloids)2021 not gold as the enhancement medium These researchers also assumed the primary interaction of PABA with silver was through the carboxylate anion and made assignments accordingly For example they assigned the 1359 cm-1 to a COO- stretch not to the amine group as we have They also favor the ring stretching mode assignment for the 1582 cm-1 band Finally it should be said that other researchers have argued that the most dominant band in the SER spectra at 1450 cm-1 a ring vibration mode suggests that PABA lies flat on the surface and the π-orbitals dominate the surface interaction22
Figure 3 SER spectra of A) PABA using polar-negative and B) polar-positive sol-gels and C) PA using weakly polar-negative and D) weakly polar-positive sol-gels PABA is 1 mgmL PA is 1 vv Spectral conditions 75 mw 1064 nm 100 scans (15 min) 8 cm-1 resolution Non-polar PA passes through the non-polar sol-gels and is also enhanced by both metals The spectra are easily understood For electropositive silver PA interacts through the cylindrical triple bond π electron cloud and a -CequivC- doublet occurs near 2000 cm-1 The interaction is reasonably strong since this band appears at 2112 cm-1 in the normal Raman spectrum For electronegative gold this interaction is unlikely and only very weak bands occur near 2000 cm-1 The remaining bands are at 1000 cm-1 1200 cm-1 doublet and 1595 cm-1 all appear in the normal Raman spectra at virtually the same frequencies and are assigned to the symmetric ring-breathing mode CH bending modes and the trigonal ring-breathing mode respectively The polarnon-polar selectivity of the polar-negative and weakly polar-negative sol-gels was tested by adding a 11 molar mixture of PABA and PA The selective enhancement is quite good (Figure 4) The spectrum obtained using the polar sol-gel represents 78 PABA and 22 PA while the spectrum obtained using the weakly polar sol-gel represents 9 PABA and 91 PA The band peak intensities at 2000 cm-1 for PA and 1450 cm-1 for PABA were used for these calculations and are expanded in Figure 4 for clarity
C A
Wavenumber (∆cm-1) Wavenumber (∆cm-1)
2NH COOH
D B
C CH
Proc SPIE Vol 4577
170
Figure 4 SERS of 11 MM of PABA and PA in A) polar-negative and C) weakly polar-negative sol-gels The lower traces compare the pure chemicals B) 1 mgml PABA in polar-negative sol-gel and D) 1 PA in weakly polar-negative sol-gel while the insets magnify the minority species for clarity (x5 in A and x10 in B) Spectral conditions as in Figure 3 Following this development of selective sol-gels that maintained SER activity we measured cyanide and MPA (Figure 5) Not surprisingly the best sensitivity for both hydrolysis products was obtained using the polar-negative sol-gel The interaction of the cyanide anion with the silver surface is sufficient to shift the CequivN stretch observed at 2080 cm-1 in the normal Raman spectrum to 2145 cm-1 in the surface-enhanced Raman spectrum Furthermore the band is substantially broadened This anion has been extensively studied by electrolytic SERS and this shift and broadening have been attributed to the formation of a tetrahedral Ag(CN)3
2- surface structure23 Figure 5 Surface-enhanced (upper traces) and normal Raman spectra (bottom traces) of A) CN- and B) MPA in silver-doped TMOS SERS conditions as in Figure 3 and 1 mgmL Note MPA yields two distinct spectra for neutral (top) and acidic pH (middle) The normal Raman spectra employed pure powders 500 scans and 900 mW of 1064 nm SER measurements of MPA with the polar-negative sol-gel yielded two unique spectral signatures that depended on solution pH (Figure 5) For more neutral solutions the P-C stretch of MPA at 762 cm-1 dominates and the CH2 stretch at 2922 cm-1 appears The SN is sufficiently high that the anti-Stokes Raman shift at -762 cm-1 is observed For deprotonated MPA an oxygen-surface mode appears at 325 cm-1 (as well as its anti-Stokes complement) suggesting a strong interaction This results in substantial enhancement of the P-O-C mode at 1051 cm-1 (upper trace) Others report that this mode dominates the infrared spectra of nerve agents measured in water8 Comparison of the two spectra suggests the following molecule-to-surface orientations The appearance of the oxygen-surface and P-O-C modes in the upper spectrum of Figure 5B indicates that the tetrahedral molecule interacts with the silver surface through the deprotonated oxygen and is oriented end-on The
A B
Wavenumber (∆cm-1)
C A
Wavenumber (∆cm-1) Wavenumber (∆cm-1)
D B
Wavenumber (∆cm-1)
Proc SPIE Vol 4577
171
dominance of the P-C and the CH2 stretches and the disappearance of the P-O-C mode in the upper spectrum suggest the molecule is oriented with the methyl-hydroxyl-hydroxyl face towards the surface However considerably more research must be performed to verify these points Both chemical forms were measured as a function of concentration to determine the onset of monolayer coverage and estimate expected detection limits (Figure 6) Below monolayer coverage the signal to concentration dependence should be linear and the SN of any spectral measurement in this range can be used to predict the detection limit In the spectra presented here the peak height was used as the signal while the noise as root-mean-squared (RMS) was measured between 4400-4600 cm-1 Since noise is distributed evenly throughout the spectrum when transformed this region was used since it does not have any contributions from signals or baseline offsets Figure 6 shows a series of spectra for MPA along with a plot of the 1050 cm-1 band intensity (the noise was identical for this series) as a function of sample concentration A clear discontinuity appears in the vicinity of 01 mgmL (19 ppm) indicating the onset of monolayer coverage A detection limit defined as a SN of 3 was calculated for the 01 and 005gmL samples at 24x10-4 and 25x10-4 gL respectively A more modest detection limit of 101x10-4 gL was obtained using the 760 cm-1 band in the second series of concentration measurements These detection limits correspond to 46 and 190 ppb for a 3-min scan and 75 mW of incident laser power
Figure 6 A) Concentration dependence of MPA SERS measured in silver-doped TMOS) B) Concentrations are 001 005 01 05 1 gL (188 94 188 94 188 ppm) I760 series (bull) and I1050 series (∆)
Enhancement factors for cyanide and methyl phosphonic acid can be determined by estimating the number of molecules contributing to the surface-enhanced and normal Raman spectra The enhancement factor EF can be defined by the following equation
EF = (ISERSINR)bull(MNRMSERS) bull(PNRPSERS) bull(TNRTSERS)12
where I is the spectral band intensity M is the sample mass P is the incident laser power and T is the measurement time (or number of scans) for the two measurements For the normal Raman spectra a cylindrical scattering volume is assumed based on the laser area (28x10-7m2 6x10-4m diameter spot) and the penetration depth (1x10-3 m)24 The density of KCN and MPA as powders were measured at 0572 and 0516 gcm3 indicating that 16x10-4 and 144x10-4 g produced the normal Raman signals in Figure 5 respectively The number of molecules contributing to the SER spectra are those on the silver particles embedded in the sol-gel The total silver surface area can be determined from the average particle size concentration and the scattering volume Previous scanning electron micrographs showed the average silver particle size to be 40 nm in diameter (335x10-23m3)17 The silver concentration is 012M based on the reactant molar concentrations and dilution factors And the scattering volume is 76x10-
11m3 again based on a cylindrical scattering volume defined by a laser area of 28x10-7m2 and a sol-gel thickness of 27x10-
4m This volume contains 123x10-6g of silver equivalent to 35x109 silver particles with a collective surface area of 18x10-
5m2 However it may reasonably be assumed that at least half of the silver surface is in contact with the sol-gel matrix and unavailable for analyte interaction If we assume monolayer coverage and that each CN molecule occupies 15x10-20m2 then approximately 62x1014 molecules or 27x10-8g of CN contribute to the SER spectrum (20x10-19m2 46x1013 molecules 74x10-9g for MPA) Accordingly the EF for cyanide equals 48x104 ((180599) bull(16x10-427x10-8) bull(90075) bull(500100)12) The EF for MPA is considerably higher at 87x106 ((60326) bull(144x10-474x10-9) bull(90075) bull(500200)12)
0
20
40
60
80
100
120
140
0 02 04 06 08 1 12[MPA] (mgmL)
I (76
0)
0
100
200
300
400
500
600
I (10
50)
Wavenumber (∆cm-1)
A B
Proc SPIE Vol 4577
172
4 CONCLUSIONS Here we present for the first time surface-enhanced Raman spectra of Tabun and Sarin obtained using silver-doped sol-gels However the inferior enhancement suggested employing SER-active sol-gels varying in composition to improve both detection limits and selectivity of the target analytes To this end we successfully demonstrated the capabilities of four sol-gels that select for 1) polar-positive 2) polar-negative 3) weakly polar-positive and 4) weakly polar-negative chemical species p-aminobenzoic acid was used to show that silver could be used to attract polar-negative chemicals or functional groups (carboxylate anion) and that gold could be used to attract polar-positive chemicals (amine) while a mixture of p-aminobenzoic acid and phenylacetylene was used to show that tetramethyl orthosilicate preferentially solvates polar chemicals and that tetraethyl orthosilicate preferentially solvates weakly polar chemicals This increased sample control was applied to cyanide and methyl phosphonic acid two hydrolysis products of chemical warfare agents Exceptional results were obtained for methyl phosphonic acid allowing measurement of 1x10-2 gL for a 3-min scan and 75 mW of incident laser power with an estimated detection limit of 245x10-4 gL and an enhancement factor of 87x106 However this detection limit is 76 times less sensitive than required for the JSAWM (32x10-6gL for the G-agents) Finally we note that the measurements performed here employed an FT-Raman spectrometer This instrumentation was chosen over dispersive Raman instrumentation for the high wavelength accuracy afforded by the HeNe reference laser (Connes Advantage25) which would allow reliable spectral subtraction matching of observed spectra to stored library spectra and confident use of chemometric approaches Such data analysis is likely to be required to identify the chemical agents as well as distinguish hydrolysis products However this instrumentation which employs 1064 nm excitation and InGaAs detection sacrifices sensitivity We believe that the measurement sensitivity can be improved by at least two orders of magnitude by using shorter laser excitation wavelengths (eg 532 nm) This would provide more efficient Raman scattering (fourth power dependence on laser excitation wavelength) more efficient generation of plasmon modes and allow using more efficient detector material (Si vs InGaAs) These modifications are underway
5 ACKNOWLEDGEMENTS The authors would like to thank Dr R Yin and J Jensen of the US Army for supporting this work (Contract Number DAAD13-01-C-0019) They would also like to thank Advanced Fuel Research for making their laboratory facilities available
6 REFERENCES 1 Tu Anthony ldquoOverview of Sarin Terrorist Incidents in Japan in 1994 and 1995rdquo 6th CBW Protection Symposium
Stockholm Sweden 10-15 May 1998 2 Jensen JL NC Wong W R Loerop SPIE 4775-03 (2001) 3 JSAWM Requirements at wwwsbccomapgeaarmymilRDAecbcrtPRODSERJSAWMjsawmhtml 4 ldquoThe Chemical Weapons Convention Redefines Analytical Challengerdquo Analytical Chemistry News amp Features June 1
397A (1998) 5 Johnston RL Hoefler CM Fargo JC and Moberley B ldquoThe Defense Nuclear Agencyrsquos ChemicalBiochemical
Weapons Agreements Verification Technology Research Development Test and Evaluation Program and its Requirements for On-Site Analysisrdquo AT-ONSITE 5-8 (1994)
6 Black RM Clarke RJ Read RW and Reid MT ldquoApplication of gas chromatography-mass spectrometry and gas chromatography-tandem mass spectrometry to the analysis of chemical warfare samples found to contain residues of the nerve agent sarin sulphur mustard and their degradation productsrdquo J Chromatography 662 301-321 (1994)
7 Hoffland LD Piffath RJ Bouck JBrdquoSpectral signatures of chemical agents and simulantsrdquo Optical Engineering 24 982-984 (1985)
8 Braue EHJ Pannella MGrdquoCIRCLE CELL FT-IR Analysis of Chemical Warfare Agents in Aqueous Solutionsrdquo Applied Spectroscopy 44 1513-1520 (1990)
9 Tseng C-H Mann CK and Vickers TJ ldquoDetermination of Organics on Metal Surfaces by Raman Spectroscopyrdquo Applied Spectroscopy 47 1767-1771 (1993)
10 Staff Reporter ldquoUS Army Tests Lidar to Detect Biological Toxinsrdquo Photonic Spectra pg 50 December 1998
Proc SPIE Vol 4577
Proc SPIE Vol 4577
173
11 Christesen SD Raman cross sections of chemical agents and simulants Applied Spectroscopy 42 318-321 (1988) 12 Alak AM and Vo-Dinh T ldquoSERS of Organophosphorous Chemical Agentsrdquo Analytical Chemistry 59 2149-2153
(1987) 13 Norrod KL Sudnik LM Rousell D and Rowlen KL ldquoQuantitative Comparison of Five SERS Substrates
Sensitivity and Detection Limitrdquo Applied Spectroscopy 51 994-1001 (1997) 14 Lee Y and Farquharson S ldquoSERS Sample Vials Based on Sol-Gel Process for Trace Pesticide Analysisrdquo SPIE 4206
140-146 (2000) 15 Farquharson S and Lee Y ldquoTrace Drug Analysis by Surface-Enhanced Raman Spectroscopyrdquo SPIE 4200-16 (2000) 16 Lee Y Farquharson S and Rainey P M Surface-Enhanced Raman Sensor for Trace Chemical Detection in Water
SPIE 3857 76-84 (1999) 17 Lee Y Farquharson S Kwong H and Shahriari M ldquoSurface-Enhanced Raman Sensor for Surface-Enhanced Raman
Spectroscopyrdquo SPIE 3537 252-260 (1998) 18 Farquharson S Smith W Carangelo R C and Brouillette C ldquoIndustrial Raman Providing Easy Immediate Cost
Effective Chemical Analysis Anywhererdquo SPIE 3859 14-23 (1999) 19 Narayanan VA JM Bello JD Stokes and T Vo-Dinh Analusis 19 307-310 (1991) 20 Laserna JJ E L Torres and JD Winefordner Analytica Chemica Acta 469-480 (1987) 21 Torres EL and JD Winefordner Analytical Chemistry 59 1626-1632 (1987) 22 Suh JS DP DiLella M Moskovits J Phys Chem 87 1540-1544 (1983) 23 Benner RE R Dornhaus R Chang and BL Laube Correlations in the Raman spectra of cyanide complexes adsorbed
at silver electrodes with voltammograms Surface Science 101 341 (1980) 24 Chase D B and JF Rabolt Fourier Transform Raman Spectroscopy Acad Press Ch1 p 131 (1994) 25 Connes J Rev Opt Theor Instrum 40 45 (1961)
APPLIED SPECTROSCOPY 351
focusing the 488 nm laser beam 2 mm deep into thebulk of the crystals to avoid contributions from potentialdepletion layers As shown in Fig 1b the dependence ofthe Raman shift on the degree of deuteration is almostperfectly linear and ts very well with D 5 22684cmR1 24526 where D is the degree of deuteration (in )and R is the spectral mean of the PO4 vibration in cm21A linear correlation coef cient of 0998 indicates an ex-cellent linear dependence of the Raman peak shift withdegree of deuteration This result shows that the shift ofthe PO 4 peak is simply caused by the linear increase inatomic mass due to isotope substitution which decreasesthe length of hydrogen-like bonds
This excellent linear dependence allows us to map thepro le of the DH exchange layer at the surface of DKDPcrystals by acquiring Raman spectra and determining theposition of the PO 4 peak for various depths This methodis preferable over other methods such as determining thestrength of the OD vibration directly (eg at 715 cm21)because the position of the most intense peak in the Ra-man spectrum can be measured more precisely than theintensity of some of the weakest peaks in the spectrumThis is demonstrated in Fig 2 where depth-dependentRaman spectra (Fig 2a) and the resulting exchange layerpro les for two DKDP crystals are shown (Fig 2b) Thespectra in Fig 2a were obtained from a depth scan of aDKDP crystal with 75 degree of deuteration in thebulk grown at 45 8C The spectra start out as DKDP with30 deuteration close to the surface and approach thebulk DKDP spectrum within a few micrometers of depthThe fact that the relative degree of deuteration does notextend to 0 D is due to the limited depth resolution ofthe Raman microprobe which averages over 4 mm indepth Figure 2b depicts the resulting DH exchange layerpro les for this and a second crystal grown at 63 8Crespectively Both crystals had the same exposure to am-bient conditions and their main difference is the temper-ature at which they were grown The different exchangelayer pro les indicate that crystals grown at differenttemperatures have differen t proton conductiv ities 13
which leads to a difference in their rate of deuteriumdepletion The parameters controlling this behavior arecurrently the objective of a detailed study the results ofwhich will be reported elsewhere
CONCLUSION
In conclusion we have shown that the shift of the to-tally symmetric PO 4 stretch mode in the Raman spectrumof DKDP crystals scales linearly with degree of deuter-ation This allows us to correlate Raman peak positionsto deuteration levels in these crystals We have presenteda new technique to determine DH diffusion pro les inDKDP frequency conversion crystals based on micro-Ra-man spectroscopy This technique is fast inexpensiveand works under various environmental conditionswhich will allow us to better understand and control deu-terium depletion in DKDP crystals
ACKNOWLEDGMENTS
We would like to thank M Runkel for rst discovering DKDP crack-ing R Floyd for providing DKDP crystals and L Chase and A Burn-ham for their support and helpful discussions This work was performedunder the auspices of the US Department of Energy by the University
of California Lawrence Livermore National Laboratory under contract W-7405-Eng-48
1 J J De Yoreo A K Burnham and P K Whitman Int Mat Rev47 113 (2002)
2 C E Barker R A Sacks B M Van Wonterghern J A Caird JR Murray J H Campbell K Kyle R B Ehrlich and N DNielsen Proc SPIE-Int Soc Opt Eng 2633 501 (1995)
3 T Suratwala paper to be published4 Cleveland Crystals Inc httpwwwclevelandcrystalscom
KDPshtmltable5 E A Popova I T Savatinova and I A Velichko Sov Phys Solid
State 12 1543 (1971)6 I P Kaminow R C C Leite and S P S Porto J Phys Chem
Solids 26 2085 (1965)7 J A Subramony B J Marquardt J W Macklin and B Kahr
Chem Mat 11 1312 (1999)8 H Tanaka M Tokunaga and I Tatsuzaki Solid State Commun
49 153 (1984)9 R J Nelmes G M Meyer and J E Tibballs J Phys C 15 59
(1982)10 M A Yakshin D W Kim Y S Kim Y Y Broslavets O E
Sidoryuk and S Goldstein Laser Physics 7 941 (1997)11 I Takenaga Y Tominaga S Endo and M Kobayashi Solid State
Commun 84 931 (1992)12 C Krenn personal communication13 M Sharon and A K Kalia J Solid State Chem 21 171 (1977)
Rapid Dipicolinic Acid Extractionfrom Bacillus Spores Detectedby Surface-EnhancedRaman Spectroscopy
STUART FARQUHARSON ALAND GIFT PAUL MAKSYMIUK andFRANK E INSCOREReal-Time Analyzers Inc East Hartford Connecticut06108
Index Headings Dipicolinic acid Bacillus spores Anthrax Surface-enhanced Raman spectroscopy
INTRODUCTION
The anxiety caused by the distribution of anthrax en-dospores through the US postal system in October 2001was exacerbated by the long time required for positiveidenti cation of the Bacillus anthracis spores and the un-known extent of their distribution Since that time manymethods capable of rapid eld analysis have been inves-tigated to augment or replace the laboratory method ofgrowing microorganisms in culture media which takesdays to perform12 Prominent among these approachesare polymerase chain reactions (PCR)3 immunoassaysand detection of calcium dipicolinate as a biochemicalsignature PCR employs primers to separate organism-
Received 10 October 2003 accepted 14 November 2003 Author to whom correspondence should be sent
352 Volume 58 Number 3 2004
speci c nucleic acid sequences (eg capsular protein en-coding gene for Bacillus anthracis)4 and polymerases toamplify the segment until it is detectable Recently am-pli cation times have been substantially reduced andcomplete analysis can now be performed in an hour orless Immunoassay methods are also being developed thatuse competitive binding of the bioagent (as an antigen)and its labeled conjugate for a limited number of anti-bodies Although analyses can be performed in under 30minutes a well-de ned anthrax antigen has not yet beenidenti ed5ndash7 and consequently the false-positive rate isunacceptably high8
A number of other methods are being developed witha focus on the detection of calcium dipicolinate (CaDPA)and its derivatives as a B anthracis signature This is sobecause only spore-forming bacteria contain CaDPA andthe most common potentially interfering spores such aspollen and mold spores do not Relatively fast methodshave been developed to chemically extract CaDPA andthen detect it directly by uorescence9 or indirectly byluminescence1011 In the latter case hot dodecylamine(DDA) has been used to extract dipicolinic acid (DPA)and terbium has been utilized to form a highly lumines-cent DPA complex11 Although measurements have beenperformed in as little as ve minutes it was found thatas many as three concentration-dependent complexes canform each with different lifetimes This coupled withthe fact that the Tb31 cation produces the same lumines-cence spectrum makes determinations of low spore con-centrations problematic
It has been long known that Raman spectra of Bacillispores are dominated by bands associated with CaDPA12
and that these spectra may provide a suitable anthrax sig-nature at the genus level13 Since that time considerableimprovements in Raman instrumentation have led to lab-oratory measurements of single Bacilli spores14 and to eld measurements of spores captured from a mail-sort-ing system15 However the single spore measurementsrequired complex instrumentation that is not ruggedwhile the eld measurements required milligrams of sam-ple Furthermore the Raman spectra of both measure-ments contained uorescence contributions that would in-crease uncertainty in quanti cation
In related research we demonstrated that nanogramquantities of DPA could be detected by uorescence-freesurface-enhanced Raman spectroscopy (SERS)16 We alsodemonstrated that microliter volumes of chemicals canbe detected by SERS using metal-doped sol-gel-packedglass capillaries17 Towards the goal of developing a rap-id eld SERS-based anthrax spore detector we havecombined our previous research and we now report thatDPA can be extracted from a 10 mg B cereus spore sam-ple using DDA in 1 minute and can be detected by SERSin an additional 1 minute
EXPERIMENTAL
Dipicolinic acid (26-pyridinedicarboxylic acid DPA)and dodecylamine (DDA) were used as received fromSigma-Aldrich (Milwaukee WI) Lyophilized B cereusspores prepared according to the literature13 were sup-plied by the University of Rhode Island and used as re-ceived Multiple particles approximately 01 mm3 each
were separated and weighed at 5 to 15 mg representing05 to 15 million spores The sample masses were con-sistent with a previous determination of spore density at0081 gmL that indicated a high degree of entrained air
All chemicals used to prepare the silver-doped sol-gelcoated capillaries were also obtained and used as receivedfrom Sigma-Aldrich According to previously publishedprocedures17 two precursor solutions were preparedmixed and then drawn into 1-mm-diameter glass capil-laries The silver amine precursor consisted of a 51 vvratio of 1 N AgNO 3 to 28 NH3OH while the alkoxideprecursor consisted of a 21 vv ratio of methanol to te-tramethyl orthosilicate The alkoxide precursors weremixed with silver amine precursor in an 81 vv ratioApproximately 015 mL was drawn into the capillarycoating a 15-mm length After sol-gel formation the in-corporated silver ions were reduced with dilute sodiumborohydride which was followed by a water wash to re-move residual reducing agent
A 100 mL drop of a 50 mM DDA solution in ethanolpre-heated to 78 8C was added to each of the B cereusparticles to digest the spore coat After 1 minute the re-sultant solution was drawn into a SER-active capillarythat was immediately xed horizontally to an XY posi-tioning stage (Conix Research Spring eld OR) just in-side the focal point of an f 07 aspheric lens The lensfocused the beam into the sample and collected the scat-tered radiation back along the same axis A dichroic lter(Omega Optical Brattleborough VT) was used to re ectthe excitation laser to the lens and pass the Raman scat-tered radiation collected by the lens An f 2 achromatwas used to collimate the laser beam exiting a 200-mm-core-diameter source ber optic while a second f 2 ach-romat was used to focus the scattered radiation into a 365mm ber optic (Spectran Avon CT) A short-pass lterwas placed in the excitation beam path to block the sil-icon Raman scattering generated in the source ber fromre ecting off sampling optics and reaching the detectorA long-pass lter was placed in the collection beam pathto block the sample Rayleigh scattering from reachingthe detector A 785 nm diode laser (Process InstrumentsInc model 785-600 Salt Lake City UT) was used todeliver 100 to 150 mW of power to the sample A Fouriertransform Raman spectrometer (Real-Time Analyzersmodel IRA-785 East Hartford CT) and a silicon photo-avalanche detector (Perkin Elmer model C30902S Stam-ford CT) were used to acquire the SER spectra
RESULTS AND DISCUSSION
As an initial experiment the SER spectrum of 1 gLof DPA in water was measured using the newly devel-oped silver-doped sol-gel-coated capillaries (Fig 1A) Atthis concentration a high signal-to-noise ratio (SN) isobtained in 1 min In fact a reasonable spectrum is ob-tained in the same time frame for 1 mgL (Fig 1B) TheSER spectra are reasonably similar to the normal Raman(NR) spectrum obtained for a saturated solution of DPAin 1 N KOH (Fig 1C) and the following band shifts areobserved (NR to SER) 647 to 657 cm21 817 to 815cm21 998 to 1008 cm21 1384 to 1382 cm21 1434 to1428 cm21 and 1569 to 1567 cm21 Many of these bandshave been previously assigned1213 such as 998 cm21 to
APPLIED SPECTROSCOPY 353
FIG 1 SERS of DPA in water using silver-doped sol-gel-coated glasscapillary for (A) 1 gL and (B) 1 mgL (C ) NR of saturated DPA in 1N KOH in a glass capillary Spectral conditions (A) and (B) 150 mWof 785 nm 1-min acquisition time (C) 450 mW of 785 nm 5-minacquisition time both 8 cm 21 resolution
FIG 2 SERS of DPA extracted from 10 mg B cereus particle using100 mL of 50 mM hot DDA acquired in (A) 1 minute and (B) 2 seconds(C ) Attempted SERS of 50 mM hot DDA in ethanol using silver-dopedsol-gel-coated glass capillary acquired in 1 min Spectral conditions150 mW of 785 nm 8 cm21 resolution
the symmetric ring stretch 1384 cm21 to the OndashCndashOsymmetric stretch 1428 cm21 to the symmetric ring CndashH bend and 1569 cm21 to the asymmetric OndashCndashOstretch
The rst B cereus samples consisted of 2 mg of sporesin 2 mL of 5 mM hot DDA The samples were main-tained at 78 8C for 40 min and while hot approximately10 mL was drawn into a SER-active capillary Since spec-tra of DPA were obtained for these initial samples small-er spore masses higher DDA concentrations and shorterheating periods were examined In due course it wasfound that 10 mg of spores could be digested by 100 mLof 50 mM hot DDA in one minute and detected (Fig2A) In fact the signal was suf ciently intense that it canbe observed in as little as two seconds (Fig 2B) Theamount of DPA that was extracted was estimated to bebetween 5 and 10 mgL by comparing the signal intensityof the 1008 cm21 band to that measured for DPA in waterThis is consistent with previous research that found thatthe majority of the DPA is extracted from spores usingDDA11 and that B cereus spores contain approximately10 DPA by weight18 The SN of 127 for the 1008 cm21
band in the 1-minute SER spectrum suggests a limit ofdetection of approximately 250 ng of B cereus sporesbased on a SN of 3 Finally it should be noted that DDAdid not produce a detectable SER spectrum as shown inFig 2C
CONCLUSION
We have demonstrated that by combining rapid extrac-tion of dipicolinic acid from Bacillus cereus spores withchemical identi cation by surface-enhanced Raman spec-troscopy as little as 10 mg of spores can be detected Infact the entire measurement from the time of adding hotdodecylamine to the spores to the time when the dipi-colinic acid SER spectrum is acquired and analyzedcould be performed in less than two minutes The abilityof this method to distinguish between spore-forming bac-teria such as Bacillus anthracis and non-DPA containingpowders could help prevent costly shutdowns associated
with the appearance of suspicious material or intentionalmailing of common substances as an anthrax hoax Thismethod could also prove useful in detecting the locationof anthrax endospores in mail distribution facilities if an-other veri ed attack should occur
Research continues to fully characterize the surface-enhanced Raman spectroscopy signal intensities as afunction of sample concentration and to explore otherextractants that do not require the use of elevated tem-perature
ACKNOWLEDGMENTS
The authors are grateful for the support of the National Science Foun-dation (DMI-0296116 and DMI-0215819) and the US Army(DAAD13-02-C-0015 Joint Service Agent Water Monitor program)The authors are indebted to Chetan Shende for preparing the sol-gelcapillaries The authors also thank James Gillespie Nicholas Fell andAugustus Fountain for providing important background informationMark Farquharson for laboratory support and Professor Jay Sperry ofthe University of Rhode Island for supplying B cereus spores
1 V A Pasechnik C C Shone and P Hambleton Bioseparations 3267 (1993)
2 P J Jackson M E Hugh-Jones D M Adair G Green K K HillC R Kuske L M Grinberg F A Abramova and P Keim ProcNatl Acad Sci USA 95 1224 (1998)
3 B R Glick and J J Pasternak Molecular Biology Principles andApplications of Recombinant DNA (ASM Press Washington DC1994)
4 C A Bell J R Uhl T L Had eld J C David R F Meyer TF Smith and F R Cockerill III J Clin Microbiol 40 2897(2002)
5 D L Gatto-Menking H Yu J G Bruno M T Goode M Millerand A W Zulich Biosens Bioelectron 10 501 (1995)
6 J J Quinlan and P M Foegeding J Rapid Methods AutomationMicrobiol 6 1 (1998)
7 A A Hindle and E A H Hall Analyst (Cambridge UK) 1241599 (1999)
8 M S Ascher US Department of Health amp Human Services(httpwwwhhsgovophppresentationsAscherdoc)
9 R Nudelman B V Bronk and S Efrima Appl Spectrosc 54 445(2000)
10 D L Rosen C Sharpless and L B McBrown Anal Chem 691082 (1997)
354 Volume 58 Number 3 2004
11 P M Pellegrino N F Fell Jr and J B Gillespie Anal ChimActa 455 167 (2002)
12 W H Woodruff T G Spiro and C Gilvarg Biochem BiophysRes Commun 58 197 (1974)
13 E Ghiamati R S Manoharan W H Nelson and J F SperryAppl Spectrosc 46 357 (1992)
14 A P Esposito C E Talley T Huser C W Hollars C M Schal-dach and S M Lane Appl Spectrosc 57 868 (2003)
15 S Farquharson L Grigely V Khitrov W W Smith J F Sperryand G Fenerty J Raman Spectrosc paper accep ted (2003)
16 S Farquharson W W Smith S Elliott and J F Sperry SPIE-IntSoc Opt Eng 3855 110 (1999)
17 S Farquharson and P Maksymiuk Appl Spectrosc 57 479(2003)
18 F W Janssen A J Lund and L E Anderson Science (Washing-ton DC) 127 26 (1958)
SPIE -2003-5269 117
pH dependence of methyl phosphonic acid dipicolinic acid and cyanide by surface-enhanced Raman spectroscopy
Stuart Farquharson Alan Gift Paul Maksymiuk Frank Inscore and Wayne Smith
Real-Time Analyzers 87 Church Street East Hartford CT 06108
ABSTRACT US and Coalition forces fighting terrorism in Afghanistan and Iraq must consider a wide range of attack scenarios in addition to car bombings Among these is the intentional poisoning of water supplies to obstruct military operations To counter such attacks the military is developing portable analyzers that can identify and quantify potential chemical agents in water supplies at microgram per liter concentrations within 10 minutes To aid this effort we have been investigating the value of a surface-enhanced Raman spectroscopy based portable analyzer In particular we have been developing silver-doped sol-gels to generate SER spectra of chemical agents and their hydrolysis products Here we present SER spectra of methyl phosphonic acid and cyanide as a function of pH an important factor affecting quantitation measurements which to our knowledge has not been examined In addition dipicolinic acid a chemical signature associated with anthrax-causing spores is also presented Keywords Chemical warfare agents agent detection agent hydrolysis SERS Raman spectroscopy homeland security
1 INTRODUCTION In the past decade the Unites States and its allies have been challenged by a different kind of warfare exemplified by the terrorist attacks of September 11 2001 Suicide bombings and the use of chemical agents are the norm and military personnel must consider a wide range of attack scenarios Among these is the intentional poisoning of water supplies to obstruct military operations in Afghanistan and Iraq Currently colorimetric paper is used to detect agents on-site while gas chromatography combined with mass spectrometry (GCMS) is used in mobile support laboratories However both methods have severe drawbacks The paper changes color in response to contact with many chemicals besides CWAs causing a high incidence of false positives1 while GCMS although very chemically specific requires hours to perform and constant re-calibration234 Military operations would be greatly aided by a portable analyzer that can identify and quantify potential chemical agents at concentrations that impact safety This includes the analysis of drinking water supplies distribution and storage systems To meet this goal the Department of Defense has been investigating numerous approaches under the auspices of the Joint Service Agent Water Monitor (JSAWM) program5 The JSAWM must be capable of identifying the chemical agents with no false-positives and quantifying the agents at microgram per liter concentrations within 10 minutes678 Recently we and others have been investigating the ability of surface-enhanced Raman spectroscopy (SERS) to measure chemical agents9-12 bioagents13-17 and their hydrolysis products in water SERS employs the interaction of surface plasmon modes of metal particles with target analytes to increase scattering efficiency by as much as 1 million times18 In 1987 the potential of this approach to measure CWAs was demonstrated by measuring a series of pesticides19 Several of these organophosphonates have chemical structures similar to CWAs in particular P=O functional groups In our studies we have been employing silver-doped sol-gels to promote the SER effect The porous silica network of the sol-gel matrix offers a unique environment for immobilizing and stabilizing SER-active metal particles20-23 The sol-gels have been coated on the internal walls of standard glass vials to allow rapid SER analysis of small sample quantities without preparation We have measured over 100 chemicals with enhancements of 104 to 106 demonstrated reversible measurements in a flowing system reproducible measurements from vial-to-vial and measurements in multiple solvents including water20-23 Previously we used these vials to perform preliminary measurements of cyanide (CN) methylphosphonic acid (MPA) and dipicolinic acid (DPA) MPA is a hydrolysis product of the nerve agents (eg sarin Reaction 1) and may be a valuable indicator of nerve agent usage particularly since the alkyl methylphosphonic acids are relatively more stable than their corresponding parent complexes24 DPA is
SPIE -2003-5269 118
a chemical signature of spore forming bacteria such as Bacillus anthracis And in light of the inability to rapidly detect the anthrax spores distributed through the US mail in October 2001 a number of methods are being developed to extract and analyze this signature Reaction 1 Stepwise hydrolysis of Sarin to form hydrofluoric acid (HF) isopropyl methylphosphonic acid (IMPA) then methyl phosphonic acid (MPA) and 2-propanol In our previous SERS investigations MPA and DPA were measured at 50 and 100 mgL respectively In both cases limits of detection (LOD) were estimated at 100 microgL providing encouragement in that SERS may satisfy the needs of the JSAWM Since it has been shown that pH can substantially influence the intensity of SER bands25 which would clearly influence quantitative analysis we undertook the present study to determine the severity of these effects for cyanide methyl phosphonic acid and dipicolinic acid Furthermore we previously observed a band at 1050 cm-1 for MPA6 possibly due to an anion formed at basic pH Here we investigate the source of this spectral anomaly
2 EXPERIMENTAL All chemicals including potassium cyanide methyl phosphonic acid dipicolinic acid and those used to prepare the silver-doped and gold-doped sol-gel coated vials were obtained and used as received from Sigma-Aldrich All samples were prepared in HPLC grade water (Fischer Scientific Fair Lawn NJ) for SERS measurements The pH of these samples was adjusted using dilute nitric acid or potassium hydroxide and verified using a pH electrode (Corning Inc Corning NY) that had been calibrated with pH 400 700 and 1000 buffered standards from Fischer Scientific Once prepared the samples were transferred into the silver-doped sol-gel vials (Real-Time Analyzers Inc East Hartford CT) The vials were coated in a manner similar to that previously reported by adding ammonium hydroxide to a solution of silver nitrate tetramethyl orthosilicate (TMOS) and methanol20 Gold-vials were coated by adding nitric acid to a solution of gold tetrachloride TMOS and methanol The two precursor solutions were prepared mixed and transferred to 2-mL glass vials dried and heated After sol-gel formation the incorporated metal ions were reduced with dilute sodium borohydride (1mgmL) which was followed by a water wash to remove residual reducing agent After the resultant analyte solution was introduced the SER-active vial was immediately fixed horizontally to an XY positioning stage (Conix Research Springfield OR) just inside the focal point of an f07 aspheric lens The lens focused the beam into the sample and collected the scattered radiation back along the same axis A dichroic filter (Omega Optical Brattleborough VT) was used to reflect the excitation laser to the lens and pass the Raman scattered radiation collected by the lens An f2 achromat was used to collimate the laser beam exiting a 200 microm core diameter source fiber optic while a second f2 achromat was used to focus the scattered radiation into a 365 microm fiber optic (Spectran Avon CT) A short pass filter was placed in the excitation beam path to block the silicon Raman scattering generated in the source fiber from reflecting off sampling optics and reaching the detector A long pass filter was placed in the collection beam path to block the sample Rayleigh scattering from reaching the detector A 785 nm diode laser (Process Instruments Inc model 785-600 Salt Lake City UT) was used to deliver 100 to 150 mW of power to the sample A Fourier transform Raman spectrometer (Real-Time Analyzers model IRA-785 East Hartford CT) and a silicon photo-avalanche detector (Perkin Elmer model C30902S Stamford CT) were used to acquire the SER spectra
3 RESULTS AND DISCUSSION In a previous study of MPA6 aimed at developing a concentration calibration curve and determining limits of detection (LOD) we observed an anomaly at 1050 cm-1 Since it was found that the band intensity changed as a function of concentration the band must be associated with a sample parameter Two possible parameters photon flux and pH are examined here The first parameter was investigated by irradiating a 1mgmL MPA sample in a SER-active vial with laser powers of 200 mW and above and monitoring spectral changes It was immediately found that the 1050 cm-1 band
2O+ H HF + +OH
OH3H C
OP
OF
CH
CH3
3
3H C
OCP
OOH
CH
CH3
3
3H C
O
CPHO
CH
CH
3
3
C
Sarin IMPA MPA 2-propanol
SPIE -2003-5269 119
appeared and grew as a function of time and that the higher the laser power the faster the growth Figure 1 shows the growth of the 1050 cm-1 band over the course of 30 minutes when using 150 mW of 785 nm excitation while Figure 2A shows that the growth can be fit with a first order exponential equation namely I1050 = 03+05e-013t Furthermore the 760 cm-1 band could be fit with a first order decay equation with an identical exponential rate constant ie I760 = 08-
08e-013t The rates represent classical first order kinetics and their correspondence allows one to conclude that MPA is being transformed one-for-one into a photo-generated product At this time the photoproduct has not been positively identified but phosphonic (phosphorous) acid and phosphonate are likely candidates since the symmetric P(OH)3 stretch occurs at ~1050cm-1 Our studies show that a reduction in laser power to 100 mW at the sample for MPA essentially eliminates this degradation process This laser power or lower was used for further measurements
Methyl phosphonic acid is a diprotic acid that stepwise dissociates into two anions MPA- and MPA= according to the following reactions26
MPA MPA- + H+ pKa1 = 212 Reaction 2
MPA- MPA= + H+ pKa2 = 729 Reaction 3 The relative concentrations of MPA MPA- and MPA= can be determined at any pH by expressing [MPA] and [MPA=] in terms of [MPA-] using Reactions 2 and 3 and summing all three to equal the total starting concentration here 2 mgmL (0021M MW = 9602) viz
[MPA] + [MPA-] + [MPA=] = 0021M Equation 1 substituting from Reactions 1 and 2
([H+][MPA-])K1a + [MPA-] + (K2a[MPA-])[H+] = 0021M Equation 2 rearranging [MPA-] = 0021M(1+[H+]K1a + K2a[H+]) Equation 3 The relative concentrations of MPA MPA- and MPA= as a function of pH are shown in Figure 3 It is worth noting that near neutral pH both MPA- and MPA= will be present To confirm that the SER signal followed this pH dependence a starting solution consisting of 20 mg of MPA in 10 mL HPLC grade water was prepared and brought to pH of 20 using dilute nitric acid From this solution 2 mL were added to a SER-active vial and the SER spectrum recorded At this pH a peak at 760 cm-1 was barely discernable The 2 mL solution was returned to the starting solution and the pH was re-measured to correct for any changes that the silver-doped sol-gel vials might cause In most cases the change was less than 02 pH units and the pH is reported as the before and after average Next the pH of the
Figure 2 A) Exponential growth of 1050 cm-1 band and B) exponential decay of 760 cm-1 band for spectral series in Figure 1
Figure 1 Growth of 1050 cm-1 band as a function of time due to exposure to 150 mW of 785 nm Spectra are 5 sec each collected every 100-sec from 0 to 30-min
B
570 770 970 1170 1370Raman Shift (cm-1)
Arbit
rary
Unit
s
MPA Photodegradation
600 800 1000 1200 1400 Wavenumber (cm-1)
A
0 10 20 30 0 10 20 30 time (min) time (min)
Ram
an In
tens
ity (r
elat
ive)
30
min
0
SPIE -2003-5269 120
starting solution was adjusted to 325 using dilute KOH Again 2 mL were added to a vial and the SER spectrum recorded At this pH a reasonably strong 760 cm-1 band was observed This process was repeated as spectra were recorded at pHs of 70 74 75 79 85 and 100 A total of 1 mL of KOH was added diluting the total concentration by 10 Next the pH of the starting solution was made acidic by adding dilute nitric acid dropwise This time spectra were recorded at pHs of 72 69 64 and 37 Figure 4 shows the SER spectra for representative pHs (spectra were left out to simplify the figure) while Figure 3 shows the 760 cm-1 peak intensities as a function of pH (The band intensities were adjusted to compensate for dilution effects caused by the addition of HNO3 and KOH then normalized to 0021 M for the most intense band observed at pH 37) It is clear from Figure 3 that the 760 cm-1 band follows the MPA- concentration as a function of pH and must be assigned to this anion No bands were observed that corresponded to MPA or MPA= The lack of an MPA SER spectrum may be due to the absence of an attraction between the neutral analyte and the electropositive silver surface The same reasoning suggests that a strong SER spectrum should be observed for MPA= but it is not and a satisfactory explanation has not been found
0000
0005
0010
0015
0020
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14pH
Con
cent
ratio
n [M
]
pK1 = 212 pK2 = 729
MPA- MPA=
MPA
Figure 4 SER spectra of 002M MPA as a function of pH Conditions 100 mW of 785 nm 36 scans (1 min) 8 cm-1 recorded 2 min after sample introduction pH 19 and 20 not apparent on this scale pH 69 and 74 near identical to 70 and 75 and not shown for clarity
00
02
04
06
08
10
12
14
16
18
0 5 10 15 20 25 30Measurement Number
Ram
an In
tens
ity (7
60 c
m-1
)
Figure 3 Concentration dependence of MPA MPA- and MPA= as a function of pH for a 002M sample Intensity of 760 cm-1
band from Figure 6 as a function of pH ( for increasing basic adjustment for increasing acidic adjustment error was measured at ~10 for pH 64)
Figure 5 SER spectra of 002M MPA at pH 64 measured around a vial at three heights (9 points per height) Conditions as in Figure 4 but 10-sec scans
Since these measurements involved the removal and replacement of the SER-active vial in the sample holder to remove and add sample variation in the intensity as a function of vial position was minimized by illuminating the exact same height along the vial wall But this does not account for variability of the SERS response of the sol-gel coating around the vial To analyze this effect a vial containing MPA at pH 64 was rotated at ~ 40o
intervals at the original height and 18rdquo above and below this value Figure 5 shows the intensity of the 760 cm-1
band for the 27 positions It was found that the average value was 137plusmn014 an RSD of 10 overall and 5 for each height An error bar is included in Figure 4 for the pH 64 measurement
SPIE -2003-5269 121
With the above analyses in mind a preliminary investigation of the SER spectral response for MPA (as MPA-) as a function of concentration was performed A single vial was used for these measurements beginning with 1 mgL followed by measurements of 10 100 and 1000 mgL In all cases the pH was ~7 and 3 positions around the vial were measured per concentration Since the 760 cm-1 band was not observed for concentrations of 1 or 10 mgL using 100 mW of 785 nm the laser power at the sample was raised to 200 mW beginning with the 10 mgL concentration Photo-degradation was largely avoided (and not observed) by exposing the sample for only 33 seconds per spectral acquisition Representative spectra for 10 100 and 1000 mgL are shown in Figure 6 while a plot of the 760 cm-1 band intensity as a function of concentration is shown in Figure 7 These values were also used to estimate limits of detection based on the signal-to-noise ratio (SN) of the 760 cm-1 band using 100 mW of 785 nm laser excitation and a 10-min acquisition time As summarized in the Figure 7 inset the lower the measured concentration the lower the predicted LOD Even if as estimated 210 microgL could be measured using the silver-doped sol-gel vials an improvement of a factor of 70 is still required to meet JSAWM goals of measuring 3 microgL in 10-minutes Similar to MPA DPA is a diprotic acid (pKa1 = 216 and pKa2 = 692) and variations in pH will effect the relative concentrations of DPA DPA- and DPA= and possibly the SER spectra and band intensities This could prove significant if an acid or base is used to denature anthrax spores with the goal of extracting and analyzing DPA The MPA pH study described above was mimicked for DPA except that the starting solution consisted of 20 mg of DPA in 20 mL HPLC grade water (60x10-3M MW = 1671) The initial solution had a pH of 245 which was made basic by dilute KOH to pHs of 355 433 487 559 1069 and 1166 SER spectra were recorded at each pH using 100 mW of 785 nm and a 44-sec acquisition time Next one drop of concentrated nitric acid was used to remake the solution acidic at a pH of 200 Again sequential pH measurements were performed at 383 510 735 and 822 The solution pH was made acidic a third time but to pH 219 171 then 135 Throughout this process no more than 20 drops of acid or base were added and therefore the concentration was diluted by no more than 10 Most of the spectral bands showed a minor decrease in intensity as a function of increasing pH values However the bands at 525 and 795 cm-1 showed the most dramatic changes which occurred at acid pH Figure 8 shows the SER spectra of DPA for the spectral region and pH range of interest The identity of the DPA species was determined by plotting the normalized peak intensities with the lowest value set to 0 and the highest to 0006 M as a function of pH and overlaying these values on a plot of the relative concentrations for DPA DPA- and DPA= as previously done for MPA (Figure 9) As can be seen the 525 cm-1 band clearly corresponds to DPA The correspondence of the 795 cm-1 band to this species is less clear as the band retains intensity until pH of 55 This can be attributed to contribution to the overlapping band at 810 cm-1 which does not change as a function of pH The fact that most bands are observed at all pHs suggest that the primary interaction with silver is through the ring nitrogen This is supported by the fact that the most intense band occurs at 1008 cm-1 attributed to a symmetric ring breathing mode and that this interaction has been characterized for pyridine in numerous papers27
0
1
2
3
4
5
6
7
0 200 400 600 800 1000 1200
MPA Concentration (mgL)
760
Ban
d In
tens
ity
Figure 6 SER spectra of MPA in water at A) 1000 B) 100 and C) 10 mgL Conditions pH of 7 silver-doped sol-gel coated vial 200 mW of 785 nm 33-sec 8 cm-1 resolution
Figure 7 Plot of SER intensity of 760 cm-1 band of MPA as a function of concentration using 200 mW of 785 nm Inset table includes average intensity LOD standard deviation and percent deviation for each concentration but for 100 mW and 10 min
conc (mgL) LOD ave stddev dev10 021 002 1021100 072 010 14301000 312 040 1280
A
B C
SPIE -2003-5269 122
The ability to detect dipicolinic acid (DPA) by SERS is demonstrated in Figure 10 Even at 1 mgL the primary bands are visible The intensity of the dominant ring breathing mode observed at 1008 cm-1 for DPA in water (pH = 25 to 55) is plotted as a function of concentration in Figure 11 These values were also used to estimate limits of detection based on the SN of the 1008 cm-1 band using 100 mW of 785 nm laser excitation and a 10-min acquisition time Again the lower the measured concentration the lower the predicted LOD (see Figure 11 inset) and detection of 160 microgL is possible Gastrointestinal anthrax requires significant more spores than inhalation anthrax28 and a limit of detection might be placed at 1 million spores in 1 liter of water or 10 microgL Since each spore contains ~10 CaDPA by weight29 a goal for DPA might be 1 microgL indicating that the present measurements must be improved by nearly two orders of magnitude Due to the increased hazards of handling HCN gas KCN salt was used for these experiments Nevertheless all sample preparations were performed in a chemical hood KCN completely dissolves in water but its conjugate acid HCN is formed and has a Ka of 615x10-1030 viz
HCN CN- + H+ pKa = 921 Reaction 4
Figure 8 SER spectra of 1 mgmL DPA as a function of pH Conditions 100 mW 785 nm 100 scans (44-sec)
Figure 9 Concentration dependence of DPA DPA- and DPA= as a function of pH for a 0006M sample Intensity of 525 () and 795 (diams) cm-1 bands from Fig 8 as a function of pH
pH
135 171 219 383
0
0001
0002
0003
0004
0005
0006
0007
0 2 4 6 8 10 12 14pH
Con
cent
ratio
n [M
]
DPADPA-DPA=795525
DPA DPA=DPA-
pK2 = 692pK1 = 216
795 525
0
05
1
15
2
25
0 200 400 600 800 1000 1200
DPA Concentration (mgL)
1008
Ban
d In
tens
ity
conc (mgL) lod-10min-100mw1 017
10 016100 103
1000 355
Figure 10 SER spectra of DPA in water at A) 1000 B) 100 C) 10 and D) 1 mgL Conditions pH of 25-55 silver-doped sol-gel coated vial 175 mW of 785 nm 1-min 8 cm-1 D) has been multiplied by x10 to make bands visible
A
B C
D
Figure 11 Plot of SER intensity of 1008 cm-1 band of DPA as a function of concentration using 175 mW of 785 nm Inset table includes LOD in mgL for each concentration but for 100 mW and 10 min
SPIE -2003-5269 123
Consequently the cyanide concentration must be determined for each initial KCN concentration Specifically the samples prepared with concentrations of 01 1 10 100 and 1000 mgL of KCN produced CN- concentrations of 63x10-3 033 69 89 and 964 mgL at pHs of 816 90 967 102 and 107 respectively The pH dependence for the HCN and CN- concentrations are shown in Figure 12 Thus as the amount of KCN added to the solution decreases so does the pH of the solution (becomes less basic) and according to Reaction 4 the relative amount of CN- to HCN also decreases For example in the preparation of a 01 mgL solution of KCN the pH is shifted from 7 for pure water to only 816 and only 63 of the starting material becomes CN- or 63x10-3 mgL In comparison for a solution of 1000 mgL the pH is shifted from 7 to 107 and 96 of the starting material becomes CN- This is significant since the cyanide ion is better able to adsorb onto the silver particles and become SERS active SER spectra of 10 100 and 1000 mgL of KCN samples are shown in Figure 13 and a corresponding plot of the 2100 cm-1 band intensity as a function of concentration is shown in Figure 14
000102030405060708091011
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14pH
Con
cent
ratio
n [m
gm
L]
pKa = 921
CN -HCN
The SER spectra of cyanide are dominated by the single CN stretch at 2100 cm-1 which occurs in normal Raman spectra of solutions at 2080 cm-1 However a low frequency mode occurs at 135 cm-1 due to a Ag-CN stretch (not shown) indicative of a strong surface interaction It is also observed that as the concentration decreases the CN stretch shifts to 2140 cm-1 This shift has been attribute to the formation of a tetrahedral Ag(CN)3
2- surface structure31 as well
Figure 14 Concentration dependence of KCN SERS measured under conditions in Fig 11 Concentrations are 1 01 and 001 mgml Intensities are measured for the CN stretch at 2100 cm-1 Inset table includes LOD in mgL for each concentration in Figs 13 and 15 but for 100 mW and 10 min
Figure 13 SER spectra of KCN in water at A) 1000 B) 100 and C) 10 mgL Conditions pHs of 107 102 and 97 silver-doped sol-gel coated vial 100 mW of 785 nm 1-min 8 cm-1 C) has been multiplied by x10 to make band visible
Figure 12 Concentration dependence of HCN and CN- as a function of pH for a 1 mgmL sample Calculated intensity of 2100 () cm-1 band for a 1 mgml sample at pHs of 816 90 967 102 and 107
A
B
C
0
20
40
60
80
100
120
140
0 200 400 600 800 1000 1200CN Concentration (mgL)
2100
Ban
d In
tens
ity
Figure 15 SER spectra of KCN in water at A) 10 B) 1 and C) 01 mgL Conditions pHs of 97 12 and 12 gold-doped sol-gel coated vial 100 mW of 785 nm 1-min 8 cm-1 C) has been multiplied by x10 to make band visible
A
B
C
conc (mgL) lod-10min-100mw Condition01 001 Au-pH 121 007 Au-pH 12
69 003 Au-pH 9769 016 Ag-pH 9789 022 Ag-pH 102946 113 Ag-pH 107
SPIE -2003-5269 124
as to CN adsorbed to two different surface sites32 Alternatively the 2140 cm-1 band could be attributed to HCN since this species dominates at lower concentrations However it is unlikely that this species would be attracted to the electropositive silver surface Further both peaks should be present at pHs between 85 and 105 but this is not observed It has also been suggested that at concentrations near and above monolayer coverage the CN- species is forced to adsorb end-on due to crowding and at lower concentrations the molecule can reorient to lie flat33 This suggests that the 2100 and 2140 cm-1 bands correspond to the end-on and flat orientations respectively As Figures 13 and 14 show the intensity of the CN stretch for the 89 mgL sample is nearly as intense as the 964 mgmL band This suggests that the Raman signal for the flat orientation is more enhanced However more extensive measurements are required to verify this point Since resent research has suggested that cyanide may be more effectively detected on gold measurements of KCN solutions were also performed using gold-doped sol-gel vials Preliminary measurements are shown in Figure 15 for samples prepared from 01 1 and 10 mgL KCN Since the pHs are 816 90 967 the resultant CN- concentrations are 63x10-3 033 and 69 mgL Initially only the highest concentration was observed and the signal intensity was significantly better than the equivalent concentration measured using silver In an effort to shift Reaction 4 to the left transforming HCN to CN- (Le Chatelierrsquos principle) KOH was added to the lower concentration samples producing solutions with pH 12 This effectively forces all of the cyanide in solution to be CN- or 01 and 10 mgL respectively More importantly the CN stretch is now observed in the SER spectra The band appears at 2125 cm-1 as has been previously reported for gold12 As calculated for MPA and DPA LODs can be estimated from this data For the three concentrations of cyanide on silver the LODs are 016 to 11 mgL for 100 mW of 785 nm laser excitation and a 10-min acquisition time For gold pH adjusted the LODs are10 to 70 microgL an improvement of more than 10 times silver Nevertheless either substrate is sufficient to meet the JSAWM goals of measuring 3 mgL in 10-minutes as the requirements form cyanide are much less stringent than the nerve agents
4 CONCLUSIONS Here we examined the surface-enhanced Raman spectral response for methyl phosphonic acid dipicolinic acid and cyanide as a function of pH It was determined that the most prominent peak at 760 cm-1 reaches a maximum intensity between pH 3 and 7 and corresponds to the MPA- species Neither the MPA nor MPA= species appear to generate a SER spectrum and consequently no spectra were observed below pH 2 or above pH 8 In this study we also found that higher laser powers could cause photodegradation of MPA signified by the exponential growth of a band at 1050 cm-1 which is tentatively assigned to phosphorous acid Unlike MPA DPA was observed at all pHs This is attributed to the dominant interaction of the pyridine functional group with silver Minor spectral changes were observed at acid pHs and were assigned to neutral DPA Like MPA SER spectra of cyanide were pH dependent No spectra were observed for the HCN species while CN- was best observed at pHs more basic than 8 Preliminary concentration studies for the three analytes allowed estimating limits of detection for MPA DPA and CN using 100 mW of 785 nm and a 10-min acquisition time of 210 165 and 70 microgL respectively Although the latter value suggests that cyanide can be measured at sufficiently low concentrations to meet JSAWM goal improvements by 100 to 200 times are required for MPA and DPA It is clear from this study that pH of the sample is important and must be taken into account when developing concentration calibration curves as well as fieldable analyzers Future work will include tailoring the sol-gel with specific functional groups to dictate sample pH and thereby optimize sensitivity
ACKNOWLEDGMENTS The authors are grateful for the support of the US Army (DAAD13-02-C-0015 Joint Service Agent Water Monitor program) The authors would also like to thank Dr Steve Christensen of the US Army SBCCOM for helpful discussions and Mr Chetan Shende of Real-Time Analyzers for assistance in development of the gold-doped sol-gels
REFERENCES 1 Erickson B Analytical Chemistry News amp Features June 1 397A (1998)
SPIE -2003-5269 125
2 Johnston RL Hoefler CM Fargo JC and Moberley B AT-ONSITE 5-8 (1994) 3 Black RM RJ Clarke RW Read and MT Reid J Chromatography 662 301-321 (1994) 4 USA Army Center for Health Promotion and Preventive Medicine (USACHPPM) Technical Guideline 230A 5 Jensen JL NC Wong W R Loerop SPIE 4775-03 (2001) 6 Hoenig SL Handbook of Chemical Warfare and Terrorism Greenwood Press Wesport CT (2002) 7 Munro NB SS Talmage GD Griffin LC Waters AP Watson JF King and V Hauschild Env Health
Persp 107 933-974 (1999) 8 Holstege CP Kirk M Sidell FR Crit Care Clin13 923-42 (1997) 9 Farquharson S P Maksymiuk K Ong and S Christesen SPIE 4577 166-173 (2001) 10 Lee Y and S Farquharson SPIE 4378 21-26 (2001) 11 Spencer KM J Sylvia S Clauson and J Janni SPIE 4577 158-165 (2001) 12 Tessier P S Christesen K Ong E Clemente A Lenhoff E Kaler and O Velev Applied Spectroscopy 56
1524-1530 (2002) 13 Farquharson S WW Smith S Elliott and JF Sperry SPIE 3533 207-214 (1998) 14 Farquharson S WW Smith S Elliott and JF Sperry SPIE 3855110-116 (1999) 15 Farquharson S WW Smith YH Lee S Elliott and JF Sperry SPIE 4575 62-72 (2002) 16 Guzelian AA J Sylvia J Janni S Clauson and KM Spencer SPIE 4577 182-192 (2001) 17 Shende C F Inscore A Gift P Maksymiuk and S Farquharson in press 18 Weaver MJ S Farquharson and MA Tadayyoni J Chem Phys 82 4867-4874 (1985) 19 Alak AM and T Vo-Dinh Analytical Chemistry 59 2149-2153 (1987) 20 Lee Y and S Farquharson SPIE 4206 140-146 (2000) 21 Farquharson S and Y Lee SPIE 4200-16 (2000) 22 Lee Y S Farquharson and P M Rainey SPIE 3857 76-84 (1999) 23 Lee Y S Farquharson H Kwong and M Shahriari SPIE 3537 252-260 (1998) 24 Wang J M Pumera G Collins and A Mulchandani Analytical Chemistry 74 6121-6125 (2002) 25 Dou X YM Jung Z-Q Cao and Y Ozaki Applied Spectroscopy 53 1440-1447 (1999) 26 Data supplied by S Christesen and K Ewing 27 Kerker M and B Thompson Eds SPIE MS 10 (1990) 28 Inglesby TV DA Henderson JG Bartlett JAMA 287 2236 (2002) 29 FW Janssen AJ Lund and LE Anderson Science 127 26 (1958) 30 Lide DR Ed Handbook of Chemistry and Physics CRC Press 77th Ed (1996-7) 31 Billmann J G Kovacs and A Otto Surf Sci 92 153 (1980) 32 Murray CA and S Bodoff Phys Rev B 32 671 (1985) 33 Kellogg D and J Pemberton J Phys Chem 91 1120 (1987)
SPIE-2003-5269 16
Chemical agent detection by surface-enhanced Raman spectroscopy
Stuart Farquharson Alan Gift Paul Maksymiuk Frank Inscore and Wayne Smith
Real-Time Analyzers 87 Church Street East Hartford CT 06108
Kevin Morrisey and Steven D Christesen US Army SBCCOM Aberdeen Proving Ground MD 21010
ABSTRACT
In the past decade the Unites States and its allies have been challenged by a different kind of warfare exemplified by the terrorist attacks of September 11 2001 Although suicide bombings are the most often used form of terror military personnel must consider a wide range of attack scenarios Among these is the intentional poisoning of water supplies to obstruct military operations in Afghanistan and Iraq To counter such attacks the military is developing portable analyzers that can identify and quantify potential chemical agents in water supplies at microgram per liter concentrations within 10 minutes To aid this effort we have been investigating the value of a surface-enhanced Raman spectroscopy based portable analyzer In particular we have been developing silver-doped sol-gels to generate SER spectra of chemical agents and their hydrolysis products Here we present SER spectra of several chemical agents measured in a generic tap water Repeat measurements were performed to establish statistical error associated with SERS obtained using the sol-gel coated vials Keywords Chemical agents chemical agent detection SERS Raman spectroscopy
1 INTRODUCTION
In the autumn of 2001 terrorism within US borders became a sobering reality While extensive efforts are being implemented to secure the homeland US and Coalition forces in Afghanistan and Iraq are constantly faced with terrorist attacks In addition to car-bombings the military has identified several non-traditional attack scenarios including poisoning of water supplies by chemical warfare agents (CWAs) To counter this threat the Department of Defense has been investigating numerous approaches to detect such attacks under the auspices of the Joint Service Agent Water Monitor (JSAWM) program1 The JSAWM must be capable of identifying the chemical agents with no false-positives and quantifying the agents at microgram per liter concentrations within 10 minutes2 This includes the analysis of drinking water supplies distribution and storage systems Currently colorimetric paper is used to detect agents on-site while gas chromatography combined with mass spectrometry (GCMS) is used in mobile support laboratories However both methods have severe drawbacks The paper changes color in response to contact with many chemicals besides CWAs causing a high incidence of false positives3 while GCMS although very chemically specific requires up to an hour to perform and regular re-calibration456 Vibrational spectroscopy has also been investigated as a potential method of rapidly detecting CWAs7-
11 as early as 1985 in the case of infrared spectroscopy8 and 1988 in the case of Raman spectroscopy11 Again however these techniques also have limitations when it comes to measuring trace poisons in water Infrared spectra would be dominated by the very strong absorption of water which would obscure absorptions by most other chemicals present Whereas Raman spectroscopy is simply not a very sensitive technique and detection limits are typically grams per liter Surface-enhanced Raman spectroscopy (SERS) offers several advantages over conventional vibrational methods and that may provide the necessary sensitivity required for detecting trace quantities of chemical agents in water SERS employs the interaction of surface plasmon modes of metal particles with target analytes to increase scattering efficiency by as much as 1 million times12 In 1987 the potential of this approach to measure CWAs was demonstrated by measuring a series of pesticides13 Several of these organophosphonates have chemical structures similar to CWAs in particular P=O functional groups In the past few years we and others have further explored the ability of SERS to detect CWAs14-17 and even bioagents 18-21 We have been employing silver-doped sol-gels to promote the SER effect
SPIE-2003-5269 17
in these studies The porous silica network of the sol-gel matrix offers a unique environment for immobilizing and stabilizing SER-active metal particles22-25 The silver-doped sol-gels have been coated on the internal walls of standard glass vials to allow rapid SER analysis of small sample quantities without preparation We have measured over 100 chemicals with enhancements of 104 to 106 demonstrated reversible measurements in a flowing system reproducible measurements from vial-to-vial and measurements in multiple solvents including water21-25 Previously we used these vials to perform preliminary measurements of cyanide (CN) and methylphosphonic acid (MPA) Most of the nerve agents form MPA during hydrolysis while Tabun forms CN a chemical agent in its own right In another paper including in these proceedings we examined the limits of detection (LOD) for MPA by measuring a series of concentrations down to 50 mgL and estimated a limit of detection of 100 microgL26 These measurements provide encouragement in that SERS may satisfy the needs of the JSAWM To further establish the viability of SERS in particular silver-doped sol-gels here we present analysis of cyanide mustard and VX in tap water The measurements performed at the US Armyrsquos Edgewood Chemical Biological Center Aberdeen MD also included numerous repeat measurements to establish reproducibility
2 EXPERIMENTAL 2a General All chemicals including potassium cyanide 2-chloroethylethyl sulfide and those used to prepare the silver-doped and gold-doped sol-gel coated vials were obtained and used as received from Sigma-Aldrich All samples were prepared in a chemical hood using HPLC grade water unless otherwise noted (Fischer Scientific Fair Lawn NJ) for SERS measurements Once prepared the samples were transferred into the silver-doped sol-gel vials (Real-Time Analyzers Inc East Hartford CT) The vials were coated in a manner similar to that previously reported by adding ammonium hydroxide to a solution of silver nitrate tetramethyl orthosilicate (TMOS) and methanol22 The two precursor solutions were prepared mixed and transferred to 2-mL glass vials dried and heated After sol-gel formation the incorporated metal ions were reduced with dilute sodium borohydride (1mgmL) which is followed by a water wash to remove residual reducing agent After the resultant analyte solution was introduced the SER-active vial was fixed horizontally to an XY positioning stage (Conix Research Springfield OR) just inside the focal point of an f07 aspheric lens The lens focused the beam into the sample and collected the scattered radiation back along the same axis A dichroic filter (Omega Optical Brattleborough VT) was used to reflect the excitation laser to the lens and pass the Raman scattered radiation collected by the lens An f2 achromat was used to collimate the laser beam exiting a 200 microm core diameter source fiber optic while a second f2 achromat was used to focus the scattered radiation into a 365 microm fiber optic (Spectran Avon CT) A short pass filter was placed in the excitation beam path to block the silicon Raman scattering generated in the source fiber from reflecting off sampling optics and reaching the detector A long pass filter was placed in the collection beam path to block the sample Rayleigh scattering from reaching the detector A 785 nm diode laser (Process Instruments Inc model 785-600 Salt Lake City UT) was used to deliver 100 to 150 mW of power to the sample A Fourier transform Raman spectrometer (Real-Time Analyzers model IRA-785 East Hartford CT) and a silicon photo-avalanche detector (Perkin Elmer model C30902S Stamford CT) were used to acquire the SER spectra
2b Edgewood Chemical Biological Center The surface-enhanced Raman spectral measurements at the US Armyrsquos Chemical Biological center presented here were all performed on September 12 2003 To expedite measurements a plate was machined to hold up to 12 SER-active sample vials (Figure 1) The plate fit a standard XY plate reader that could be programmed Pure KCN bis-(2-chloroethyl)sulfide (distilled mustard gas HD) and ethyl S-2-diisopropylamino ethyl methylphosphonothioate (VX) were obtained on-site and used to prepare 1 gL tap water solutions in a chemical hood with appropriate safety equipment Simulated tap water was prepared by adding 10 chemicals most often found in tap water at appropriate concentrations to distilled water (Table 1) SER measurements were also performed in a chemical hood For added safety the FT-Raman instrument was placed outside the laboratory and 30 foot fiber optic and electrical cables were used to allow remote SERS measurements and plate manipulation For each experiment 1gL samples were prepared and added to 9 individual vials which were then loaded on the plate In some cases a tenth vial was included as a blank
Table 1 Generic Tap Water Composition
Compound mgL NaHCO3 100 CaSO4 27 MgSO4bull7H2O 67 NaNO3 1 Fulvic Acid 1 K2HPO4 07 KH2PO4 03 (NH4)2HSO4 001 NaCl 001 FeSO4 0001 pH 76-78
SPIE-2003-5269 18
A software program was written that allowed selecting the sequence that the vials were measured the number of positions along the length of the vials to measure (1 to 5) and the number of scans to co-add During sample analysis the program displayed the vial being analyzed the point being analyzed and the spectrum as it was being acquired Once all the data was collected a second software program was written to rapidly analyze the data The spectra collected for all the vials on a plate could be loaded at one time and then the spectra for each point could be displayed simultaneously or separately The user could then select the Raman peak to analyze in terms of peak height or area This was accomplished by selecting points on either side of the peak to define a baseline of zero The peak height or area could then be computed for all of the spectra loaded and then exported to a spreadsheet for statistical analysis
Figure 1 A) Vial Holder 6 slots to hold 2 vials each end-to-end B) Measurement Configuration Program user selects vials to measure sequence number of points per vial (1 to 5) and number of scans per point C) Spectral Acquisition Program shows spectrum being collected which vial and position D) Spectra Analysis Program user selects spectra to analyze by plate vial and point (s) as well as two wavenumbers defining the peak and the baseline to subtract The image is of 5 repeat measurements of 10 mgL KCN in generic tap water 16 sec each 100 mW of 785 nm
3 RESULTS AND DISCUSSION Raman and surface-enhanced Raman spectra were obtained for potassium cyanide bis-(2-chloroethyl)sulfide and ethyl S-2-diisopropylamino ethyl methylphosphonothioate representing three classes of chemical agents cyanides mustards and nerve agents respectively Spectra were also obtained for 2-chloroethyl ethyl sulfide (CEES) a structural analogue to HD which was included in the study to aid in assigning spectral bands KCN salt was used for cyanide experiments to avoid the increased hazards of handling HCN gas KCN completely dissolves in water forming its conjugate acid HCN according to its Ka of 615x10-1027 and at a concentration of 1 mgmL results in a pH 107 solution This is significant in that only CN- appears to interact sufficiently with silver to produce a SER spectrum and no spectral signal is observed below pH 726 Figure 2 shows the SER and normal Raman spectra for KCN The SERS spectrum of 1mgml KCN in water shows a single intense somewhat broad feature at
A B
C D
SPIE-2003-5269 19
2100 cm-1 assigned to the single CequivN stretch The band is much sharper in the normal Raman spectra of the solid KCN salt at 2074 cm-1 This band does broaden and shift to 2080 cm-1 in solution (not shown) However the observed SERS frequency is attributed to interaction with silver and low frequency mode at 135 cm-1 attributed to a Ag-CN stretch (not shown) supports this conclusion
cm-1 band to a C-S stretch but the authors concede that it is in fact more likely a C-Cl stretch28 It appears that the most intense Raman bands at 648 692 and 747 cm-1 shift to 620 660 and 720 cm-1 in the SER spectra and are tentatively assigned as above The width of these bands suggests that they overlap underlying spectral features Additional bands in the Raman spectra occur at 972 1034 1049 1263 1286 1423 1442 2865 2935 and 2960 cm-1 Corresponding bands occur in the SER spectra at 964 1015 1054 1286 1410 1447 2865 and 2935 cm-1 Most of these bands are associated with alkane modes specifically the bands at approximately 1040 cm-1 to a C-C stretch 1290 cm-1 to a CH2 in-phase twist 1440 cm-1 to a CH2 wag 2865 cm-1 to a symmetric CH2 stretch and 2965 cm-1 to an asymmetric CH2 stretch The Raman and SER spectra of sulfur mustard were measured at the Edgewood center (Figure 4) Both spectra are largely similar to CEES The C-Cl and C-S bands in the Raman spectrum of HD now occur at 640 655 700 739 and 760 cm-1 and are more resolved possibly due to the increased molecular symmetry Theoretical calculations indicate that the first three bands are due to C-Cl stretching modes and the latter two to C-S stretching modes28 Only the C-Cl bands maintain significant intensity in the SER spectra occurring at 624 and 643 cm-1 which is attributed to the
Prior to measurements of HD CEES was examined by Raman and SER spectroscopy (Figure 3) CEES also known as half-mustard is essentially identical to HD except one of the chlorine end atoms is replaced by a hydrogen atom Again although not as toxic as HD CEES is a blister agent and dilute aqueous samples were prepared using appropriate safety equipment Both the Raman and SER spectra of CEES are similar and dominated by bands between 600 and 800 cm-1 These are associated with C-Cl and C-S stretching modes which are tentatively assigned to 648 and 747 cm-1 in the Raman spectra respectively The shoulder at 630 cm-1 the overlapped band at 660 cm-1 and the strong band at 692 cm-1 could also be due to these modes or their asymmetric counterparts It is worth noting that theoretical calculations assign the 692
Figure 2 A) SER and B) NR spectra of KCN Conditions A) 1 mgml in tap water 100 mW of 785 nm at sample 1-min acquisition time B) solid 300 mW of 785 nm 5-min All spectra are 8 cm-1 resolution
A
B
Figure 3 A) SER and B) NR spectra of CEES Conditions A) 1 vv (10 mgml) in MeOH 100 mW of 785 nm 1-min acquisition time B) neat 300 mW of 785 nm 5-min
A
B
Cl-CH2-CH2-S-CH2-CH3
A
B
Figure 4 A) SER and B) NR of HD Conditions A) 1mgml in tap water B) pure both 100 mW of 785 nm 1-min
Cl-CH2-CH2-S-CH2-CH2-Cl
SPIE-2003-5269 20
expected strong interaction between chlorine and silver and adds support to the assignment of this band to a C-Cl stretch Weaker overlapping bands occur at 670 692 and 724 cm-1 the latter possibly due to C-S stretching modes Again the alkane modes are apparent in the normal Raman spectra of HD but only a broad feature at 1300 to 1450 cm-1 suggests CH2 contributions in the SER spectrum Although the observed bands in the VX spectrum have not been assigned (Figure 4) a computer generated Raman spectrum29 predicts many of the same features with surprising accuracy and are used here Two intense bands at 460 and 530 cm-1 closely match predicted bands at 463 and 546 cm-1 assigned to a CH3-P=O bend and a PO2CS wag Three highly overlapped bands occur at 694 745 and 771 cm-1 matching predicted bands at 713 730 and 760 cm-1 The first
Table 2 Measured SER peak heights for the CN stretch at 2100 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
Figure 5 A) SERS and B) NR spectra of VX Conditions A) 1 vv (10 mgml) in MeOH B) pure sample both 100 mW of 785 nm 1-min acquisition time
A
B
two have been assigned to a C-S stretch and CH2bend respectively while the latter has been attributed to either a P-C stretch or an O-C-C stretch Although the 745 cm-1 band may alternatively be assigned to a C-S stretch based on the previous measurements of CEES and HD The relatively intense bands at 890 1106 1218 1445 and 1465 cm-1 also match predicted bands at 880 1108 1216 1440 and 1464 cm-1 that are assigned to a C-C stretch CH3 rock N-C3 stretch various C-H3 bends and C-H bends respectively Both the computer generated and the measured spectra contain numerous other less intense bands One is worth mentioning A unique band appears at 370 cm-1 that is predicted at 368 cm-1 and corresponds to an O-P=O bend The surface-enhanced Raman spectrum of VX is also rich with spectral features It has the unique low frequency band at 370 cm-1 as well as a second band at 380 cm-1 that is assigned tothe S-P-O bend predicted in the normal Raman spectrum at 388 cm-1 Based on the measured and predicted normal Raman spectra the following SERS assignments are given 460 cm-1 to the CH3-P=O bend 544 cm-1 to the PO2CS wag 738 cm-1 to a C-S stretch (based on arguments above) 890 cm-1
to a C-C stretch 1101 cm-1 to a CH3 rock and 1456
cm-1 to a C-H bend The ability of SERS to measure chemical agents in water containing real-world chemical interferents was tested by using the generic tap water described in Table 1 The ability to reproduce measurements was accomplished by preparing three separate water stock solutions which were used to prepare three sample solutions each of 1 mgmL KCN in the generic tap water The samples defined as vials 1-9 were then measured at 4 points per vial in a semi-random fashion such that errors associated with stock solution preparation and errors associated with instrument drift could be identified No trends were apparent that signified such systematic errors Each spectrum collected consisted of 20 averaged scans taking 16 seconds at 8 cm-1 resolution The laser power at the sample was measured periodically during the day and it ranged from 102 to 105 mW spacccc
spot Vial 1 Vial 2 Vial 41 13754 13004 128192 13519 12692 12909
1 3 13541 12721 126394 13462 12648 12651
ave 13569 12766 12755Vial 6 Vial 3 Vial 5
1 11586 14068 10757 2 11236 14402 115122 3 11376 14546 11559
4 10894 11763 11214ave 11273 13695 11261
Vial 7 Vial8 Vial 91 11198 15804 11023
2 11243 15716 88753 3 11562 15216 11218
4 11655 15084 11077ave 11415 15455 10548
CN AVG STDEV ERR1 mgmL pk ht 12526 1560 1245
stock solution
SPIE-2003-5269 21
The CN measurements consisting of 36 data points produced an average height of 125 for the 2100 cm-1 peak with a standard deviation of 156 or 125 (Table 2) The HD and VX measurements were performed precisely the same way (Tables 3 and 4) For HD the 624 cm-1 peak was used for analysis and it had an average height of 53 with a standard deviation of 068 or 129 while for VX the 544 cm-1 peak was used for analysis and it had an average height of 1051 with a standard deviation of 308 or 293 The greater error in the VX measurements can be somewhat attributed to Vial 4 which produced lower SER signal intensities But removing this vial from the data set changes the standard deviation to 233 only a modest improvement
spot Vial 1 Vial 2 Vial 41 525 609 4682 45 675 484
1 3 527 644 5094 559 807 669
ave 51525 68375 5325Vial 6 Vial 3 Vial 5
1 521 574 575 2 536 53 4492 3 509 508 451
4 565 594 379ave 53275 5515 4635
Vial 7 Vial 8 Vial 91 457 589 413
2 583 497 4273 3 544 505 4
4 528 5 509ave 528 52275 43725
HD AVG STDEV ERR1 mgmL pk ht 530 068 1291
stock solution
spot Vial 1 Vial 2 Vial 41 1464 1034 4672 1485 99 61
1 3 1491 989 5684 1041 777 553
ave 137025 9475 5495Vial 6 Vial 3 Vial 5
1 1058 942 1293 2 697 121 965
2 3 727 1206 11124 689 1254 746
ave 79275 1153 1029Vial 7 Vial 8 Vial 9
1 1127 783 163 2 1358 812 1614
3 3 1371 908 15124 1097 875 1397
ave 123825 8445 153825
VX AVG STDEV ERR1 mgmL pk ht 1051 308 2925
stock solution
4 CONCLUSIONS In this paper we examined the ability of surface-enhanced Raman spectroscopy to reproducibly measure CN HD and VX in tap water without chemical interference Both normal and surface-enhanced Raman spectra were examined to select unique bands suitable to identify and quantify these chemical agents For SER measurements the 2100 cm-1 C-N stretch was used for CN the 624 cm-1 C-Cl stretch was used for HD and the 544 cm-1 PO2CS wag was used for VX It was determined that 1 mgmL samples of each of these chemicals measured 36 times in glass vials coated with a silver-doped sol-gel reproduced measurements with standard deviations of 125 129 and 293 It was further found that the 10 chemicals added to simulate generic tap water did not interfere with or alter the SER spectra It should be noted that the concentrations used in this study were considerably greater than those required by the JSAWM program Current work involves improving SER sensitivity and designing sampling systems with better reproducibility This includes the development of fractal silver and gold structures within the sol-gel matrix and the development of chemically selective sol-gels
ACKNOWLEDGMENTS The authors are grateful for the support of the US Army (DAAD13-02-C-0015 Joint Service Agent Water Monitor program) The authors would also like to thank Janet Jensen Ronald Crosier and Kristina Gonser for helpful discussions
Table 4 Measured SER peak heights for the VX band at 544 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
Table 3 Measured SER peak heights for the HD band at 624 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
SPIE-2003-5269 22
REFERENCES 1 Jensen JL NC Wong W R Loerop SPIE 4775-03 (2001) 2 JSAWM Requirements at wwwsbccomapgeaarmymilRDAecbcrtPRODSERJSAWMjsawmhtml 3 Erickson B ldquoThe Chemical Weapons Convention Redefines Analytical Challengerdquo Anal Chem 70 397A (1998) 4 Johnston RL Hoefler CM Fargo JC and Moberley B ldquoThe Defense Nuclear Agencyrsquos
ChemicalBiochemical Weapons Agreements Verification Technology Research Development Test and Evaluation Program and its Requirements for On-Site Analysisrdquo AT-ONSITE 5-8 (1994)
5 Black RM Clarke RJ Read RW and Reid MT ldquoApplication of GC-MS and GC-tandem MS to the analysis of chemical warfare samples found to contain residues of the nerve agent sarin sulphur mustard and their degradation productsrdquo J Chrom 662 301-321 (1994)
6 USA Army Center for Health Promotion and Preventive Medicine (USACHPPM) Technical Guideline 230A 7 Hoffland LD Piffath RJ Bouck JB rdquoSpectral signatures of chemical agents and simulantsrdquo Optical
Engineering 24 982-984 (1985) 8 Braue EHJ Pannella MGrdquoCIRCLE CELL FT-IR Analysis of Chemical Warfare Agents in Aqueous Solutionsrdquo
App Spectrosc 44 1513-1520 (1990) 9 Tseng C-H Mann CK and Vickers TJ ldquoDetermination of Organics on Metal Surfaces by Raman
Spectroscopyrdquo App Spectrosc 47 1767-1771 (1993) 10 Staff Reporter ldquoUS Army Tests Lidar to Detect Biological Toxinsrdquo Photonic Spectra pg 50 December 1998 11 Christesen SD Raman cross sections of chemical agents and simulants App Spectrosc 42 318-321 (1988) 12 Weaver MJ Farquharson S Tadayyoni MA ldquoSurface-enhancement factors for Raman scattering at silver electrodesrdquo J Chem Phys 82 4867-4874 (1985) 13 Alak AM and Vo-Dinh T ldquoSERS of Organophosphorous Chemical Agentsrdquo Anal Chem 59 2149-2153 (1987) 14 Lee Y Farquharson S ldquoRapid chemical agent identification by SERSrdquo SPIE 4378 21-26 (2001) 15 Farquharson S Maksymiuk P Ong K Christesen S ldquoChemical agent identification by surface-enhanced Raman spectroscopyrdquo SPIE 4577 166-173 (2001) 16 Spencer KM Sylvia J Clauson S and Janni J ldquoSurface Enhanced Raman as a Water Monitor for Warfare
Agents in Waterrdquo SPIE 4577 158-165 (2001) 17 Tessier P Christesen S Ong K Clemente E Lenhoff A Kaler E Velev O ldquoOn-line spectroscopic characterization of sodium cyanide with nanostructured gold surface-enhanced Raman spectroscopy substratesrdquo
App Spectrosc 56 1524-1530 (2002) 18 Farquharson S W W Smith S Elliott and J F Sperry Rapid biological agent identification by surface- enhanced Raman spectroscopy SPIE 3855110-116 (1999) 19 Farquharson S WW Smith YH Lee S Elliott and J F Sperry Detection of biological signatures A
comparison of electrolytic and metal-doped sol-gel surface-enhanced Raman media SPIE 4575 62-72 (2002) 20 Guzelian AA Sylvia J Janni J Clauson S and Spencer KM ldquoSERS of whole cell bacteria and trace levels of biological moleculesrdquo SPIE 4577 182-192 (2001) 21 Shende C Inscore F Gift A Maksymiuk P Farquharson S ldquoRapid dipicolinic acid extraction from Bacillus spores detected by surface-enhanced Raman spectroscopyrdquo App Spectrosc 58 accepted 22 Lee Y and Farquharson S ldquoSERS Sample Vials Based on Sol-Gel Process for Trace Pesticide Analysisrdquo SPIE
4206 140-146 (2000) 23 Farquharson S and Lee Y ldquoTrace Drug Analysis by SERSrdquo SPIE 4200-16 (2000) 24 Lee Y Farquharson S and Rainey P M Surface-Enhanced Raman Sensor for Trace Chemical Detection in
Water SPIE 3857 76-84 (1999) 25 Lee Y Farquharson S Kwong H and Shahriari M ldquoSol-Gel Chemical Sensor for Surface-Enhanced Raman
Spectroscopyrdquo SPIE 3537 252-260 (1998) 26 Farquharson S A Gift P Maksymiuk F Inscore and W Smith ldquopH dependence of methyl phosphonic acid
dipicolinic acid and cyanide by surface-enhanced Raman spectroscopyrdquo SPIE 5269-19 (2003) 27 Lide DR Ed Handbook of Chemistry and Physics CRC Press 77th Ed (1996-7) 28 Sosa C RJ Bartlett K KuBulat and WB Person ldquoA theoretical study of harmonic vibrational frequencies and
infrared intensities of XCH2CH2SCH2CH2X and XCH2CH2SH (= H Cl)rdquo J Phys Chem 93 577-588 (1993) 29 Hameka HF and JO Jenson ldquoComputer-generated predictions of the structure and the IR and Raman spectra of
VXrdquo ERDEC-TR-065 May 1993
- Chemagents Appendicespdf
-
- SPIE2001-4575-Bioagent-SERSapdf
-
- Detection of bioagent signatures A comparison of electrolytic and metal-doped sol-gel surface-enhanced Raman media
- Stuart Farquharson Wayne Smith and Yuan Lee
-
- Real-Time Analyzers 87 Church Street East Hartford CT 06108
-
- ABSTRACT
- 4 CONCLUSIONS
- 5 ACKNOWLEDGEMENTS
- 6 REFERENCES
-
SPIE 2001-4575
71
intensities for RNA extracted from E coli are very similar to those measured by normal Raman spectroscopy Although the relative percent that each of the NA bases contributed to each RNA sample was not determined reproducible band intensities allowed noting the following trends The percent adenine decreases while the combined percent cytosine and guanine increase for both B subtilis and S aureus compared to E coli Quality spectra were also obtained for the RNA samples by SG-SERS but only a few bands were readily identified Calculations of NA base concentrations by SG-SERS will require further research In light of recent events we summarize the capabilities of these SERS media in terms of rapid detection of B anthracis and dipicolinic acid However these capabilities must be qualified First and foremost the level to which SERS can distinguish bacteria or viruses has not yet been determined Development of a database of both DNA and RNA base concentrations for BWAs and common bacteria to establish the level of taxonomic identification is ongoing Second rapid collection of aerosol water or surface samples is being addressed by others who report trapping particles on filters from 100 liters of air per minute Third although not presented here we have developed methods to extract RNA or DNA from cells and spores for SER analysis within 10 minutes Finally we assume a detection limit of 3600 spores per 100 liters of air is required although a 50 lethal dosage of anthrax has not been established With these qualifications a mass detection limit for RNA using SERS is estimated as follows A single measurement is performed in ca 20 minutes (140 liters collected in 14 min RNA extracted in 8 min spectral acquisition and analysis in 10 min) The average human breaths 7 liters per minute therefore the analyzer must at the very minimum detect 5000 spores in 140 liters of air One spore is approximately 2x10-18m3 (1x1x2 microm) and if a density of 075 gcm3 is assumed this corresponds to a mass of 15x10-12g Each spore contains 4-12 RNA or 12x10-13g RNA for 8 If we assume 23 of the RNA can be isolated for analysis during lysis then the proposed instrument must be able to detect 4x10-10g RNA from 5000 spores per 70 liters of air within 10 minutes As noted above the mass detection limits for adenine were estimated at 16x10-11g and 12x10-10g for E-SERS and SG-SERS respectively Although these detection limits suggests that RNA from 5000 spores is detectable with the current instrumentation it is highly likely that only a portion of an RNA segment (eg 120-nucleotide 5S rRNA) is in contact with the metal surface and will contribute to the SER effect The SN for the RNA spectra were 110th of the average SN for the four individual RNA bases suggesting a 10 contribution Furthermore effective taxonomy will likely require knowing the NA base concentrations to 1 of the value (eg 25plusmn025) Again the average measurement error for the bases is 12 These values suggest that the E-SERS is within a factor of 4 of the required detection limit whereas the SG-SERS detection limit must be improved by 25 times The same arguments can be applied to the detection of dipicolinic acid If we assume a spore releases 10 by weight DPA during germination then the proposed instrument must be able to detect 75x10-10g DPA from 5000 spores per 70 liters of air within 10 minutes The detection limit for SG-SERS was estimated at 20x10-10g and suggest that the vials are suitable to perform a rapid screen for anthrax A series of concentration dependent measurements are currently being performed to verify this assertion Finally we note that the measurements performed here employed an FT-Raman spectrometer This instrumentation was chosen over dispersive Raman instrumentation for the high wavelength accuracy afforded by the HeNe reference laser (Connes Advantage43) which would allow reliable spectral subtraction matching of observed spectra to stored library spectra and confident use of chemometric approaches Such data analysis is likely to be required to enhance BWA identification However this instrumentation sacrifices sensitivity due to a less efficient detector (InGaAs vs Si) less efficient Raman scattering and less efficient generation of plasmon modes Substantial improvements can be obtained using visible excitation and Si detection and these measurements are underway
5 ACKNOWLEDGEMENTS The authors are grateful to Drs D Cookmeyer and S Tove of the US Army Research Office (Contract Number DAAH04-96-C-0078) for their interest and support of this research The authors would also like top acknowledge Dr R Yin and J Jensen for supporting development of the metal-doped sol-gels (Contract Number DAAD13-01-C-0019) They also thank Dr Wilfred H Nelson for assistance in spectral interpretations
SPIE 2001-4575
72
6 REFERENCES 1 Roberts WL and Rainey PM Clin Chem 39 1872-1877 (1993) 2 Pasechnik VA CC Shone and P Hambleton Bioseperations 3 267-283 (1993) 3 Jackson PJ ME Hugh-Jones DM Adair G Green KK Hill CR Kuske LM Grinberg FA Abramova and P
Keim Proc Natl Acad Sci 95 1224-1229 (1998) 4 Jensen JL NC Wong W R Loerop SPIE 4775-03 (2001) 5 Snyder A Pet al SPIE 3853-15 (1999) 6 Danosky T R and McFadden P N in press (1997) 7 Woodruff WH Farquharson S Science 201 831-833 (1978) 8 Lord RC and Thomas GJJr Spectrochemica Acta 23A 2551-2591 (1967) 9 Thomas GJJr Biochim Biophys Acta 213 417-423 (1970) 10 Placzek G Handbuch der Radiologie 2 EMarx ed Akademische Verlagagescellschatt Liepzig 1934 UCRL
Trans No 526 (1959) 11 Jeanmaire DL and RP Van Duyne J Electroanalytical Chem 84 1-20 (1977) 12 Chada S Manoharan R Moenne-Loccoz P Nelson WH Peticolas WL and Sperry JF Applied Spectroscopy
47 38-43 (1993) 13 Manoharan R Ghiamati E Chada S Nelson WH and Sperry JF Applied Spectroscopy 47 2145-2150 (1993) 14 Todd EA Morris MD Applied Spectroscopy 48 545-548 (1994) 15 Wentrup-Byrne E Sarinas S and Fredericks PM Applied Spectroscopy 47 1192-1197 (1993) 16 Weldon MK VR Zhelyaskov and MD Morris Applied Spectroscopy 52 265-269 (1998) 17 Kneipp K and J Fleming J Mol Structure 145 173-179 (1986) 18 Pothier NJ and Force RK Applied Spectroscopy 46 147-151 (1992) 19 Ervin KM E Koglin JM Sequaris P Valenta and HW Nurnberg J Electroanal Chem 114 179-194 (1980) 20 Kim SK TH Joo SW Suh and MS Kim J Raman Spectrosc 17 381-386 (1986) 21 Pothier NJ and Force RK Analytical Chemistry 62 678-680 (1990) 22 Kneipp K Y Wang RR Dasari and MS Feld Applied Spectroscopy 49 780-784 (1995) 23 Nie S and Emory SR Science 275 1102 (1997) 24 Graham D WE Smith AMT Linacre CH Munro ND Watson and PC White Analytical Chemistry 69 4703-
4707 (1997) 25 Fleischmann M PJ Hendra and AJ McQuillan Chem Phys Lett 26 163-166 (1974) 26 Pettinger B J Chemical Phys 85 7442-7451 (1986) 27 Surface-Enhanced Raman Scattering Section Four Theory SPIE MS 10 M Kerker and B Thompson Eds (1990) 28 Wang D-S and Kerker M SPIE (M Kerker and B Thompson Eds) MS 10 417-429 (1990) 29 Weaver MJ Farquharson S Tadayyoni MA J Chem Phys 82 4867-4874 (1985) 30 Norrod KL Sudnik LM Rousell D and Rowlen KL Applied Spectroscopy 51 994-1001 (1997) 31 Farquharson S Weaver WJ Lay PA Magnuson RH and Taube H J Am Chem Soc 105 3350-3351 (1983) 32 Lee PC and Meisel D ldquoAdsorption and Surface-Enhanced Raman of Dyes on Silver and gold Solsrdquo J Phys Chem
86 3391-3395 (1982) 33 Li Y-S and Wang Y Applied Spectroscopy 46 142-146 (1992) 34 Bello JM DL Stokes and T Vo-Dinh Analytical Chemistry 61 1779-1783 (1989) 35 van Duyne RP J C Hulteen D A Treichel M T Smith M L Duval and T R Jensen J Phys Chem B1033854-
3863 (1999) 36 Farquharson S and W W Smith W H Nelson and J F Sperry SPIE 3533-27 207-214 (1998) 37 Lee Y H W Smith S Farquharson H C Kwon M R Shahriari and P M Rainey SPIE 3537 252-260 (1998) 38 Lee Y-H S Farquharson and P M Rainey SPIE 3857 76-84 (1999) 39 Current Protocols in Molecular Biology Wiley Interscience1003-1006 (1987) 40 Farquharson S Smith W Carangelo R C and Brouillette C SPIE 3859 14-23 (1999) 41 Chase D B and JF Rabolt Fourier Transform Raman Spectroscopy Acad Press Ch1 p 131 (1994) 42 Brock TD MT Madigan JM Martinko and J Parker Biology of Microorganisms 7th Ed Prentice Hall p 76-80
(1994) 43 Connes J Rev Opt Theor Instrum 40 45 (1961)
166
Chemical agent identification by surface-enhanced Raman spectroscopy
Stuart Farquharson and Paul Maksymiuk
Real-Time Analyzers 87 Church Street East Hartford CT 06108
Kate Ong and Steven D Christesen US Army SBCCOM Aberdeen Proving Ground MD 21010
ABSTRACT The recent distribution of anthrax through the US postal system and the subsequent infection and death of several postal and national media employees amplifies the need for methods to rapidly detect identify and quantify this and other chemical and biological warfare agents The US military has also identified water supplies as a likely method of warfare agent deployment and is funding the development of a Joint Service Agent Water Monitor (JSAWM) In an effort to aid military personnel and the public at large we are developing a portable analyzer capable of identifying and quantifying chemical agents rapidly either on-demand or continuously The approach is based on the ability of Raman spectroscopy to identify molecular structure through the abundant vibration information provided in spectra and the ability of surface-enhanced Raman spectroscopy (SERS) to detect extremely low concentrations (eg part-per-billion) through the enhancement of Raman scattering by six orders of magnitude or more A key element to the analyzer design is a new SER active medium that is capable of quantitative reversible measurements The medium consists of silver or gold nanoparticles incorporated into a sol-gel matrix The porous silica network offers a unique environment for stabilizing SER active metals and the high surface area increases the interaction between the analyte and metal particles Here we present the use of new sol-gels that also selectively enhance chemicals based on polarity and charge Base-line measurements of chemical agents and their hydrolysis products are presented and compared to the JSAWM goal of 30 micrograms per liter detection Keywords Chemical warfare agent hydrolysis product SERS Raman spectroscopy sol-gel nanoparticle
1 INTRODUCTION Since September 11 2001 the threat of terrorist attacks and biological warfare within US borders has become a sobering reality The simplicity in manufacturing ease of deployment and the relatively low cost of chemical warfare agents (CWAs) raises public concern that they may also be used by terrorists Indeed terrorists released sarin (GB) in the Tokyo subway in 19951 Countering terrorism and terrorist attacks requires recognizing likely deployment scenarios and having the required technology to rapidly detect the deployment event One method of deployment has been long identified by the US military distribution through water supplies To counter this threat the Department of Defense is funding or monitoring the capabilities of 200 technologies with the goal of developing a Joint Service Agent Water Monitor (JSAWM) that is field portable2 The JSAWM must be capable of identifying the chemical agents with no false-positives and quantifying the agents at microgram per liter concentrations within 10 minutes (Table 1)3 This includes the analysis of drinking water supplies distribution and storage systems as well as potable water supplies The earliest technologies developed for CWA detection were based on electrochemical ionization or colorimetric analysis (eg phosgene tape) Although these analyzers were easy to use they were not generally agent specific and suffered from false-positives4 More traditional laboratory methods have also been investigated and in particular combined gas chromatography and mass spectrometry (GCMS) has been very successful at eliminating false-positives56 However GCMS requires extraction repeated calibration and long analysis times (typically 20 to 60 minutes)6 making it labor intensive and less than desirable for field use More rapid analysis of agents in the solid liquid and gas phase has been To whom correspondence should be addressed emailfarqureal-time-analyzerscom
Vibrational Spectroscopy-based Sensor Systems Steven D Christesen Arthur J Sedlacek III EditorsProceedings of SPIE Vol 4557 (2002) copy 2002 SPIE middot0277-786X02$1500
167
Table 1 Chemical Agent Structures Hydrolysis Half-lives and JSAWM Thresholds Agent Short-Hand Chemical Structure Hydrolysis
Half-Life JSAWM
Thresholds
Sarin (GB) F-[O=P-CH3]-O-CH(CH3)2 213 hours 32 microgL Soman (GD) F-[O=P-CH3]-O-CH(CH3)-(C-(CH3)3) 23 hours 32 microgL Tabun (GA) (CH3)2-N-[O=P-CN]-O-C2H5 41 hours 32 microgL VX C2H5O-[O=P-CH3]-S-(CH2)2-N-(CH(CH2)2)2 821 hours 32 microgL EA2192 HO-[O=P-CH3]-S-(CH2)2-N-(CH(CH2)2)2 gt9 years 32 microgL Mustard (H) ClCH2CH2-S-CH2CH2Cl encapsulates 47 microgL Lewisite (L) ClCH=CH-As-Cl2 rapid 27 microgL HCN HCN rapid 20 mgL BZ C7NH12-O-[C=O]-COH(C6H5)2 23 microgL T-2 Toxin 87 microgL at pH 7 to 75 and 20 to 25 oC
demonstrated by vibrational spectroscopy7-10 Hoffland et al7 reported infrared absorbance spectra and absolute Raman cross sections for several chemical agents while Christesen measured Raman cross sections for sarin tabun mustard gas and VX (ethyl S-2-diisopropylamino ethyl methylphosphonothioate)11 Again however these techniques also have limitations Raman spectroscopy is simply not a very sensitive technique and detection limits are typically 01 (1000 ppm) While infrared spectroscopy would have limited value in analyzing poisoned water since the very strong infrared absorption of water would obscure most other chemicals present Nevertheless efforts to overcome these limitations have been demonstrated Braue and Pannella8 quantified the G-series nerve agents (tabun sarin and soman) in terms of infrared attenuated total reflectance using a circle-cell And Alak and Vo-Dinh demonstrated the possibility of surface-enhanced Raman spectroscopy (SERS) to identify CWAs by measuring several organophosphonates that simulate the nerve agents12 However quantitative measurements have not been demonstrated for the SER-active material used (silver coated on alumina particles) or other SER-active media13 Recently we developed silver-doped sol-gels to promote the SER effect14-17 The porous silica network of the sol-gel matrix offers a unique environment for stabilizing SER-active metal particles and the sol-gel provides a high surface area that effectively increases the number of molecules observed within the Raman scattering volume The silver-doped sol-gels have been coated on the internal walls of standard glass vials to allow rapid SER analysis of small sample quantities (lt 01 mL) without preparation We have used p-aminobenzoic acid (PABA) as a test chemical to demonstrate surface enhancements greater than 106 reversible measurements in a flowing system reproducible measurements from vial-to-vial and batch-to-batch and measurements in multiple solvents including water14-17 Recently we used these vials to measure Tabun (GB) and Sarin and several hydrolysis products pinacolyl methyl phosphonate (PMP from Soman) and methyl phosphonic acid (MPA from all G-agents Figure 1) Although a number of unique vibrational bands are observed (eg C-N stretch doublet and P-C stretch) the G-agents were only observed for 5 concentrations and all spectra required baseline corrections Figure 1 Surface enhanced Raman spectra of ~5 vv A) Tabun and B) Sarin C) 1 vv PMP and D) 10 ppm MPA using sol-gel sample vials 785 nm excitation 1-min scan and CCD detection Performed at Aberdeen Proving Ground
C A
Wavenumber (∆cm-1) Wavenumber (∆cm-1)
D B
790
545 C-N
2135 2190 P-C 770
760 1290
Proc SPIE Vol 4577
168
Nevertheless MPA was readily observed for a 10 ppm sample with an estimated detection limit of 04 ppm (based on a signal-to-noise ratio of 3 for the 760 cm-1 band intensity) This measurement provides encouragement in that SERS may satisfy the needs of a JSAWM Furthermore MPA is also a hydrolysis product of VX and V-gas and EA2192 (Figure 2) and may prove a valuable indicator of agent usage Figure 2 Hydrolysis of Sarin to form hydrofluoric acid (HF) methylphosphonic acid 1-methylethyl ester (MPAMME) methyl phosphonic acid (MPA) and 2-propanol With this initial albeit modest success we began analyzing chemicals with various sol-gel compositions that we have been developing Here we describe four sol-gel compositions that select for 1) polar-positive 2) polar-negative 3) weakly polar-positive and 4) weakly polar-negative chemical species The ability of these sol-gels to select and enhance Raman scattering is described for several test chemicals and MPA
2 EXPERIMENTAL The chemicals analyzed as well as all chemicals used to prepare the metal-doped sol-gels were obtained at their purest commercially available grade from Aldrich (Milwaukee WI) The sol-gel designed to select for polar-negative species was prepared from a silver amine complex tetramethyl orthosilicate (TMOS) and methanol After mixing 02 mL of the sol-gel solution was transferred into a glass vial (2 mL) dried and heated The incorporated silver ions were then reduced using dilute sodium borohydride The vials were washed and dried prior to the addition of a sample solution In a similar manner the sol-gel designed to select for polar-positive species was prepared from a gold salt TMOS and methanol The sol-gel designed to select for weakly polar-negative species was prepared from a silver amine complex tetraethyl orthosilicate (TEOS) and methanol And the last sol-gel designed to select for weakly polar-positive species was prepared from a gold salt TEOS and methanol All samples were prepared in a chemical hood and transferred into plain or SER-active vials for analysis Normal Raman spectral measurements employed 1-mL pure samples that were placed in a 1-cm3 cuvette and weighed This yielded a powder density that allowed accurate calculation of molecules in the optical collection field SERS measurements employed 1-mg sample per mL water concentrations unless otherwise stated Once prepared a 01 mL sample was placed into one of the four selective sample vials which in turn was placed into the sample compartment of a Raman spectrometer for analysis A prototype Fourier transform Raman spectrometer (RTA) was used for these measurements18 The system consisted of a NdYAG laser (Brimrose) for excitation at 1064 nm an interferometer built by On-Line Technologies (OLT East Hartford CT) for frequency separation an uncooled InGaAs detector for signal detection (RTA) and an Intel 400 MHz Pentium II based laptop computer (Dell Round Rock TX) for interferometric control data acquisition (OLT) and analysis (LabVIEW by National Instruments Austin TX) Additional components included a Notch filter (Kaiser Ann Arbor MI) and interferometer entrance and exit optics (Edmund Scientific Barrington NJ) Fiber optics were used to deliver the excitation beam to the sample and the scattered radiation to the interferometer (1 meter lengths of 200 and 365 micron core diameter respectively Spectran Avon CT) A second Notch filter (Kaiser) was used as a beam splitter to direct the excitation beam along the same axis as the collected radiation A microscope object (20x04 Newport Irvine CA) was used to focus the beam into the sample and to collect the scattered radiation back along the same axis In this co-axial backscattering arrangement the excitation beam was passed through the outside of a glass vial and focused onto the silver-doped sol-gel film (01-03 mm thickness) containing the sample
3 RESULTS AND DISCUSSION p-aminobenzoic acid (PABA) and phenyl acetylene (PA) and were used to refine the selectivity and SER-activity of the four different metal-doped sol-gels PABA is a popular chemical used to evaluate the performance of SER-active media Here the polar end groups can be used to test selectivity of the polar-negative and polar-positive sol-gels PA is essentially non-
2O+ H HF + +OH
OH3H C
OP
OF
CH
CH3
3
3H C
OCP
OOH
CH
CH3
3
3H C
O
CPHO
CH
CH
3
3
C
Sarin MPAMME MPA 2-propanol
Proc SPIE Vol 4577
169
polar but a high electron density in the cylindrical π cloud around the carbon-carbon triple bond allows testing the selectivity of the weakly polar-negative and weakly polar-positive sol-gels As Figure 3 illustrates PABA passes through the polar sol-gel and is enhanced by either the silver or gold particles At 1 mgml the concentration of neutral PABA is ca 20 times that of the ionized form (pKa = 48) For electropositive silver the PABA anion is expected to interact through the carboxylate group and the associated vibrational modes are expected to dominate the spectrum Conversely for electronegative gold either form of PABA is expected to interact through the amine group The clear differences in our spectra support this expectation Furthermore bands at 840 and 1405 cm-1 assigned to a COO- bend and stretch respectively are significantly more intense for silver than gold Additional bands at 1140 and 1195 cm-1 are assigned to CH bending modes while bands at 1450 1500 and 1605 cm-1 are assigned to ring vibrational modes A very similar SER spectrum for PABA on a silver-coated alumina substrate has previously been reported with similar assignments19 For the gold-doped sol-gel new bands appear at 690 1355 and 1585 cm-1 The first band is assigned to a ring-H bending mode the second band to a ring-N- stretching mode and the third band to a possible NH2 scissors mode or ring mode The second band is not observed in the normal Raman spectrum but infrared bands occur at this frequency for aromatic ring-secondary amine stretching modes The scissors mode occurs at this frequency in Raman spectra for several chemicals but is absent in the PABA Raman spectrum Alternatively this mode may be the1600 cm-1 ring mode that has been shifted by the gold interaction Again a very similar SER spectrum of PABA has been reported but surprisingly using silver (colloids)2021 not gold as the enhancement medium These researchers also assumed the primary interaction of PABA with silver was through the carboxylate anion and made assignments accordingly For example they assigned the 1359 cm-1 to a COO- stretch not to the amine group as we have They also favor the ring stretching mode assignment for the 1582 cm-1 band Finally it should be said that other researchers have argued that the most dominant band in the SER spectra at 1450 cm-1 a ring vibration mode suggests that PABA lies flat on the surface and the π-orbitals dominate the surface interaction22
Figure 3 SER spectra of A) PABA using polar-negative and B) polar-positive sol-gels and C) PA using weakly polar-negative and D) weakly polar-positive sol-gels PABA is 1 mgmL PA is 1 vv Spectral conditions 75 mw 1064 nm 100 scans (15 min) 8 cm-1 resolution Non-polar PA passes through the non-polar sol-gels and is also enhanced by both metals The spectra are easily understood For electropositive silver PA interacts through the cylindrical triple bond π electron cloud and a -CequivC- doublet occurs near 2000 cm-1 The interaction is reasonably strong since this band appears at 2112 cm-1 in the normal Raman spectrum For electronegative gold this interaction is unlikely and only very weak bands occur near 2000 cm-1 The remaining bands are at 1000 cm-1 1200 cm-1 doublet and 1595 cm-1 all appear in the normal Raman spectra at virtually the same frequencies and are assigned to the symmetric ring-breathing mode CH bending modes and the trigonal ring-breathing mode respectively The polarnon-polar selectivity of the polar-negative and weakly polar-negative sol-gels was tested by adding a 11 molar mixture of PABA and PA The selective enhancement is quite good (Figure 4) The spectrum obtained using the polar sol-gel represents 78 PABA and 22 PA while the spectrum obtained using the weakly polar sol-gel represents 9 PABA and 91 PA The band peak intensities at 2000 cm-1 for PA and 1450 cm-1 for PABA were used for these calculations and are expanded in Figure 4 for clarity
C A
Wavenumber (∆cm-1) Wavenumber (∆cm-1)
2NH COOH
D B
C CH
Proc SPIE Vol 4577
170
Figure 4 SERS of 11 MM of PABA and PA in A) polar-negative and C) weakly polar-negative sol-gels The lower traces compare the pure chemicals B) 1 mgml PABA in polar-negative sol-gel and D) 1 PA in weakly polar-negative sol-gel while the insets magnify the minority species for clarity (x5 in A and x10 in B) Spectral conditions as in Figure 3 Following this development of selective sol-gels that maintained SER activity we measured cyanide and MPA (Figure 5) Not surprisingly the best sensitivity for both hydrolysis products was obtained using the polar-negative sol-gel The interaction of the cyanide anion with the silver surface is sufficient to shift the CequivN stretch observed at 2080 cm-1 in the normal Raman spectrum to 2145 cm-1 in the surface-enhanced Raman spectrum Furthermore the band is substantially broadened This anion has been extensively studied by electrolytic SERS and this shift and broadening have been attributed to the formation of a tetrahedral Ag(CN)3
2- surface structure23 Figure 5 Surface-enhanced (upper traces) and normal Raman spectra (bottom traces) of A) CN- and B) MPA in silver-doped TMOS SERS conditions as in Figure 3 and 1 mgmL Note MPA yields two distinct spectra for neutral (top) and acidic pH (middle) The normal Raman spectra employed pure powders 500 scans and 900 mW of 1064 nm SER measurements of MPA with the polar-negative sol-gel yielded two unique spectral signatures that depended on solution pH (Figure 5) For more neutral solutions the P-C stretch of MPA at 762 cm-1 dominates and the CH2 stretch at 2922 cm-1 appears The SN is sufficiently high that the anti-Stokes Raman shift at -762 cm-1 is observed For deprotonated MPA an oxygen-surface mode appears at 325 cm-1 (as well as its anti-Stokes complement) suggesting a strong interaction This results in substantial enhancement of the P-O-C mode at 1051 cm-1 (upper trace) Others report that this mode dominates the infrared spectra of nerve agents measured in water8 Comparison of the two spectra suggests the following molecule-to-surface orientations The appearance of the oxygen-surface and P-O-C modes in the upper spectrum of Figure 5B indicates that the tetrahedral molecule interacts with the silver surface through the deprotonated oxygen and is oriented end-on The
A B
Wavenumber (∆cm-1)
C A
Wavenumber (∆cm-1) Wavenumber (∆cm-1)
D B
Wavenumber (∆cm-1)
Proc SPIE Vol 4577
171
dominance of the P-C and the CH2 stretches and the disappearance of the P-O-C mode in the upper spectrum suggest the molecule is oriented with the methyl-hydroxyl-hydroxyl face towards the surface However considerably more research must be performed to verify these points Both chemical forms were measured as a function of concentration to determine the onset of monolayer coverage and estimate expected detection limits (Figure 6) Below monolayer coverage the signal to concentration dependence should be linear and the SN of any spectral measurement in this range can be used to predict the detection limit In the spectra presented here the peak height was used as the signal while the noise as root-mean-squared (RMS) was measured between 4400-4600 cm-1 Since noise is distributed evenly throughout the spectrum when transformed this region was used since it does not have any contributions from signals or baseline offsets Figure 6 shows a series of spectra for MPA along with a plot of the 1050 cm-1 band intensity (the noise was identical for this series) as a function of sample concentration A clear discontinuity appears in the vicinity of 01 mgmL (19 ppm) indicating the onset of monolayer coverage A detection limit defined as a SN of 3 was calculated for the 01 and 005gmL samples at 24x10-4 and 25x10-4 gL respectively A more modest detection limit of 101x10-4 gL was obtained using the 760 cm-1 band in the second series of concentration measurements These detection limits correspond to 46 and 190 ppb for a 3-min scan and 75 mW of incident laser power
Figure 6 A) Concentration dependence of MPA SERS measured in silver-doped TMOS) B) Concentrations are 001 005 01 05 1 gL (188 94 188 94 188 ppm) I760 series (bull) and I1050 series (∆)
Enhancement factors for cyanide and methyl phosphonic acid can be determined by estimating the number of molecules contributing to the surface-enhanced and normal Raman spectra The enhancement factor EF can be defined by the following equation
EF = (ISERSINR)bull(MNRMSERS) bull(PNRPSERS) bull(TNRTSERS)12
where I is the spectral band intensity M is the sample mass P is the incident laser power and T is the measurement time (or number of scans) for the two measurements For the normal Raman spectra a cylindrical scattering volume is assumed based on the laser area (28x10-7m2 6x10-4m diameter spot) and the penetration depth (1x10-3 m)24 The density of KCN and MPA as powders were measured at 0572 and 0516 gcm3 indicating that 16x10-4 and 144x10-4 g produced the normal Raman signals in Figure 5 respectively The number of molecules contributing to the SER spectra are those on the silver particles embedded in the sol-gel The total silver surface area can be determined from the average particle size concentration and the scattering volume Previous scanning electron micrographs showed the average silver particle size to be 40 nm in diameter (335x10-23m3)17 The silver concentration is 012M based on the reactant molar concentrations and dilution factors And the scattering volume is 76x10-
11m3 again based on a cylindrical scattering volume defined by a laser area of 28x10-7m2 and a sol-gel thickness of 27x10-
4m This volume contains 123x10-6g of silver equivalent to 35x109 silver particles with a collective surface area of 18x10-
5m2 However it may reasonably be assumed that at least half of the silver surface is in contact with the sol-gel matrix and unavailable for analyte interaction If we assume monolayer coverage and that each CN molecule occupies 15x10-20m2 then approximately 62x1014 molecules or 27x10-8g of CN contribute to the SER spectrum (20x10-19m2 46x1013 molecules 74x10-9g for MPA) Accordingly the EF for cyanide equals 48x104 ((180599) bull(16x10-427x10-8) bull(90075) bull(500100)12) The EF for MPA is considerably higher at 87x106 ((60326) bull(144x10-474x10-9) bull(90075) bull(500200)12)
0
20
40
60
80
100
120
140
0 02 04 06 08 1 12[MPA] (mgmL)
I (76
0)
0
100
200
300
400
500
600
I (10
50)
Wavenumber (∆cm-1)
A B
Proc SPIE Vol 4577
172
4 CONCLUSIONS Here we present for the first time surface-enhanced Raman spectra of Tabun and Sarin obtained using silver-doped sol-gels However the inferior enhancement suggested employing SER-active sol-gels varying in composition to improve both detection limits and selectivity of the target analytes To this end we successfully demonstrated the capabilities of four sol-gels that select for 1) polar-positive 2) polar-negative 3) weakly polar-positive and 4) weakly polar-negative chemical species p-aminobenzoic acid was used to show that silver could be used to attract polar-negative chemicals or functional groups (carboxylate anion) and that gold could be used to attract polar-positive chemicals (amine) while a mixture of p-aminobenzoic acid and phenylacetylene was used to show that tetramethyl orthosilicate preferentially solvates polar chemicals and that tetraethyl orthosilicate preferentially solvates weakly polar chemicals This increased sample control was applied to cyanide and methyl phosphonic acid two hydrolysis products of chemical warfare agents Exceptional results were obtained for methyl phosphonic acid allowing measurement of 1x10-2 gL for a 3-min scan and 75 mW of incident laser power with an estimated detection limit of 245x10-4 gL and an enhancement factor of 87x106 However this detection limit is 76 times less sensitive than required for the JSAWM (32x10-6gL for the G-agents) Finally we note that the measurements performed here employed an FT-Raman spectrometer This instrumentation was chosen over dispersive Raman instrumentation for the high wavelength accuracy afforded by the HeNe reference laser (Connes Advantage25) which would allow reliable spectral subtraction matching of observed spectra to stored library spectra and confident use of chemometric approaches Such data analysis is likely to be required to identify the chemical agents as well as distinguish hydrolysis products However this instrumentation which employs 1064 nm excitation and InGaAs detection sacrifices sensitivity We believe that the measurement sensitivity can be improved by at least two orders of magnitude by using shorter laser excitation wavelengths (eg 532 nm) This would provide more efficient Raman scattering (fourth power dependence on laser excitation wavelength) more efficient generation of plasmon modes and allow using more efficient detector material (Si vs InGaAs) These modifications are underway
5 ACKNOWLEDGEMENTS The authors would like to thank Dr R Yin and J Jensen of the US Army for supporting this work (Contract Number DAAD13-01-C-0019) They would also like to thank Advanced Fuel Research for making their laboratory facilities available
6 REFERENCES 1 Tu Anthony ldquoOverview of Sarin Terrorist Incidents in Japan in 1994 and 1995rdquo 6th CBW Protection Symposium
Stockholm Sweden 10-15 May 1998 2 Jensen JL NC Wong W R Loerop SPIE 4775-03 (2001) 3 JSAWM Requirements at wwwsbccomapgeaarmymilRDAecbcrtPRODSERJSAWMjsawmhtml 4 ldquoThe Chemical Weapons Convention Redefines Analytical Challengerdquo Analytical Chemistry News amp Features June 1
397A (1998) 5 Johnston RL Hoefler CM Fargo JC and Moberley B ldquoThe Defense Nuclear Agencyrsquos ChemicalBiochemical
Weapons Agreements Verification Technology Research Development Test and Evaluation Program and its Requirements for On-Site Analysisrdquo AT-ONSITE 5-8 (1994)
6 Black RM Clarke RJ Read RW and Reid MT ldquoApplication of gas chromatography-mass spectrometry and gas chromatography-tandem mass spectrometry to the analysis of chemical warfare samples found to contain residues of the nerve agent sarin sulphur mustard and their degradation productsrdquo J Chromatography 662 301-321 (1994)
7 Hoffland LD Piffath RJ Bouck JBrdquoSpectral signatures of chemical agents and simulantsrdquo Optical Engineering 24 982-984 (1985)
8 Braue EHJ Pannella MGrdquoCIRCLE CELL FT-IR Analysis of Chemical Warfare Agents in Aqueous Solutionsrdquo Applied Spectroscopy 44 1513-1520 (1990)
9 Tseng C-H Mann CK and Vickers TJ ldquoDetermination of Organics on Metal Surfaces by Raman Spectroscopyrdquo Applied Spectroscopy 47 1767-1771 (1993)
10 Staff Reporter ldquoUS Army Tests Lidar to Detect Biological Toxinsrdquo Photonic Spectra pg 50 December 1998
Proc SPIE Vol 4577
Proc SPIE Vol 4577
173
11 Christesen SD Raman cross sections of chemical agents and simulants Applied Spectroscopy 42 318-321 (1988) 12 Alak AM and Vo-Dinh T ldquoSERS of Organophosphorous Chemical Agentsrdquo Analytical Chemistry 59 2149-2153
(1987) 13 Norrod KL Sudnik LM Rousell D and Rowlen KL ldquoQuantitative Comparison of Five SERS Substrates
Sensitivity and Detection Limitrdquo Applied Spectroscopy 51 994-1001 (1997) 14 Lee Y and Farquharson S ldquoSERS Sample Vials Based on Sol-Gel Process for Trace Pesticide Analysisrdquo SPIE 4206
140-146 (2000) 15 Farquharson S and Lee Y ldquoTrace Drug Analysis by Surface-Enhanced Raman Spectroscopyrdquo SPIE 4200-16 (2000) 16 Lee Y Farquharson S and Rainey P M Surface-Enhanced Raman Sensor for Trace Chemical Detection in Water
SPIE 3857 76-84 (1999) 17 Lee Y Farquharson S Kwong H and Shahriari M ldquoSurface-Enhanced Raman Sensor for Surface-Enhanced Raman
Spectroscopyrdquo SPIE 3537 252-260 (1998) 18 Farquharson S Smith W Carangelo R C and Brouillette C ldquoIndustrial Raman Providing Easy Immediate Cost
Effective Chemical Analysis Anywhererdquo SPIE 3859 14-23 (1999) 19 Narayanan VA JM Bello JD Stokes and T Vo-Dinh Analusis 19 307-310 (1991) 20 Laserna JJ E L Torres and JD Winefordner Analytica Chemica Acta 469-480 (1987) 21 Torres EL and JD Winefordner Analytical Chemistry 59 1626-1632 (1987) 22 Suh JS DP DiLella M Moskovits J Phys Chem 87 1540-1544 (1983) 23 Benner RE R Dornhaus R Chang and BL Laube Correlations in the Raman spectra of cyanide complexes adsorbed
at silver electrodes with voltammograms Surface Science 101 341 (1980) 24 Chase D B and JF Rabolt Fourier Transform Raman Spectroscopy Acad Press Ch1 p 131 (1994) 25 Connes J Rev Opt Theor Instrum 40 45 (1961)
APPLIED SPECTROSCOPY 351
focusing the 488 nm laser beam 2 mm deep into thebulk of the crystals to avoid contributions from potentialdepletion layers As shown in Fig 1b the dependence ofthe Raman shift on the degree of deuteration is almostperfectly linear and ts very well with D 5 22684cmR1 24526 where D is the degree of deuteration (in )and R is the spectral mean of the PO4 vibration in cm21A linear correlation coef cient of 0998 indicates an ex-cellent linear dependence of the Raman peak shift withdegree of deuteration This result shows that the shift ofthe PO 4 peak is simply caused by the linear increase inatomic mass due to isotope substitution which decreasesthe length of hydrogen-like bonds
This excellent linear dependence allows us to map thepro le of the DH exchange layer at the surface of DKDPcrystals by acquiring Raman spectra and determining theposition of the PO 4 peak for various depths This methodis preferable over other methods such as determining thestrength of the OD vibration directly (eg at 715 cm21)because the position of the most intense peak in the Ra-man spectrum can be measured more precisely than theintensity of some of the weakest peaks in the spectrumThis is demonstrated in Fig 2 where depth-dependentRaman spectra (Fig 2a) and the resulting exchange layerpro les for two DKDP crystals are shown (Fig 2b) Thespectra in Fig 2a were obtained from a depth scan of aDKDP crystal with 75 degree of deuteration in thebulk grown at 45 8C The spectra start out as DKDP with30 deuteration close to the surface and approach thebulk DKDP spectrum within a few micrometers of depthThe fact that the relative degree of deuteration does notextend to 0 D is due to the limited depth resolution ofthe Raman microprobe which averages over 4 mm indepth Figure 2b depicts the resulting DH exchange layerpro les for this and a second crystal grown at 63 8Crespectively Both crystals had the same exposure to am-bient conditions and their main difference is the temper-ature at which they were grown The different exchangelayer pro les indicate that crystals grown at differenttemperatures have differen t proton conductiv ities 13
which leads to a difference in their rate of deuteriumdepletion The parameters controlling this behavior arecurrently the objective of a detailed study the results ofwhich will be reported elsewhere
CONCLUSION
In conclusion we have shown that the shift of the to-tally symmetric PO 4 stretch mode in the Raman spectrumof DKDP crystals scales linearly with degree of deuter-ation This allows us to correlate Raman peak positionsto deuteration levels in these crystals We have presenteda new technique to determine DH diffusion pro les inDKDP frequency conversion crystals based on micro-Ra-man spectroscopy This technique is fast inexpensiveand works under various environmental conditionswhich will allow us to better understand and control deu-terium depletion in DKDP crystals
ACKNOWLEDGMENTS
We would like to thank M Runkel for rst discovering DKDP crack-ing R Floyd for providing DKDP crystals and L Chase and A Burn-ham for their support and helpful discussions This work was performedunder the auspices of the US Department of Energy by the University
of California Lawrence Livermore National Laboratory under contract W-7405-Eng-48
1 J J De Yoreo A K Burnham and P K Whitman Int Mat Rev47 113 (2002)
2 C E Barker R A Sacks B M Van Wonterghern J A Caird JR Murray J H Campbell K Kyle R B Ehrlich and N DNielsen Proc SPIE-Int Soc Opt Eng 2633 501 (1995)
3 T Suratwala paper to be published4 Cleveland Crystals Inc httpwwwclevelandcrystalscom
KDPshtmltable5 E A Popova I T Savatinova and I A Velichko Sov Phys Solid
State 12 1543 (1971)6 I P Kaminow R C C Leite and S P S Porto J Phys Chem
Solids 26 2085 (1965)7 J A Subramony B J Marquardt J W Macklin and B Kahr
Chem Mat 11 1312 (1999)8 H Tanaka M Tokunaga and I Tatsuzaki Solid State Commun
49 153 (1984)9 R J Nelmes G M Meyer and J E Tibballs J Phys C 15 59
(1982)10 M A Yakshin D W Kim Y S Kim Y Y Broslavets O E
Sidoryuk and S Goldstein Laser Physics 7 941 (1997)11 I Takenaga Y Tominaga S Endo and M Kobayashi Solid State
Commun 84 931 (1992)12 C Krenn personal communication13 M Sharon and A K Kalia J Solid State Chem 21 171 (1977)
Rapid Dipicolinic Acid Extractionfrom Bacillus Spores Detectedby Surface-EnhancedRaman Spectroscopy
STUART FARQUHARSON ALAND GIFT PAUL MAKSYMIUK andFRANK E INSCOREReal-Time Analyzers Inc East Hartford Connecticut06108
Index Headings Dipicolinic acid Bacillus spores Anthrax Surface-enhanced Raman spectroscopy
INTRODUCTION
The anxiety caused by the distribution of anthrax en-dospores through the US postal system in October 2001was exacerbated by the long time required for positiveidenti cation of the Bacillus anthracis spores and the un-known extent of their distribution Since that time manymethods capable of rapid eld analysis have been inves-tigated to augment or replace the laboratory method ofgrowing microorganisms in culture media which takesdays to perform12 Prominent among these approachesare polymerase chain reactions (PCR)3 immunoassaysand detection of calcium dipicolinate as a biochemicalsignature PCR employs primers to separate organism-
Received 10 October 2003 accepted 14 November 2003 Author to whom correspondence should be sent
352 Volume 58 Number 3 2004
speci c nucleic acid sequences (eg capsular protein en-coding gene for Bacillus anthracis)4 and polymerases toamplify the segment until it is detectable Recently am-pli cation times have been substantially reduced andcomplete analysis can now be performed in an hour orless Immunoassay methods are also being developed thatuse competitive binding of the bioagent (as an antigen)and its labeled conjugate for a limited number of anti-bodies Although analyses can be performed in under 30minutes a well-de ned anthrax antigen has not yet beenidenti ed5ndash7 and consequently the false-positive rate isunacceptably high8
A number of other methods are being developed witha focus on the detection of calcium dipicolinate (CaDPA)and its derivatives as a B anthracis signature This is sobecause only spore-forming bacteria contain CaDPA andthe most common potentially interfering spores such aspollen and mold spores do not Relatively fast methodshave been developed to chemically extract CaDPA andthen detect it directly by uorescence9 or indirectly byluminescence1011 In the latter case hot dodecylamine(DDA) has been used to extract dipicolinic acid (DPA)and terbium has been utilized to form a highly lumines-cent DPA complex11 Although measurements have beenperformed in as little as ve minutes it was found thatas many as three concentration-dependent complexes canform each with different lifetimes This coupled withthe fact that the Tb31 cation produces the same lumines-cence spectrum makes determinations of low spore con-centrations problematic
It has been long known that Raman spectra of Bacillispores are dominated by bands associated with CaDPA12
and that these spectra may provide a suitable anthrax sig-nature at the genus level13 Since that time considerableimprovements in Raman instrumentation have led to lab-oratory measurements of single Bacilli spores14 and to eld measurements of spores captured from a mail-sort-ing system15 However the single spore measurementsrequired complex instrumentation that is not ruggedwhile the eld measurements required milligrams of sam-ple Furthermore the Raman spectra of both measure-ments contained uorescence contributions that would in-crease uncertainty in quanti cation
In related research we demonstrated that nanogramquantities of DPA could be detected by uorescence-freesurface-enhanced Raman spectroscopy (SERS)16 We alsodemonstrated that microliter volumes of chemicals canbe detected by SERS using metal-doped sol-gel-packedglass capillaries17 Towards the goal of developing a rap-id eld SERS-based anthrax spore detector we havecombined our previous research and we now report thatDPA can be extracted from a 10 mg B cereus spore sam-ple using DDA in 1 minute and can be detected by SERSin an additional 1 minute
EXPERIMENTAL
Dipicolinic acid (26-pyridinedicarboxylic acid DPA)and dodecylamine (DDA) were used as received fromSigma-Aldrich (Milwaukee WI) Lyophilized B cereusspores prepared according to the literature13 were sup-plied by the University of Rhode Island and used as re-ceived Multiple particles approximately 01 mm3 each
were separated and weighed at 5 to 15 mg representing05 to 15 million spores The sample masses were con-sistent with a previous determination of spore density at0081 gmL that indicated a high degree of entrained air
All chemicals used to prepare the silver-doped sol-gelcoated capillaries were also obtained and used as receivedfrom Sigma-Aldrich According to previously publishedprocedures17 two precursor solutions were preparedmixed and then drawn into 1-mm-diameter glass capil-laries The silver amine precursor consisted of a 51 vvratio of 1 N AgNO 3 to 28 NH3OH while the alkoxideprecursor consisted of a 21 vv ratio of methanol to te-tramethyl orthosilicate The alkoxide precursors weremixed with silver amine precursor in an 81 vv ratioApproximately 015 mL was drawn into the capillarycoating a 15-mm length After sol-gel formation the in-corporated silver ions were reduced with dilute sodiumborohydride which was followed by a water wash to re-move residual reducing agent
A 100 mL drop of a 50 mM DDA solution in ethanolpre-heated to 78 8C was added to each of the B cereusparticles to digest the spore coat After 1 minute the re-sultant solution was drawn into a SER-active capillarythat was immediately xed horizontally to an XY posi-tioning stage (Conix Research Spring eld OR) just in-side the focal point of an f 07 aspheric lens The lensfocused the beam into the sample and collected the scat-tered radiation back along the same axis A dichroic lter(Omega Optical Brattleborough VT) was used to re ectthe excitation laser to the lens and pass the Raman scat-tered radiation collected by the lens An f 2 achromatwas used to collimate the laser beam exiting a 200-mm-core-diameter source ber optic while a second f 2 ach-romat was used to focus the scattered radiation into a 365mm ber optic (Spectran Avon CT) A short-pass lterwas placed in the excitation beam path to block the sil-icon Raman scattering generated in the source ber fromre ecting off sampling optics and reaching the detectorA long-pass lter was placed in the collection beam pathto block the sample Rayleigh scattering from reachingthe detector A 785 nm diode laser (Process InstrumentsInc model 785-600 Salt Lake City UT) was used todeliver 100 to 150 mW of power to the sample A Fouriertransform Raman spectrometer (Real-Time Analyzersmodel IRA-785 East Hartford CT) and a silicon photo-avalanche detector (Perkin Elmer model C30902S Stam-ford CT) were used to acquire the SER spectra
RESULTS AND DISCUSSION
As an initial experiment the SER spectrum of 1 gLof DPA in water was measured using the newly devel-oped silver-doped sol-gel-coated capillaries (Fig 1A) Atthis concentration a high signal-to-noise ratio (SN) isobtained in 1 min In fact a reasonable spectrum is ob-tained in the same time frame for 1 mgL (Fig 1B) TheSER spectra are reasonably similar to the normal Raman(NR) spectrum obtained for a saturated solution of DPAin 1 N KOH (Fig 1C) and the following band shifts areobserved (NR to SER) 647 to 657 cm21 817 to 815cm21 998 to 1008 cm21 1384 to 1382 cm21 1434 to1428 cm21 and 1569 to 1567 cm21 Many of these bandshave been previously assigned1213 such as 998 cm21 to
APPLIED SPECTROSCOPY 353
FIG 1 SERS of DPA in water using silver-doped sol-gel-coated glasscapillary for (A) 1 gL and (B) 1 mgL (C ) NR of saturated DPA in 1N KOH in a glass capillary Spectral conditions (A) and (B) 150 mWof 785 nm 1-min acquisition time (C) 450 mW of 785 nm 5-minacquisition time both 8 cm 21 resolution
FIG 2 SERS of DPA extracted from 10 mg B cereus particle using100 mL of 50 mM hot DDA acquired in (A) 1 minute and (B) 2 seconds(C ) Attempted SERS of 50 mM hot DDA in ethanol using silver-dopedsol-gel-coated glass capillary acquired in 1 min Spectral conditions150 mW of 785 nm 8 cm21 resolution
the symmetric ring stretch 1384 cm21 to the OndashCndashOsymmetric stretch 1428 cm21 to the symmetric ring CndashH bend and 1569 cm21 to the asymmetric OndashCndashOstretch
The rst B cereus samples consisted of 2 mg of sporesin 2 mL of 5 mM hot DDA The samples were main-tained at 78 8C for 40 min and while hot approximately10 mL was drawn into a SER-active capillary Since spec-tra of DPA were obtained for these initial samples small-er spore masses higher DDA concentrations and shorterheating periods were examined In due course it wasfound that 10 mg of spores could be digested by 100 mLof 50 mM hot DDA in one minute and detected (Fig2A) In fact the signal was suf ciently intense that it canbe observed in as little as two seconds (Fig 2B) Theamount of DPA that was extracted was estimated to bebetween 5 and 10 mgL by comparing the signal intensityof the 1008 cm21 band to that measured for DPA in waterThis is consistent with previous research that found thatthe majority of the DPA is extracted from spores usingDDA11 and that B cereus spores contain approximately10 DPA by weight18 The SN of 127 for the 1008 cm21
band in the 1-minute SER spectrum suggests a limit ofdetection of approximately 250 ng of B cereus sporesbased on a SN of 3 Finally it should be noted that DDAdid not produce a detectable SER spectrum as shown inFig 2C
CONCLUSION
We have demonstrated that by combining rapid extrac-tion of dipicolinic acid from Bacillus cereus spores withchemical identi cation by surface-enhanced Raman spec-troscopy as little as 10 mg of spores can be detected Infact the entire measurement from the time of adding hotdodecylamine to the spores to the time when the dipi-colinic acid SER spectrum is acquired and analyzedcould be performed in less than two minutes The abilityof this method to distinguish between spore-forming bac-teria such as Bacillus anthracis and non-DPA containingpowders could help prevent costly shutdowns associated
with the appearance of suspicious material or intentionalmailing of common substances as an anthrax hoax Thismethod could also prove useful in detecting the locationof anthrax endospores in mail distribution facilities if an-other veri ed attack should occur
Research continues to fully characterize the surface-enhanced Raman spectroscopy signal intensities as afunction of sample concentration and to explore otherextractants that do not require the use of elevated tem-perature
ACKNOWLEDGMENTS
The authors are grateful for the support of the National Science Foun-dation (DMI-0296116 and DMI-0215819) and the US Army(DAAD13-02-C-0015 Joint Service Agent Water Monitor program)The authors are indebted to Chetan Shende for preparing the sol-gelcapillaries The authors also thank James Gillespie Nicholas Fell andAugustus Fountain for providing important background informationMark Farquharson for laboratory support and Professor Jay Sperry ofthe University of Rhode Island for supplying B cereus spores
1 V A Pasechnik C C Shone and P Hambleton Bioseparations 3267 (1993)
2 P J Jackson M E Hugh-Jones D M Adair G Green K K HillC R Kuske L M Grinberg F A Abramova and P Keim ProcNatl Acad Sci USA 95 1224 (1998)
3 B R Glick and J J Pasternak Molecular Biology Principles andApplications of Recombinant DNA (ASM Press Washington DC1994)
4 C A Bell J R Uhl T L Had eld J C David R F Meyer TF Smith and F R Cockerill III J Clin Microbiol 40 2897(2002)
5 D L Gatto-Menking H Yu J G Bruno M T Goode M Millerand A W Zulich Biosens Bioelectron 10 501 (1995)
6 J J Quinlan and P M Foegeding J Rapid Methods AutomationMicrobiol 6 1 (1998)
7 A A Hindle and E A H Hall Analyst (Cambridge UK) 1241599 (1999)
8 M S Ascher US Department of Health amp Human Services(httpwwwhhsgovophppresentationsAscherdoc)
9 R Nudelman B V Bronk and S Efrima Appl Spectrosc 54 445(2000)
10 D L Rosen C Sharpless and L B McBrown Anal Chem 691082 (1997)
354 Volume 58 Number 3 2004
11 P M Pellegrino N F Fell Jr and J B Gillespie Anal ChimActa 455 167 (2002)
12 W H Woodruff T G Spiro and C Gilvarg Biochem BiophysRes Commun 58 197 (1974)
13 E Ghiamati R S Manoharan W H Nelson and J F SperryAppl Spectrosc 46 357 (1992)
14 A P Esposito C E Talley T Huser C W Hollars C M Schal-dach and S M Lane Appl Spectrosc 57 868 (2003)
15 S Farquharson L Grigely V Khitrov W W Smith J F Sperryand G Fenerty J Raman Spectrosc paper accep ted (2003)
16 S Farquharson W W Smith S Elliott and J F Sperry SPIE-IntSoc Opt Eng 3855 110 (1999)
17 S Farquharson and P Maksymiuk Appl Spectrosc 57 479(2003)
18 F W Janssen A J Lund and L E Anderson Science (Washing-ton DC) 127 26 (1958)
SPIE -2003-5269 117
pH dependence of methyl phosphonic acid dipicolinic acid and cyanide by surface-enhanced Raman spectroscopy
Stuart Farquharson Alan Gift Paul Maksymiuk Frank Inscore and Wayne Smith
Real-Time Analyzers 87 Church Street East Hartford CT 06108
ABSTRACT US and Coalition forces fighting terrorism in Afghanistan and Iraq must consider a wide range of attack scenarios in addition to car bombings Among these is the intentional poisoning of water supplies to obstruct military operations To counter such attacks the military is developing portable analyzers that can identify and quantify potential chemical agents in water supplies at microgram per liter concentrations within 10 minutes To aid this effort we have been investigating the value of a surface-enhanced Raman spectroscopy based portable analyzer In particular we have been developing silver-doped sol-gels to generate SER spectra of chemical agents and their hydrolysis products Here we present SER spectra of methyl phosphonic acid and cyanide as a function of pH an important factor affecting quantitation measurements which to our knowledge has not been examined In addition dipicolinic acid a chemical signature associated with anthrax-causing spores is also presented Keywords Chemical warfare agents agent detection agent hydrolysis SERS Raman spectroscopy homeland security
1 INTRODUCTION In the past decade the Unites States and its allies have been challenged by a different kind of warfare exemplified by the terrorist attacks of September 11 2001 Suicide bombings and the use of chemical agents are the norm and military personnel must consider a wide range of attack scenarios Among these is the intentional poisoning of water supplies to obstruct military operations in Afghanistan and Iraq Currently colorimetric paper is used to detect agents on-site while gas chromatography combined with mass spectrometry (GCMS) is used in mobile support laboratories However both methods have severe drawbacks The paper changes color in response to contact with many chemicals besides CWAs causing a high incidence of false positives1 while GCMS although very chemically specific requires hours to perform and constant re-calibration234 Military operations would be greatly aided by a portable analyzer that can identify and quantify potential chemical agents at concentrations that impact safety This includes the analysis of drinking water supplies distribution and storage systems To meet this goal the Department of Defense has been investigating numerous approaches under the auspices of the Joint Service Agent Water Monitor (JSAWM) program5 The JSAWM must be capable of identifying the chemical agents with no false-positives and quantifying the agents at microgram per liter concentrations within 10 minutes678 Recently we and others have been investigating the ability of surface-enhanced Raman spectroscopy (SERS) to measure chemical agents9-12 bioagents13-17 and their hydrolysis products in water SERS employs the interaction of surface plasmon modes of metal particles with target analytes to increase scattering efficiency by as much as 1 million times18 In 1987 the potential of this approach to measure CWAs was demonstrated by measuring a series of pesticides19 Several of these organophosphonates have chemical structures similar to CWAs in particular P=O functional groups In our studies we have been employing silver-doped sol-gels to promote the SER effect The porous silica network of the sol-gel matrix offers a unique environment for immobilizing and stabilizing SER-active metal particles20-23 The sol-gels have been coated on the internal walls of standard glass vials to allow rapid SER analysis of small sample quantities without preparation We have measured over 100 chemicals with enhancements of 104 to 106 demonstrated reversible measurements in a flowing system reproducible measurements from vial-to-vial and measurements in multiple solvents including water20-23 Previously we used these vials to perform preliminary measurements of cyanide (CN) methylphosphonic acid (MPA) and dipicolinic acid (DPA) MPA is a hydrolysis product of the nerve agents (eg sarin Reaction 1) and may be a valuable indicator of nerve agent usage particularly since the alkyl methylphosphonic acids are relatively more stable than their corresponding parent complexes24 DPA is
SPIE -2003-5269 118
a chemical signature of spore forming bacteria such as Bacillus anthracis And in light of the inability to rapidly detect the anthrax spores distributed through the US mail in October 2001 a number of methods are being developed to extract and analyze this signature Reaction 1 Stepwise hydrolysis of Sarin to form hydrofluoric acid (HF) isopropyl methylphosphonic acid (IMPA) then methyl phosphonic acid (MPA) and 2-propanol In our previous SERS investigations MPA and DPA were measured at 50 and 100 mgL respectively In both cases limits of detection (LOD) were estimated at 100 microgL providing encouragement in that SERS may satisfy the needs of the JSAWM Since it has been shown that pH can substantially influence the intensity of SER bands25 which would clearly influence quantitative analysis we undertook the present study to determine the severity of these effects for cyanide methyl phosphonic acid and dipicolinic acid Furthermore we previously observed a band at 1050 cm-1 for MPA6 possibly due to an anion formed at basic pH Here we investigate the source of this spectral anomaly
2 EXPERIMENTAL All chemicals including potassium cyanide methyl phosphonic acid dipicolinic acid and those used to prepare the silver-doped and gold-doped sol-gel coated vials were obtained and used as received from Sigma-Aldrich All samples were prepared in HPLC grade water (Fischer Scientific Fair Lawn NJ) for SERS measurements The pH of these samples was adjusted using dilute nitric acid or potassium hydroxide and verified using a pH electrode (Corning Inc Corning NY) that had been calibrated with pH 400 700 and 1000 buffered standards from Fischer Scientific Once prepared the samples were transferred into the silver-doped sol-gel vials (Real-Time Analyzers Inc East Hartford CT) The vials were coated in a manner similar to that previously reported by adding ammonium hydroxide to a solution of silver nitrate tetramethyl orthosilicate (TMOS) and methanol20 Gold-vials were coated by adding nitric acid to a solution of gold tetrachloride TMOS and methanol The two precursor solutions were prepared mixed and transferred to 2-mL glass vials dried and heated After sol-gel formation the incorporated metal ions were reduced with dilute sodium borohydride (1mgmL) which was followed by a water wash to remove residual reducing agent After the resultant analyte solution was introduced the SER-active vial was immediately fixed horizontally to an XY positioning stage (Conix Research Springfield OR) just inside the focal point of an f07 aspheric lens The lens focused the beam into the sample and collected the scattered radiation back along the same axis A dichroic filter (Omega Optical Brattleborough VT) was used to reflect the excitation laser to the lens and pass the Raman scattered radiation collected by the lens An f2 achromat was used to collimate the laser beam exiting a 200 microm core diameter source fiber optic while a second f2 achromat was used to focus the scattered radiation into a 365 microm fiber optic (Spectran Avon CT) A short pass filter was placed in the excitation beam path to block the silicon Raman scattering generated in the source fiber from reflecting off sampling optics and reaching the detector A long pass filter was placed in the collection beam path to block the sample Rayleigh scattering from reaching the detector A 785 nm diode laser (Process Instruments Inc model 785-600 Salt Lake City UT) was used to deliver 100 to 150 mW of power to the sample A Fourier transform Raman spectrometer (Real-Time Analyzers model IRA-785 East Hartford CT) and a silicon photo-avalanche detector (Perkin Elmer model C30902S Stamford CT) were used to acquire the SER spectra
3 RESULTS AND DISCUSSION In a previous study of MPA6 aimed at developing a concentration calibration curve and determining limits of detection (LOD) we observed an anomaly at 1050 cm-1 Since it was found that the band intensity changed as a function of concentration the band must be associated with a sample parameter Two possible parameters photon flux and pH are examined here The first parameter was investigated by irradiating a 1mgmL MPA sample in a SER-active vial with laser powers of 200 mW and above and monitoring spectral changes It was immediately found that the 1050 cm-1 band
2O+ H HF + +OH
OH3H C
OP
OF
CH
CH3
3
3H C
OCP
OOH
CH
CH3
3
3H C
O
CPHO
CH
CH
3
3
C
Sarin IMPA MPA 2-propanol
SPIE -2003-5269 119
appeared and grew as a function of time and that the higher the laser power the faster the growth Figure 1 shows the growth of the 1050 cm-1 band over the course of 30 minutes when using 150 mW of 785 nm excitation while Figure 2A shows that the growth can be fit with a first order exponential equation namely I1050 = 03+05e-013t Furthermore the 760 cm-1 band could be fit with a first order decay equation with an identical exponential rate constant ie I760 = 08-
08e-013t The rates represent classical first order kinetics and their correspondence allows one to conclude that MPA is being transformed one-for-one into a photo-generated product At this time the photoproduct has not been positively identified but phosphonic (phosphorous) acid and phosphonate are likely candidates since the symmetric P(OH)3 stretch occurs at ~1050cm-1 Our studies show that a reduction in laser power to 100 mW at the sample for MPA essentially eliminates this degradation process This laser power or lower was used for further measurements
Methyl phosphonic acid is a diprotic acid that stepwise dissociates into two anions MPA- and MPA= according to the following reactions26
MPA MPA- + H+ pKa1 = 212 Reaction 2
MPA- MPA= + H+ pKa2 = 729 Reaction 3 The relative concentrations of MPA MPA- and MPA= can be determined at any pH by expressing [MPA] and [MPA=] in terms of [MPA-] using Reactions 2 and 3 and summing all three to equal the total starting concentration here 2 mgmL (0021M MW = 9602) viz
[MPA] + [MPA-] + [MPA=] = 0021M Equation 1 substituting from Reactions 1 and 2
([H+][MPA-])K1a + [MPA-] + (K2a[MPA-])[H+] = 0021M Equation 2 rearranging [MPA-] = 0021M(1+[H+]K1a + K2a[H+]) Equation 3 The relative concentrations of MPA MPA- and MPA= as a function of pH are shown in Figure 3 It is worth noting that near neutral pH both MPA- and MPA= will be present To confirm that the SER signal followed this pH dependence a starting solution consisting of 20 mg of MPA in 10 mL HPLC grade water was prepared and brought to pH of 20 using dilute nitric acid From this solution 2 mL were added to a SER-active vial and the SER spectrum recorded At this pH a peak at 760 cm-1 was barely discernable The 2 mL solution was returned to the starting solution and the pH was re-measured to correct for any changes that the silver-doped sol-gel vials might cause In most cases the change was less than 02 pH units and the pH is reported as the before and after average Next the pH of the
Figure 2 A) Exponential growth of 1050 cm-1 band and B) exponential decay of 760 cm-1 band for spectral series in Figure 1
Figure 1 Growth of 1050 cm-1 band as a function of time due to exposure to 150 mW of 785 nm Spectra are 5 sec each collected every 100-sec from 0 to 30-min
B
570 770 970 1170 1370Raman Shift (cm-1)
Arbit
rary
Unit
s
MPA Photodegradation
600 800 1000 1200 1400 Wavenumber (cm-1)
A
0 10 20 30 0 10 20 30 time (min) time (min)
Ram
an In
tens
ity (r
elat
ive)
30
min
0
SPIE -2003-5269 120
starting solution was adjusted to 325 using dilute KOH Again 2 mL were added to a vial and the SER spectrum recorded At this pH a reasonably strong 760 cm-1 band was observed This process was repeated as spectra were recorded at pHs of 70 74 75 79 85 and 100 A total of 1 mL of KOH was added diluting the total concentration by 10 Next the pH of the starting solution was made acidic by adding dilute nitric acid dropwise This time spectra were recorded at pHs of 72 69 64 and 37 Figure 4 shows the SER spectra for representative pHs (spectra were left out to simplify the figure) while Figure 3 shows the 760 cm-1 peak intensities as a function of pH (The band intensities were adjusted to compensate for dilution effects caused by the addition of HNO3 and KOH then normalized to 0021 M for the most intense band observed at pH 37) It is clear from Figure 3 that the 760 cm-1 band follows the MPA- concentration as a function of pH and must be assigned to this anion No bands were observed that corresponded to MPA or MPA= The lack of an MPA SER spectrum may be due to the absence of an attraction between the neutral analyte and the electropositive silver surface The same reasoning suggests that a strong SER spectrum should be observed for MPA= but it is not and a satisfactory explanation has not been found
0000
0005
0010
0015
0020
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14pH
Con
cent
ratio
n [M
]
pK1 = 212 pK2 = 729
MPA- MPA=
MPA
Figure 4 SER spectra of 002M MPA as a function of pH Conditions 100 mW of 785 nm 36 scans (1 min) 8 cm-1 recorded 2 min after sample introduction pH 19 and 20 not apparent on this scale pH 69 and 74 near identical to 70 and 75 and not shown for clarity
00
02
04
06
08
10
12
14
16
18
0 5 10 15 20 25 30Measurement Number
Ram
an In
tens
ity (7
60 c
m-1
)
Figure 3 Concentration dependence of MPA MPA- and MPA= as a function of pH for a 002M sample Intensity of 760 cm-1
band from Figure 6 as a function of pH ( for increasing basic adjustment for increasing acidic adjustment error was measured at ~10 for pH 64)
Figure 5 SER spectra of 002M MPA at pH 64 measured around a vial at three heights (9 points per height) Conditions as in Figure 4 but 10-sec scans
Since these measurements involved the removal and replacement of the SER-active vial in the sample holder to remove and add sample variation in the intensity as a function of vial position was minimized by illuminating the exact same height along the vial wall But this does not account for variability of the SERS response of the sol-gel coating around the vial To analyze this effect a vial containing MPA at pH 64 was rotated at ~ 40o
intervals at the original height and 18rdquo above and below this value Figure 5 shows the intensity of the 760 cm-1
band for the 27 positions It was found that the average value was 137plusmn014 an RSD of 10 overall and 5 for each height An error bar is included in Figure 4 for the pH 64 measurement
SPIE -2003-5269 121
With the above analyses in mind a preliminary investigation of the SER spectral response for MPA (as MPA-) as a function of concentration was performed A single vial was used for these measurements beginning with 1 mgL followed by measurements of 10 100 and 1000 mgL In all cases the pH was ~7 and 3 positions around the vial were measured per concentration Since the 760 cm-1 band was not observed for concentrations of 1 or 10 mgL using 100 mW of 785 nm the laser power at the sample was raised to 200 mW beginning with the 10 mgL concentration Photo-degradation was largely avoided (and not observed) by exposing the sample for only 33 seconds per spectral acquisition Representative spectra for 10 100 and 1000 mgL are shown in Figure 6 while a plot of the 760 cm-1 band intensity as a function of concentration is shown in Figure 7 These values were also used to estimate limits of detection based on the signal-to-noise ratio (SN) of the 760 cm-1 band using 100 mW of 785 nm laser excitation and a 10-min acquisition time As summarized in the Figure 7 inset the lower the measured concentration the lower the predicted LOD Even if as estimated 210 microgL could be measured using the silver-doped sol-gel vials an improvement of a factor of 70 is still required to meet JSAWM goals of measuring 3 microgL in 10-minutes Similar to MPA DPA is a diprotic acid (pKa1 = 216 and pKa2 = 692) and variations in pH will effect the relative concentrations of DPA DPA- and DPA= and possibly the SER spectra and band intensities This could prove significant if an acid or base is used to denature anthrax spores with the goal of extracting and analyzing DPA The MPA pH study described above was mimicked for DPA except that the starting solution consisted of 20 mg of DPA in 20 mL HPLC grade water (60x10-3M MW = 1671) The initial solution had a pH of 245 which was made basic by dilute KOH to pHs of 355 433 487 559 1069 and 1166 SER spectra were recorded at each pH using 100 mW of 785 nm and a 44-sec acquisition time Next one drop of concentrated nitric acid was used to remake the solution acidic at a pH of 200 Again sequential pH measurements were performed at 383 510 735 and 822 The solution pH was made acidic a third time but to pH 219 171 then 135 Throughout this process no more than 20 drops of acid or base were added and therefore the concentration was diluted by no more than 10 Most of the spectral bands showed a minor decrease in intensity as a function of increasing pH values However the bands at 525 and 795 cm-1 showed the most dramatic changes which occurred at acid pH Figure 8 shows the SER spectra of DPA for the spectral region and pH range of interest The identity of the DPA species was determined by plotting the normalized peak intensities with the lowest value set to 0 and the highest to 0006 M as a function of pH and overlaying these values on a plot of the relative concentrations for DPA DPA- and DPA= as previously done for MPA (Figure 9) As can be seen the 525 cm-1 band clearly corresponds to DPA The correspondence of the 795 cm-1 band to this species is less clear as the band retains intensity until pH of 55 This can be attributed to contribution to the overlapping band at 810 cm-1 which does not change as a function of pH The fact that most bands are observed at all pHs suggest that the primary interaction with silver is through the ring nitrogen This is supported by the fact that the most intense band occurs at 1008 cm-1 attributed to a symmetric ring breathing mode and that this interaction has been characterized for pyridine in numerous papers27
0
1
2
3
4
5
6
7
0 200 400 600 800 1000 1200
MPA Concentration (mgL)
760
Ban
d In
tens
ity
Figure 6 SER spectra of MPA in water at A) 1000 B) 100 and C) 10 mgL Conditions pH of 7 silver-doped sol-gel coated vial 200 mW of 785 nm 33-sec 8 cm-1 resolution
Figure 7 Plot of SER intensity of 760 cm-1 band of MPA as a function of concentration using 200 mW of 785 nm Inset table includes average intensity LOD standard deviation and percent deviation for each concentration but for 100 mW and 10 min
conc (mgL) LOD ave stddev dev10 021 002 1021100 072 010 14301000 312 040 1280
A
B C
SPIE -2003-5269 122
The ability to detect dipicolinic acid (DPA) by SERS is demonstrated in Figure 10 Even at 1 mgL the primary bands are visible The intensity of the dominant ring breathing mode observed at 1008 cm-1 for DPA in water (pH = 25 to 55) is plotted as a function of concentration in Figure 11 These values were also used to estimate limits of detection based on the SN of the 1008 cm-1 band using 100 mW of 785 nm laser excitation and a 10-min acquisition time Again the lower the measured concentration the lower the predicted LOD (see Figure 11 inset) and detection of 160 microgL is possible Gastrointestinal anthrax requires significant more spores than inhalation anthrax28 and a limit of detection might be placed at 1 million spores in 1 liter of water or 10 microgL Since each spore contains ~10 CaDPA by weight29 a goal for DPA might be 1 microgL indicating that the present measurements must be improved by nearly two orders of magnitude Due to the increased hazards of handling HCN gas KCN salt was used for these experiments Nevertheless all sample preparations were performed in a chemical hood KCN completely dissolves in water but its conjugate acid HCN is formed and has a Ka of 615x10-1030 viz
HCN CN- + H+ pKa = 921 Reaction 4
Figure 8 SER spectra of 1 mgmL DPA as a function of pH Conditions 100 mW 785 nm 100 scans (44-sec)
Figure 9 Concentration dependence of DPA DPA- and DPA= as a function of pH for a 0006M sample Intensity of 525 () and 795 (diams) cm-1 bands from Fig 8 as a function of pH
pH
135 171 219 383
0
0001
0002
0003
0004
0005
0006
0007
0 2 4 6 8 10 12 14pH
Con
cent
ratio
n [M
]
DPADPA-DPA=795525
DPA DPA=DPA-
pK2 = 692pK1 = 216
795 525
0
05
1
15
2
25
0 200 400 600 800 1000 1200
DPA Concentration (mgL)
1008
Ban
d In
tens
ity
conc (mgL) lod-10min-100mw1 017
10 016100 103
1000 355
Figure 10 SER spectra of DPA in water at A) 1000 B) 100 C) 10 and D) 1 mgL Conditions pH of 25-55 silver-doped sol-gel coated vial 175 mW of 785 nm 1-min 8 cm-1 D) has been multiplied by x10 to make bands visible
A
B C
D
Figure 11 Plot of SER intensity of 1008 cm-1 band of DPA as a function of concentration using 175 mW of 785 nm Inset table includes LOD in mgL for each concentration but for 100 mW and 10 min
SPIE -2003-5269 123
Consequently the cyanide concentration must be determined for each initial KCN concentration Specifically the samples prepared with concentrations of 01 1 10 100 and 1000 mgL of KCN produced CN- concentrations of 63x10-3 033 69 89 and 964 mgL at pHs of 816 90 967 102 and 107 respectively The pH dependence for the HCN and CN- concentrations are shown in Figure 12 Thus as the amount of KCN added to the solution decreases so does the pH of the solution (becomes less basic) and according to Reaction 4 the relative amount of CN- to HCN also decreases For example in the preparation of a 01 mgL solution of KCN the pH is shifted from 7 for pure water to only 816 and only 63 of the starting material becomes CN- or 63x10-3 mgL In comparison for a solution of 1000 mgL the pH is shifted from 7 to 107 and 96 of the starting material becomes CN- This is significant since the cyanide ion is better able to adsorb onto the silver particles and become SERS active SER spectra of 10 100 and 1000 mgL of KCN samples are shown in Figure 13 and a corresponding plot of the 2100 cm-1 band intensity as a function of concentration is shown in Figure 14
000102030405060708091011
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14pH
Con
cent
ratio
n [m
gm
L]
pKa = 921
CN -HCN
The SER spectra of cyanide are dominated by the single CN stretch at 2100 cm-1 which occurs in normal Raman spectra of solutions at 2080 cm-1 However a low frequency mode occurs at 135 cm-1 due to a Ag-CN stretch (not shown) indicative of a strong surface interaction It is also observed that as the concentration decreases the CN stretch shifts to 2140 cm-1 This shift has been attribute to the formation of a tetrahedral Ag(CN)3
2- surface structure31 as well
Figure 14 Concentration dependence of KCN SERS measured under conditions in Fig 11 Concentrations are 1 01 and 001 mgml Intensities are measured for the CN stretch at 2100 cm-1 Inset table includes LOD in mgL for each concentration in Figs 13 and 15 but for 100 mW and 10 min
Figure 13 SER spectra of KCN in water at A) 1000 B) 100 and C) 10 mgL Conditions pHs of 107 102 and 97 silver-doped sol-gel coated vial 100 mW of 785 nm 1-min 8 cm-1 C) has been multiplied by x10 to make band visible
Figure 12 Concentration dependence of HCN and CN- as a function of pH for a 1 mgmL sample Calculated intensity of 2100 () cm-1 band for a 1 mgml sample at pHs of 816 90 967 102 and 107
A
B
C
0
20
40
60
80
100
120
140
0 200 400 600 800 1000 1200CN Concentration (mgL)
2100
Ban
d In
tens
ity
Figure 15 SER spectra of KCN in water at A) 10 B) 1 and C) 01 mgL Conditions pHs of 97 12 and 12 gold-doped sol-gel coated vial 100 mW of 785 nm 1-min 8 cm-1 C) has been multiplied by x10 to make band visible
A
B
C
conc (mgL) lod-10min-100mw Condition01 001 Au-pH 121 007 Au-pH 12
69 003 Au-pH 9769 016 Ag-pH 9789 022 Ag-pH 102946 113 Ag-pH 107
SPIE -2003-5269 124
as to CN adsorbed to two different surface sites32 Alternatively the 2140 cm-1 band could be attributed to HCN since this species dominates at lower concentrations However it is unlikely that this species would be attracted to the electropositive silver surface Further both peaks should be present at pHs between 85 and 105 but this is not observed It has also been suggested that at concentrations near and above monolayer coverage the CN- species is forced to adsorb end-on due to crowding and at lower concentrations the molecule can reorient to lie flat33 This suggests that the 2100 and 2140 cm-1 bands correspond to the end-on and flat orientations respectively As Figures 13 and 14 show the intensity of the CN stretch for the 89 mgL sample is nearly as intense as the 964 mgmL band This suggests that the Raman signal for the flat orientation is more enhanced However more extensive measurements are required to verify this point Since resent research has suggested that cyanide may be more effectively detected on gold measurements of KCN solutions were also performed using gold-doped sol-gel vials Preliminary measurements are shown in Figure 15 for samples prepared from 01 1 and 10 mgL KCN Since the pHs are 816 90 967 the resultant CN- concentrations are 63x10-3 033 and 69 mgL Initially only the highest concentration was observed and the signal intensity was significantly better than the equivalent concentration measured using silver In an effort to shift Reaction 4 to the left transforming HCN to CN- (Le Chatelierrsquos principle) KOH was added to the lower concentration samples producing solutions with pH 12 This effectively forces all of the cyanide in solution to be CN- or 01 and 10 mgL respectively More importantly the CN stretch is now observed in the SER spectra The band appears at 2125 cm-1 as has been previously reported for gold12 As calculated for MPA and DPA LODs can be estimated from this data For the three concentrations of cyanide on silver the LODs are 016 to 11 mgL for 100 mW of 785 nm laser excitation and a 10-min acquisition time For gold pH adjusted the LODs are10 to 70 microgL an improvement of more than 10 times silver Nevertheless either substrate is sufficient to meet the JSAWM goals of measuring 3 mgL in 10-minutes as the requirements form cyanide are much less stringent than the nerve agents
4 CONCLUSIONS Here we examined the surface-enhanced Raman spectral response for methyl phosphonic acid dipicolinic acid and cyanide as a function of pH It was determined that the most prominent peak at 760 cm-1 reaches a maximum intensity between pH 3 and 7 and corresponds to the MPA- species Neither the MPA nor MPA= species appear to generate a SER spectrum and consequently no spectra were observed below pH 2 or above pH 8 In this study we also found that higher laser powers could cause photodegradation of MPA signified by the exponential growth of a band at 1050 cm-1 which is tentatively assigned to phosphorous acid Unlike MPA DPA was observed at all pHs This is attributed to the dominant interaction of the pyridine functional group with silver Minor spectral changes were observed at acid pHs and were assigned to neutral DPA Like MPA SER spectra of cyanide were pH dependent No spectra were observed for the HCN species while CN- was best observed at pHs more basic than 8 Preliminary concentration studies for the three analytes allowed estimating limits of detection for MPA DPA and CN using 100 mW of 785 nm and a 10-min acquisition time of 210 165 and 70 microgL respectively Although the latter value suggests that cyanide can be measured at sufficiently low concentrations to meet JSAWM goal improvements by 100 to 200 times are required for MPA and DPA It is clear from this study that pH of the sample is important and must be taken into account when developing concentration calibration curves as well as fieldable analyzers Future work will include tailoring the sol-gel with specific functional groups to dictate sample pH and thereby optimize sensitivity
ACKNOWLEDGMENTS The authors are grateful for the support of the US Army (DAAD13-02-C-0015 Joint Service Agent Water Monitor program) The authors would also like to thank Dr Steve Christensen of the US Army SBCCOM for helpful discussions and Mr Chetan Shende of Real-Time Analyzers for assistance in development of the gold-doped sol-gels
REFERENCES 1 Erickson B Analytical Chemistry News amp Features June 1 397A (1998)
SPIE -2003-5269 125
2 Johnston RL Hoefler CM Fargo JC and Moberley B AT-ONSITE 5-8 (1994) 3 Black RM RJ Clarke RW Read and MT Reid J Chromatography 662 301-321 (1994) 4 USA Army Center for Health Promotion and Preventive Medicine (USACHPPM) Technical Guideline 230A 5 Jensen JL NC Wong W R Loerop SPIE 4775-03 (2001) 6 Hoenig SL Handbook of Chemical Warfare and Terrorism Greenwood Press Wesport CT (2002) 7 Munro NB SS Talmage GD Griffin LC Waters AP Watson JF King and V Hauschild Env Health
Persp 107 933-974 (1999) 8 Holstege CP Kirk M Sidell FR Crit Care Clin13 923-42 (1997) 9 Farquharson S P Maksymiuk K Ong and S Christesen SPIE 4577 166-173 (2001) 10 Lee Y and S Farquharson SPIE 4378 21-26 (2001) 11 Spencer KM J Sylvia S Clauson and J Janni SPIE 4577 158-165 (2001) 12 Tessier P S Christesen K Ong E Clemente A Lenhoff E Kaler and O Velev Applied Spectroscopy 56
1524-1530 (2002) 13 Farquharson S WW Smith S Elliott and JF Sperry SPIE 3533 207-214 (1998) 14 Farquharson S WW Smith S Elliott and JF Sperry SPIE 3855110-116 (1999) 15 Farquharson S WW Smith YH Lee S Elliott and JF Sperry SPIE 4575 62-72 (2002) 16 Guzelian AA J Sylvia J Janni S Clauson and KM Spencer SPIE 4577 182-192 (2001) 17 Shende C F Inscore A Gift P Maksymiuk and S Farquharson in press 18 Weaver MJ S Farquharson and MA Tadayyoni J Chem Phys 82 4867-4874 (1985) 19 Alak AM and T Vo-Dinh Analytical Chemistry 59 2149-2153 (1987) 20 Lee Y and S Farquharson SPIE 4206 140-146 (2000) 21 Farquharson S and Y Lee SPIE 4200-16 (2000) 22 Lee Y S Farquharson and P M Rainey SPIE 3857 76-84 (1999) 23 Lee Y S Farquharson H Kwong and M Shahriari SPIE 3537 252-260 (1998) 24 Wang J M Pumera G Collins and A Mulchandani Analytical Chemistry 74 6121-6125 (2002) 25 Dou X YM Jung Z-Q Cao and Y Ozaki Applied Spectroscopy 53 1440-1447 (1999) 26 Data supplied by S Christesen and K Ewing 27 Kerker M and B Thompson Eds SPIE MS 10 (1990) 28 Inglesby TV DA Henderson JG Bartlett JAMA 287 2236 (2002) 29 FW Janssen AJ Lund and LE Anderson Science 127 26 (1958) 30 Lide DR Ed Handbook of Chemistry and Physics CRC Press 77th Ed (1996-7) 31 Billmann J G Kovacs and A Otto Surf Sci 92 153 (1980) 32 Murray CA and S Bodoff Phys Rev B 32 671 (1985) 33 Kellogg D and J Pemberton J Phys Chem 91 1120 (1987)
SPIE-2003-5269 16
Chemical agent detection by surface-enhanced Raman spectroscopy
Stuart Farquharson Alan Gift Paul Maksymiuk Frank Inscore and Wayne Smith
Real-Time Analyzers 87 Church Street East Hartford CT 06108
Kevin Morrisey and Steven D Christesen US Army SBCCOM Aberdeen Proving Ground MD 21010
ABSTRACT
In the past decade the Unites States and its allies have been challenged by a different kind of warfare exemplified by the terrorist attacks of September 11 2001 Although suicide bombings are the most often used form of terror military personnel must consider a wide range of attack scenarios Among these is the intentional poisoning of water supplies to obstruct military operations in Afghanistan and Iraq To counter such attacks the military is developing portable analyzers that can identify and quantify potential chemical agents in water supplies at microgram per liter concentrations within 10 minutes To aid this effort we have been investigating the value of a surface-enhanced Raman spectroscopy based portable analyzer In particular we have been developing silver-doped sol-gels to generate SER spectra of chemical agents and their hydrolysis products Here we present SER spectra of several chemical agents measured in a generic tap water Repeat measurements were performed to establish statistical error associated with SERS obtained using the sol-gel coated vials Keywords Chemical agents chemical agent detection SERS Raman spectroscopy
1 INTRODUCTION
In the autumn of 2001 terrorism within US borders became a sobering reality While extensive efforts are being implemented to secure the homeland US and Coalition forces in Afghanistan and Iraq are constantly faced with terrorist attacks In addition to car-bombings the military has identified several non-traditional attack scenarios including poisoning of water supplies by chemical warfare agents (CWAs) To counter this threat the Department of Defense has been investigating numerous approaches to detect such attacks under the auspices of the Joint Service Agent Water Monitor (JSAWM) program1 The JSAWM must be capable of identifying the chemical agents with no false-positives and quantifying the agents at microgram per liter concentrations within 10 minutes2 This includes the analysis of drinking water supplies distribution and storage systems Currently colorimetric paper is used to detect agents on-site while gas chromatography combined with mass spectrometry (GCMS) is used in mobile support laboratories However both methods have severe drawbacks The paper changes color in response to contact with many chemicals besides CWAs causing a high incidence of false positives3 while GCMS although very chemically specific requires up to an hour to perform and regular re-calibration456 Vibrational spectroscopy has also been investigated as a potential method of rapidly detecting CWAs7-
11 as early as 1985 in the case of infrared spectroscopy8 and 1988 in the case of Raman spectroscopy11 Again however these techniques also have limitations when it comes to measuring trace poisons in water Infrared spectra would be dominated by the very strong absorption of water which would obscure absorptions by most other chemicals present Whereas Raman spectroscopy is simply not a very sensitive technique and detection limits are typically grams per liter Surface-enhanced Raman spectroscopy (SERS) offers several advantages over conventional vibrational methods and that may provide the necessary sensitivity required for detecting trace quantities of chemical agents in water SERS employs the interaction of surface plasmon modes of metal particles with target analytes to increase scattering efficiency by as much as 1 million times12 In 1987 the potential of this approach to measure CWAs was demonstrated by measuring a series of pesticides13 Several of these organophosphonates have chemical structures similar to CWAs in particular P=O functional groups In the past few years we and others have further explored the ability of SERS to detect CWAs14-17 and even bioagents 18-21 We have been employing silver-doped sol-gels to promote the SER effect
SPIE-2003-5269 17
in these studies The porous silica network of the sol-gel matrix offers a unique environment for immobilizing and stabilizing SER-active metal particles22-25 The silver-doped sol-gels have been coated on the internal walls of standard glass vials to allow rapid SER analysis of small sample quantities without preparation We have measured over 100 chemicals with enhancements of 104 to 106 demonstrated reversible measurements in a flowing system reproducible measurements from vial-to-vial and measurements in multiple solvents including water21-25 Previously we used these vials to perform preliminary measurements of cyanide (CN) and methylphosphonic acid (MPA) Most of the nerve agents form MPA during hydrolysis while Tabun forms CN a chemical agent in its own right In another paper including in these proceedings we examined the limits of detection (LOD) for MPA by measuring a series of concentrations down to 50 mgL and estimated a limit of detection of 100 microgL26 These measurements provide encouragement in that SERS may satisfy the needs of the JSAWM To further establish the viability of SERS in particular silver-doped sol-gels here we present analysis of cyanide mustard and VX in tap water The measurements performed at the US Armyrsquos Edgewood Chemical Biological Center Aberdeen MD also included numerous repeat measurements to establish reproducibility
2 EXPERIMENTAL 2a General All chemicals including potassium cyanide 2-chloroethylethyl sulfide and those used to prepare the silver-doped and gold-doped sol-gel coated vials were obtained and used as received from Sigma-Aldrich All samples were prepared in a chemical hood using HPLC grade water unless otherwise noted (Fischer Scientific Fair Lawn NJ) for SERS measurements Once prepared the samples were transferred into the silver-doped sol-gel vials (Real-Time Analyzers Inc East Hartford CT) The vials were coated in a manner similar to that previously reported by adding ammonium hydroxide to a solution of silver nitrate tetramethyl orthosilicate (TMOS) and methanol22 The two precursor solutions were prepared mixed and transferred to 2-mL glass vials dried and heated After sol-gel formation the incorporated metal ions were reduced with dilute sodium borohydride (1mgmL) which is followed by a water wash to remove residual reducing agent After the resultant analyte solution was introduced the SER-active vial was fixed horizontally to an XY positioning stage (Conix Research Springfield OR) just inside the focal point of an f07 aspheric lens The lens focused the beam into the sample and collected the scattered radiation back along the same axis A dichroic filter (Omega Optical Brattleborough VT) was used to reflect the excitation laser to the lens and pass the Raman scattered radiation collected by the lens An f2 achromat was used to collimate the laser beam exiting a 200 microm core diameter source fiber optic while a second f2 achromat was used to focus the scattered radiation into a 365 microm fiber optic (Spectran Avon CT) A short pass filter was placed in the excitation beam path to block the silicon Raman scattering generated in the source fiber from reflecting off sampling optics and reaching the detector A long pass filter was placed in the collection beam path to block the sample Rayleigh scattering from reaching the detector A 785 nm diode laser (Process Instruments Inc model 785-600 Salt Lake City UT) was used to deliver 100 to 150 mW of power to the sample A Fourier transform Raman spectrometer (Real-Time Analyzers model IRA-785 East Hartford CT) and a silicon photo-avalanche detector (Perkin Elmer model C30902S Stamford CT) were used to acquire the SER spectra
2b Edgewood Chemical Biological Center The surface-enhanced Raman spectral measurements at the US Armyrsquos Chemical Biological center presented here were all performed on September 12 2003 To expedite measurements a plate was machined to hold up to 12 SER-active sample vials (Figure 1) The plate fit a standard XY plate reader that could be programmed Pure KCN bis-(2-chloroethyl)sulfide (distilled mustard gas HD) and ethyl S-2-diisopropylamino ethyl methylphosphonothioate (VX) were obtained on-site and used to prepare 1 gL tap water solutions in a chemical hood with appropriate safety equipment Simulated tap water was prepared by adding 10 chemicals most often found in tap water at appropriate concentrations to distilled water (Table 1) SER measurements were also performed in a chemical hood For added safety the FT-Raman instrument was placed outside the laboratory and 30 foot fiber optic and electrical cables were used to allow remote SERS measurements and plate manipulation For each experiment 1gL samples were prepared and added to 9 individual vials which were then loaded on the plate In some cases a tenth vial was included as a blank
Table 1 Generic Tap Water Composition
Compound mgL NaHCO3 100 CaSO4 27 MgSO4bull7H2O 67 NaNO3 1 Fulvic Acid 1 K2HPO4 07 KH2PO4 03 (NH4)2HSO4 001 NaCl 001 FeSO4 0001 pH 76-78
SPIE-2003-5269 18
A software program was written that allowed selecting the sequence that the vials were measured the number of positions along the length of the vials to measure (1 to 5) and the number of scans to co-add During sample analysis the program displayed the vial being analyzed the point being analyzed and the spectrum as it was being acquired Once all the data was collected a second software program was written to rapidly analyze the data The spectra collected for all the vials on a plate could be loaded at one time and then the spectra for each point could be displayed simultaneously or separately The user could then select the Raman peak to analyze in terms of peak height or area This was accomplished by selecting points on either side of the peak to define a baseline of zero The peak height or area could then be computed for all of the spectra loaded and then exported to a spreadsheet for statistical analysis
Figure 1 A) Vial Holder 6 slots to hold 2 vials each end-to-end B) Measurement Configuration Program user selects vials to measure sequence number of points per vial (1 to 5) and number of scans per point C) Spectral Acquisition Program shows spectrum being collected which vial and position D) Spectra Analysis Program user selects spectra to analyze by plate vial and point (s) as well as two wavenumbers defining the peak and the baseline to subtract The image is of 5 repeat measurements of 10 mgL KCN in generic tap water 16 sec each 100 mW of 785 nm
3 RESULTS AND DISCUSSION Raman and surface-enhanced Raman spectra were obtained for potassium cyanide bis-(2-chloroethyl)sulfide and ethyl S-2-diisopropylamino ethyl methylphosphonothioate representing three classes of chemical agents cyanides mustards and nerve agents respectively Spectra were also obtained for 2-chloroethyl ethyl sulfide (CEES) a structural analogue to HD which was included in the study to aid in assigning spectral bands KCN salt was used for cyanide experiments to avoid the increased hazards of handling HCN gas KCN completely dissolves in water forming its conjugate acid HCN according to its Ka of 615x10-1027 and at a concentration of 1 mgmL results in a pH 107 solution This is significant in that only CN- appears to interact sufficiently with silver to produce a SER spectrum and no spectral signal is observed below pH 726 Figure 2 shows the SER and normal Raman spectra for KCN The SERS spectrum of 1mgml KCN in water shows a single intense somewhat broad feature at
A B
C D
SPIE-2003-5269 19
2100 cm-1 assigned to the single CequivN stretch The band is much sharper in the normal Raman spectra of the solid KCN salt at 2074 cm-1 This band does broaden and shift to 2080 cm-1 in solution (not shown) However the observed SERS frequency is attributed to interaction with silver and low frequency mode at 135 cm-1 attributed to a Ag-CN stretch (not shown) supports this conclusion
cm-1 band to a C-S stretch but the authors concede that it is in fact more likely a C-Cl stretch28 It appears that the most intense Raman bands at 648 692 and 747 cm-1 shift to 620 660 and 720 cm-1 in the SER spectra and are tentatively assigned as above The width of these bands suggests that they overlap underlying spectral features Additional bands in the Raman spectra occur at 972 1034 1049 1263 1286 1423 1442 2865 2935 and 2960 cm-1 Corresponding bands occur in the SER spectra at 964 1015 1054 1286 1410 1447 2865 and 2935 cm-1 Most of these bands are associated with alkane modes specifically the bands at approximately 1040 cm-1 to a C-C stretch 1290 cm-1 to a CH2 in-phase twist 1440 cm-1 to a CH2 wag 2865 cm-1 to a symmetric CH2 stretch and 2965 cm-1 to an asymmetric CH2 stretch The Raman and SER spectra of sulfur mustard were measured at the Edgewood center (Figure 4) Both spectra are largely similar to CEES The C-Cl and C-S bands in the Raman spectrum of HD now occur at 640 655 700 739 and 760 cm-1 and are more resolved possibly due to the increased molecular symmetry Theoretical calculations indicate that the first three bands are due to C-Cl stretching modes and the latter two to C-S stretching modes28 Only the C-Cl bands maintain significant intensity in the SER spectra occurring at 624 and 643 cm-1 which is attributed to the
Prior to measurements of HD CEES was examined by Raman and SER spectroscopy (Figure 3) CEES also known as half-mustard is essentially identical to HD except one of the chlorine end atoms is replaced by a hydrogen atom Again although not as toxic as HD CEES is a blister agent and dilute aqueous samples were prepared using appropriate safety equipment Both the Raman and SER spectra of CEES are similar and dominated by bands between 600 and 800 cm-1 These are associated with C-Cl and C-S stretching modes which are tentatively assigned to 648 and 747 cm-1 in the Raman spectra respectively The shoulder at 630 cm-1 the overlapped band at 660 cm-1 and the strong band at 692 cm-1 could also be due to these modes or their asymmetric counterparts It is worth noting that theoretical calculations assign the 692
Figure 2 A) SER and B) NR spectra of KCN Conditions A) 1 mgml in tap water 100 mW of 785 nm at sample 1-min acquisition time B) solid 300 mW of 785 nm 5-min All spectra are 8 cm-1 resolution
A
B
Figure 3 A) SER and B) NR spectra of CEES Conditions A) 1 vv (10 mgml) in MeOH 100 mW of 785 nm 1-min acquisition time B) neat 300 mW of 785 nm 5-min
A
B
Cl-CH2-CH2-S-CH2-CH3
A
B
Figure 4 A) SER and B) NR of HD Conditions A) 1mgml in tap water B) pure both 100 mW of 785 nm 1-min
Cl-CH2-CH2-S-CH2-CH2-Cl
SPIE-2003-5269 20
expected strong interaction between chlorine and silver and adds support to the assignment of this band to a C-Cl stretch Weaker overlapping bands occur at 670 692 and 724 cm-1 the latter possibly due to C-S stretching modes Again the alkane modes are apparent in the normal Raman spectra of HD but only a broad feature at 1300 to 1450 cm-1 suggests CH2 contributions in the SER spectrum Although the observed bands in the VX spectrum have not been assigned (Figure 4) a computer generated Raman spectrum29 predicts many of the same features with surprising accuracy and are used here Two intense bands at 460 and 530 cm-1 closely match predicted bands at 463 and 546 cm-1 assigned to a CH3-P=O bend and a PO2CS wag Three highly overlapped bands occur at 694 745 and 771 cm-1 matching predicted bands at 713 730 and 760 cm-1 The first
Table 2 Measured SER peak heights for the CN stretch at 2100 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
Figure 5 A) SERS and B) NR spectra of VX Conditions A) 1 vv (10 mgml) in MeOH B) pure sample both 100 mW of 785 nm 1-min acquisition time
A
B
two have been assigned to a C-S stretch and CH2bend respectively while the latter has been attributed to either a P-C stretch or an O-C-C stretch Although the 745 cm-1 band may alternatively be assigned to a C-S stretch based on the previous measurements of CEES and HD The relatively intense bands at 890 1106 1218 1445 and 1465 cm-1 also match predicted bands at 880 1108 1216 1440 and 1464 cm-1 that are assigned to a C-C stretch CH3 rock N-C3 stretch various C-H3 bends and C-H bends respectively Both the computer generated and the measured spectra contain numerous other less intense bands One is worth mentioning A unique band appears at 370 cm-1 that is predicted at 368 cm-1 and corresponds to an O-P=O bend The surface-enhanced Raman spectrum of VX is also rich with spectral features It has the unique low frequency band at 370 cm-1 as well as a second band at 380 cm-1 that is assigned tothe S-P-O bend predicted in the normal Raman spectrum at 388 cm-1 Based on the measured and predicted normal Raman spectra the following SERS assignments are given 460 cm-1 to the CH3-P=O bend 544 cm-1 to the PO2CS wag 738 cm-1 to a C-S stretch (based on arguments above) 890 cm-1
to a C-C stretch 1101 cm-1 to a CH3 rock and 1456
cm-1 to a C-H bend The ability of SERS to measure chemical agents in water containing real-world chemical interferents was tested by using the generic tap water described in Table 1 The ability to reproduce measurements was accomplished by preparing three separate water stock solutions which were used to prepare three sample solutions each of 1 mgmL KCN in the generic tap water The samples defined as vials 1-9 were then measured at 4 points per vial in a semi-random fashion such that errors associated with stock solution preparation and errors associated with instrument drift could be identified No trends were apparent that signified such systematic errors Each spectrum collected consisted of 20 averaged scans taking 16 seconds at 8 cm-1 resolution The laser power at the sample was measured periodically during the day and it ranged from 102 to 105 mW spacccc
spot Vial 1 Vial 2 Vial 41 13754 13004 128192 13519 12692 12909
1 3 13541 12721 126394 13462 12648 12651
ave 13569 12766 12755Vial 6 Vial 3 Vial 5
1 11586 14068 10757 2 11236 14402 115122 3 11376 14546 11559
4 10894 11763 11214ave 11273 13695 11261
Vial 7 Vial8 Vial 91 11198 15804 11023
2 11243 15716 88753 3 11562 15216 11218
4 11655 15084 11077ave 11415 15455 10548
CN AVG STDEV ERR1 mgmL pk ht 12526 1560 1245
stock solution
SPIE-2003-5269 21
The CN measurements consisting of 36 data points produced an average height of 125 for the 2100 cm-1 peak with a standard deviation of 156 or 125 (Table 2) The HD and VX measurements were performed precisely the same way (Tables 3 and 4) For HD the 624 cm-1 peak was used for analysis and it had an average height of 53 with a standard deviation of 068 or 129 while for VX the 544 cm-1 peak was used for analysis and it had an average height of 1051 with a standard deviation of 308 or 293 The greater error in the VX measurements can be somewhat attributed to Vial 4 which produced lower SER signal intensities But removing this vial from the data set changes the standard deviation to 233 only a modest improvement
spot Vial 1 Vial 2 Vial 41 525 609 4682 45 675 484
1 3 527 644 5094 559 807 669
ave 51525 68375 5325Vial 6 Vial 3 Vial 5
1 521 574 575 2 536 53 4492 3 509 508 451
4 565 594 379ave 53275 5515 4635
Vial 7 Vial 8 Vial 91 457 589 413
2 583 497 4273 3 544 505 4
4 528 5 509ave 528 52275 43725
HD AVG STDEV ERR1 mgmL pk ht 530 068 1291
stock solution
spot Vial 1 Vial 2 Vial 41 1464 1034 4672 1485 99 61
1 3 1491 989 5684 1041 777 553
ave 137025 9475 5495Vial 6 Vial 3 Vial 5
1 1058 942 1293 2 697 121 965
2 3 727 1206 11124 689 1254 746
ave 79275 1153 1029Vial 7 Vial 8 Vial 9
1 1127 783 163 2 1358 812 1614
3 3 1371 908 15124 1097 875 1397
ave 123825 8445 153825
VX AVG STDEV ERR1 mgmL pk ht 1051 308 2925
stock solution
4 CONCLUSIONS In this paper we examined the ability of surface-enhanced Raman spectroscopy to reproducibly measure CN HD and VX in tap water without chemical interference Both normal and surface-enhanced Raman spectra were examined to select unique bands suitable to identify and quantify these chemical agents For SER measurements the 2100 cm-1 C-N stretch was used for CN the 624 cm-1 C-Cl stretch was used for HD and the 544 cm-1 PO2CS wag was used for VX It was determined that 1 mgmL samples of each of these chemicals measured 36 times in glass vials coated with a silver-doped sol-gel reproduced measurements with standard deviations of 125 129 and 293 It was further found that the 10 chemicals added to simulate generic tap water did not interfere with or alter the SER spectra It should be noted that the concentrations used in this study were considerably greater than those required by the JSAWM program Current work involves improving SER sensitivity and designing sampling systems with better reproducibility This includes the development of fractal silver and gold structures within the sol-gel matrix and the development of chemically selective sol-gels
ACKNOWLEDGMENTS The authors are grateful for the support of the US Army (DAAD13-02-C-0015 Joint Service Agent Water Monitor program) The authors would also like to thank Janet Jensen Ronald Crosier and Kristina Gonser for helpful discussions
Table 4 Measured SER peak heights for the VX band at 544 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
Table 3 Measured SER peak heights for the HD band at 624 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
SPIE-2003-5269 22
REFERENCES 1 Jensen JL NC Wong W R Loerop SPIE 4775-03 (2001) 2 JSAWM Requirements at wwwsbccomapgeaarmymilRDAecbcrtPRODSERJSAWMjsawmhtml 3 Erickson B ldquoThe Chemical Weapons Convention Redefines Analytical Challengerdquo Anal Chem 70 397A (1998) 4 Johnston RL Hoefler CM Fargo JC and Moberley B ldquoThe Defense Nuclear Agencyrsquos
ChemicalBiochemical Weapons Agreements Verification Technology Research Development Test and Evaluation Program and its Requirements for On-Site Analysisrdquo AT-ONSITE 5-8 (1994)
5 Black RM Clarke RJ Read RW and Reid MT ldquoApplication of GC-MS and GC-tandem MS to the analysis of chemical warfare samples found to contain residues of the nerve agent sarin sulphur mustard and their degradation productsrdquo J Chrom 662 301-321 (1994)
6 USA Army Center for Health Promotion and Preventive Medicine (USACHPPM) Technical Guideline 230A 7 Hoffland LD Piffath RJ Bouck JB rdquoSpectral signatures of chemical agents and simulantsrdquo Optical
Engineering 24 982-984 (1985) 8 Braue EHJ Pannella MGrdquoCIRCLE CELL FT-IR Analysis of Chemical Warfare Agents in Aqueous Solutionsrdquo
App Spectrosc 44 1513-1520 (1990) 9 Tseng C-H Mann CK and Vickers TJ ldquoDetermination of Organics on Metal Surfaces by Raman
Spectroscopyrdquo App Spectrosc 47 1767-1771 (1993) 10 Staff Reporter ldquoUS Army Tests Lidar to Detect Biological Toxinsrdquo Photonic Spectra pg 50 December 1998 11 Christesen SD Raman cross sections of chemical agents and simulants App Spectrosc 42 318-321 (1988) 12 Weaver MJ Farquharson S Tadayyoni MA ldquoSurface-enhancement factors for Raman scattering at silver electrodesrdquo J Chem Phys 82 4867-4874 (1985) 13 Alak AM and Vo-Dinh T ldquoSERS of Organophosphorous Chemical Agentsrdquo Anal Chem 59 2149-2153 (1987) 14 Lee Y Farquharson S ldquoRapid chemical agent identification by SERSrdquo SPIE 4378 21-26 (2001) 15 Farquharson S Maksymiuk P Ong K Christesen S ldquoChemical agent identification by surface-enhanced Raman spectroscopyrdquo SPIE 4577 166-173 (2001) 16 Spencer KM Sylvia J Clauson S and Janni J ldquoSurface Enhanced Raman as a Water Monitor for Warfare
Agents in Waterrdquo SPIE 4577 158-165 (2001) 17 Tessier P Christesen S Ong K Clemente E Lenhoff A Kaler E Velev O ldquoOn-line spectroscopic characterization of sodium cyanide with nanostructured gold surface-enhanced Raman spectroscopy substratesrdquo
App Spectrosc 56 1524-1530 (2002) 18 Farquharson S W W Smith S Elliott and J F Sperry Rapid biological agent identification by surface- enhanced Raman spectroscopy SPIE 3855110-116 (1999) 19 Farquharson S WW Smith YH Lee S Elliott and J F Sperry Detection of biological signatures A
comparison of electrolytic and metal-doped sol-gel surface-enhanced Raman media SPIE 4575 62-72 (2002) 20 Guzelian AA Sylvia J Janni J Clauson S and Spencer KM ldquoSERS of whole cell bacteria and trace levels of biological moleculesrdquo SPIE 4577 182-192 (2001) 21 Shende C Inscore F Gift A Maksymiuk P Farquharson S ldquoRapid dipicolinic acid extraction from Bacillus spores detected by surface-enhanced Raman spectroscopyrdquo App Spectrosc 58 accepted 22 Lee Y and Farquharson S ldquoSERS Sample Vials Based on Sol-Gel Process for Trace Pesticide Analysisrdquo SPIE
4206 140-146 (2000) 23 Farquharson S and Lee Y ldquoTrace Drug Analysis by SERSrdquo SPIE 4200-16 (2000) 24 Lee Y Farquharson S and Rainey P M Surface-Enhanced Raman Sensor for Trace Chemical Detection in
Water SPIE 3857 76-84 (1999) 25 Lee Y Farquharson S Kwong H and Shahriari M ldquoSol-Gel Chemical Sensor for Surface-Enhanced Raman
Spectroscopyrdquo SPIE 3537 252-260 (1998) 26 Farquharson S A Gift P Maksymiuk F Inscore and W Smith ldquopH dependence of methyl phosphonic acid
dipicolinic acid and cyanide by surface-enhanced Raman spectroscopyrdquo SPIE 5269-19 (2003) 27 Lide DR Ed Handbook of Chemistry and Physics CRC Press 77th Ed (1996-7) 28 Sosa C RJ Bartlett K KuBulat and WB Person ldquoA theoretical study of harmonic vibrational frequencies and
infrared intensities of XCH2CH2SCH2CH2X and XCH2CH2SH (= H Cl)rdquo J Phys Chem 93 577-588 (1993) 29 Hameka HF and JO Jenson ldquoComputer-generated predictions of the structure and the IR and Raman spectra of
VXrdquo ERDEC-TR-065 May 1993
- Chemagents Appendicespdf
-
- SPIE2001-4575-Bioagent-SERSapdf
-
- Detection of bioagent signatures A comparison of electrolytic and metal-doped sol-gel surface-enhanced Raman media
- Stuart Farquharson Wayne Smith and Yuan Lee
-
- Real-Time Analyzers 87 Church Street East Hartford CT 06108
-
- ABSTRACT
- 4 CONCLUSIONS
- 5 ACKNOWLEDGEMENTS
- 6 REFERENCES
-
SPIE 2001-4575
72
6 REFERENCES 1 Roberts WL and Rainey PM Clin Chem 39 1872-1877 (1993) 2 Pasechnik VA CC Shone and P Hambleton Bioseperations 3 267-283 (1993) 3 Jackson PJ ME Hugh-Jones DM Adair G Green KK Hill CR Kuske LM Grinberg FA Abramova and P
Keim Proc Natl Acad Sci 95 1224-1229 (1998) 4 Jensen JL NC Wong W R Loerop SPIE 4775-03 (2001) 5 Snyder A Pet al SPIE 3853-15 (1999) 6 Danosky T R and McFadden P N in press (1997) 7 Woodruff WH Farquharson S Science 201 831-833 (1978) 8 Lord RC and Thomas GJJr Spectrochemica Acta 23A 2551-2591 (1967) 9 Thomas GJJr Biochim Biophys Acta 213 417-423 (1970) 10 Placzek G Handbuch der Radiologie 2 EMarx ed Akademische Verlagagescellschatt Liepzig 1934 UCRL
Trans No 526 (1959) 11 Jeanmaire DL and RP Van Duyne J Electroanalytical Chem 84 1-20 (1977) 12 Chada S Manoharan R Moenne-Loccoz P Nelson WH Peticolas WL and Sperry JF Applied Spectroscopy
47 38-43 (1993) 13 Manoharan R Ghiamati E Chada S Nelson WH and Sperry JF Applied Spectroscopy 47 2145-2150 (1993) 14 Todd EA Morris MD Applied Spectroscopy 48 545-548 (1994) 15 Wentrup-Byrne E Sarinas S and Fredericks PM Applied Spectroscopy 47 1192-1197 (1993) 16 Weldon MK VR Zhelyaskov and MD Morris Applied Spectroscopy 52 265-269 (1998) 17 Kneipp K and J Fleming J Mol Structure 145 173-179 (1986) 18 Pothier NJ and Force RK Applied Spectroscopy 46 147-151 (1992) 19 Ervin KM E Koglin JM Sequaris P Valenta and HW Nurnberg J Electroanal Chem 114 179-194 (1980) 20 Kim SK TH Joo SW Suh and MS Kim J Raman Spectrosc 17 381-386 (1986) 21 Pothier NJ and Force RK Analytical Chemistry 62 678-680 (1990) 22 Kneipp K Y Wang RR Dasari and MS Feld Applied Spectroscopy 49 780-784 (1995) 23 Nie S and Emory SR Science 275 1102 (1997) 24 Graham D WE Smith AMT Linacre CH Munro ND Watson and PC White Analytical Chemistry 69 4703-
4707 (1997) 25 Fleischmann M PJ Hendra and AJ McQuillan Chem Phys Lett 26 163-166 (1974) 26 Pettinger B J Chemical Phys 85 7442-7451 (1986) 27 Surface-Enhanced Raman Scattering Section Four Theory SPIE MS 10 M Kerker and B Thompson Eds (1990) 28 Wang D-S and Kerker M SPIE (M Kerker and B Thompson Eds) MS 10 417-429 (1990) 29 Weaver MJ Farquharson S Tadayyoni MA J Chem Phys 82 4867-4874 (1985) 30 Norrod KL Sudnik LM Rousell D and Rowlen KL Applied Spectroscopy 51 994-1001 (1997) 31 Farquharson S Weaver WJ Lay PA Magnuson RH and Taube H J Am Chem Soc 105 3350-3351 (1983) 32 Lee PC and Meisel D ldquoAdsorption and Surface-Enhanced Raman of Dyes on Silver and gold Solsrdquo J Phys Chem
86 3391-3395 (1982) 33 Li Y-S and Wang Y Applied Spectroscopy 46 142-146 (1992) 34 Bello JM DL Stokes and T Vo-Dinh Analytical Chemistry 61 1779-1783 (1989) 35 van Duyne RP J C Hulteen D A Treichel M T Smith M L Duval and T R Jensen J Phys Chem B1033854-
3863 (1999) 36 Farquharson S and W W Smith W H Nelson and J F Sperry SPIE 3533-27 207-214 (1998) 37 Lee Y H W Smith S Farquharson H C Kwon M R Shahriari and P M Rainey SPIE 3537 252-260 (1998) 38 Lee Y-H S Farquharson and P M Rainey SPIE 3857 76-84 (1999) 39 Current Protocols in Molecular Biology Wiley Interscience1003-1006 (1987) 40 Farquharson S Smith W Carangelo R C and Brouillette C SPIE 3859 14-23 (1999) 41 Chase D B and JF Rabolt Fourier Transform Raman Spectroscopy Acad Press Ch1 p 131 (1994) 42 Brock TD MT Madigan JM Martinko and J Parker Biology of Microorganisms 7th Ed Prentice Hall p 76-80
(1994) 43 Connes J Rev Opt Theor Instrum 40 45 (1961)
166
Chemical agent identification by surface-enhanced Raman spectroscopy
Stuart Farquharson and Paul Maksymiuk
Real-Time Analyzers 87 Church Street East Hartford CT 06108
Kate Ong and Steven D Christesen US Army SBCCOM Aberdeen Proving Ground MD 21010
ABSTRACT The recent distribution of anthrax through the US postal system and the subsequent infection and death of several postal and national media employees amplifies the need for methods to rapidly detect identify and quantify this and other chemical and biological warfare agents The US military has also identified water supplies as a likely method of warfare agent deployment and is funding the development of a Joint Service Agent Water Monitor (JSAWM) In an effort to aid military personnel and the public at large we are developing a portable analyzer capable of identifying and quantifying chemical agents rapidly either on-demand or continuously The approach is based on the ability of Raman spectroscopy to identify molecular structure through the abundant vibration information provided in spectra and the ability of surface-enhanced Raman spectroscopy (SERS) to detect extremely low concentrations (eg part-per-billion) through the enhancement of Raman scattering by six orders of magnitude or more A key element to the analyzer design is a new SER active medium that is capable of quantitative reversible measurements The medium consists of silver or gold nanoparticles incorporated into a sol-gel matrix The porous silica network offers a unique environment for stabilizing SER active metals and the high surface area increases the interaction between the analyte and metal particles Here we present the use of new sol-gels that also selectively enhance chemicals based on polarity and charge Base-line measurements of chemical agents and their hydrolysis products are presented and compared to the JSAWM goal of 30 micrograms per liter detection Keywords Chemical warfare agent hydrolysis product SERS Raman spectroscopy sol-gel nanoparticle
1 INTRODUCTION Since September 11 2001 the threat of terrorist attacks and biological warfare within US borders has become a sobering reality The simplicity in manufacturing ease of deployment and the relatively low cost of chemical warfare agents (CWAs) raises public concern that they may also be used by terrorists Indeed terrorists released sarin (GB) in the Tokyo subway in 19951 Countering terrorism and terrorist attacks requires recognizing likely deployment scenarios and having the required technology to rapidly detect the deployment event One method of deployment has been long identified by the US military distribution through water supplies To counter this threat the Department of Defense is funding or monitoring the capabilities of 200 technologies with the goal of developing a Joint Service Agent Water Monitor (JSAWM) that is field portable2 The JSAWM must be capable of identifying the chemical agents with no false-positives and quantifying the agents at microgram per liter concentrations within 10 minutes (Table 1)3 This includes the analysis of drinking water supplies distribution and storage systems as well as potable water supplies The earliest technologies developed for CWA detection were based on electrochemical ionization or colorimetric analysis (eg phosgene tape) Although these analyzers were easy to use they were not generally agent specific and suffered from false-positives4 More traditional laboratory methods have also been investigated and in particular combined gas chromatography and mass spectrometry (GCMS) has been very successful at eliminating false-positives56 However GCMS requires extraction repeated calibration and long analysis times (typically 20 to 60 minutes)6 making it labor intensive and less than desirable for field use More rapid analysis of agents in the solid liquid and gas phase has been To whom correspondence should be addressed emailfarqureal-time-analyzerscom
Vibrational Spectroscopy-based Sensor Systems Steven D Christesen Arthur J Sedlacek III EditorsProceedings of SPIE Vol 4557 (2002) copy 2002 SPIE middot0277-786X02$1500
167
Table 1 Chemical Agent Structures Hydrolysis Half-lives and JSAWM Thresholds Agent Short-Hand Chemical Structure Hydrolysis
Half-Life JSAWM
Thresholds
Sarin (GB) F-[O=P-CH3]-O-CH(CH3)2 213 hours 32 microgL Soman (GD) F-[O=P-CH3]-O-CH(CH3)-(C-(CH3)3) 23 hours 32 microgL Tabun (GA) (CH3)2-N-[O=P-CN]-O-C2H5 41 hours 32 microgL VX C2H5O-[O=P-CH3]-S-(CH2)2-N-(CH(CH2)2)2 821 hours 32 microgL EA2192 HO-[O=P-CH3]-S-(CH2)2-N-(CH(CH2)2)2 gt9 years 32 microgL Mustard (H) ClCH2CH2-S-CH2CH2Cl encapsulates 47 microgL Lewisite (L) ClCH=CH-As-Cl2 rapid 27 microgL HCN HCN rapid 20 mgL BZ C7NH12-O-[C=O]-COH(C6H5)2 23 microgL T-2 Toxin 87 microgL at pH 7 to 75 and 20 to 25 oC
demonstrated by vibrational spectroscopy7-10 Hoffland et al7 reported infrared absorbance spectra and absolute Raman cross sections for several chemical agents while Christesen measured Raman cross sections for sarin tabun mustard gas and VX (ethyl S-2-diisopropylamino ethyl methylphosphonothioate)11 Again however these techniques also have limitations Raman spectroscopy is simply not a very sensitive technique and detection limits are typically 01 (1000 ppm) While infrared spectroscopy would have limited value in analyzing poisoned water since the very strong infrared absorption of water would obscure most other chemicals present Nevertheless efforts to overcome these limitations have been demonstrated Braue and Pannella8 quantified the G-series nerve agents (tabun sarin and soman) in terms of infrared attenuated total reflectance using a circle-cell And Alak and Vo-Dinh demonstrated the possibility of surface-enhanced Raman spectroscopy (SERS) to identify CWAs by measuring several organophosphonates that simulate the nerve agents12 However quantitative measurements have not been demonstrated for the SER-active material used (silver coated on alumina particles) or other SER-active media13 Recently we developed silver-doped sol-gels to promote the SER effect14-17 The porous silica network of the sol-gel matrix offers a unique environment for stabilizing SER-active metal particles and the sol-gel provides a high surface area that effectively increases the number of molecules observed within the Raman scattering volume The silver-doped sol-gels have been coated on the internal walls of standard glass vials to allow rapid SER analysis of small sample quantities (lt 01 mL) without preparation We have used p-aminobenzoic acid (PABA) as a test chemical to demonstrate surface enhancements greater than 106 reversible measurements in a flowing system reproducible measurements from vial-to-vial and batch-to-batch and measurements in multiple solvents including water14-17 Recently we used these vials to measure Tabun (GB) and Sarin and several hydrolysis products pinacolyl methyl phosphonate (PMP from Soman) and methyl phosphonic acid (MPA from all G-agents Figure 1) Although a number of unique vibrational bands are observed (eg C-N stretch doublet and P-C stretch) the G-agents were only observed for 5 concentrations and all spectra required baseline corrections Figure 1 Surface enhanced Raman spectra of ~5 vv A) Tabun and B) Sarin C) 1 vv PMP and D) 10 ppm MPA using sol-gel sample vials 785 nm excitation 1-min scan and CCD detection Performed at Aberdeen Proving Ground
C A
Wavenumber (∆cm-1) Wavenumber (∆cm-1)
D B
790
545 C-N
2135 2190 P-C 770
760 1290
Proc SPIE Vol 4577
168
Nevertheless MPA was readily observed for a 10 ppm sample with an estimated detection limit of 04 ppm (based on a signal-to-noise ratio of 3 for the 760 cm-1 band intensity) This measurement provides encouragement in that SERS may satisfy the needs of a JSAWM Furthermore MPA is also a hydrolysis product of VX and V-gas and EA2192 (Figure 2) and may prove a valuable indicator of agent usage Figure 2 Hydrolysis of Sarin to form hydrofluoric acid (HF) methylphosphonic acid 1-methylethyl ester (MPAMME) methyl phosphonic acid (MPA) and 2-propanol With this initial albeit modest success we began analyzing chemicals with various sol-gel compositions that we have been developing Here we describe four sol-gel compositions that select for 1) polar-positive 2) polar-negative 3) weakly polar-positive and 4) weakly polar-negative chemical species The ability of these sol-gels to select and enhance Raman scattering is described for several test chemicals and MPA
2 EXPERIMENTAL The chemicals analyzed as well as all chemicals used to prepare the metal-doped sol-gels were obtained at their purest commercially available grade from Aldrich (Milwaukee WI) The sol-gel designed to select for polar-negative species was prepared from a silver amine complex tetramethyl orthosilicate (TMOS) and methanol After mixing 02 mL of the sol-gel solution was transferred into a glass vial (2 mL) dried and heated The incorporated silver ions were then reduced using dilute sodium borohydride The vials were washed and dried prior to the addition of a sample solution In a similar manner the sol-gel designed to select for polar-positive species was prepared from a gold salt TMOS and methanol The sol-gel designed to select for weakly polar-negative species was prepared from a silver amine complex tetraethyl orthosilicate (TEOS) and methanol And the last sol-gel designed to select for weakly polar-positive species was prepared from a gold salt TEOS and methanol All samples were prepared in a chemical hood and transferred into plain or SER-active vials for analysis Normal Raman spectral measurements employed 1-mL pure samples that were placed in a 1-cm3 cuvette and weighed This yielded a powder density that allowed accurate calculation of molecules in the optical collection field SERS measurements employed 1-mg sample per mL water concentrations unless otherwise stated Once prepared a 01 mL sample was placed into one of the four selective sample vials which in turn was placed into the sample compartment of a Raman spectrometer for analysis A prototype Fourier transform Raman spectrometer (RTA) was used for these measurements18 The system consisted of a NdYAG laser (Brimrose) for excitation at 1064 nm an interferometer built by On-Line Technologies (OLT East Hartford CT) for frequency separation an uncooled InGaAs detector for signal detection (RTA) and an Intel 400 MHz Pentium II based laptop computer (Dell Round Rock TX) for interferometric control data acquisition (OLT) and analysis (LabVIEW by National Instruments Austin TX) Additional components included a Notch filter (Kaiser Ann Arbor MI) and interferometer entrance and exit optics (Edmund Scientific Barrington NJ) Fiber optics were used to deliver the excitation beam to the sample and the scattered radiation to the interferometer (1 meter lengths of 200 and 365 micron core diameter respectively Spectran Avon CT) A second Notch filter (Kaiser) was used as a beam splitter to direct the excitation beam along the same axis as the collected radiation A microscope object (20x04 Newport Irvine CA) was used to focus the beam into the sample and to collect the scattered radiation back along the same axis In this co-axial backscattering arrangement the excitation beam was passed through the outside of a glass vial and focused onto the silver-doped sol-gel film (01-03 mm thickness) containing the sample
3 RESULTS AND DISCUSSION p-aminobenzoic acid (PABA) and phenyl acetylene (PA) and were used to refine the selectivity and SER-activity of the four different metal-doped sol-gels PABA is a popular chemical used to evaluate the performance of SER-active media Here the polar end groups can be used to test selectivity of the polar-negative and polar-positive sol-gels PA is essentially non-
2O+ H HF + +OH
OH3H C
OP
OF
CH
CH3
3
3H C
OCP
OOH
CH
CH3
3
3H C
O
CPHO
CH
CH
3
3
C
Sarin MPAMME MPA 2-propanol
Proc SPIE Vol 4577
169
polar but a high electron density in the cylindrical π cloud around the carbon-carbon triple bond allows testing the selectivity of the weakly polar-negative and weakly polar-positive sol-gels As Figure 3 illustrates PABA passes through the polar sol-gel and is enhanced by either the silver or gold particles At 1 mgml the concentration of neutral PABA is ca 20 times that of the ionized form (pKa = 48) For electropositive silver the PABA anion is expected to interact through the carboxylate group and the associated vibrational modes are expected to dominate the spectrum Conversely for electronegative gold either form of PABA is expected to interact through the amine group The clear differences in our spectra support this expectation Furthermore bands at 840 and 1405 cm-1 assigned to a COO- bend and stretch respectively are significantly more intense for silver than gold Additional bands at 1140 and 1195 cm-1 are assigned to CH bending modes while bands at 1450 1500 and 1605 cm-1 are assigned to ring vibrational modes A very similar SER spectrum for PABA on a silver-coated alumina substrate has previously been reported with similar assignments19 For the gold-doped sol-gel new bands appear at 690 1355 and 1585 cm-1 The first band is assigned to a ring-H bending mode the second band to a ring-N- stretching mode and the third band to a possible NH2 scissors mode or ring mode The second band is not observed in the normal Raman spectrum but infrared bands occur at this frequency for aromatic ring-secondary amine stretching modes The scissors mode occurs at this frequency in Raman spectra for several chemicals but is absent in the PABA Raman spectrum Alternatively this mode may be the1600 cm-1 ring mode that has been shifted by the gold interaction Again a very similar SER spectrum of PABA has been reported but surprisingly using silver (colloids)2021 not gold as the enhancement medium These researchers also assumed the primary interaction of PABA with silver was through the carboxylate anion and made assignments accordingly For example they assigned the 1359 cm-1 to a COO- stretch not to the amine group as we have They also favor the ring stretching mode assignment for the 1582 cm-1 band Finally it should be said that other researchers have argued that the most dominant band in the SER spectra at 1450 cm-1 a ring vibration mode suggests that PABA lies flat on the surface and the π-orbitals dominate the surface interaction22
Figure 3 SER spectra of A) PABA using polar-negative and B) polar-positive sol-gels and C) PA using weakly polar-negative and D) weakly polar-positive sol-gels PABA is 1 mgmL PA is 1 vv Spectral conditions 75 mw 1064 nm 100 scans (15 min) 8 cm-1 resolution Non-polar PA passes through the non-polar sol-gels and is also enhanced by both metals The spectra are easily understood For electropositive silver PA interacts through the cylindrical triple bond π electron cloud and a -CequivC- doublet occurs near 2000 cm-1 The interaction is reasonably strong since this band appears at 2112 cm-1 in the normal Raman spectrum For electronegative gold this interaction is unlikely and only very weak bands occur near 2000 cm-1 The remaining bands are at 1000 cm-1 1200 cm-1 doublet and 1595 cm-1 all appear in the normal Raman spectra at virtually the same frequencies and are assigned to the symmetric ring-breathing mode CH bending modes and the trigonal ring-breathing mode respectively The polarnon-polar selectivity of the polar-negative and weakly polar-negative sol-gels was tested by adding a 11 molar mixture of PABA and PA The selective enhancement is quite good (Figure 4) The spectrum obtained using the polar sol-gel represents 78 PABA and 22 PA while the spectrum obtained using the weakly polar sol-gel represents 9 PABA and 91 PA The band peak intensities at 2000 cm-1 for PA and 1450 cm-1 for PABA were used for these calculations and are expanded in Figure 4 for clarity
C A
Wavenumber (∆cm-1) Wavenumber (∆cm-1)
2NH COOH
D B
C CH
Proc SPIE Vol 4577
170
Figure 4 SERS of 11 MM of PABA and PA in A) polar-negative and C) weakly polar-negative sol-gels The lower traces compare the pure chemicals B) 1 mgml PABA in polar-negative sol-gel and D) 1 PA in weakly polar-negative sol-gel while the insets magnify the minority species for clarity (x5 in A and x10 in B) Spectral conditions as in Figure 3 Following this development of selective sol-gels that maintained SER activity we measured cyanide and MPA (Figure 5) Not surprisingly the best sensitivity for both hydrolysis products was obtained using the polar-negative sol-gel The interaction of the cyanide anion with the silver surface is sufficient to shift the CequivN stretch observed at 2080 cm-1 in the normal Raman spectrum to 2145 cm-1 in the surface-enhanced Raman spectrum Furthermore the band is substantially broadened This anion has been extensively studied by electrolytic SERS and this shift and broadening have been attributed to the formation of a tetrahedral Ag(CN)3
2- surface structure23 Figure 5 Surface-enhanced (upper traces) and normal Raman spectra (bottom traces) of A) CN- and B) MPA in silver-doped TMOS SERS conditions as in Figure 3 and 1 mgmL Note MPA yields two distinct spectra for neutral (top) and acidic pH (middle) The normal Raman spectra employed pure powders 500 scans and 900 mW of 1064 nm SER measurements of MPA with the polar-negative sol-gel yielded two unique spectral signatures that depended on solution pH (Figure 5) For more neutral solutions the P-C stretch of MPA at 762 cm-1 dominates and the CH2 stretch at 2922 cm-1 appears The SN is sufficiently high that the anti-Stokes Raman shift at -762 cm-1 is observed For deprotonated MPA an oxygen-surface mode appears at 325 cm-1 (as well as its anti-Stokes complement) suggesting a strong interaction This results in substantial enhancement of the P-O-C mode at 1051 cm-1 (upper trace) Others report that this mode dominates the infrared spectra of nerve agents measured in water8 Comparison of the two spectra suggests the following molecule-to-surface orientations The appearance of the oxygen-surface and P-O-C modes in the upper spectrum of Figure 5B indicates that the tetrahedral molecule interacts with the silver surface through the deprotonated oxygen and is oriented end-on The
A B
Wavenumber (∆cm-1)
C A
Wavenumber (∆cm-1) Wavenumber (∆cm-1)
D B
Wavenumber (∆cm-1)
Proc SPIE Vol 4577
171
dominance of the P-C and the CH2 stretches and the disappearance of the P-O-C mode in the upper spectrum suggest the molecule is oriented with the methyl-hydroxyl-hydroxyl face towards the surface However considerably more research must be performed to verify these points Both chemical forms were measured as a function of concentration to determine the onset of monolayer coverage and estimate expected detection limits (Figure 6) Below monolayer coverage the signal to concentration dependence should be linear and the SN of any spectral measurement in this range can be used to predict the detection limit In the spectra presented here the peak height was used as the signal while the noise as root-mean-squared (RMS) was measured between 4400-4600 cm-1 Since noise is distributed evenly throughout the spectrum when transformed this region was used since it does not have any contributions from signals or baseline offsets Figure 6 shows a series of spectra for MPA along with a plot of the 1050 cm-1 band intensity (the noise was identical for this series) as a function of sample concentration A clear discontinuity appears in the vicinity of 01 mgmL (19 ppm) indicating the onset of monolayer coverage A detection limit defined as a SN of 3 was calculated for the 01 and 005gmL samples at 24x10-4 and 25x10-4 gL respectively A more modest detection limit of 101x10-4 gL was obtained using the 760 cm-1 band in the second series of concentration measurements These detection limits correspond to 46 and 190 ppb for a 3-min scan and 75 mW of incident laser power
Figure 6 A) Concentration dependence of MPA SERS measured in silver-doped TMOS) B) Concentrations are 001 005 01 05 1 gL (188 94 188 94 188 ppm) I760 series (bull) and I1050 series (∆)
Enhancement factors for cyanide and methyl phosphonic acid can be determined by estimating the number of molecules contributing to the surface-enhanced and normal Raman spectra The enhancement factor EF can be defined by the following equation
EF = (ISERSINR)bull(MNRMSERS) bull(PNRPSERS) bull(TNRTSERS)12
where I is the spectral band intensity M is the sample mass P is the incident laser power and T is the measurement time (or number of scans) for the two measurements For the normal Raman spectra a cylindrical scattering volume is assumed based on the laser area (28x10-7m2 6x10-4m diameter spot) and the penetration depth (1x10-3 m)24 The density of KCN and MPA as powders were measured at 0572 and 0516 gcm3 indicating that 16x10-4 and 144x10-4 g produced the normal Raman signals in Figure 5 respectively The number of molecules contributing to the SER spectra are those on the silver particles embedded in the sol-gel The total silver surface area can be determined from the average particle size concentration and the scattering volume Previous scanning electron micrographs showed the average silver particle size to be 40 nm in diameter (335x10-23m3)17 The silver concentration is 012M based on the reactant molar concentrations and dilution factors And the scattering volume is 76x10-
11m3 again based on a cylindrical scattering volume defined by a laser area of 28x10-7m2 and a sol-gel thickness of 27x10-
4m This volume contains 123x10-6g of silver equivalent to 35x109 silver particles with a collective surface area of 18x10-
5m2 However it may reasonably be assumed that at least half of the silver surface is in contact with the sol-gel matrix and unavailable for analyte interaction If we assume monolayer coverage and that each CN molecule occupies 15x10-20m2 then approximately 62x1014 molecules or 27x10-8g of CN contribute to the SER spectrum (20x10-19m2 46x1013 molecules 74x10-9g for MPA) Accordingly the EF for cyanide equals 48x104 ((180599) bull(16x10-427x10-8) bull(90075) bull(500100)12) The EF for MPA is considerably higher at 87x106 ((60326) bull(144x10-474x10-9) bull(90075) bull(500200)12)
0
20
40
60
80
100
120
140
0 02 04 06 08 1 12[MPA] (mgmL)
I (76
0)
0
100
200
300
400
500
600
I (10
50)
Wavenumber (∆cm-1)
A B
Proc SPIE Vol 4577
172
4 CONCLUSIONS Here we present for the first time surface-enhanced Raman spectra of Tabun and Sarin obtained using silver-doped sol-gels However the inferior enhancement suggested employing SER-active sol-gels varying in composition to improve both detection limits and selectivity of the target analytes To this end we successfully demonstrated the capabilities of four sol-gels that select for 1) polar-positive 2) polar-negative 3) weakly polar-positive and 4) weakly polar-negative chemical species p-aminobenzoic acid was used to show that silver could be used to attract polar-negative chemicals or functional groups (carboxylate anion) and that gold could be used to attract polar-positive chemicals (amine) while a mixture of p-aminobenzoic acid and phenylacetylene was used to show that tetramethyl orthosilicate preferentially solvates polar chemicals and that tetraethyl orthosilicate preferentially solvates weakly polar chemicals This increased sample control was applied to cyanide and methyl phosphonic acid two hydrolysis products of chemical warfare agents Exceptional results were obtained for methyl phosphonic acid allowing measurement of 1x10-2 gL for a 3-min scan and 75 mW of incident laser power with an estimated detection limit of 245x10-4 gL and an enhancement factor of 87x106 However this detection limit is 76 times less sensitive than required for the JSAWM (32x10-6gL for the G-agents) Finally we note that the measurements performed here employed an FT-Raman spectrometer This instrumentation was chosen over dispersive Raman instrumentation for the high wavelength accuracy afforded by the HeNe reference laser (Connes Advantage25) which would allow reliable spectral subtraction matching of observed spectra to stored library spectra and confident use of chemometric approaches Such data analysis is likely to be required to identify the chemical agents as well as distinguish hydrolysis products However this instrumentation which employs 1064 nm excitation and InGaAs detection sacrifices sensitivity We believe that the measurement sensitivity can be improved by at least two orders of magnitude by using shorter laser excitation wavelengths (eg 532 nm) This would provide more efficient Raman scattering (fourth power dependence on laser excitation wavelength) more efficient generation of plasmon modes and allow using more efficient detector material (Si vs InGaAs) These modifications are underway
5 ACKNOWLEDGEMENTS The authors would like to thank Dr R Yin and J Jensen of the US Army for supporting this work (Contract Number DAAD13-01-C-0019) They would also like to thank Advanced Fuel Research for making their laboratory facilities available
6 REFERENCES 1 Tu Anthony ldquoOverview of Sarin Terrorist Incidents in Japan in 1994 and 1995rdquo 6th CBW Protection Symposium
Stockholm Sweden 10-15 May 1998 2 Jensen JL NC Wong W R Loerop SPIE 4775-03 (2001) 3 JSAWM Requirements at wwwsbccomapgeaarmymilRDAecbcrtPRODSERJSAWMjsawmhtml 4 ldquoThe Chemical Weapons Convention Redefines Analytical Challengerdquo Analytical Chemistry News amp Features June 1
397A (1998) 5 Johnston RL Hoefler CM Fargo JC and Moberley B ldquoThe Defense Nuclear Agencyrsquos ChemicalBiochemical
Weapons Agreements Verification Technology Research Development Test and Evaluation Program and its Requirements for On-Site Analysisrdquo AT-ONSITE 5-8 (1994)
6 Black RM Clarke RJ Read RW and Reid MT ldquoApplication of gas chromatography-mass spectrometry and gas chromatography-tandem mass spectrometry to the analysis of chemical warfare samples found to contain residues of the nerve agent sarin sulphur mustard and their degradation productsrdquo J Chromatography 662 301-321 (1994)
7 Hoffland LD Piffath RJ Bouck JBrdquoSpectral signatures of chemical agents and simulantsrdquo Optical Engineering 24 982-984 (1985)
8 Braue EHJ Pannella MGrdquoCIRCLE CELL FT-IR Analysis of Chemical Warfare Agents in Aqueous Solutionsrdquo Applied Spectroscopy 44 1513-1520 (1990)
9 Tseng C-H Mann CK and Vickers TJ ldquoDetermination of Organics on Metal Surfaces by Raman Spectroscopyrdquo Applied Spectroscopy 47 1767-1771 (1993)
10 Staff Reporter ldquoUS Army Tests Lidar to Detect Biological Toxinsrdquo Photonic Spectra pg 50 December 1998
Proc SPIE Vol 4577
Proc SPIE Vol 4577
173
11 Christesen SD Raman cross sections of chemical agents and simulants Applied Spectroscopy 42 318-321 (1988) 12 Alak AM and Vo-Dinh T ldquoSERS of Organophosphorous Chemical Agentsrdquo Analytical Chemistry 59 2149-2153
(1987) 13 Norrod KL Sudnik LM Rousell D and Rowlen KL ldquoQuantitative Comparison of Five SERS Substrates
Sensitivity and Detection Limitrdquo Applied Spectroscopy 51 994-1001 (1997) 14 Lee Y and Farquharson S ldquoSERS Sample Vials Based on Sol-Gel Process for Trace Pesticide Analysisrdquo SPIE 4206
140-146 (2000) 15 Farquharson S and Lee Y ldquoTrace Drug Analysis by Surface-Enhanced Raman Spectroscopyrdquo SPIE 4200-16 (2000) 16 Lee Y Farquharson S and Rainey P M Surface-Enhanced Raman Sensor for Trace Chemical Detection in Water
SPIE 3857 76-84 (1999) 17 Lee Y Farquharson S Kwong H and Shahriari M ldquoSurface-Enhanced Raman Sensor for Surface-Enhanced Raman
Spectroscopyrdquo SPIE 3537 252-260 (1998) 18 Farquharson S Smith W Carangelo R C and Brouillette C ldquoIndustrial Raman Providing Easy Immediate Cost
Effective Chemical Analysis Anywhererdquo SPIE 3859 14-23 (1999) 19 Narayanan VA JM Bello JD Stokes and T Vo-Dinh Analusis 19 307-310 (1991) 20 Laserna JJ E L Torres and JD Winefordner Analytica Chemica Acta 469-480 (1987) 21 Torres EL and JD Winefordner Analytical Chemistry 59 1626-1632 (1987) 22 Suh JS DP DiLella M Moskovits J Phys Chem 87 1540-1544 (1983) 23 Benner RE R Dornhaus R Chang and BL Laube Correlations in the Raman spectra of cyanide complexes adsorbed
at silver electrodes with voltammograms Surface Science 101 341 (1980) 24 Chase D B and JF Rabolt Fourier Transform Raman Spectroscopy Acad Press Ch1 p 131 (1994) 25 Connes J Rev Opt Theor Instrum 40 45 (1961)
APPLIED SPECTROSCOPY 351
focusing the 488 nm laser beam 2 mm deep into thebulk of the crystals to avoid contributions from potentialdepletion layers As shown in Fig 1b the dependence ofthe Raman shift on the degree of deuteration is almostperfectly linear and ts very well with D 5 22684cmR1 24526 where D is the degree of deuteration (in )and R is the spectral mean of the PO4 vibration in cm21A linear correlation coef cient of 0998 indicates an ex-cellent linear dependence of the Raman peak shift withdegree of deuteration This result shows that the shift ofthe PO 4 peak is simply caused by the linear increase inatomic mass due to isotope substitution which decreasesthe length of hydrogen-like bonds
This excellent linear dependence allows us to map thepro le of the DH exchange layer at the surface of DKDPcrystals by acquiring Raman spectra and determining theposition of the PO 4 peak for various depths This methodis preferable over other methods such as determining thestrength of the OD vibration directly (eg at 715 cm21)because the position of the most intense peak in the Ra-man spectrum can be measured more precisely than theintensity of some of the weakest peaks in the spectrumThis is demonstrated in Fig 2 where depth-dependentRaman spectra (Fig 2a) and the resulting exchange layerpro les for two DKDP crystals are shown (Fig 2b) Thespectra in Fig 2a were obtained from a depth scan of aDKDP crystal with 75 degree of deuteration in thebulk grown at 45 8C The spectra start out as DKDP with30 deuteration close to the surface and approach thebulk DKDP spectrum within a few micrometers of depthThe fact that the relative degree of deuteration does notextend to 0 D is due to the limited depth resolution ofthe Raman microprobe which averages over 4 mm indepth Figure 2b depicts the resulting DH exchange layerpro les for this and a second crystal grown at 63 8Crespectively Both crystals had the same exposure to am-bient conditions and their main difference is the temper-ature at which they were grown The different exchangelayer pro les indicate that crystals grown at differenttemperatures have differen t proton conductiv ities 13
which leads to a difference in their rate of deuteriumdepletion The parameters controlling this behavior arecurrently the objective of a detailed study the results ofwhich will be reported elsewhere
CONCLUSION
In conclusion we have shown that the shift of the to-tally symmetric PO 4 stretch mode in the Raman spectrumof DKDP crystals scales linearly with degree of deuter-ation This allows us to correlate Raman peak positionsto deuteration levels in these crystals We have presenteda new technique to determine DH diffusion pro les inDKDP frequency conversion crystals based on micro-Ra-man spectroscopy This technique is fast inexpensiveand works under various environmental conditionswhich will allow us to better understand and control deu-terium depletion in DKDP crystals
ACKNOWLEDGMENTS
We would like to thank M Runkel for rst discovering DKDP crack-ing R Floyd for providing DKDP crystals and L Chase and A Burn-ham for their support and helpful discussions This work was performedunder the auspices of the US Department of Energy by the University
of California Lawrence Livermore National Laboratory under contract W-7405-Eng-48
1 J J De Yoreo A K Burnham and P K Whitman Int Mat Rev47 113 (2002)
2 C E Barker R A Sacks B M Van Wonterghern J A Caird JR Murray J H Campbell K Kyle R B Ehrlich and N DNielsen Proc SPIE-Int Soc Opt Eng 2633 501 (1995)
3 T Suratwala paper to be published4 Cleveland Crystals Inc httpwwwclevelandcrystalscom
KDPshtmltable5 E A Popova I T Savatinova and I A Velichko Sov Phys Solid
State 12 1543 (1971)6 I P Kaminow R C C Leite and S P S Porto J Phys Chem
Solids 26 2085 (1965)7 J A Subramony B J Marquardt J W Macklin and B Kahr
Chem Mat 11 1312 (1999)8 H Tanaka M Tokunaga and I Tatsuzaki Solid State Commun
49 153 (1984)9 R J Nelmes G M Meyer and J E Tibballs J Phys C 15 59
(1982)10 M A Yakshin D W Kim Y S Kim Y Y Broslavets O E
Sidoryuk and S Goldstein Laser Physics 7 941 (1997)11 I Takenaga Y Tominaga S Endo and M Kobayashi Solid State
Commun 84 931 (1992)12 C Krenn personal communication13 M Sharon and A K Kalia J Solid State Chem 21 171 (1977)
Rapid Dipicolinic Acid Extractionfrom Bacillus Spores Detectedby Surface-EnhancedRaman Spectroscopy
STUART FARQUHARSON ALAND GIFT PAUL MAKSYMIUK andFRANK E INSCOREReal-Time Analyzers Inc East Hartford Connecticut06108
Index Headings Dipicolinic acid Bacillus spores Anthrax Surface-enhanced Raman spectroscopy
INTRODUCTION
The anxiety caused by the distribution of anthrax en-dospores through the US postal system in October 2001was exacerbated by the long time required for positiveidenti cation of the Bacillus anthracis spores and the un-known extent of their distribution Since that time manymethods capable of rapid eld analysis have been inves-tigated to augment or replace the laboratory method ofgrowing microorganisms in culture media which takesdays to perform12 Prominent among these approachesare polymerase chain reactions (PCR)3 immunoassaysand detection of calcium dipicolinate as a biochemicalsignature PCR employs primers to separate organism-
Received 10 October 2003 accepted 14 November 2003 Author to whom correspondence should be sent
352 Volume 58 Number 3 2004
speci c nucleic acid sequences (eg capsular protein en-coding gene for Bacillus anthracis)4 and polymerases toamplify the segment until it is detectable Recently am-pli cation times have been substantially reduced andcomplete analysis can now be performed in an hour orless Immunoassay methods are also being developed thatuse competitive binding of the bioagent (as an antigen)and its labeled conjugate for a limited number of anti-bodies Although analyses can be performed in under 30minutes a well-de ned anthrax antigen has not yet beenidenti ed5ndash7 and consequently the false-positive rate isunacceptably high8
A number of other methods are being developed witha focus on the detection of calcium dipicolinate (CaDPA)and its derivatives as a B anthracis signature This is sobecause only spore-forming bacteria contain CaDPA andthe most common potentially interfering spores such aspollen and mold spores do not Relatively fast methodshave been developed to chemically extract CaDPA andthen detect it directly by uorescence9 or indirectly byluminescence1011 In the latter case hot dodecylamine(DDA) has been used to extract dipicolinic acid (DPA)and terbium has been utilized to form a highly lumines-cent DPA complex11 Although measurements have beenperformed in as little as ve minutes it was found thatas many as three concentration-dependent complexes canform each with different lifetimes This coupled withthe fact that the Tb31 cation produces the same lumines-cence spectrum makes determinations of low spore con-centrations problematic
It has been long known that Raman spectra of Bacillispores are dominated by bands associated with CaDPA12
and that these spectra may provide a suitable anthrax sig-nature at the genus level13 Since that time considerableimprovements in Raman instrumentation have led to lab-oratory measurements of single Bacilli spores14 and to eld measurements of spores captured from a mail-sort-ing system15 However the single spore measurementsrequired complex instrumentation that is not ruggedwhile the eld measurements required milligrams of sam-ple Furthermore the Raman spectra of both measure-ments contained uorescence contributions that would in-crease uncertainty in quanti cation
In related research we demonstrated that nanogramquantities of DPA could be detected by uorescence-freesurface-enhanced Raman spectroscopy (SERS)16 We alsodemonstrated that microliter volumes of chemicals canbe detected by SERS using metal-doped sol-gel-packedglass capillaries17 Towards the goal of developing a rap-id eld SERS-based anthrax spore detector we havecombined our previous research and we now report thatDPA can be extracted from a 10 mg B cereus spore sam-ple using DDA in 1 minute and can be detected by SERSin an additional 1 minute
EXPERIMENTAL
Dipicolinic acid (26-pyridinedicarboxylic acid DPA)and dodecylamine (DDA) were used as received fromSigma-Aldrich (Milwaukee WI) Lyophilized B cereusspores prepared according to the literature13 were sup-plied by the University of Rhode Island and used as re-ceived Multiple particles approximately 01 mm3 each
were separated and weighed at 5 to 15 mg representing05 to 15 million spores The sample masses were con-sistent with a previous determination of spore density at0081 gmL that indicated a high degree of entrained air
All chemicals used to prepare the silver-doped sol-gelcoated capillaries were also obtained and used as receivedfrom Sigma-Aldrich According to previously publishedprocedures17 two precursor solutions were preparedmixed and then drawn into 1-mm-diameter glass capil-laries The silver amine precursor consisted of a 51 vvratio of 1 N AgNO 3 to 28 NH3OH while the alkoxideprecursor consisted of a 21 vv ratio of methanol to te-tramethyl orthosilicate The alkoxide precursors weremixed with silver amine precursor in an 81 vv ratioApproximately 015 mL was drawn into the capillarycoating a 15-mm length After sol-gel formation the in-corporated silver ions were reduced with dilute sodiumborohydride which was followed by a water wash to re-move residual reducing agent
A 100 mL drop of a 50 mM DDA solution in ethanolpre-heated to 78 8C was added to each of the B cereusparticles to digest the spore coat After 1 minute the re-sultant solution was drawn into a SER-active capillarythat was immediately xed horizontally to an XY posi-tioning stage (Conix Research Spring eld OR) just in-side the focal point of an f 07 aspheric lens The lensfocused the beam into the sample and collected the scat-tered radiation back along the same axis A dichroic lter(Omega Optical Brattleborough VT) was used to re ectthe excitation laser to the lens and pass the Raman scat-tered radiation collected by the lens An f 2 achromatwas used to collimate the laser beam exiting a 200-mm-core-diameter source ber optic while a second f 2 ach-romat was used to focus the scattered radiation into a 365mm ber optic (Spectran Avon CT) A short-pass lterwas placed in the excitation beam path to block the sil-icon Raman scattering generated in the source ber fromre ecting off sampling optics and reaching the detectorA long-pass lter was placed in the collection beam pathto block the sample Rayleigh scattering from reachingthe detector A 785 nm diode laser (Process InstrumentsInc model 785-600 Salt Lake City UT) was used todeliver 100 to 150 mW of power to the sample A Fouriertransform Raman spectrometer (Real-Time Analyzersmodel IRA-785 East Hartford CT) and a silicon photo-avalanche detector (Perkin Elmer model C30902S Stam-ford CT) were used to acquire the SER spectra
RESULTS AND DISCUSSION
As an initial experiment the SER spectrum of 1 gLof DPA in water was measured using the newly devel-oped silver-doped sol-gel-coated capillaries (Fig 1A) Atthis concentration a high signal-to-noise ratio (SN) isobtained in 1 min In fact a reasonable spectrum is ob-tained in the same time frame for 1 mgL (Fig 1B) TheSER spectra are reasonably similar to the normal Raman(NR) spectrum obtained for a saturated solution of DPAin 1 N KOH (Fig 1C) and the following band shifts areobserved (NR to SER) 647 to 657 cm21 817 to 815cm21 998 to 1008 cm21 1384 to 1382 cm21 1434 to1428 cm21 and 1569 to 1567 cm21 Many of these bandshave been previously assigned1213 such as 998 cm21 to
APPLIED SPECTROSCOPY 353
FIG 1 SERS of DPA in water using silver-doped sol-gel-coated glasscapillary for (A) 1 gL and (B) 1 mgL (C ) NR of saturated DPA in 1N KOH in a glass capillary Spectral conditions (A) and (B) 150 mWof 785 nm 1-min acquisition time (C) 450 mW of 785 nm 5-minacquisition time both 8 cm 21 resolution
FIG 2 SERS of DPA extracted from 10 mg B cereus particle using100 mL of 50 mM hot DDA acquired in (A) 1 minute and (B) 2 seconds(C ) Attempted SERS of 50 mM hot DDA in ethanol using silver-dopedsol-gel-coated glass capillary acquired in 1 min Spectral conditions150 mW of 785 nm 8 cm21 resolution
the symmetric ring stretch 1384 cm21 to the OndashCndashOsymmetric stretch 1428 cm21 to the symmetric ring CndashH bend and 1569 cm21 to the asymmetric OndashCndashOstretch
The rst B cereus samples consisted of 2 mg of sporesin 2 mL of 5 mM hot DDA The samples were main-tained at 78 8C for 40 min and while hot approximately10 mL was drawn into a SER-active capillary Since spec-tra of DPA were obtained for these initial samples small-er spore masses higher DDA concentrations and shorterheating periods were examined In due course it wasfound that 10 mg of spores could be digested by 100 mLof 50 mM hot DDA in one minute and detected (Fig2A) In fact the signal was suf ciently intense that it canbe observed in as little as two seconds (Fig 2B) Theamount of DPA that was extracted was estimated to bebetween 5 and 10 mgL by comparing the signal intensityof the 1008 cm21 band to that measured for DPA in waterThis is consistent with previous research that found thatthe majority of the DPA is extracted from spores usingDDA11 and that B cereus spores contain approximately10 DPA by weight18 The SN of 127 for the 1008 cm21
band in the 1-minute SER spectrum suggests a limit ofdetection of approximately 250 ng of B cereus sporesbased on a SN of 3 Finally it should be noted that DDAdid not produce a detectable SER spectrum as shown inFig 2C
CONCLUSION
We have demonstrated that by combining rapid extrac-tion of dipicolinic acid from Bacillus cereus spores withchemical identi cation by surface-enhanced Raman spec-troscopy as little as 10 mg of spores can be detected Infact the entire measurement from the time of adding hotdodecylamine to the spores to the time when the dipi-colinic acid SER spectrum is acquired and analyzedcould be performed in less than two minutes The abilityof this method to distinguish between spore-forming bac-teria such as Bacillus anthracis and non-DPA containingpowders could help prevent costly shutdowns associated
with the appearance of suspicious material or intentionalmailing of common substances as an anthrax hoax Thismethod could also prove useful in detecting the locationof anthrax endospores in mail distribution facilities if an-other veri ed attack should occur
Research continues to fully characterize the surface-enhanced Raman spectroscopy signal intensities as afunction of sample concentration and to explore otherextractants that do not require the use of elevated tem-perature
ACKNOWLEDGMENTS
The authors are grateful for the support of the National Science Foun-dation (DMI-0296116 and DMI-0215819) and the US Army(DAAD13-02-C-0015 Joint Service Agent Water Monitor program)The authors are indebted to Chetan Shende for preparing the sol-gelcapillaries The authors also thank James Gillespie Nicholas Fell andAugustus Fountain for providing important background informationMark Farquharson for laboratory support and Professor Jay Sperry ofthe University of Rhode Island for supplying B cereus spores
1 V A Pasechnik C C Shone and P Hambleton Bioseparations 3267 (1993)
2 P J Jackson M E Hugh-Jones D M Adair G Green K K HillC R Kuske L M Grinberg F A Abramova and P Keim ProcNatl Acad Sci USA 95 1224 (1998)
3 B R Glick and J J Pasternak Molecular Biology Principles andApplications of Recombinant DNA (ASM Press Washington DC1994)
4 C A Bell J R Uhl T L Had eld J C David R F Meyer TF Smith and F R Cockerill III J Clin Microbiol 40 2897(2002)
5 D L Gatto-Menking H Yu J G Bruno M T Goode M Millerand A W Zulich Biosens Bioelectron 10 501 (1995)
6 J J Quinlan and P M Foegeding J Rapid Methods AutomationMicrobiol 6 1 (1998)
7 A A Hindle and E A H Hall Analyst (Cambridge UK) 1241599 (1999)
8 M S Ascher US Department of Health amp Human Services(httpwwwhhsgovophppresentationsAscherdoc)
9 R Nudelman B V Bronk and S Efrima Appl Spectrosc 54 445(2000)
10 D L Rosen C Sharpless and L B McBrown Anal Chem 691082 (1997)
354 Volume 58 Number 3 2004
11 P M Pellegrino N F Fell Jr and J B Gillespie Anal ChimActa 455 167 (2002)
12 W H Woodruff T G Spiro and C Gilvarg Biochem BiophysRes Commun 58 197 (1974)
13 E Ghiamati R S Manoharan W H Nelson and J F SperryAppl Spectrosc 46 357 (1992)
14 A P Esposito C E Talley T Huser C W Hollars C M Schal-dach and S M Lane Appl Spectrosc 57 868 (2003)
15 S Farquharson L Grigely V Khitrov W W Smith J F Sperryand G Fenerty J Raman Spectrosc paper accep ted (2003)
16 S Farquharson W W Smith S Elliott and J F Sperry SPIE-IntSoc Opt Eng 3855 110 (1999)
17 S Farquharson and P Maksymiuk Appl Spectrosc 57 479(2003)
18 F W Janssen A J Lund and L E Anderson Science (Washing-ton DC) 127 26 (1958)
SPIE -2003-5269 117
pH dependence of methyl phosphonic acid dipicolinic acid and cyanide by surface-enhanced Raman spectroscopy
Stuart Farquharson Alan Gift Paul Maksymiuk Frank Inscore and Wayne Smith
Real-Time Analyzers 87 Church Street East Hartford CT 06108
ABSTRACT US and Coalition forces fighting terrorism in Afghanistan and Iraq must consider a wide range of attack scenarios in addition to car bombings Among these is the intentional poisoning of water supplies to obstruct military operations To counter such attacks the military is developing portable analyzers that can identify and quantify potential chemical agents in water supplies at microgram per liter concentrations within 10 minutes To aid this effort we have been investigating the value of a surface-enhanced Raman spectroscopy based portable analyzer In particular we have been developing silver-doped sol-gels to generate SER spectra of chemical agents and their hydrolysis products Here we present SER spectra of methyl phosphonic acid and cyanide as a function of pH an important factor affecting quantitation measurements which to our knowledge has not been examined In addition dipicolinic acid a chemical signature associated with anthrax-causing spores is also presented Keywords Chemical warfare agents agent detection agent hydrolysis SERS Raman spectroscopy homeland security
1 INTRODUCTION In the past decade the Unites States and its allies have been challenged by a different kind of warfare exemplified by the terrorist attacks of September 11 2001 Suicide bombings and the use of chemical agents are the norm and military personnel must consider a wide range of attack scenarios Among these is the intentional poisoning of water supplies to obstruct military operations in Afghanistan and Iraq Currently colorimetric paper is used to detect agents on-site while gas chromatography combined with mass spectrometry (GCMS) is used in mobile support laboratories However both methods have severe drawbacks The paper changes color in response to contact with many chemicals besides CWAs causing a high incidence of false positives1 while GCMS although very chemically specific requires hours to perform and constant re-calibration234 Military operations would be greatly aided by a portable analyzer that can identify and quantify potential chemical agents at concentrations that impact safety This includes the analysis of drinking water supplies distribution and storage systems To meet this goal the Department of Defense has been investigating numerous approaches under the auspices of the Joint Service Agent Water Monitor (JSAWM) program5 The JSAWM must be capable of identifying the chemical agents with no false-positives and quantifying the agents at microgram per liter concentrations within 10 minutes678 Recently we and others have been investigating the ability of surface-enhanced Raman spectroscopy (SERS) to measure chemical agents9-12 bioagents13-17 and their hydrolysis products in water SERS employs the interaction of surface plasmon modes of metal particles with target analytes to increase scattering efficiency by as much as 1 million times18 In 1987 the potential of this approach to measure CWAs was demonstrated by measuring a series of pesticides19 Several of these organophosphonates have chemical structures similar to CWAs in particular P=O functional groups In our studies we have been employing silver-doped sol-gels to promote the SER effect The porous silica network of the sol-gel matrix offers a unique environment for immobilizing and stabilizing SER-active metal particles20-23 The sol-gels have been coated on the internal walls of standard glass vials to allow rapid SER analysis of small sample quantities without preparation We have measured over 100 chemicals with enhancements of 104 to 106 demonstrated reversible measurements in a flowing system reproducible measurements from vial-to-vial and measurements in multiple solvents including water20-23 Previously we used these vials to perform preliminary measurements of cyanide (CN) methylphosphonic acid (MPA) and dipicolinic acid (DPA) MPA is a hydrolysis product of the nerve agents (eg sarin Reaction 1) and may be a valuable indicator of nerve agent usage particularly since the alkyl methylphosphonic acids are relatively more stable than their corresponding parent complexes24 DPA is
SPIE -2003-5269 118
a chemical signature of spore forming bacteria such as Bacillus anthracis And in light of the inability to rapidly detect the anthrax spores distributed through the US mail in October 2001 a number of methods are being developed to extract and analyze this signature Reaction 1 Stepwise hydrolysis of Sarin to form hydrofluoric acid (HF) isopropyl methylphosphonic acid (IMPA) then methyl phosphonic acid (MPA) and 2-propanol In our previous SERS investigations MPA and DPA were measured at 50 and 100 mgL respectively In both cases limits of detection (LOD) were estimated at 100 microgL providing encouragement in that SERS may satisfy the needs of the JSAWM Since it has been shown that pH can substantially influence the intensity of SER bands25 which would clearly influence quantitative analysis we undertook the present study to determine the severity of these effects for cyanide methyl phosphonic acid and dipicolinic acid Furthermore we previously observed a band at 1050 cm-1 for MPA6 possibly due to an anion formed at basic pH Here we investigate the source of this spectral anomaly
2 EXPERIMENTAL All chemicals including potassium cyanide methyl phosphonic acid dipicolinic acid and those used to prepare the silver-doped and gold-doped sol-gel coated vials were obtained and used as received from Sigma-Aldrich All samples were prepared in HPLC grade water (Fischer Scientific Fair Lawn NJ) for SERS measurements The pH of these samples was adjusted using dilute nitric acid or potassium hydroxide and verified using a pH electrode (Corning Inc Corning NY) that had been calibrated with pH 400 700 and 1000 buffered standards from Fischer Scientific Once prepared the samples were transferred into the silver-doped sol-gel vials (Real-Time Analyzers Inc East Hartford CT) The vials were coated in a manner similar to that previously reported by adding ammonium hydroxide to a solution of silver nitrate tetramethyl orthosilicate (TMOS) and methanol20 Gold-vials were coated by adding nitric acid to a solution of gold tetrachloride TMOS and methanol The two precursor solutions were prepared mixed and transferred to 2-mL glass vials dried and heated After sol-gel formation the incorporated metal ions were reduced with dilute sodium borohydride (1mgmL) which was followed by a water wash to remove residual reducing agent After the resultant analyte solution was introduced the SER-active vial was immediately fixed horizontally to an XY positioning stage (Conix Research Springfield OR) just inside the focal point of an f07 aspheric lens The lens focused the beam into the sample and collected the scattered radiation back along the same axis A dichroic filter (Omega Optical Brattleborough VT) was used to reflect the excitation laser to the lens and pass the Raman scattered radiation collected by the lens An f2 achromat was used to collimate the laser beam exiting a 200 microm core diameter source fiber optic while a second f2 achromat was used to focus the scattered radiation into a 365 microm fiber optic (Spectran Avon CT) A short pass filter was placed in the excitation beam path to block the silicon Raman scattering generated in the source fiber from reflecting off sampling optics and reaching the detector A long pass filter was placed in the collection beam path to block the sample Rayleigh scattering from reaching the detector A 785 nm diode laser (Process Instruments Inc model 785-600 Salt Lake City UT) was used to deliver 100 to 150 mW of power to the sample A Fourier transform Raman spectrometer (Real-Time Analyzers model IRA-785 East Hartford CT) and a silicon photo-avalanche detector (Perkin Elmer model C30902S Stamford CT) were used to acquire the SER spectra
3 RESULTS AND DISCUSSION In a previous study of MPA6 aimed at developing a concentration calibration curve and determining limits of detection (LOD) we observed an anomaly at 1050 cm-1 Since it was found that the band intensity changed as a function of concentration the band must be associated with a sample parameter Two possible parameters photon flux and pH are examined here The first parameter was investigated by irradiating a 1mgmL MPA sample in a SER-active vial with laser powers of 200 mW and above and monitoring spectral changes It was immediately found that the 1050 cm-1 band
2O+ H HF + +OH
OH3H C
OP
OF
CH
CH3
3
3H C
OCP
OOH
CH
CH3
3
3H C
O
CPHO
CH
CH
3
3
C
Sarin IMPA MPA 2-propanol
SPIE -2003-5269 119
appeared and grew as a function of time and that the higher the laser power the faster the growth Figure 1 shows the growth of the 1050 cm-1 band over the course of 30 minutes when using 150 mW of 785 nm excitation while Figure 2A shows that the growth can be fit with a first order exponential equation namely I1050 = 03+05e-013t Furthermore the 760 cm-1 band could be fit with a first order decay equation with an identical exponential rate constant ie I760 = 08-
08e-013t The rates represent classical first order kinetics and their correspondence allows one to conclude that MPA is being transformed one-for-one into a photo-generated product At this time the photoproduct has not been positively identified but phosphonic (phosphorous) acid and phosphonate are likely candidates since the symmetric P(OH)3 stretch occurs at ~1050cm-1 Our studies show that a reduction in laser power to 100 mW at the sample for MPA essentially eliminates this degradation process This laser power or lower was used for further measurements
Methyl phosphonic acid is a diprotic acid that stepwise dissociates into two anions MPA- and MPA= according to the following reactions26
MPA MPA- + H+ pKa1 = 212 Reaction 2
MPA- MPA= + H+ pKa2 = 729 Reaction 3 The relative concentrations of MPA MPA- and MPA= can be determined at any pH by expressing [MPA] and [MPA=] in terms of [MPA-] using Reactions 2 and 3 and summing all three to equal the total starting concentration here 2 mgmL (0021M MW = 9602) viz
[MPA] + [MPA-] + [MPA=] = 0021M Equation 1 substituting from Reactions 1 and 2
([H+][MPA-])K1a + [MPA-] + (K2a[MPA-])[H+] = 0021M Equation 2 rearranging [MPA-] = 0021M(1+[H+]K1a + K2a[H+]) Equation 3 The relative concentrations of MPA MPA- and MPA= as a function of pH are shown in Figure 3 It is worth noting that near neutral pH both MPA- and MPA= will be present To confirm that the SER signal followed this pH dependence a starting solution consisting of 20 mg of MPA in 10 mL HPLC grade water was prepared and brought to pH of 20 using dilute nitric acid From this solution 2 mL were added to a SER-active vial and the SER spectrum recorded At this pH a peak at 760 cm-1 was barely discernable The 2 mL solution was returned to the starting solution and the pH was re-measured to correct for any changes that the silver-doped sol-gel vials might cause In most cases the change was less than 02 pH units and the pH is reported as the before and after average Next the pH of the
Figure 2 A) Exponential growth of 1050 cm-1 band and B) exponential decay of 760 cm-1 band for spectral series in Figure 1
Figure 1 Growth of 1050 cm-1 band as a function of time due to exposure to 150 mW of 785 nm Spectra are 5 sec each collected every 100-sec from 0 to 30-min
B
570 770 970 1170 1370Raman Shift (cm-1)
Arbit
rary
Unit
s
MPA Photodegradation
600 800 1000 1200 1400 Wavenumber (cm-1)
A
0 10 20 30 0 10 20 30 time (min) time (min)
Ram
an In
tens
ity (r
elat
ive)
30
min
0
SPIE -2003-5269 120
starting solution was adjusted to 325 using dilute KOH Again 2 mL were added to a vial and the SER spectrum recorded At this pH a reasonably strong 760 cm-1 band was observed This process was repeated as spectra were recorded at pHs of 70 74 75 79 85 and 100 A total of 1 mL of KOH was added diluting the total concentration by 10 Next the pH of the starting solution was made acidic by adding dilute nitric acid dropwise This time spectra were recorded at pHs of 72 69 64 and 37 Figure 4 shows the SER spectra for representative pHs (spectra were left out to simplify the figure) while Figure 3 shows the 760 cm-1 peak intensities as a function of pH (The band intensities were adjusted to compensate for dilution effects caused by the addition of HNO3 and KOH then normalized to 0021 M for the most intense band observed at pH 37) It is clear from Figure 3 that the 760 cm-1 band follows the MPA- concentration as a function of pH and must be assigned to this anion No bands were observed that corresponded to MPA or MPA= The lack of an MPA SER spectrum may be due to the absence of an attraction between the neutral analyte and the electropositive silver surface The same reasoning suggests that a strong SER spectrum should be observed for MPA= but it is not and a satisfactory explanation has not been found
0000
0005
0010
0015
0020
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14pH
Con
cent
ratio
n [M
]
pK1 = 212 pK2 = 729
MPA- MPA=
MPA
Figure 4 SER spectra of 002M MPA as a function of pH Conditions 100 mW of 785 nm 36 scans (1 min) 8 cm-1 recorded 2 min after sample introduction pH 19 and 20 not apparent on this scale pH 69 and 74 near identical to 70 and 75 and not shown for clarity
00
02
04
06
08
10
12
14
16
18
0 5 10 15 20 25 30Measurement Number
Ram
an In
tens
ity (7
60 c
m-1
)
Figure 3 Concentration dependence of MPA MPA- and MPA= as a function of pH for a 002M sample Intensity of 760 cm-1
band from Figure 6 as a function of pH ( for increasing basic adjustment for increasing acidic adjustment error was measured at ~10 for pH 64)
Figure 5 SER spectra of 002M MPA at pH 64 measured around a vial at three heights (9 points per height) Conditions as in Figure 4 but 10-sec scans
Since these measurements involved the removal and replacement of the SER-active vial in the sample holder to remove and add sample variation in the intensity as a function of vial position was minimized by illuminating the exact same height along the vial wall But this does not account for variability of the SERS response of the sol-gel coating around the vial To analyze this effect a vial containing MPA at pH 64 was rotated at ~ 40o
intervals at the original height and 18rdquo above and below this value Figure 5 shows the intensity of the 760 cm-1
band for the 27 positions It was found that the average value was 137plusmn014 an RSD of 10 overall and 5 for each height An error bar is included in Figure 4 for the pH 64 measurement
SPIE -2003-5269 121
With the above analyses in mind a preliminary investigation of the SER spectral response for MPA (as MPA-) as a function of concentration was performed A single vial was used for these measurements beginning with 1 mgL followed by measurements of 10 100 and 1000 mgL In all cases the pH was ~7 and 3 positions around the vial were measured per concentration Since the 760 cm-1 band was not observed for concentrations of 1 or 10 mgL using 100 mW of 785 nm the laser power at the sample was raised to 200 mW beginning with the 10 mgL concentration Photo-degradation was largely avoided (and not observed) by exposing the sample for only 33 seconds per spectral acquisition Representative spectra for 10 100 and 1000 mgL are shown in Figure 6 while a plot of the 760 cm-1 band intensity as a function of concentration is shown in Figure 7 These values were also used to estimate limits of detection based on the signal-to-noise ratio (SN) of the 760 cm-1 band using 100 mW of 785 nm laser excitation and a 10-min acquisition time As summarized in the Figure 7 inset the lower the measured concentration the lower the predicted LOD Even if as estimated 210 microgL could be measured using the silver-doped sol-gel vials an improvement of a factor of 70 is still required to meet JSAWM goals of measuring 3 microgL in 10-minutes Similar to MPA DPA is a diprotic acid (pKa1 = 216 and pKa2 = 692) and variations in pH will effect the relative concentrations of DPA DPA- and DPA= and possibly the SER spectra and band intensities This could prove significant if an acid or base is used to denature anthrax spores with the goal of extracting and analyzing DPA The MPA pH study described above was mimicked for DPA except that the starting solution consisted of 20 mg of DPA in 20 mL HPLC grade water (60x10-3M MW = 1671) The initial solution had a pH of 245 which was made basic by dilute KOH to pHs of 355 433 487 559 1069 and 1166 SER spectra were recorded at each pH using 100 mW of 785 nm and a 44-sec acquisition time Next one drop of concentrated nitric acid was used to remake the solution acidic at a pH of 200 Again sequential pH measurements were performed at 383 510 735 and 822 The solution pH was made acidic a third time but to pH 219 171 then 135 Throughout this process no more than 20 drops of acid or base were added and therefore the concentration was diluted by no more than 10 Most of the spectral bands showed a minor decrease in intensity as a function of increasing pH values However the bands at 525 and 795 cm-1 showed the most dramatic changes which occurred at acid pH Figure 8 shows the SER spectra of DPA for the spectral region and pH range of interest The identity of the DPA species was determined by plotting the normalized peak intensities with the lowest value set to 0 and the highest to 0006 M as a function of pH and overlaying these values on a plot of the relative concentrations for DPA DPA- and DPA= as previously done for MPA (Figure 9) As can be seen the 525 cm-1 band clearly corresponds to DPA The correspondence of the 795 cm-1 band to this species is less clear as the band retains intensity until pH of 55 This can be attributed to contribution to the overlapping band at 810 cm-1 which does not change as a function of pH The fact that most bands are observed at all pHs suggest that the primary interaction with silver is through the ring nitrogen This is supported by the fact that the most intense band occurs at 1008 cm-1 attributed to a symmetric ring breathing mode and that this interaction has been characterized for pyridine in numerous papers27
0
1
2
3
4
5
6
7
0 200 400 600 800 1000 1200
MPA Concentration (mgL)
760
Ban
d In
tens
ity
Figure 6 SER spectra of MPA in water at A) 1000 B) 100 and C) 10 mgL Conditions pH of 7 silver-doped sol-gel coated vial 200 mW of 785 nm 33-sec 8 cm-1 resolution
Figure 7 Plot of SER intensity of 760 cm-1 band of MPA as a function of concentration using 200 mW of 785 nm Inset table includes average intensity LOD standard deviation and percent deviation for each concentration but for 100 mW and 10 min
conc (mgL) LOD ave stddev dev10 021 002 1021100 072 010 14301000 312 040 1280
A
B C
SPIE -2003-5269 122
The ability to detect dipicolinic acid (DPA) by SERS is demonstrated in Figure 10 Even at 1 mgL the primary bands are visible The intensity of the dominant ring breathing mode observed at 1008 cm-1 for DPA in water (pH = 25 to 55) is plotted as a function of concentration in Figure 11 These values were also used to estimate limits of detection based on the SN of the 1008 cm-1 band using 100 mW of 785 nm laser excitation and a 10-min acquisition time Again the lower the measured concentration the lower the predicted LOD (see Figure 11 inset) and detection of 160 microgL is possible Gastrointestinal anthrax requires significant more spores than inhalation anthrax28 and a limit of detection might be placed at 1 million spores in 1 liter of water or 10 microgL Since each spore contains ~10 CaDPA by weight29 a goal for DPA might be 1 microgL indicating that the present measurements must be improved by nearly two orders of magnitude Due to the increased hazards of handling HCN gas KCN salt was used for these experiments Nevertheless all sample preparations were performed in a chemical hood KCN completely dissolves in water but its conjugate acid HCN is formed and has a Ka of 615x10-1030 viz
HCN CN- + H+ pKa = 921 Reaction 4
Figure 8 SER spectra of 1 mgmL DPA as a function of pH Conditions 100 mW 785 nm 100 scans (44-sec)
Figure 9 Concentration dependence of DPA DPA- and DPA= as a function of pH for a 0006M sample Intensity of 525 () and 795 (diams) cm-1 bands from Fig 8 as a function of pH
pH
135 171 219 383
0
0001
0002
0003
0004
0005
0006
0007
0 2 4 6 8 10 12 14pH
Con
cent
ratio
n [M
]
DPADPA-DPA=795525
DPA DPA=DPA-
pK2 = 692pK1 = 216
795 525
0
05
1
15
2
25
0 200 400 600 800 1000 1200
DPA Concentration (mgL)
1008
Ban
d In
tens
ity
conc (mgL) lod-10min-100mw1 017
10 016100 103
1000 355
Figure 10 SER spectra of DPA in water at A) 1000 B) 100 C) 10 and D) 1 mgL Conditions pH of 25-55 silver-doped sol-gel coated vial 175 mW of 785 nm 1-min 8 cm-1 D) has been multiplied by x10 to make bands visible
A
B C
D
Figure 11 Plot of SER intensity of 1008 cm-1 band of DPA as a function of concentration using 175 mW of 785 nm Inset table includes LOD in mgL for each concentration but for 100 mW and 10 min
SPIE -2003-5269 123
Consequently the cyanide concentration must be determined for each initial KCN concentration Specifically the samples prepared with concentrations of 01 1 10 100 and 1000 mgL of KCN produced CN- concentrations of 63x10-3 033 69 89 and 964 mgL at pHs of 816 90 967 102 and 107 respectively The pH dependence for the HCN and CN- concentrations are shown in Figure 12 Thus as the amount of KCN added to the solution decreases so does the pH of the solution (becomes less basic) and according to Reaction 4 the relative amount of CN- to HCN also decreases For example in the preparation of a 01 mgL solution of KCN the pH is shifted from 7 for pure water to only 816 and only 63 of the starting material becomes CN- or 63x10-3 mgL In comparison for a solution of 1000 mgL the pH is shifted from 7 to 107 and 96 of the starting material becomes CN- This is significant since the cyanide ion is better able to adsorb onto the silver particles and become SERS active SER spectra of 10 100 and 1000 mgL of KCN samples are shown in Figure 13 and a corresponding plot of the 2100 cm-1 band intensity as a function of concentration is shown in Figure 14
000102030405060708091011
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14pH
Con
cent
ratio
n [m
gm
L]
pKa = 921
CN -HCN
The SER spectra of cyanide are dominated by the single CN stretch at 2100 cm-1 which occurs in normal Raman spectra of solutions at 2080 cm-1 However a low frequency mode occurs at 135 cm-1 due to a Ag-CN stretch (not shown) indicative of a strong surface interaction It is also observed that as the concentration decreases the CN stretch shifts to 2140 cm-1 This shift has been attribute to the formation of a tetrahedral Ag(CN)3
2- surface structure31 as well
Figure 14 Concentration dependence of KCN SERS measured under conditions in Fig 11 Concentrations are 1 01 and 001 mgml Intensities are measured for the CN stretch at 2100 cm-1 Inset table includes LOD in mgL for each concentration in Figs 13 and 15 but for 100 mW and 10 min
Figure 13 SER spectra of KCN in water at A) 1000 B) 100 and C) 10 mgL Conditions pHs of 107 102 and 97 silver-doped sol-gel coated vial 100 mW of 785 nm 1-min 8 cm-1 C) has been multiplied by x10 to make band visible
Figure 12 Concentration dependence of HCN and CN- as a function of pH for a 1 mgmL sample Calculated intensity of 2100 () cm-1 band for a 1 mgml sample at pHs of 816 90 967 102 and 107
A
B
C
0
20
40
60
80
100
120
140
0 200 400 600 800 1000 1200CN Concentration (mgL)
2100
Ban
d In
tens
ity
Figure 15 SER spectra of KCN in water at A) 10 B) 1 and C) 01 mgL Conditions pHs of 97 12 and 12 gold-doped sol-gel coated vial 100 mW of 785 nm 1-min 8 cm-1 C) has been multiplied by x10 to make band visible
A
B
C
conc (mgL) lod-10min-100mw Condition01 001 Au-pH 121 007 Au-pH 12
69 003 Au-pH 9769 016 Ag-pH 9789 022 Ag-pH 102946 113 Ag-pH 107
SPIE -2003-5269 124
as to CN adsorbed to two different surface sites32 Alternatively the 2140 cm-1 band could be attributed to HCN since this species dominates at lower concentrations However it is unlikely that this species would be attracted to the electropositive silver surface Further both peaks should be present at pHs between 85 and 105 but this is not observed It has also been suggested that at concentrations near and above monolayer coverage the CN- species is forced to adsorb end-on due to crowding and at lower concentrations the molecule can reorient to lie flat33 This suggests that the 2100 and 2140 cm-1 bands correspond to the end-on and flat orientations respectively As Figures 13 and 14 show the intensity of the CN stretch for the 89 mgL sample is nearly as intense as the 964 mgmL band This suggests that the Raman signal for the flat orientation is more enhanced However more extensive measurements are required to verify this point Since resent research has suggested that cyanide may be more effectively detected on gold measurements of KCN solutions were also performed using gold-doped sol-gel vials Preliminary measurements are shown in Figure 15 for samples prepared from 01 1 and 10 mgL KCN Since the pHs are 816 90 967 the resultant CN- concentrations are 63x10-3 033 and 69 mgL Initially only the highest concentration was observed and the signal intensity was significantly better than the equivalent concentration measured using silver In an effort to shift Reaction 4 to the left transforming HCN to CN- (Le Chatelierrsquos principle) KOH was added to the lower concentration samples producing solutions with pH 12 This effectively forces all of the cyanide in solution to be CN- or 01 and 10 mgL respectively More importantly the CN stretch is now observed in the SER spectra The band appears at 2125 cm-1 as has been previously reported for gold12 As calculated for MPA and DPA LODs can be estimated from this data For the three concentrations of cyanide on silver the LODs are 016 to 11 mgL for 100 mW of 785 nm laser excitation and a 10-min acquisition time For gold pH adjusted the LODs are10 to 70 microgL an improvement of more than 10 times silver Nevertheless either substrate is sufficient to meet the JSAWM goals of measuring 3 mgL in 10-minutes as the requirements form cyanide are much less stringent than the nerve agents
4 CONCLUSIONS Here we examined the surface-enhanced Raman spectral response for methyl phosphonic acid dipicolinic acid and cyanide as a function of pH It was determined that the most prominent peak at 760 cm-1 reaches a maximum intensity between pH 3 and 7 and corresponds to the MPA- species Neither the MPA nor MPA= species appear to generate a SER spectrum and consequently no spectra were observed below pH 2 or above pH 8 In this study we also found that higher laser powers could cause photodegradation of MPA signified by the exponential growth of a band at 1050 cm-1 which is tentatively assigned to phosphorous acid Unlike MPA DPA was observed at all pHs This is attributed to the dominant interaction of the pyridine functional group with silver Minor spectral changes were observed at acid pHs and were assigned to neutral DPA Like MPA SER spectra of cyanide were pH dependent No spectra were observed for the HCN species while CN- was best observed at pHs more basic than 8 Preliminary concentration studies for the three analytes allowed estimating limits of detection for MPA DPA and CN using 100 mW of 785 nm and a 10-min acquisition time of 210 165 and 70 microgL respectively Although the latter value suggests that cyanide can be measured at sufficiently low concentrations to meet JSAWM goal improvements by 100 to 200 times are required for MPA and DPA It is clear from this study that pH of the sample is important and must be taken into account when developing concentration calibration curves as well as fieldable analyzers Future work will include tailoring the sol-gel with specific functional groups to dictate sample pH and thereby optimize sensitivity
ACKNOWLEDGMENTS The authors are grateful for the support of the US Army (DAAD13-02-C-0015 Joint Service Agent Water Monitor program) The authors would also like to thank Dr Steve Christensen of the US Army SBCCOM for helpful discussions and Mr Chetan Shende of Real-Time Analyzers for assistance in development of the gold-doped sol-gels
REFERENCES 1 Erickson B Analytical Chemistry News amp Features June 1 397A (1998)
SPIE -2003-5269 125
2 Johnston RL Hoefler CM Fargo JC and Moberley B AT-ONSITE 5-8 (1994) 3 Black RM RJ Clarke RW Read and MT Reid J Chromatography 662 301-321 (1994) 4 USA Army Center for Health Promotion and Preventive Medicine (USACHPPM) Technical Guideline 230A 5 Jensen JL NC Wong W R Loerop SPIE 4775-03 (2001) 6 Hoenig SL Handbook of Chemical Warfare and Terrorism Greenwood Press Wesport CT (2002) 7 Munro NB SS Talmage GD Griffin LC Waters AP Watson JF King and V Hauschild Env Health
Persp 107 933-974 (1999) 8 Holstege CP Kirk M Sidell FR Crit Care Clin13 923-42 (1997) 9 Farquharson S P Maksymiuk K Ong and S Christesen SPIE 4577 166-173 (2001) 10 Lee Y and S Farquharson SPIE 4378 21-26 (2001) 11 Spencer KM J Sylvia S Clauson and J Janni SPIE 4577 158-165 (2001) 12 Tessier P S Christesen K Ong E Clemente A Lenhoff E Kaler and O Velev Applied Spectroscopy 56
1524-1530 (2002) 13 Farquharson S WW Smith S Elliott and JF Sperry SPIE 3533 207-214 (1998) 14 Farquharson S WW Smith S Elliott and JF Sperry SPIE 3855110-116 (1999) 15 Farquharson S WW Smith YH Lee S Elliott and JF Sperry SPIE 4575 62-72 (2002) 16 Guzelian AA J Sylvia J Janni S Clauson and KM Spencer SPIE 4577 182-192 (2001) 17 Shende C F Inscore A Gift P Maksymiuk and S Farquharson in press 18 Weaver MJ S Farquharson and MA Tadayyoni J Chem Phys 82 4867-4874 (1985) 19 Alak AM and T Vo-Dinh Analytical Chemistry 59 2149-2153 (1987) 20 Lee Y and S Farquharson SPIE 4206 140-146 (2000) 21 Farquharson S and Y Lee SPIE 4200-16 (2000) 22 Lee Y S Farquharson and P M Rainey SPIE 3857 76-84 (1999) 23 Lee Y S Farquharson H Kwong and M Shahriari SPIE 3537 252-260 (1998) 24 Wang J M Pumera G Collins and A Mulchandani Analytical Chemistry 74 6121-6125 (2002) 25 Dou X YM Jung Z-Q Cao and Y Ozaki Applied Spectroscopy 53 1440-1447 (1999) 26 Data supplied by S Christesen and K Ewing 27 Kerker M and B Thompson Eds SPIE MS 10 (1990) 28 Inglesby TV DA Henderson JG Bartlett JAMA 287 2236 (2002) 29 FW Janssen AJ Lund and LE Anderson Science 127 26 (1958) 30 Lide DR Ed Handbook of Chemistry and Physics CRC Press 77th Ed (1996-7) 31 Billmann J G Kovacs and A Otto Surf Sci 92 153 (1980) 32 Murray CA and S Bodoff Phys Rev B 32 671 (1985) 33 Kellogg D and J Pemberton J Phys Chem 91 1120 (1987)
SPIE-2003-5269 16
Chemical agent detection by surface-enhanced Raman spectroscopy
Stuart Farquharson Alan Gift Paul Maksymiuk Frank Inscore and Wayne Smith
Real-Time Analyzers 87 Church Street East Hartford CT 06108
Kevin Morrisey and Steven D Christesen US Army SBCCOM Aberdeen Proving Ground MD 21010
ABSTRACT
In the past decade the Unites States and its allies have been challenged by a different kind of warfare exemplified by the terrorist attacks of September 11 2001 Although suicide bombings are the most often used form of terror military personnel must consider a wide range of attack scenarios Among these is the intentional poisoning of water supplies to obstruct military operations in Afghanistan and Iraq To counter such attacks the military is developing portable analyzers that can identify and quantify potential chemical agents in water supplies at microgram per liter concentrations within 10 minutes To aid this effort we have been investigating the value of a surface-enhanced Raman spectroscopy based portable analyzer In particular we have been developing silver-doped sol-gels to generate SER spectra of chemical agents and their hydrolysis products Here we present SER spectra of several chemical agents measured in a generic tap water Repeat measurements were performed to establish statistical error associated with SERS obtained using the sol-gel coated vials Keywords Chemical agents chemical agent detection SERS Raman spectroscopy
1 INTRODUCTION
In the autumn of 2001 terrorism within US borders became a sobering reality While extensive efforts are being implemented to secure the homeland US and Coalition forces in Afghanistan and Iraq are constantly faced with terrorist attacks In addition to car-bombings the military has identified several non-traditional attack scenarios including poisoning of water supplies by chemical warfare agents (CWAs) To counter this threat the Department of Defense has been investigating numerous approaches to detect such attacks under the auspices of the Joint Service Agent Water Monitor (JSAWM) program1 The JSAWM must be capable of identifying the chemical agents with no false-positives and quantifying the agents at microgram per liter concentrations within 10 minutes2 This includes the analysis of drinking water supplies distribution and storage systems Currently colorimetric paper is used to detect agents on-site while gas chromatography combined with mass spectrometry (GCMS) is used in mobile support laboratories However both methods have severe drawbacks The paper changes color in response to contact with many chemicals besides CWAs causing a high incidence of false positives3 while GCMS although very chemically specific requires up to an hour to perform and regular re-calibration456 Vibrational spectroscopy has also been investigated as a potential method of rapidly detecting CWAs7-
11 as early as 1985 in the case of infrared spectroscopy8 and 1988 in the case of Raman spectroscopy11 Again however these techniques also have limitations when it comes to measuring trace poisons in water Infrared spectra would be dominated by the very strong absorption of water which would obscure absorptions by most other chemicals present Whereas Raman spectroscopy is simply not a very sensitive technique and detection limits are typically grams per liter Surface-enhanced Raman spectroscopy (SERS) offers several advantages over conventional vibrational methods and that may provide the necessary sensitivity required for detecting trace quantities of chemical agents in water SERS employs the interaction of surface plasmon modes of metal particles with target analytes to increase scattering efficiency by as much as 1 million times12 In 1987 the potential of this approach to measure CWAs was demonstrated by measuring a series of pesticides13 Several of these organophosphonates have chemical structures similar to CWAs in particular P=O functional groups In the past few years we and others have further explored the ability of SERS to detect CWAs14-17 and even bioagents 18-21 We have been employing silver-doped sol-gels to promote the SER effect
SPIE-2003-5269 17
in these studies The porous silica network of the sol-gel matrix offers a unique environment for immobilizing and stabilizing SER-active metal particles22-25 The silver-doped sol-gels have been coated on the internal walls of standard glass vials to allow rapid SER analysis of small sample quantities without preparation We have measured over 100 chemicals with enhancements of 104 to 106 demonstrated reversible measurements in a flowing system reproducible measurements from vial-to-vial and measurements in multiple solvents including water21-25 Previously we used these vials to perform preliminary measurements of cyanide (CN) and methylphosphonic acid (MPA) Most of the nerve agents form MPA during hydrolysis while Tabun forms CN a chemical agent in its own right In another paper including in these proceedings we examined the limits of detection (LOD) for MPA by measuring a series of concentrations down to 50 mgL and estimated a limit of detection of 100 microgL26 These measurements provide encouragement in that SERS may satisfy the needs of the JSAWM To further establish the viability of SERS in particular silver-doped sol-gels here we present analysis of cyanide mustard and VX in tap water The measurements performed at the US Armyrsquos Edgewood Chemical Biological Center Aberdeen MD also included numerous repeat measurements to establish reproducibility
2 EXPERIMENTAL 2a General All chemicals including potassium cyanide 2-chloroethylethyl sulfide and those used to prepare the silver-doped and gold-doped sol-gel coated vials were obtained and used as received from Sigma-Aldrich All samples were prepared in a chemical hood using HPLC grade water unless otherwise noted (Fischer Scientific Fair Lawn NJ) for SERS measurements Once prepared the samples were transferred into the silver-doped sol-gel vials (Real-Time Analyzers Inc East Hartford CT) The vials were coated in a manner similar to that previously reported by adding ammonium hydroxide to a solution of silver nitrate tetramethyl orthosilicate (TMOS) and methanol22 The two precursor solutions were prepared mixed and transferred to 2-mL glass vials dried and heated After sol-gel formation the incorporated metal ions were reduced with dilute sodium borohydride (1mgmL) which is followed by a water wash to remove residual reducing agent After the resultant analyte solution was introduced the SER-active vial was fixed horizontally to an XY positioning stage (Conix Research Springfield OR) just inside the focal point of an f07 aspheric lens The lens focused the beam into the sample and collected the scattered radiation back along the same axis A dichroic filter (Omega Optical Brattleborough VT) was used to reflect the excitation laser to the lens and pass the Raman scattered radiation collected by the lens An f2 achromat was used to collimate the laser beam exiting a 200 microm core diameter source fiber optic while a second f2 achromat was used to focus the scattered radiation into a 365 microm fiber optic (Spectran Avon CT) A short pass filter was placed in the excitation beam path to block the silicon Raman scattering generated in the source fiber from reflecting off sampling optics and reaching the detector A long pass filter was placed in the collection beam path to block the sample Rayleigh scattering from reaching the detector A 785 nm diode laser (Process Instruments Inc model 785-600 Salt Lake City UT) was used to deliver 100 to 150 mW of power to the sample A Fourier transform Raman spectrometer (Real-Time Analyzers model IRA-785 East Hartford CT) and a silicon photo-avalanche detector (Perkin Elmer model C30902S Stamford CT) were used to acquire the SER spectra
2b Edgewood Chemical Biological Center The surface-enhanced Raman spectral measurements at the US Armyrsquos Chemical Biological center presented here were all performed on September 12 2003 To expedite measurements a plate was machined to hold up to 12 SER-active sample vials (Figure 1) The plate fit a standard XY plate reader that could be programmed Pure KCN bis-(2-chloroethyl)sulfide (distilled mustard gas HD) and ethyl S-2-diisopropylamino ethyl methylphosphonothioate (VX) were obtained on-site and used to prepare 1 gL tap water solutions in a chemical hood with appropriate safety equipment Simulated tap water was prepared by adding 10 chemicals most often found in tap water at appropriate concentrations to distilled water (Table 1) SER measurements were also performed in a chemical hood For added safety the FT-Raman instrument was placed outside the laboratory and 30 foot fiber optic and electrical cables were used to allow remote SERS measurements and plate manipulation For each experiment 1gL samples were prepared and added to 9 individual vials which were then loaded on the plate In some cases a tenth vial was included as a blank
Table 1 Generic Tap Water Composition
Compound mgL NaHCO3 100 CaSO4 27 MgSO4bull7H2O 67 NaNO3 1 Fulvic Acid 1 K2HPO4 07 KH2PO4 03 (NH4)2HSO4 001 NaCl 001 FeSO4 0001 pH 76-78
SPIE-2003-5269 18
A software program was written that allowed selecting the sequence that the vials were measured the number of positions along the length of the vials to measure (1 to 5) and the number of scans to co-add During sample analysis the program displayed the vial being analyzed the point being analyzed and the spectrum as it was being acquired Once all the data was collected a second software program was written to rapidly analyze the data The spectra collected for all the vials on a plate could be loaded at one time and then the spectra for each point could be displayed simultaneously or separately The user could then select the Raman peak to analyze in terms of peak height or area This was accomplished by selecting points on either side of the peak to define a baseline of zero The peak height or area could then be computed for all of the spectra loaded and then exported to a spreadsheet for statistical analysis
Figure 1 A) Vial Holder 6 slots to hold 2 vials each end-to-end B) Measurement Configuration Program user selects vials to measure sequence number of points per vial (1 to 5) and number of scans per point C) Spectral Acquisition Program shows spectrum being collected which vial and position D) Spectra Analysis Program user selects spectra to analyze by plate vial and point (s) as well as two wavenumbers defining the peak and the baseline to subtract The image is of 5 repeat measurements of 10 mgL KCN in generic tap water 16 sec each 100 mW of 785 nm
3 RESULTS AND DISCUSSION Raman and surface-enhanced Raman spectra were obtained for potassium cyanide bis-(2-chloroethyl)sulfide and ethyl S-2-diisopropylamino ethyl methylphosphonothioate representing three classes of chemical agents cyanides mustards and nerve agents respectively Spectra were also obtained for 2-chloroethyl ethyl sulfide (CEES) a structural analogue to HD which was included in the study to aid in assigning spectral bands KCN salt was used for cyanide experiments to avoid the increased hazards of handling HCN gas KCN completely dissolves in water forming its conjugate acid HCN according to its Ka of 615x10-1027 and at a concentration of 1 mgmL results in a pH 107 solution This is significant in that only CN- appears to interact sufficiently with silver to produce a SER spectrum and no spectral signal is observed below pH 726 Figure 2 shows the SER and normal Raman spectra for KCN The SERS spectrum of 1mgml KCN in water shows a single intense somewhat broad feature at
A B
C D
SPIE-2003-5269 19
2100 cm-1 assigned to the single CequivN stretch The band is much sharper in the normal Raman spectra of the solid KCN salt at 2074 cm-1 This band does broaden and shift to 2080 cm-1 in solution (not shown) However the observed SERS frequency is attributed to interaction with silver and low frequency mode at 135 cm-1 attributed to a Ag-CN stretch (not shown) supports this conclusion
cm-1 band to a C-S stretch but the authors concede that it is in fact more likely a C-Cl stretch28 It appears that the most intense Raman bands at 648 692 and 747 cm-1 shift to 620 660 and 720 cm-1 in the SER spectra and are tentatively assigned as above The width of these bands suggests that they overlap underlying spectral features Additional bands in the Raman spectra occur at 972 1034 1049 1263 1286 1423 1442 2865 2935 and 2960 cm-1 Corresponding bands occur in the SER spectra at 964 1015 1054 1286 1410 1447 2865 and 2935 cm-1 Most of these bands are associated with alkane modes specifically the bands at approximately 1040 cm-1 to a C-C stretch 1290 cm-1 to a CH2 in-phase twist 1440 cm-1 to a CH2 wag 2865 cm-1 to a symmetric CH2 stretch and 2965 cm-1 to an asymmetric CH2 stretch The Raman and SER spectra of sulfur mustard were measured at the Edgewood center (Figure 4) Both spectra are largely similar to CEES The C-Cl and C-S bands in the Raman spectrum of HD now occur at 640 655 700 739 and 760 cm-1 and are more resolved possibly due to the increased molecular symmetry Theoretical calculations indicate that the first three bands are due to C-Cl stretching modes and the latter two to C-S stretching modes28 Only the C-Cl bands maintain significant intensity in the SER spectra occurring at 624 and 643 cm-1 which is attributed to the
Prior to measurements of HD CEES was examined by Raman and SER spectroscopy (Figure 3) CEES also known as half-mustard is essentially identical to HD except one of the chlorine end atoms is replaced by a hydrogen atom Again although not as toxic as HD CEES is a blister agent and dilute aqueous samples were prepared using appropriate safety equipment Both the Raman and SER spectra of CEES are similar and dominated by bands between 600 and 800 cm-1 These are associated with C-Cl and C-S stretching modes which are tentatively assigned to 648 and 747 cm-1 in the Raman spectra respectively The shoulder at 630 cm-1 the overlapped band at 660 cm-1 and the strong band at 692 cm-1 could also be due to these modes or their asymmetric counterparts It is worth noting that theoretical calculations assign the 692
Figure 2 A) SER and B) NR spectra of KCN Conditions A) 1 mgml in tap water 100 mW of 785 nm at sample 1-min acquisition time B) solid 300 mW of 785 nm 5-min All spectra are 8 cm-1 resolution
A
B
Figure 3 A) SER and B) NR spectra of CEES Conditions A) 1 vv (10 mgml) in MeOH 100 mW of 785 nm 1-min acquisition time B) neat 300 mW of 785 nm 5-min
A
B
Cl-CH2-CH2-S-CH2-CH3
A
B
Figure 4 A) SER and B) NR of HD Conditions A) 1mgml in tap water B) pure both 100 mW of 785 nm 1-min
Cl-CH2-CH2-S-CH2-CH2-Cl
SPIE-2003-5269 20
expected strong interaction between chlorine and silver and adds support to the assignment of this band to a C-Cl stretch Weaker overlapping bands occur at 670 692 and 724 cm-1 the latter possibly due to C-S stretching modes Again the alkane modes are apparent in the normal Raman spectra of HD but only a broad feature at 1300 to 1450 cm-1 suggests CH2 contributions in the SER spectrum Although the observed bands in the VX spectrum have not been assigned (Figure 4) a computer generated Raman spectrum29 predicts many of the same features with surprising accuracy and are used here Two intense bands at 460 and 530 cm-1 closely match predicted bands at 463 and 546 cm-1 assigned to a CH3-P=O bend and a PO2CS wag Three highly overlapped bands occur at 694 745 and 771 cm-1 matching predicted bands at 713 730 and 760 cm-1 The first
Table 2 Measured SER peak heights for the CN stretch at 2100 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
Figure 5 A) SERS and B) NR spectra of VX Conditions A) 1 vv (10 mgml) in MeOH B) pure sample both 100 mW of 785 nm 1-min acquisition time
A
B
two have been assigned to a C-S stretch and CH2bend respectively while the latter has been attributed to either a P-C stretch or an O-C-C stretch Although the 745 cm-1 band may alternatively be assigned to a C-S stretch based on the previous measurements of CEES and HD The relatively intense bands at 890 1106 1218 1445 and 1465 cm-1 also match predicted bands at 880 1108 1216 1440 and 1464 cm-1 that are assigned to a C-C stretch CH3 rock N-C3 stretch various C-H3 bends and C-H bends respectively Both the computer generated and the measured spectra contain numerous other less intense bands One is worth mentioning A unique band appears at 370 cm-1 that is predicted at 368 cm-1 and corresponds to an O-P=O bend The surface-enhanced Raman spectrum of VX is also rich with spectral features It has the unique low frequency band at 370 cm-1 as well as a second band at 380 cm-1 that is assigned tothe S-P-O bend predicted in the normal Raman spectrum at 388 cm-1 Based on the measured and predicted normal Raman spectra the following SERS assignments are given 460 cm-1 to the CH3-P=O bend 544 cm-1 to the PO2CS wag 738 cm-1 to a C-S stretch (based on arguments above) 890 cm-1
to a C-C stretch 1101 cm-1 to a CH3 rock and 1456
cm-1 to a C-H bend The ability of SERS to measure chemical agents in water containing real-world chemical interferents was tested by using the generic tap water described in Table 1 The ability to reproduce measurements was accomplished by preparing three separate water stock solutions which were used to prepare three sample solutions each of 1 mgmL KCN in the generic tap water The samples defined as vials 1-9 were then measured at 4 points per vial in a semi-random fashion such that errors associated with stock solution preparation and errors associated with instrument drift could be identified No trends were apparent that signified such systematic errors Each spectrum collected consisted of 20 averaged scans taking 16 seconds at 8 cm-1 resolution The laser power at the sample was measured periodically during the day and it ranged from 102 to 105 mW spacccc
spot Vial 1 Vial 2 Vial 41 13754 13004 128192 13519 12692 12909
1 3 13541 12721 126394 13462 12648 12651
ave 13569 12766 12755Vial 6 Vial 3 Vial 5
1 11586 14068 10757 2 11236 14402 115122 3 11376 14546 11559
4 10894 11763 11214ave 11273 13695 11261
Vial 7 Vial8 Vial 91 11198 15804 11023
2 11243 15716 88753 3 11562 15216 11218
4 11655 15084 11077ave 11415 15455 10548
CN AVG STDEV ERR1 mgmL pk ht 12526 1560 1245
stock solution
SPIE-2003-5269 21
The CN measurements consisting of 36 data points produced an average height of 125 for the 2100 cm-1 peak with a standard deviation of 156 or 125 (Table 2) The HD and VX measurements were performed precisely the same way (Tables 3 and 4) For HD the 624 cm-1 peak was used for analysis and it had an average height of 53 with a standard deviation of 068 or 129 while for VX the 544 cm-1 peak was used for analysis and it had an average height of 1051 with a standard deviation of 308 or 293 The greater error in the VX measurements can be somewhat attributed to Vial 4 which produced lower SER signal intensities But removing this vial from the data set changes the standard deviation to 233 only a modest improvement
spot Vial 1 Vial 2 Vial 41 525 609 4682 45 675 484
1 3 527 644 5094 559 807 669
ave 51525 68375 5325Vial 6 Vial 3 Vial 5
1 521 574 575 2 536 53 4492 3 509 508 451
4 565 594 379ave 53275 5515 4635
Vial 7 Vial 8 Vial 91 457 589 413
2 583 497 4273 3 544 505 4
4 528 5 509ave 528 52275 43725
HD AVG STDEV ERR1 mgmL pk ht 530 068 1291
stock solution
spot Vial 1 Vial 2 Vial 41 1464 1034 4672 1485 99 61
1 3 1491 989 5684 1041 777 553
ave 137025 9475 5495Vial 6 Vial 3 Vial 5
1 1058 942 1293 2 697 121 965
2 3 727 1206 11124 689 1254 746
ave 79275 1153 1029Vial 7 Vial 8 Vial 9
1 1127 783 163 2 1358 812 1614
3 3 1371 908 15124 1097 875 1397
ave 123825 8445 153825
VX AVG STDEV ERR1 mgmL pk ht 1051 308 2925
stock solution
4 CONCLUSIONS In this paper we examined the ability of surface-enhanced Raman spectroscopy to reproducibly measure CN HD and VX in tap water without chemical interference Both normal and surface-enhanced Raman spectra were examined to select unique bands suitable to identify and quantify these chemical agents For SER measurements the 2100 cm-1 C-N stretch was used for CN the 624 cm-1 C-Cl stretch was used for HD and the 544 cm-1 PO2CS wag was used for VX It was determined that 1 mgmL samples of each of these chemicals measured 36 times in glass vials coated with a silver-doped sol-gel reproduced measurements with standard deviations of 125 129 and 293 It was further found that the 10 chemicals added to simulate generic tap water did not interfere with or alter the SER spectra It should be noted that the concentrations used in this study were considerably greater than those required by the JSAWM program Current work involves improving SER sensitivity and designing sampling systems with better reproducibility This includes the development of fractal silver and gold structures within the sol-gel matrix and the development of chemically selective sol-gels
ACKNOWLEDGMENTS The authors are grateful for the support of the US Army (DAAD13-02-C-0015 Joint Service Agent Water Monitor program) The authors would also like to thank Janet Jensen Ronald Crosier and Kristina Gonser for helpful discussions
Table 4 Measured SER peak heights for the VX band at 544 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
Table 3 Measured SER peak heights for the HD band at 624 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
SPIE-2003-5269 22
REFERENCES 1 Jensen JL NC Wong W R Loerop SPIE 4775-03 (2001) 2 JSAWM Requirements at wwwsbccomapgeaarmymilRDAecbcrtPRODSERJSAWMjsawmhtml 3 Erickson B ldquoThe Chemical Weapons Convention Redefines Analytical Challengerdquo Anal Chem 70 397A (1998) 4 Johnston RL Hoefler CM Fargo JC and Moberley B ldquoThe Defense Nuclear Agencyrsquos
ChemicalBiochemical Weapons Agreements Verification Technology Research Development Test and Evaluation Program and its Requirements for On-Site Analysisrdquo AT-ONSITE 5-8 (1994)
5 Black RM Clarke RJ Read RW and Reid MT ldquoApplication of GC-MS and GC-tandem MS to the analysis of chemical warfare samples found to contain residues of the nerve agent sarin sulphur mustard and their degradation productsrdquo J Chrom 662 301-321 (1994)
6 USA Army Center for Health Promotion and Preventive Medicine (USACHPPM) Technical Guideline 230A 7 Hoffland LD Piffath RJ Bouck JB rdquoSpectral signatures of chemical agents and simulantsrdquo Optical
Engineering 24 982-984 (1985) 8 Braue EHJ Pannella MGrdquoCIRCLE CELL FT-IR Analysis of Chemical Warfare Agents in Aqueous Solutionsrdquo
App Spectrosc 44 1513-1520 (1990) 9 Tseng C-H Mann CK and Vickers TJ ldquoDetermination of Organics on Metal Surfaces by Raman
Spectroscopyrdquo App Spectrosc 47 1767-1771 (1993) 10 Staff Reporter ldquoUS Army Tests Lidar to Detect Biological Toxinsrdquo Photonic Spectra pg 50 December 1998 11 Christesen SD Raman cross sections of chemical agents and simulants App Spectrosc 42 318-321 (1988) 12 Weaver MJ Farquharson S Tadayyoni MA ldquoSurface-enhancement factors for Raman scattering at silver electrodesrdquo J Chem Phys 82 4867-4874 (1985) 13 Alak AM and Vo-Dinh T ldquoSERS of Organophosphorous Chemical Agentsrdquo Anal Chem 59 2149-2153 (1987) 14 Lee Y Farquharson S ldquoRapid chemical agent identification by SERSrdquo SPIE 4378 21-26 (2001) 15 Farquharson S Maksymiuk P Ong K Christesen S ldquoChemical agent identification by surface-enhanced Raman spectroscopyrdquo SPIE 4577 166-173 (2001) 16 Spencer KM Sylvia J Clauson S and Janni J ldquoSurface Enhanced Raman as a Water Monitor for Warfare
Agents in Waterrdquo SPIE 4577 158-165 (2001) 17 Tessier P Christesen S Ong K Clemente E Lenhoff A Kaler E Velev O ldquoOn-line spectroscopic characterization of sodium cyanide with nanostructured gold surface-enhanced Raman spectroscopy substratesrdquo
App Spectrosc 56 1524-1530 (2002) 18 Farquharson S W W Smith S Elliott and J F Sperry Rapid biological agent identification by surface- enhanced Raman spectroscopy SPIE 3855110-116 (1999) 19 Farquharson S WW Smith YH Lee S Elliott and J F Sperry Detection of biological signatures A
comparison of electrolytic and metal-doped sol-gel surface-enhanced Raman media SPIE 4575 62-72 (2002) 20 Guzelian AA Sylvia J Janni J Clauson S and Spencer KM ldquoSERS of whole cell bacteria and trace levels of biological moleculesrdquo SPIE 4577 182-192 (2001) 21 Shende C Inscore F Gift A Maksymiuk P Farquharson S ldquoRapid dipicolinic acid extraction from Bacillus spores detected by surface-enhanced Raman spectroscopyrdquo App Spectrosc 58 accepted 22 Lee Y and Farquharson S ldquoSERS Sample Vials Based on Sol-Gel Process for Trace Pesticide Analysisrdquo SPIE
4206 140-146 (2000) 23 Farquharson S and Lee Y ldquoTrace Drug Analysis by SERSrdquo SPIE 4200-16 (2000) 24 Lee Y Farquharson S and Rainey P M Surface-Enhanced Raman Sensor for Trace Chemical Detection in
Water SPIE 3857 76-84 (1999) 25 Lee Y Farquharson S Kwong H and Shahriari M ldquoSol-Gel Chemical Sensor for Surface-Enhanced Raman
Spectroscopyrdquo SPIE 3537 252-260 (1998) 26 Farquharson S A Gift P Maksymiuk F Inscore and W Smith ldquopH dependence of methyl phosphonic acid
dipicolinic acid and cyanide by surface-enhanced Raman spectroscopyrdquo SPIE 5269-19 (2003) 27 Lide DR Ed Handbook of Chemistry and Physics CRC Press 77th Ed (1996-7) 28 Sosa C RJ Bartlett K KuBulat and WB Person ldquoA theoretical study of harmonic vibrational frequencies and
infrared intensities of XCH2CH2SCH2CH2X and XCH2CH2SH (= H Cl)rdquo J Phys Chem 93 577-588 (1993) 29 Hameka HF and JO Jenson ldquoComputer-generated predictions of the structure and the IR and Raman spectra of
VXrdquo ERDEC-TR-065 May 1993
- Chemagents Appendicespdf
-
- SPIE2001-4575-Bioagent-SERSapdf
-
- Detection of bioagent signatures A comparison of electrolytic and metal-doped sol-gel surface-enhanced Raman media
- Stuart Farquharson Wayne Smith and Yuan Lee
-
- Real-Time Analyzers 87 Church Street East Hartford CT 06108
-
- ABSTRACT
- 4 CONCLUSIONS
- 5 ACKNOWLEDGEMENTS
- 6 REFERENCES
-
166
Chemical agent identification by surface-enhanced Raman spectroscopy
Stuart Farquharson and Paul Maksymiuk
Real-Time Analyzers 87 Church Street East Hartford CT 06108
Kate Ong and Steven D Christesen US Army SBCCOM Aberdeen Proving Ground MD 21010
ABSTRACT The recent distribution of anthrax through the US postal system and the subsequent infection and death of several postal and national media employees amplifies the need for methods to rapidly detect identify and quantify this and other chemical and biological warfare agents The US military has also identified water supplies as a likely method of warfare agent deployment and is funding the development of a Joint Service Agent Water Monitor (JSAWM) In an effort to aid military personnel and the public at large we are developing a portable analyzer capable of identifying and quantifying chemical agents rapidly either on-demand or continuously The approach is based on the ability of Raman spectroscopy to identify molecular structure through the abundant vibration information provided in spectra and the ability of surface-enhanced Raman spectroscopy (SERS) to detect extremely low concentrations (eg part-per-billion) through the enhancement of Raman scattering by six orders of magnitude or more A key element to the analyzer design is a new SER active medium that is capable of quantitative reversible measurements The medium consists of silver or gold nanoparticles incorporated into a sol-gel matrix The porous silica network offers a unique environment for stabilizing SER active metals and the high surface area increases the interaction between the analyte and metal particles Here we present the use of new sol-gels that also selectively enhance chemicals based on polarity and charge Base-line measurements of chemical agents and their hydrolysis products are presented and compared to the JSAWM goal of 30 micrograms per liter detection Keywords Chemical warfare agent hydrolysis product SERS Raman spectroscopy sol-gel nanoparticle
1 INTRODUCTION Since September 11 2001 the threat of terrorist attacks and biological warfare within US borders has become a sobering reality The simplicity in manufacturing ease of deployment and the relatively low cost of chemical warfare agents (CWAs) raises public concern that they may also be used by terrorists Indeed terrorists released sarin (GB) in the Tokyo subway in 19951 Countering terrorism and terrorist attacks requires recognizing likely deployment scenarios and having the required technology to rapidly detect the deployment event One method of deployment has been long identified by the US military distribution through water supplies To counter this threat the Department of Defense is funding or monitoring the capabilities of 200 technologies with the goal of developing a Joint Service Agent Water Monitor (JSAWM) that is field portable2 The JSAWM must be capable of identifying the chemical agents with no false-positives and quantifying the agents at microgram per liter concentrations within 10 minutes (Table 1)3 This includes the analysis of drinking water supplies distribution and storage systems as well as potable water supplies The earliest technologies developed for CWA detection were based on electrochemical ionization or colorimetric analysis (eg phosgene tape) Although these analyzers were easy to use they were not generally agent specific and suffered from false-positives4 More traditional laboratory methods have also been investigated and in particular combined gas chromatography and mass spectrometry (GCMS) has been very successful at eliminating false-positives56 However GCMS requires extraction repeated calibration and long analysis times (typically 20 to 60 minutes)6 making it labor intensive and less than desirable for field use More rapid analysis of agents in the solid liquid and gas phase has been To whom correspondence should be addressed emailfarqureal-time-analyzerscom
Vibrational Spectroscopy-based Sensor Systems Steven D Christesen Arthur J Sedlacek III EditorsProceedings of SPIE Vol 4557 (2002) copy 2002 SPIE middot0277-786X02$1500
167
Table 1 Chemical Agent Structures Hydrolysis Half-lives and JSAWM Thresholds Agent Short-Hand Chemical Structure Hydrolysis
Half-Life JSAWM
Thresholds
Sarin (GB) F-[O=P-CH3]-O-CH(CH3)2 213 hours 32 microgL Soman (GD) F-[O=P-CH3]-O-CH(CH3)-(C-(CH3)3) 23 hours 32 microgL Tabun (GA) (CH3)2-N-[O=P-CN]-O-C2H5 41 hours 32 microgL VX C2H5O-[O=P-CH3]-S-(CH2)2-N-(CH(CH2)2)2 821 hours 32 microgL EA2192 HO-[O=P-CH3]-S-(CH2)2-N-(CH(CH2)2)2 gt9 years 32 microgL Mustard (H) ClCH2CH2-S-CH2CH2Cl encapsulates 47 microgL Lewisite (L) ClCH=CH-As-Cl2 rapid 27 microgL HCN HCN rapid 20 mgL BZ C7NH12-O-[C=O]-COH(C6H5)2 23 microgL T-2 Toxin 87 microgL at pH 7 to 75 and 20 to 25 oC
demonstrated by vibrational spectroscopy7-10 Hoffland et al7 reported infrared absorbance spectra and absolute Raman cross sections for several chemical agents while Christesen measured Raman cross sections for sarin tabun mustard gas and VX (ethyl S-2-diisopropylamino ethyl methylphosphonothioate)11 Again however these techniques also have limitations Raman spectroscopy is simply not a very sensitive technique and detection limits are typically 01 (1000 ppm) While infrared spectroscopy would have limited value in analyzing poisoned water since the very strong infrared absorption of water would obscure most other chemicals present Nevertheless efforts to overcome these limitations have been demonstrated Braue and Pannella8 quantified the G-series nerve agents (tabun sarin and soman) in terms of infrared attenuated total reflectance using a circle-cell And Alak and Vo-Dinh demonstrated the possibility of surface-enhanced Raman spectroscopy (SERS) to identify CWAs by measuring several organophosphonates that simulate the nerve agents12 However quantitative measurements have not been demonstrated for the SER-active material used (silver coated on alumina particles) or other SER-active media13 Recently we developed silver-doped sol-gels to promote the SER effect14-17 The porous silica network of the sol-gel matrix offers a unique environment for stabilizing SER-active metal particles and the sol-gel provides a high surface area that effectively increases the number of molecules observed within the Raman scattering volume The silver-doped sol-gels have been coated on the internal walls of standard glass vials to allow rapid SER analysis of small sample quantities (lt 01 mL) without preparation We have used p-aminobenzoic acid (PABA) as a test chemical to demonstrate surface enhancements greater than 106 reversible measurements in a flowing system reproducible measurements from vial-to-vial and batch-to-batch and measurements in multiple solvents including water14-17 Recently we used these vials to measure Tabun (GB) and Sarin and several hydrolysis products pinacolyl methyl phosphonate (PMP from Soman) and methyl phosphonic acid (MPA from all G-agents Figure 1) Although a number of unique vibrational bands are observed (eg C-N stretch doublet and P-C stretch) the G-agents were only observed for 5 concentrations and all spectra required baseline corrections Figure 1 Surface enhanced Raman spectra of ~5 vv A) Tabun and B) Sarin C) 1 vv PMP and D) 10 ppm MPA using sol-gel sample vials 785 nm excitation 1-min scan and CCD detection Performed at Aberdeen Proving Ground
C A
Wavenumber (∆cm-1) Wavenumber (∆cm-1)
D B
790
545 C-N
2135 2190 P-C 770
760 1290
Proc SPIE Vol 4577
168
Nevertheless MPA was readily observed for a 10 ppm sample with an estimated detection limit of 04 ppm (based on a signal-to-noise ratio of 3 for the 760 cm-1 band intensity) This measurement provides encouragement in that SERS may satisfy the needs of a JSAWM Furthermore MPA is also a hydrolysis product of VX and V-gas and EA2192 (Figure 2) and may prove a valuable indicator of agent usage Figure 2 Hydrolysis of Sarin to form hydrofluoric acid (HF) methylphosphonic acid 1-methylethyl ester (MPAMME) methyl phosphonic acid (MPA) and 2-propanol With this initial albeit modest success we began analyzing chemicals with various sol-gel compositions that we have been developing Here we describe four sol-gel compositions that select for 1) polar-positive 2) polar-negative 3) weakly polar-positive and 4) weakly polar-negative chemical species The ability of these sol-gels to select and enhance Raman scattering is described for several test chemicals and MPA
2 EXPERIMENTAL The chemicals analyzed as well as all chemicals used to prepare the metal-doped sol-gels were obtained at their purest commercially available grade from Aldrich (Milwaukee WI) The sol-gel designed to select for polar-negative species was prepared from a silver amine complex tetramethyl orthosilicate (TMOS) and methanol After mixing 02 mL of the sol-gel solution was transferred into a glass vial (2 mL) dried and heated The incorporated silver ions were then reduced using dilute sodium borohydride The vials were washed and dried prior to the addition of a sample solution In a similar manner the sol-gel designed to select for polar-positive species was prepared from a gold salt TMOS and methanol The sol-gel designed to select for weakly polar-negative species was prepared from a silver amine complex tetraethyl orthosilicate (TEOS) and methanol And the last sol-gel designed to select for weakly polar-positive species was prepared from a gold salt TEOS and methanol All samples were prepared in a chemical hood and transferred into plain or SER-active vials for analysis Normal Raman spectral measurements employed 1-mL pure samples that were placed in a 1-cm3 cuvette and weighed This yielded a powder density that allowed accurate calculation of molecules in the optical collection field SERS measurements employed 1-mg sample per mL water concentrations unless otherwise stated Once prepared a 01 mL sample was placed into one of the four selective sample vials which in turn was placed into the sample compartment of a Raman spectrometer for analysis A prototype Fourier transform Raman spectrometer (RTA) was used for these measurements18 The system consisted of a NdYAG laser (Brimrose) for excitation at 1064 nm an interferometer built by On-Line Technologies (OLT East Hartford CT) for frequency separation an uncooled InGaAs detector for signal detection (RTA) and an Intel 400 MHz Pentium II based laptop computer (Dell Round Rock TX) for interferometric control data acquisition (OLT) and analysis (LabVIEW by National Instruments Austin TX) Additional components included a Notch filter (Kaiser Ann Arbor MI) and interferometer entrance and exit optics (Edmund Scientific Barrington NJ) Fiber optics were used to deliver the excitation beam to the sample and the scattered radiation to the interferometer (1 meter lengths of 200 and 365 micron core diameter respectively Spectran Avon CT) A second Notch filter (Kaiser) was used as a beam splitter to direct the excitation beam along the same axis as the collected radiation A microscope object (20x04 Newport Irvine CA) was used to focus the beam into the sample and to collect the scattered radiation back along the same axis In this co-axial backscattering arrangement the excitation beam was passed through the outside of a glass vial and focused onto the silver-doped sol-gel film (01-03 mm thickness) containing the sample
3 RESULTS AND DISCUSSION p-aminobenzoic acid (PABA) and phenyl acetylene (PA) and were used to refine the selectivity and SER-activity of the four different metal-doped sol-gels PABA is a popular chemical used to evaluate the performance of SER-active media Here the polar end groups can be used to test selectivity of the polar-negative and polar-positive sol-gels PA is essentially non-
2O+ H HF + +OH
OH3H C
OP
OF
CH
CH3
3
3H C
OCP
OOH
CH
CH3
3
3H C
O
CPHO
CH
CH
3
3
C
Sarin MPAMME MPA 2-propanol
Proc SPIE Vol 4577
169
polar but a high electron density in the cylindrical π cloud around the carbon-carbon triple bond allows testing the selectivity of the weakly polar-negative and weakly polar-positive sol-gels As Figure 3 illustrates PABA passes through the polar sol-gel and is enhanced by either the silver or gold particles At 1 mgml the concentration of neutral PABA is ca 20 times that of the ionized form (pKa = 48) For electropositive silver the PABA anion is expected to interact through the carboxylate group and the associated vibrational modes are expected to dominate the spectrum Conversely for electronegative gold either form of PABA is expected to interact through the amine group The clear differences in our spectra support this expectation Furthermore bands at 840 and 1405 cm-1 assigned to a COO- bend and stretch respectively are significantly more intense for silver than gold Additional bands at 1140 and 1195 cm-1 are assigned to CH bending modes while bands at 1450 1500 and 1605 cm-1 are assigned to ring vibrational modes A very similar SER spectrum for PABA on a silver-coated alumina substrate has previously been reported with similar assignments19 For the gold-doped sol-gel new bands appear at 690 1355 and 1585 cm-1 The first band is assigned to a ring-H bending mode the second band to a ring-N- stretching mode and the third band to a possible NH2 scissors mode or ring mode The second band is not observed in the normal Raman spectrum but infrared bands occur at this frequency for aromatic ring-secondary amine stretching modes The scissors mode occurs at this frequency in Raman spectra for several chemicals but is absent in the PABA Raman spectrum Alternatively this mode may be the1600 cm-1 ring mode that has been shifted by the gold interaction Again a very similar SER spectrum of PABA has been reported but surprisingly using silver (colloids)2021 not gold as the enhancement medium These researchers also assumed the primary interaction of PABA with silver was through the carboxylate anion and made assignments accordingly For example they assigned the 1359 cm-1 to a COO- stretch not to the amine group as we have They also favor the ring stretching mode assignment for the 1582 cm-1 band Finally it should be said that other researchers have argued that the most dominant band in the SER spectra at 1450 cm-1 a ring vibration mode suggests that PABA lies flat on the surface and the π-orbitals dominate the surface interaction22
Figure 3 SER spectra of A) PABA using polar-negative and B) polar-positive sol-gels and C) PA using weakly polar-negative and D) weakly polar-positive sol-gels PABA is 1 mgmL PA is 1 vv Spectral conditions 75 mw 1064 nm 100 scans (15 min) 8 cm-1 resolution Non-polar PA passes through the non-polar sol-gels and is also enhanced by both metals The spectra are easily understood For electropositive silver PA interacts through the cylindrical triple bond π electron cloud and a -CequivC- doublet occurs near 2000 cm-1 The interaction is reasonably strong since this band appears at 2112 cm-1 in the normal Raman spectrum For electronegative gold this interaction is unlikely and only very weak bands occur near 2000 cm-1 The remaining bands are at 1000 cm-1 1200 cm-1 doublet and 1595 cm-1 all appear in the normal Raman spectra at virtually the same frequencies and are assigned to the symmetric ring-breathing mode CH bending modes and the trigonal ring-breathing mode respectively The polarnon-polar selectivity of the polar-negative and weakly polar-negative sol-gels was tested by adding a 11 molar mixture of PABA and PA The selective enhancement is quite good (Figure 4) The spectrum obtained using the polar sol-gel represents 78 PABA and 22 PA while the spectrum obtained using the weakly polar sol-gel represents 9 PABA and 91 PA The band peak intensities at 2000 cm-1 for PA and 1450 cm-1 for PABA were used for these calculations and are expanded in Figure 4 for clarity
C A
Wavenumber (∆cm-1) Wavenumber (∆cm-1)
2NH COOH
D B
C CH
Proc SPIE Vol 4577
170
Figure 4 SERS of 11 MM of PABA and PA in A) polar-negative and C) weakly polar-negative sol-gels The lower traces compare the pure chemicals B) 1 mgml PABA in polar-negative sol-gel and D) 1 PA in weakly polar-negative sol-gel while the insets magnify the minority species for clarity (x5 in A and x10 in B) Spectral conditions as in Figure 3 Following this development of selective sol-gels that maintained SER activity we measured cyanide and MPA (Figure 5) Not surprisingly the best sensitivity for both hydrolysis products was obtained using the polar-negative sol-gel The interaction of the cyanide anion with the silver surface is sufficient to shift the CequivN stretch observed at 2080 cm-1 in the normal Raman spectrum to 2145 cm-1 in the surface-enhanced Raman spectrum Furthermore the band is substantially broadened This anion has been extensively studied by electrolytic SERS and this shift and broadening have been attributed to the formation of a tetrahedral Ag(CN)3
2- surface structure23 Figure 5 Surface-enhanced (upper traces) and normal Raman spectra (bottom traces) of A) CN- and B) MPA in silver-doped TMOS SERS conditions as in Figure 3 and 1 mgmL Note MPA yields two distinct spectra for neutral (top) and acidic pH (middle) The normal Raman spectra employed pure powders 500 scans and 900 mW of 1064 nm SER measurements of MPA with the polar-negative sol-gel yielded two unique spectral signatures that depended on solution pH (Figure 5) For more neutral solutions the P-C stretch of MPA at 762 cm-1 dominates and the CH2 stretch at 2922 cm-1 appears The SN is sufficiently high that the anti-Stokes Raman shift at -762 cm-1 is observed For deprotonated MPA an oxygen-surface mode appears at 325 cm-1 (as well as its anti-Stokes complement) suggesting a strong interaction This results in substantial enhancement of the P-O-C mode at 1051 cm-1 (upper trace) Others report that this mode dominates the infrared spectra of nerve agents measured in water8 Comparison of the two spectra suggests the following molecule-to-surface orientations The appearance of the oxygen-surface and P-O-C modes in the upper spectrum of Figure 5B indicates that the tetrahedral molecule interacts with the silver surface through the deprotonated oxygen and is oriented end-on The
A B
Wavenumber (∆cm-1)
C A
Wavenumber (∆cm-1) Wavenumber (∆cm-1)
D B
Wavenumber (∆cm-1)
Proc SPIE Vol 4577
171
dominance of the P-C and the CH2 stretches and the disappearance of the P-O-C mode in the upper spectrum suggest the molecule is oriented with the methyl-hydroxyl-hydroxyl face towards the surface However considerably more research must be performed to verify these points Both chemical forms were measured as a function of concentration to determine the onset of monolayer coverage and estimate expected detection limits (Figure 6) Below monolayer coverage the signal to concentration dependence should be linear and the SN of any spectral measurement in this range can be used to predict the detection limit In the spectra presented here the peak height was used as the signal while the noise as root-mean-squared (RMS) was measured between 4400-4600 cm-1 Since noise is distributed evenly throughout the spectrum when transformed this region was used since it does not have any contributions from signals or baseline offsets Figure 6 shows a series of spectra for MPA along with a plot of the 1050 cm-1 band intensity (the noise was identical for this series) as a function of sample concentration A clear discontinuity appears in the vicinity of 01 mgmL (19 ppm) indicating the onset of monolayer coverage A detection limit defined as a SN of 3 was calculated for the 01 and 005gmL samples at 24x10-4 and 25x10-4 gL respectively A more modest detection limit of 101x10-4 gL was obtained using the 760 cm-1 band in the second series of concentration measurements These detection limits correspond to 46 and 190 ppb for a 3-min scan and 75 mW of incident laser power
Figure 6 A) Concentration dependence of MPA SERS measured in silver-doped TMOS) B) Concentrations are 001 005 01 05 1 gL (188 94 188 94 188 ppm) I760 series (bull) and I1050 series (∆)
Enhancement factors for cyanide and methyl phosphonic acid can be determined by estimating the number of molecules contributing to the surface-enhanced and normal Raman spectra The enhancement factor EF can be defined by the following equation
EF = (ISERSINR)bull(MNRMSERS) bull(PNRPSERS) bull(TNRTSERS)12
where I is the spectral band intensity M is the sample mass P is the incident laser power and T is the measurement time (or number of scans) for the two measurements For the normal Raman spectra a cylindrical scattering volume is assumed based on the laser area (28x10-7m2 6x10-4m diameter spot) and the penetration depth (1x10-3 m)24 The density of KCN and MPA as powders were measured at 0572 and 0516 gcm3 indicating that 16x10-4 and 144x10-4 g produced the normal Raman signals in Figure 5 respectively The number of molecules contributing to the SER spectra are those on the silver particles embedded in the sol-gel The total silver surface area can be determined from the average particle size concentration and the scattering volume Previous scanning electron micrographs showed the average silver particle size to be 40 nm in diameter (335x10-23m3)17 The silver concentration is 012M based on the reactant molar concentrations and dilution factors And the scattering volume is 76x10-
11m3 again based on a cylindrical scattering volume defined by a laser area of 28x10-7m2 and a sol-gel thickness of 27x10-
4m This volume contains 123x10-6g of silver equivalent to 35x109 silver particles with a collective surface area of 18x10-
5m2 However it may reasonably be assumed that at least half of the silver surface is in contact with the sol-gel matrix and unavailable for analyte interaction If we assume monolayer coverage and that each CN molecule occupies 15x10-20m2 then approximately 62x1014 molecules or 27x10-8g of CN contribute to the SER spectrum (20x10-19m2 46x1013 molecules 74x10-9g for MPA) Accordingly the EF for cyanide equals 48x104 ((180599) bull(16x10-427x10-8) bull(90075) bull(500100)12) The EF for MPA is considerably higher at 87x106 ((60326) bull(144x10-474x10-9) bull(90075) bull(500200)12)
0
20
40
60
80
100
120
140
0 02 04 06 08 1 12[MPA] (mgmL)
I (76
0)
0
100
200
300
400
500
600
I (10
50)
Wavenumber (∆cm-1)
A B
Proc SPIE Vol 4577
172
4 CONCLUSIONS Here we present for the first time surface-enhanced Raman spectra of Tabun and Sarin obtained using silver-doped sol-gels However the inferior enhancement suggested employing SER-active sol-gels varying in composition to improve both detection limits and selectivity of the target analytes To this end we successfully demonstrated the capabilities of four sol-gels that select for 1) polar-positive 2) polar-negative 3) weakly polar-positive and 4) weakly polar-negative chemical species p-aminobenzoic acid was used to show that silver could be used to attract polar-negative chemicals or functional groups (carboxylate anion) and that gold could be used to attract polar-positive chemicals (amine) while a mixture of p-aminobenzoic acid and phenylacetylene was used to show that tetramethyl orthosilicate preferentially solvates polar chemicals and that tetraethyl orthosilicate preferentially solvates weakly polar chemicals This increased sample control was applied to cyanide and methyl phosphonic acid two hydrolysis products of chemical warfare agents Exceptional results were obtained for methyl phosphonic acid allowing measurement of 1x10-2 gL for a 3-min scan and 75 mW of incident laser power with an estimated detection limit of 245x10-4 gL and an enhancement factor of 87x106 However this detection limit is 76 times less sensitive than required for the JSAWM (32x10-6gL for the G-agents) Finally we note that the measurements performed here employed an FT-Raman spectrometer This instrumentation was chosen over dispersive Raman instrumentation for the high wavelength accuracy afforded by the HeNe reference laser (Connes Advantage25) which would allow reliable spectral subtraction matching of observed spectra to stored library spectra and confident use of chemometric approaches Such data analysis is likely to be required to identify the chemical agents as well as distinguish hydrolysis products However this instrumentation which employs 1064 nm excitation and InGaAs detection sacrifices sensitivity We believe that the measurement sensitivity can be improved by at least two orders of magnitude by using shorter laser excitation wavelengths (eg 532 nm) This would provide more efficient Raman scattering (fourth power dependence on laser excitation wavelength) more efficient generation of plasmon modes and allow using more efficient detector material (Si vs InGaAs) These modifications are underway
5 ACKNOWLEDGEMENTS The authors would like to thank Dr R Yin and J Jensen of the US Army for supporting this work (Contract Number DAAD13-01-C-0019) They would also like to thank Advanced Fuel Research for making their laboratory facilities available
6 REFERENCES 1 Tu Anthony ldquoOverview of Sarin Terrorist Incidents in Japan in 1994 and 1995rdquo 6th CBW Protection Symposium
Stockholm Sweden 10-15 May 1998 2 Jensen JL NC Wong W R Loerop SPIE 4775-03 (2001) 3 JSAWM Requirements at wwwsbccomapgeaarmymilRDAecbcrtPRODSERJSAWMjsawmhtml 4 ldquoThe Chemical Weapons Convention Redefines Analytical Challengerdquo Analytical Chemistry News amp Features June 1
397A (1998) 5 Johnston RL Hoefler CM Fargo JC and Moberley B ldquoThe Defense Nuclear Agencyrsquos ChemicalBiochemical
Weapons Agreements Verification Technology Research Development Test and Evaluation Program and its Requirements for On-Site Analysisrdquo AT-ONSITE 5-8 (1994)
6 Black RM Clarke RJ Read RW and Reid MT ldquoApplication of gas chromatography-mass spectrometry and gas chromatography-tandem mass spectrometry to the analysis of chemical warfare samples found to contain residues of the nerve agent sarin sulphur mustard and their degradation productsrdquo J Chromatography 662 301-321 (1994)
7 Hoffland LD Piffath RJ Bouck JBrdquoSpectral signatures of chemical agents and simulantsrdquo Optical Engineering 24 982-984 (1985)
8 Braue EHJ Pannella MGrdquoCIRCLE CELL FT-IR Analysis of Chemical Warfare Agents in Aqueous Solutionsrdquo Applied Spectroscopy 44 1513-1520 (1990)
9 Tseng C-H Mann CK and Vickers TJ ldquoDetermination of Organics on Metal Surfaces by Raman Spectroscopyrdquo Applied Spectroscopy 47 1767-1771 (1993)
10 Staff Reporter ldquoUS Army Tests Lidar to Detect Biological Toxinsrdquo Photonic Spectra pg 50 December 1998
Proc SPIE Vol 4577
Proc SPIE Vol 4577
173
11 Christesen SD Raman cross sections of chemical agents and simulants Applied Spectroscopy 42 318-321 (1988) 12 Alak AM and Vo-Dinh T ldquoSERS of Organophosphorous Chemical Agentsrdquo Analytical Chemistry 59 2149-2153
(1987) 13 Norrod KL Sudnik LM Rousell D and Rowlen KL ldquoQuantitative Comparison of Five SERS Substrates
Sensitivity and Detection Limitrdquo Applied Spectroscopy 51 994-1001 (1997) 14 Lee Y and Farquharson S ldquoSERS Sample Vials Based on Sol-Gel Process for Trace Pesticide Analysisrdquo SPIE 4206
140-146 (2000) 15 Farquharson S and Lee Y ldquoTrace Drug Analysis by Surface-Enhanced Raman Spectroscopyrdquo SPIE 4200-16 (2000) 16 Lee Y Farquharson S and Rainey P M Surface-Enhanced Raman Sensor for Trace Chemical Detection in Water
SPIE 3857 76-84 (1999) 17 Lee Y Farquharson S Kwong H and Shahriari M ldquoSurface-Enhanced Raman Sensor for Surface-Enhanced Raman
Spectroscopyrdquo SPIE 3537 252-260 (1998) 18 Farquharson S Smith W Carangelo R C and Brouillette C ldquoIndustrial Raman Providing Easy Immediate Cost
Effective Chemical Analysis Anywhererdquo SPIE 3859 14-23 (1999) 19 Narayanan VA JM Bello JD Stokes and T Vo-Dinh Analusis 19 307-310 (1991) 20 Laserna JJ E L Torres and JD Winefordner Analytica Chemica Acta 469-480 (1987) 21 Torres EL and JD Winefordner Analytical Chemistry 59 1626-1632 (1987) 22 Suh JS DP DiLella M Moskovits J Phys Chem 87 1540-1544 (1983) 23 Benner RE R Dornhaus R Chang and BL Laube Correlations in the Raman spectra of cyanide complexes adsorbed
at silver electrodes with voltammograms Surface Science 101 341 (1980) 24 Chase D B and JF Rabolt Fourier Transform Raman Spectroscopy Acad Press Ch1 p 131 (1994) 25 Connes J Rev Opt Theor Instrum 40 45 (1961)
APPLIED SPECTROSCOPY 351
focusing the 488 nm laser beam 2 mm deep into thebulk of the crystals to avoid contributions from potentialdepletion layers As shown in Fig 1b the dependence ofthe Raman shift on the degree of deuteration is almostperfectly linear and ts very well with D 5 22684cmR1 24526 where D is the degree of deuteration (in )and R is the spectral mean of the PO4 vibration in cm21A linear correlation coef cient of 0998 indicates an ex-cellent linear dependence of the Raman peak shift withdegree of deuteration This result shows that the shift ofthe PO 4 peak is simply caused by the linear increase inatomic mass due to isotope substitution which decreasesthe length of hydrogen-like bonds
This excellent linear dependence allows us to map thepro le of the DH exchange layer at the surface of DKDPcrystals by acquiring Raman spectra and determining theposition of the PO 4 peak for various depths This methodis preferable over other methods such as determining thestrength of the OD vibration directly (eg at 715 cm21)because the position of the most intense peak in the Ra-man spectrum can be measured more precisely than theintensity of some of the weakest peaks in the spectrumThis is demonstrated in Fig 2 where depth-dependentRaman spectra (Fig 2a) and the resulting exchange layerpro les for two DKDP crystals are shown (Fig 2b) Thespectra in Fig 2a were obtained from a depth scan of aDKDP crystal with 75 degree of deuteration in thebulk grown at 45 8C The spectra start out as DKDP with30 deuteration close to the surface and approach thebulk DKDP spectrum within a few micrometers of depthThe fact that the relative degree of deuteration does notextend to 0 D is due to the limited depth resolution ofthe Raman microprobe which averages over 4 mm indepth Figure 2b depicts the resulting DH exchange layerpro les for this and a second crystal grown at 63 8Crespectively Both crystals had the same exposure to am-bient conditions and their main difference is the temper-ature at which they were grown The different exchangelayer pro les indicate that crystals grown at differenttemperatures have differen t proton conductiv ities 13
which leads to a difference in their rate of deuteriumdepletion The parameters controlling this behavior arecurrently the objective of a detailed study the results ofwhich will be reported elsewhere
CONCLUSION
In conclusion we have shown that the shift of the to-tally symmetric PO 4 stretch mode in the Raman spectrumof DKDP crystals scales linearly with degree of deuter-ation This allows us to correlate Raman peak positionsto deuteration levels in these crystals We have presenteda new technique to determine DH diffusion pro les inDKDP frequency conversion crystals based on micro-Ra-man spectroscopy This technique is fast inexpensiveand works under various environmental conditionswhich will allow us to better understand and control deu-terium depletion in DKDP crystals
ACKNOWLEDGMENTS
We would like to thank M Runkel for rst discovering DKDP crack-ing R Floyd for providing DKDP crystals and L Chase and A Burn-ham for their support and helpful discussions This work was performedunder the auspices of the US Department of Energy by the University
of California Lawrence Livermore National Laboratory under contract W-7405-Eng-48
1 J J De Yoreo A K Burnham and P K Whitman Int Mat Rev47 113 (2002)
2 C E Barker R A Sacks B M Van Wonterghern J A Caird JR Murray J H Campbell K Kyle R B Ehrlich and N DNielsen Proc SPIE-Int Soc Opt Eng 2633 501 (1995)
3 T Suratwala paper to be published4 Cleveland Crystals Inc httpwwwclevelandcrystalscom
KDPshtmltable5 E A Popova I T Savatinova and I A Velichko Sov Phys Solid
State 12 1543 (1971)6 I P Kaminow R C C Leite and S P S Porto J Phys Chem
Solids 26 2085 (1965)7 J A Subramony B J Marquardt J W Macklin and B Kahr
Chem Mat 11 1312 (1999)8 H Tanaka M Tokunaga and I Tatsuzaki Solid State Commun
49 153 (1984)9 R J Nelmes G M Meyer and J E Tibballs J Phys C 15 59
(1982)10 M A Yakshin D W Kim Y S Kim Y Y Broslavets O E
Sidoryuk and S Goldstein Laser Physics 7 941 (1997)11 I Takenaga Y Tominaga S Endo and M Kobayashi Solid State
Commun 84 931 (1992)12 C Krenn personal communication13 M Sharon and A K Kalia J Solid State Chem 21 171 (1977)
Rapid Dipicolinic Acid Extractionfrom Bacillus Spores Detectedby Surface-EnhancedRaman Spectroscopy
STUART FARQUHARSON ALAND GIFT PAUL MAKSYMIUK andFRANK E INSCOREReal-Time Analyzers Inc East Hartford Connecticut06108
Index Headings Dipicolinic acid Bacillus spores Anthrax Surface-enhanced Raman spectroscopy
INTRODUCTION
The anxiety caused by the distribution of anthrax en-dospores through the US postal system in October 2001was exacerbated by the long time required for positiveidenti cation of the Bacillus anthracis spores and the un-known extent of their distribution Since that time manymethods capable of rapid eld analysis have been inves-tigated to augment or replace the laboratory method ofgrowing microorganisms in culture media which takesdays to perform12 Prominent among these approachesare polymerase chain reactions (PCR)3 immunoassaysand detection of calcium dipicolinate as a biochemicalsignature PCR employs primers to separate organism-
Received 10 October 2003 accepted 14 November 2003 Author to whom correspondence should be sent
352 Volume 58 Number 3 2004
speci c nucleic acid sequences (eg capsular protein en-coding gene for Bacillus anthracis)4 and polymerases toamplify the segment until it is detectable Recently am-pli cation times have been substantially reduced andcomplete analysis can now be performed in an hour orless Immunoassay methods are also being developed thatuse competitive binding of the bioagent (as an antigen)and its labeled conjugate for a limited number of anti-bodies Although analyses can be performed in under 30minutes a well-de ned anthrax antigen has not yet beenidenti ed5ndash7 and consequently the false-positive rate isunacceptably high8
A number of other methods are being developed witha focus on the detection of calcium dipicolinate (CaDPA)and its derivatives as a B anthracis signature This is sobecause only spore-forming bacteria contain CaDPA andthe most common potentially interfering spores such aspollen and mold spores do not Relatively fast methodshave been developed to chemically extract CaDPA andthen detect it directly by uorescence9 or indirectly byluminescence1011 In the latter case hot dodecylamine(DDA) has been used to extract dipicolinic acid (DPA)and terbium has been utilized to form a highly lumines-cent DPA complex11 Although measurements have beenperformed in as little as ve minutes it was found thatas many as three concentration-dependent complexes canform each with different lifetimes This coupled withthe fact that the Tb31 cation produces the same lumines-cence spectrum makes determinations of low spore con-centrations problematic
It has been long known that Raman spectra of Bacillispores are dominated by bands associated with CaDPA12
and that these spectra may provide a suitable anthrax sig-nature at the genus level13 Since that time considerableimprovements in Raman instrumentation have led to lab-oratory measurements of single Bacilli spores14 and to eld measurements of spores captured from a mail-sort-ing system15 However the single spore measurementsrequired complex instrumentation that is not ruggedwhile the eld measurements required milligrams of sam-ple Furthermore the Raman spectra of both measure-ments contained uorescence contributions that would in-crease uncertainty in quanti cation
In related research we demonstrated that nanogramquantities of DPA could be detected by uorescence-freesurface-enhanced Raman spectroscopy (SERS)16 We alsodemonstrated that microliter volumes of chemicals canbe detected by SERS using metal-doped sol-gel-packedglass capillaries17 Towards the goal of developing a rap-id eld SERS-based anthrax spore detector we havecombined our previous research and we now report thatDPA can be extracted from a 10 mg B cereus spore sam-ple using DDA in 1 minute and can be detected by SERSin an additional 1 minute
EXPERIMENTAL
Dipicolinic acid (26-pyridinedicarboxylic acid DPA)and dodecylamine (DDA) were used as received fromSigma-Aldrich (Milwaukee WI) Lyophilized B cereusspores prepared according to the literature13 were sup-plied by the University of Rhode Island and used as re-ceived Multiple particles approximately 01 mm3 each
were separated and weighed at 5 to 15 mg representing05 to 15 million spores The sample masses were con-sistent with a previous determination of spore density at0081 gmL that indicated a high degree of entrained air
All chemicals used to prepare the silver-doped sol-gelcoated capillaries were also obtained and used as receivedfrom Sigma-Aldrich According to previously publishedprocedures17 two precursor solutions were preparedmixed and then drawn into 1-mm-diameter glass capil-laries The silver amine precursor consisted of a 51 vvratio of 1 N AgNO 3 to 28 NH3OH while the alkoxideprecursor consisted of a 21 vv ratio of methanol to te-tramethyl orthosilicate The alkoxide precursors weremixed with silver amine precursor in an 81 vv ratioApproximately 015 mL was drawn into the capillarycoating a 15-mm length After sol-gel formation the in-corporated silver ions were reduced with dilute sodiumborohydride which was followed by a water wash to re-move residual reducing agent
A 100 mL drop of a 50 mM DDA solution in ethanolpre-heated to 78 8C was added to each of the B cereusparticles to digest the spore coat After 1 minute the re-sultant solution was drawn into a SER-active capillarythat was immediately xed horizontally to an XY posi-tioning stage (Conix Research Spring eld OR) just in-side the focal point of an f 07 aspheric lens The lensfocused the beam into the sample and collected the scat-tered radiation back along the same axis A dichroic lter(Omega Optical Brattleborough VT) was used to re ectthe excitation laser to the lens and pass the Raman scat-tered radiation collected by the lens An f 2 achromatwas used to collimate the laser beam exiting a 200-mm-core-diameter source ber optic while a second f 2 ach-romat was used to focus the scattered radiation into a 365mm ber optic (Spectran Avon CT) A short-pass lterwas placed in the excitation beam path to block the sil-icon Raman scattering generated in the source ber fromre ecting off sampling optics and reaching the detectorA long-pass lter was placed in the collection beam pathto block the sample Rayleigh scattering from reachingthe detector A 785 nm diode laser (Process InstrumentsInc model 785-600 Salt Lake City UT) was used todeliver 100 to 150 mW of power to the sample A Fouriertransform Raman spectrometer (Real-Time Analyzersmodel IRA-785 East Hartford CT) and a silicon photo-avalanche detector (Perkin Elmer model C30902S Stam-ford CT) were used to acquire the SER spectra
RESULTS AND DISCUSSION
As an initial experiment the SER spectrum of 1 gLof DPA in water was measured using the newly devel-oped silver-doped sol-gel-coated capillaries (Fig 1A) Atthis concentration a high signal-to-noise ratio (SN) isobtained in 1 min In fact a reasonable spectrum is ob-tained in the same time frame for 1 mgL (Fig 1B) TheSER spectra are reasonably similar to the normal Raman(NR) spectrum obtained for a saturated solution of DPAin 1 N KOH (Fig 1C) and the following band shifts areobserved (NR to SER) 647 to 657 cm21 817 to 815cm21 998 to 1008 cm21 1384 to 1382 cm21 1434 to1428 cm21 and 1569 to 1567 cm21 Many of these bandshave been previously assigned1213 such as 998 cm21 to
APPLIED SPECTROSCOPY 353
FIG 1 SERS of DPA in water using silver-doped sol-gel-coated glasscapillary for (A) 1 gL and (B) 1 mgL (C ) NR of saturated DPA in 1N KOH in a glass capillary Spectral conditions (A) and (B) 150 mWof 785 nm 1-min acquisition time (C) 450 mW of 785 nm 5-minacquisition time both 8 cm 21 resolution
FIG 2 SERS of DPA extracted from 10 mg B cereus particle using100 mL of 50 mM hot DDA acquired in (A) 1 minute and (B) 2 seconds(C ) Attempted SERS of 50 mM hot DDA in ethanol using silver-dopedsol-gel-coated glass capillary acquired in 1 min Spectral conditions150 mW of 785 nm 8 cm21 resolution
the symmetric ring stretch 1384 cm21 to the OndashCndashOsymmetric stretch 1428 cm21 to the symmetric ring CndashH bend and 1569 cm21 to the asymmetric OndashCndashOstretch
The rst B cereus samples consisted of 2 mg of sporesin 2 mL of 5 mM hot DDA The samples were main-tained at 78 8C for 40 min and while hot approximately10 mL was drawn into a SER-active capillary Since spec-tra of DPA were obtained for these initial samples small-er spore masses higher DDA concentrations and shorterheating periods were examined In due course it wasfound that 10 mg of spores could be digested by 100 mLof 50 mM hot DDA in one minute and detected (Fig2A) In fact the signal was suf ciently intense that it canbe observed in as little as two seconds (Fig 2B) Theamount of DPA that was extracted was estimated to bebetween 5 and 10 mgL by comparing the signal intensityof the 1008 cm21 band to that measured for DPA in waterThis is consistent with previous research that found thatthe majority of the DPA is extracted from spores usingDDA11 and that B cereus spores contain approximately10 DPA by weight18 The SN of 127 for the 1008 cm21
band in the 1-minute SER spectrum suggests a limit ofdetection of approximately 250 ng of B cereus sporesbased on a SN of 3 Finally it should be noted that DDAdid not produce a detectable SER spectrum as shown inFig 2C
CONCLUSION
We have demonstrated that by combining rapid extrac-tion of dipicolinic acid from Bacillus cereus spores withchemical identi cation by surface-enhanced Raman spec-troscopy as little as 10 mg of spores can be detected Infact the entire measurement from the time of adding hotdodecylamine to the spores to the time when the dipi-colinic acid SER spectrum is acquired and analyzedcould be performed in less than two minutes The abilityof this method to distinguish between spore-forming bac-teria such as Bacillus anthracis and non-DPA containingpowders could help prevent costly shutdowns associated
with the appearance of suspicious material or intentionalmailing of common substances as an anthrax hoax Thismethod could also prove useful in detecting the locationof anthrax endospores in mail distribution facilities if an-other veri ed attack should occur
Research continues to fully characterize the surface-enhanced Raman spectroscopy signal intensities as afunction of sample concentration and to explore otherextractants that do not require the use of elevated tem-perature
ACKNOWLEDGMENTS
The authors are grateful for the support of the National Science Foun-dation (DMI-0296116 and DMI-0215819) and the US Army(DAAD13-02-C-0015 Joint Service Agent Water Monitor program)The authors are indebted to Chetan Shende for preparing the sol-gelcapillaries The authors also thank James Gillespie Nicholas Fell andAugustus Fountain for providing important background informationMark Farquharson for laboratory support and Professor Jay Sperry ofthe University of Rhode Island for supplying B cereus spores
1 V A Pasechnik C C Shone and P Hambleton Bioseparations 3267 (1993)
2 P J Jackson M E Hugh-Jones D M Adair G Green K K HillC R Kuske L M Grinberg F A Abramova and P Keim ProcNatl Acad Sci USA 95 1224 (1998)
3 B R Glick and J J Pasternak Molecular Biology Principles andApplications of Recombinant DNA (ASM Press Washington DC1994)
4 C A Bell J R Uhl T L Had eld J C David R F Meyer TF Smith and F R Cockerill III J Clin Microbiol 40 2897(2002)
5 D L Gatto-Menking H Yu J G Bruno M T Goode M Millerand A W Zulich Biosens Bioelectron 10 501 (1995)
6 J J Quinlan and P M Foegeding J Rapid Methods AutomationMicrobiol 6 1 (1998)
7 A A Hindle and E A H Hall Analyst (Cambridge UK) 1241599 (1999)
8 M S Ascher US Department of Health amp Human Services(httpwwwhhsgovophppresentationsAscherdoc)
9 R Nudelman B V Bronk and S Efrima Appl Spectrosc 54 445(2000)
10 D L Rosen C Sharpless and L B McBrown Anal Chem 691082 (1997)
354 Volume 58 Number 3 2004
11 P M Pellegrino N F Fell Jr and J B Gillespie Anal ChimActa 455 167 (2002)
12 W H Woodruff T G Spiro and C Gilvarg Biochem BiophysRes Commun 58 197 (1974)
13 E Ghiamati R S Manoharan W H Nelson and J F SperryAppl Spectrosc 46 357 (1992)
14 A P Esposito C E Talley T Huser C W Hollars C M Schal-dach and S M Lane Appl Spectrosc 57 868 (2003)
15 S Farquharson L Grigely V Khitrov W W Smith J F Sperryand G Fenerty J Raman Spectrosc paper accep ted (2003)
16 S Farquharson W W Smith S Elliott and J F Sperry SPIE-IntSoc Opt Eng 3855 110 (1999)
17 S Farquharson and P Maksymiuk Appl Spectrosc 57 479(2003)
18 F W Janssen A J Lund and L E Anderson Science (Washing-ton DC) 127 26 (1958)
SPIE -2003-5269 117
pH dependence of methyl phosphonic acid dipicolinic acid and cyanide by surface-enhanced Raman spectroscopy
Stuart Farquharson Alan Gift Paul Maksymiuk Frank Inscore and Wayne Smith
Real-Time Analyzers 87 Church Street East Hartford CT 06108
ABSTRACT US and Coalition forces fighting terrorism in Afghanistan and Iraq must consider a wide range of attack scenarios in addition to car bombings Among these is the intentional poisoning of water supplies to obstruct military operations To counter such attacks the military is developing portable analyzers that can identify and quantify potential chemical agents in water supplies at microgram per liter concentrations within 10 minutes To aid this effort we have been investigating the value of a surface-enhanced Raman spectroscopy based portable analyzer In particular we have been developing silver-doped sol-gels to generate SER spectra of chemical agents and their hydrolysis products Here we present SER spectra of methyl phosphonic acid and cyanide as a function of pH an important factor affecting quantitation measurements which to our knowledge has not been examined In addition dipicolinic acid a chemical signature associated with anthrax-causing spores is also presented Keywords Chemical warfare agents agent detection agent hydrolysis SERS Raman spectroscopy homeland security
1 INTRODUCTION In the past decade the Unites States and its allies have been challenged by a different kind of warfare exemplified by the terrorist attacks of September 11 2001 Suicide bombings and the use of chemical agents are the norm and military personnel must consider a wide range of attack scenarios Among these is the intentional poisoning of water supplies to obstruct military operations in Afghanistan and Iraq Currently colorimetric paper is used to detect agents on-site while gas chromatography combined with mass spectrometry (GCMS) is used in mobile support laboratories However both methods have severe drawbacks The paper changes color in response to contact with many chemicals besides CWAs causing a high incidence of false positives1 while GCMS although very chemically specific requires hours to perform and constant re-calibration234 Military operations would be greatly aided by a portable analyzer that can identify and quantify potential chemical agents at concentrations that impact safety This includes the analysis of drinking water supplies distribution and storage systems To meet this goal the Department of Defense has been investigating numerous approaches under the auspices of the Joint Service Agent Water Monitor (JSAWM) program5 The JSAWM must be capable of identifying the chemical agents with no false-positives and quantifying the agents at microgram per liter concentrations within 10 minutes678 Recently we and others have been investigating the ability of surface-enhanced Raman spectroscopy (SERS) to measure chemical agents9-12 bioagents13-17 and their hydrolysis products in water SERS employs the interaction of surface plasmon modes of metal particles with target analytes to increase scattering efficiency by as much as 1 million times18 In 1987 the potential of this approach to measure CWAs was demonstrated by measuring a series of pesticides19 Several of these organophosphonates have chemical structures similar to CWAs in particular P=O functional groups In our studies we have been employing silver-doped sol-gels to promote the SER effect The porous silica network of the sol-gel matrix offers a unique environment for immobilizing and stabilizing SER-active metal particles20-23 The sol-gels have been coated on the internal walls of standard glass vials to allow rapid SER analysis of small sample quantities without preparation We have measured over 100 chemicals with enhancements of 104 to 106 demonstrated reversible measurements in a flowing system reproducible measurements from vial-to-vial and measurements in multiple solvents including water20-23 Previously we used these vials to perform preliminary measurements of cyanide (CN) methylphosphonic acid (MPA) and dipicolinic acid (DPA) MPA is a hydrolysis product of the nerve agents (eg sarin Reaction 1) and may be a valuable indicator of nerve agent usage particularly since the alkyl methylphosphonic acids are relatively more stable than their corresponding parent complexes24 DPA is
SPIE -2003-5269 118
a chemical signature of spore forming bacteria such as Bacillus anthracis And in light of the inability to rapidly detect the anthrax spores distributed through the US mail in October 2001 a number of methods are being developed to extract and analyze this signature Reaction 1 Stepwise hydrolysis of Sarin to form hydrofluoric acid (HF) isopropyl methylphosphonic acid (IMPA) then methyl phosphonic acid (MPA) and 2-propanol In our previous SERS investigations MPA and DPA were measured at 50 and 100 mgL respectively In both cases limits of detection (LOD) were estimated at 100 microgL providing encouragement in that SERS may satisfy the needs of the JSAWM Since it has been shown that pH can substantially influence the intensity of SER bands25 which would clearly influence quantitative analysis we undertook the present study to determine the severity of these effects for cyanide methyl phosphonic acid and dipicolinic acid Furthermore we previously observed a band at 1050 cm-1 for MPA6 possibly due to an anion formed at basic pH Here we investigate the source of this spectral anomaly
2 EXPERIMENTAL All chemicals including potassium cyanide methyl phosphonic acid dipicolinic acid and those used to prepare the silver-doped and gold-doped sol-gel coated vials were obtained and used as received from Sigma-Aldrich All samples were prepared in HPLC grade water (Fischer Scientific Fair Lawn NJ) for SERS measurements The pH of these samples was adjusted using dilute nitric acid or potassium hydroxide and verified using a pH electrode (Corning Inc Corning NY) that had been calibrated with pH 400 700 and 1000 buffered standards from Fischer Scientific Once prepared the samples were transferred into the silver-doped sol-gel vials (Real-Time Analyzers Inc East Hartford CT) The vials were coated in a manner similar to that previously reported by adding ammonium hydroxide to a solution of silver nitrate tetramethyl orthosilicate (TMOS) and methanol20 Gold-vials were coated by adding nitric acid to a solution of gold tetrachloride TMOS and methanol The two precursor solutions were prepared mixed and transferred to 2-mL glass vials dried and heated After sol-gel formation the incorporated metal ions were reduced with dilute sodium borohydride (1mgmL) which was followed by a water wash to remove residual reducing agent After the resultant analyte solution was introduced the SER-active vial was immediately fixed horizontally to an XY positioning stage (Conix Research Springfield OR) just inside the focal point of an f07 aspheric lens The lens focused the beam into the sample and collected the scattered radiation back along the same axis A dichroic filter (Omega Optical Brattleborough VT) was used to reflect the excitation laser to the lens and pass the Raman scattered radiation collected by the lens An f2 achromat was used to collimate the laser beam exiting a 200 microm core diameter source fiber optic while a second f2 achromat was used to focus the scattered radiation into a 365 microm fiber optic (Spectran Avon CT) A short pass filter was placed in the excitation beam path to block the silicon Raman scattering generated in the source fiber from reflecting off sampling optics and reaching the detector A long pass filter was placed in the collection beam path to block the sample Rayleigh scattering from reaching the detector A 785 nm diode laser (Process Instruments Inc model 785-600 Salt Lake City UT) was used to deliver 100 to 150 mW of power to the sample A Fourier transform Raman spectrometer (Real-Time Analyzers model IRA-785 East Hartford CT) and a silicon photo-avalanche detector (Perkin Elmer model C30902S Stamford CT) were used to acquire the SER spectra
3 RESULTS AND DISCUSSION In a previous study of MPA6 aimed at developing a concentration calibration curve and determining limits of detection (LOD) we observed an anomaly at 1050 cm-1 Since it was found that the band intensity changed as a function of concentration the band must be associated with a sample parameter Two possible parameters photon flux and pH are examined here The first parameter was investigated by irradiating a 1mgmL MPA sample in a SER-active vial with laser powers of 200 mW and above and monitoring spectral changes It was immediately found that the 1050 cm-1 band
2O+ H HF + +OH
OH3H C
OP
OF
CH
CH3
3
3H C
OCP
OOH
CH
CH3
3
3H C
O
CPHO
CH
CH
3
3
C
Sarin IMPA MPA 2-propanol
SPIE -2003-5269 119
appeared and grew as a function of time and that the higher the laser power the faster the growth Figure 1 shows the growth of the 1050 cm-1 band over the course of 30 minutes when using 150 mW of 785 nm excitation while Figure 2A shows that the growth can be fit with a first order exponential equation namely I1050 = 03+05e-013t Furthermore the 760 cm-1 band could be fit with a first order decay equation with an identical exponential rate constant ie I760 = 08-
08e-013t The rates represent classical first order kinetics and their correspondence allows one to conclude that MPA is being transformed one-for-one into a photo-generated product At this time the photoproduct has not been positively identified but phosphonic (phosphorous) acid and phosphonate are likely candidates since the symmetric P(OH)3 stretch occurs at ~1050cm-1 Our studies show that a reduction in laser power to 100 mW at the sample for MPA essentially eliminates this degradation process This laser power or lower was used for further measurements
Methyl phosphonic acid is a diprotic acid that stepwise dissociates into two anions MPA- and MPA= according to the following reactions26
MPA MPA- + H+ pKa1 = 212 Reaction 2
MPA- MPA= + H+ pKa2 = 729 Reaction 3 The relative concentrations of MPA MPA- and MPA= can be determined at any pH by expressing [MPA] and [MPA=] in terms of [MPA-] using Reactions 2 and 3 and summing all three to equal the total starting concentration here 2 mgmL (0021M MW = 9602) viz
[MPA] + [MPA-] + [MPA=] = 0021M Equation 1 substituting from Reactions 1 and 2
([H+][MPA-])K1a + [MPA-] + (K2a[MPA-])[H+] = 0021M Equation 2 rearranging [MPA-] = 0021M(1+[H+]K1a + K2a[H+]) Equation 3 The relative concentrations of MPA MPA- and MPA= as a function of pH are shown in Figure 3 It is worth noting that near neutral pH both MPA- and MPA= will be present To confirm that the SER signal followed this pH dependence a starting solution consisting of 20 mg of MPA in 10 mL HPLC grade water was prepared and brought to pH of 20 using dilute nitric acid From this solution 2 mL were added to a SER-active vial and the SER spectrum recorded At this pH a peak at 760 cm-1 was barely discernable The 2 mL solution was returned to the starting solution and the pH was re-measured to correct for any changes that the silver-doped sol-gel vials might cause In most cases the change was less than 02 pH units and the pH is reported as the before and after average Next the pH of the
Figure 2 A) Exponential growth of 1050 cm-1 band and B) exponential decay of 760 cm-1 band for spectral series in Figure 1
Figure 1 Growth of 1050 cm-1 band as a function of time due to exposure to 150 mW of 785 nm Spectra are 5 sec each collected every 100-sec from 0 to 30-min
B
570 770 970 1170 1370Raman Shift (cm-1)
Arbit
rary
Unit
s
MPA Photodegradation
600 800 1000 1200 1400 Wavenumber (cm-1)
A
0 10 20 30 0 10 20 30 time (min) time (min)
Ram
an In
tens
ity (r
elat
ive)
30
min
0
SPIE -2003-5269 120
starting solution was adjusted to 325 using dilute KOH Again 2 mL were added to a vial and the SER spectrum recorded At this pH a reasonably strong 760 cm-1 band was observed This process was repeated as spectra were recorded at pHs of 70 74 75 79 85 and 100 A total of 1 mL of KOH was added diluting the total concentration by 10 Next the pH of the starting solution was made acidic by adding dilute nitric acid dropwise This time spectra were recorded at pHs of 72 69 64 and 37 Figure 4 shows the SER spectra for representative pHs (spectra were left out to simplify the figure) while Figure 3 shows the 760 cm-1 peak intensities as a function of pH (The band intensities were adjusted to compensate for dilution effects caused by the addition of HNO3 and KOH then normalized to 0021 M for the most intense band observed at pH 37) It is clear from Figure 3 that the 760 cm-1 band follows the MPA- concentration as a function of pH and must be assigned to this anion No bands were observed that corresponded to MPA or MPA= The lack of an MPA SER spectrum may be due to the absence of an attraction between the neutral analyte and the electropositive silver surface The same reasoning suggests that a strong SER spectrum should be observed for MPA= but it is not and a satisfactory explanation has not been found
0000
0005
0010
0015
0020
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14pH
Con
cent
ratio
n [M
]
pK1 = 212 pK2 = 729
MPA- MPA=
MPA
Figure 4 SER spectra of 002M MPA as a function of pH Conditions 100 mW of 785 nm 36 scans (1 min) 8 cm-1 recorded 2 min after sample introduction pH 19 and 20 not apparent on this scale pH 69 and 74 near identical to 70 and 75 and not shown for clarity
00
02
04
06
08
10
12
14
16
18
0 5 10 15 20 25 30Measurement Number
Ram
an In
tens
ity (7
60 c
m-1
)
Figure 3 Concentration dependence of MPA MPA- and MPA= as a function of pH for a 002M sample Intensity of 760 cm-1
band from Figure 6 as a function of pH ( for increasing basic adjustment for increasing acidic adjustment error was measured at ~10 for pH 64)
Figure 5 SER spectra of 002M MPA at pH 64 measured around a vial at three heights (9 points per height) Conditions as in Figure 4 but 10-sec scans
Since these measurements involved the removal and replacement of the SER-active vial in the sample holder to remove and add sample variation in the intensity as a function of vial position was minimized by illuminating the exact same height along the vial wall But this does not account for variability of the SERS response of the sol-gel coating around the vial To analyze this effect a vial containing MPA at pH 64 was rotated at ~ 40o
intervals at the original height and 18rdquo above and below this value Figure 5 shows the intensity of the 760 cm-1
band for the 27 positions It was found that the average value was 137plusmn014 an RSD of 10 overall and 5 for each height An error bar is included in Figure 4 for the pH 64 measurement
SPIE -2003-5269 121
With the above analyses in mind a preliminary investigation of the SER spectral response for MPA (as MPA-) as a function of concentration was performed A single vial was used for these measurements beginning with 1 mgL followed by measurements of 10 100 and 1000 mgL In all cases the pH was ~7 and 3 positions around the vial were measured per concentration Since the 760 cm-1 band was not observed for concentrations of 1 or 10 mgL using 100 mW of 785 nm the laser power at the sample was raised to 200 mW beginning with the 10 mgL concentration Photo-degradation was largely avoided (and not observed) by exposing the sample for only 33 seconds per spectral acquisition Representative spectra for 10 100 and 1000 mgL are shown in Figure 6 while a plot of the 760 cm-1 band intensity as a function of concentration is shown in Figure 7 These values were also used to estimate limits of detection based on the signal-to-noise ratio (SN) of the 760 cm-1 band using 100 mW of 785 nm laser excitation and a 10-min acquisition time As summarized in the Figure 7 inset the lower the measured concentration the lower the predicted LOD Even if as estimated 210 microgL could be measured using the silver-doped sol-gel vials an improvement of a factor of 70 is still required to meet JSAWM goals of measuring 3 microgL in 10-minutes Similar to MPA DPA is a diprotic acid (pKa1 = 216 and pKa2 = 692) and variations in pH will effect the relative concentrations of DPA DPA- and DPA= and possibly the SER spectra and band intensities This could prove significant if an acid or base is used to denature anthrax spores with the goal of extracting and analyzing DPA The MPA pH study described above was mimicked for DPA except that the starting solution consisted of 20 mg of DPA in 20 mL HPLC grade water (60x10-3M MW = 1671) The initial solution had a pH of 245 which was made basic by dilute KOH to pHs of 355 433 487 559 1069 and 1166 SER spectra were recorded at each pH using 100 mW of 785 nm and a 44-sec acquisition time Next one drop of concentrated nitric acid was used to remake the solution acidic at a pH of 200 Again sequential pH measurements were performed at 383 510 735 and 822 The solution pH was made acidic a third time but to pH 219 171 then 135 Throughout this process no more than 20 drops of acid or base were added and therefore the concentration was diluted by no more than 10 Most of the spectral bands showed a minor decrease in intensity as a function of increasing pH values However the bands at 525 and 795 cm-1 showed the most dramatic changes which occurred at acid pH Figure 8 shows the SER spectra of DPA for the spectral region and pH range of interest The identity of the DPA species was determined by plotting the normalized peak intensities with the lowest value set to 0 and the highest to 0006 M as a function of pH and overlaying these values on a plot of the relative concentrations for DPA DPA- and DPA= as previously done for MPA (Figure 9) As can be seen the 525 cm-1 band clearly corresponds to DPA The correspondence of the 795 cm-1 band to this species is less clear as the band retains intensity until pH of 55 This can be attributed to contribution to the overlapping band at 810 cm-1 which does not change as a function of pH The fact that most bands are observed at all pHs suggest that the primary interaction with silver is through the ring nitrogen This is supported by the fact that the most intense band occurs at 1008 cm-1 attributed to a symmetric ring breathing mode and that this interaction has been characterized for pyridine in numerous papers27
0
1
2
3
4
5
6
7
0 200 400 600 800 1000 1200
MPA Concentration (mgL)
760
Ban
d In
tens
ity
Figure 6 SER spectra of MPA in water at A) 1000 B) 100 and C) 10 mgL Conditions pH of 7 silver-doped sol-gel coated vial 200 mW of 785 nm 33-sec 8 cm-1 resolution
Figure 7 Plot of SER intensity of 760 cm-1 band of MPA as a function of concentration using 200 mW of 785 nm Inset table includes average intensity LOD standard deviation and percent deviation for each concentration but for 100 mW and 10 min
conc (mgL) LOD ave stddev dev10 021 002 1021100 072 010 14301000 312 040 1280
A
B C
SPIE -2003-5269 122
The ability to detect dipicolinic acid (DPA) by SERS is demonstrated in Figure 10 Even at 1 mgL the primary bands are visible The intensity of the dominant ring breathing mode observed at 1008 cm-1 for DPA in water (pH = 25 to 55) is plotted as a function of concentration in Figure 11 These values were also used to estimate limits of detection based on the SN of the 1008 cm-1 band using 100 mW of 785 nm laser excitation and a 10-min acquisition time Again the lower the measured concentration the lower the predicted LOD (see Figure 11 inset) and detection of 160 microgL is possible Gastrointestinal anthrax requires significant more spores than inhalation anthrax28 and a limit of detection might be placed at 1 million spores in 1 liter of water or 10 microgL Since each spore contains ~10 CaDPA by weight29 a goal for DPA might be 1 microgL indicating that the present measurements must be improved by nearly two orders of magnitude Due to the increased hazards of handling HCN gas KCN salt was used for these experiments Nevertheless all sample preparations were performed in a chemical hood KCN completely dissolves in water but its conjugate acid HCN is formed and has a Ka of 615x10-1030 viz
HCN CN- + H+ pKa = 921 Reaction 4
Figure 8 SER spectra of 1 mgmL DPA as a function of pH Conditions 100 mW 785 nm 100 scans (44-sec)
Figure 9 Concentration dependence of DPA DPA- and DPA= as a function of pH for a 0006M sample Intensity of 525 () and 795 (diams) cm-1 bands from Fig 8 as a function of pH
pH
135 171 219 383
0
0001
0002
0003
0004
0005
0006
0007
0 2 4 6 8 10 12 14pH
Con
cent
ratio
n [M
]
DPADPA-DPA=795525
DPA DPA=DPA-
pK2 = 692pK1 = 216
795 525
0
05
1
15
2
25
0 200 400 600 800 1000 1200
DPA Concentration (mgL)
1008
Ban
d In
tens
ity
conc (mgL) lod-10min-100mw1 017
10 016100 103
1000 355
Figure 10 SER spectra of DPA in water at A) 1000 B) 100 C) 10 and D) 1 mgL Conditions pH of 25-55 silver-doped sol-gel coated vial 175 mW of 785 nm 1-min 8 cm-1 D) has been multiplied by x10 to make bands visible
A
B C
D
Figure 11 Plot of SER intensity of 1008 cm-1 band of DPA as a function of concentration using 175 mW of 785 nm Inset table includes LOD in mgL for each concentration but for 100 mW and 10 min
SPIE -2003-5269 123
Consequently the cyanide concentration must be determined for each initial KCN concentration Specifically the samples prepared with concentrations of 01 1 10 100 and 1000 mgL of KCN produced CN- concentrations of 63x10-3 033 69 89 and 964 mgL at pHs of 816 90 967 102 and 107 respectively The pH dependence for the HCN and CN- concentrations are shown in Figure 12 Thus as the amount of KCN added to the solution decreases so does the pH of the solution (becomes less basic) and according to Reaction 4 the relative amount of CN- to HCN also decreases For example in the preparation of a 01 mgL solution of KCN the pH is shifted from 7 for pure water to only 816 and only 63 of the starting material becomes CN- or 63x10-3 mgL In comparison for a solution of 1000 mgL the pH is shifted from 7 to 107 and 96 of the starting material becomes CN- This is significant since the cyanide ion is better able to adsorb onto the silver particles and become SERS active SER spectra of 10 100 and 1000 mgL of KCN samples are shown in Figure 13 and a corresponding plot of the 2100 cm-1 band intensity as a function of concentration is shown in Figure 14
000102030405060708091011
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14pH
Con
cent
ratio
n [m
gm
L]
pKa = 921
CN -HCN
The SER spectra of cyanide are dominated by the single CN stretch at 2100 cm-1 which occurs in normal Raman spectra of solutions at 2080 cm-1 However a low frequency mode occurs at 135 cm-1 due to a Ag-CN stretch (not shown) indicative of a strong surface interaction It is also observed that as the concentration decreases the CN stretch shifts to 2140 cm-1 This shift has been attribute to the formation of a tetrahedral Ag(CN)3
2- surface structure31 as well
Figure 14 Concentration dependence of KCN SERS measured under conditions in Fig 11 Concentrations are 1 01 and 001 mgml Intensities are measured for the CN stretch at 2100 cm-1 Inset table includes LOD in mgL for each concentration in Figs 13 and 15 but for 100 mW and 10 min
Figure 13 SER spectra of KCN in water at A) 1000 B) 100 and C) 10 mgL Conditions pHs of 107 102 and 97 silver-doped sol-gel coated vial 100 mW of 785 nm 1-min 8 cm-1 C) has been multiplied by x10 to make band visible
Figure 12 Concentration dependence of HCN and CN- as a function of pH for a 1 mgmL sample Calculated intensity of 2100 () cm-1 band for a 1 mgml sample at pHs of 816 90 967 102 and 107
A
B
C
0
20
40
60
80
100
120
140
0 200 400 600 800 1000 1200CN Concentration (mgL)
2100
Ban
d In
tens
ity
Figure 15 SER spectra of KCN in water at A) 10 B) 1 and C) 01 mgL Conditions pHs of 97 12 and 12 gold-doped sol-gel coated vial 100 mW of 785 nm 1-min 8 cm-1 C) has been multiplied by x10 to make band visible
A
B
C
conc (mgL) lod-10min-100mw Condition01 001 Au-pH 121 007 Au-pH 12
69 003 Au-pH 9769 016 Ag-pH 9789 022 Ag-pH 102946 113 Ag-pH 107
SPIE -2003-5269 124
as to CN adsorbed to two different surface sites32 Alternatively the 2140 cm-1 band could be attributed to HCN since this species dominates at lower concentrations However it is unlikely that this species would be attracted to the electropositive silver surface Further both peaks should be present at pHs between 85 and 105 but this is not observed It has also been suggested that at concentrations near and above monolayer coverage the CN- species is forced to adsorb end-on due to crowding and at lower concentrations the molecule can reorient to lie flat33 This suggests that the 2100 and 2140 cm-1 bands correspond to the end-on and flat orientations respectively As Figures 13 and 14 show the intensity of the CN stretch for the 89 mgL sample is nearly as intense as the 964 mgmL band This suggests that the Raman signal for the flat orientation is more enhanced However more extensive measurements are required to verify this point Since resent research has suggested that cyanide may be more effectively detected on gold measurements of KCN solutions were also performed using gold-doped sol-gel vials Preliminary measurements are shown in Figure 15 for samples prepared from 01 1 and 10 mgL KCN Since the pHs are 816 90 967 the resultant CN- concentrations are 63x10-3 033 and 69 mgL Initially only the highest concentration was observed and the signal intensity was significantly better than the equivalent concentration measured using silver In an effort to shift Reaction 4 to the left transforming HCN to CN- (Le Chatelierrsquos principle) KOH was added to the lower concentration samples producing solutions with pH 12 This effectively forces all of the cyanide in solution to be CN- or 01 and 10 mgL respectively More importantly the CN stretch is now observed in the SER spectra The band appears at 2125 cm-1 as has been previously reported for gold12 As calculated for MPA and DPA LODs can be estimated from this data For the three concentrations of cyanide on silver the LODs are 016 to 11 mgL for 100 mW of 785 nm laser excitation and a 10-min acquisition time For gold pH adjusted the LODs are10 to 70 microgL an improvement of more than 10 times silver Nevertheless either substrate is sufficient to meet the JSAWM goals of measuring 3 mgL in 10-minutes as the requirements form cyanide are much less stringent than the nerve agents
4 CONCLUSIONS Here we examined the surface-enhanced Raman spectral response for methyl phosphonic acid dipicolinic acid and cyanide as a function of pH It was determined that the most prominent peak at 760 cm-1 reaches a maximum intensity between pH 3 and 7 and corresponds to the MPA- species Neither the MPA nor MPA= species appear to generate a SER spectrum and consequently no spectra were observed below pH 2 or above pH 8 In this study we also found that higher laser powers could cause photodegradation of MPA signified by the exponential growth of a band at 1050 cm-1 which is tentatively assigned to phosphorous acid Unlike MPA DPA was observed at all pHs This is attributed to the dominant interaction of the pyridine functional group with silver Minor spectral changes were observed at acid pHs and were assigned to neutral DPA Like MPA SER spectra of cyanide were pH dependent No spectra were observed for the HCN species while CN- was best observed at pHs more basic than 8 Preliminary concentration studies for the three analytes allowed estimating limits of detection for MPA DPA and CN using 100 mW of 785 nm and a 10-min acquisition time of 210 165 and 70 microgL respectively Although the latter value suggests that cyanide can be measured at sufficiently low concentrations to meet JSAWM goal improvements by 100 to 200 times are required for MPA and DPA It is clear from this study that pH of the sample is important and must be taken into account when developing concentration calibration curves as well as fieldable analyzers Future work will include tailoring the sol-gel with specific functional groups to dictate sample pH and thereby optimize sensitivity
ACKNOWLEDGMENTS The authors are grateful for the support of the US Army (DAAD13-02-C-0015 Joint Service Agent Water Monitor program) The authors would also like to thank Dr Steve Christensen of the US Army SBCCOM for helpful discussions and Mr Chetan Shende of Real-Time Analyzers for assistance in development of the gold-doped sol-gels
REFERENCES 1 Erickson B Analytical Chemistry News amp Features June 1 397A (1998)
SPIE -2003-5269 125
2 Johnston RL Hoefler CM Fargo JC and Moberley B AT-ONSITE 5-8 (1994) 3 Black RM RJ Clarke RW Read and MT Reid J Chromatography 662 301-321 (1994) 4 USA Army Center for Health Promotion and Preventive Medicine (USACHPPM) Technical Guideline 230A 5 Jensen JL NC Wong W R Loerop SPIE 4775-03 (2001) 6 Hoenig SL Handbook of Chemical Warfare and Terrorism Greenwood Press Wesport CT (2002) 7 Munro NB SS Talmage GD Griffin LC Waters AP Watson JF King and V Hauschild Env Health
Persp 107 933-974 (1999) 8 Holstege CP Kirk M Sidell FR Crit Care Clin13 923-42 (1997) 9 Farquharson S P Maksymiuk K Ong and S Christesen SPIE 4577 166-173 (2001) 10 Lee Y and S Farquharson SPIE 4378 21-26 (2001) 11 Spencer KM J Sylvia S Clauson and J Janni SPIE 4577 158-165 (2001) 12 Tessier P S Christesen K Ong E Clemente A Lenhoff E Kaler and O Velev Applied Spectroscopy 56
1524-1530 (2002) 13 Farquharson S WW Smith S Elliott and JF Sperry SPIE 3533 207-214 (1998) 14 Farquharson S WW Smith S Elliott and JF Sperry SPIE 3855110-116 (1999) 15 Farquharson S WW Smith YH Lee S Elliott and JF Sperry SPIE 4575 62-72 (2002) 16 Guzelian AA J Sylvia J Janni S Clauson and KM Spencer SPIE 4577 182-192 (2001) 17 Shende C F Inscore A Gift P Maksymiuk and S Farquharson in press 18 Weaver MJ S Farquharson and MA Tadayyoni J Chem Phys 82 4867-4874 (1985) 19 Alak AM and T Vo-Dinh Analytical Chemistry 59 2149-2153 (1987) 20 Lee Y and S Farquharson SPIE 4206 140-146 (2000) 21 Farquharson S and Y Lee SPIE 4200-16 (2000) 22 Lee Y S Farquharson and P M Rainey SPIE 3857 76-84 (1999) 23 Lee Y S Farquharson H Kwong and M Shahriari SPIE 3537 252-260 (1998) 24 Wang J M Pumera G Collins and A Mulchandani Analytical Chemistry 74 6121-6125 (2002) 25 Dou X YM Jung Z-Q Cao and Y Ozaki Applied Spectroscopy 53 1440-1447 (1999) 26 Data supplied by S Christesen and K Ewing 27 Kerker M and B Thompson Eds SPIE MS 10 (1990) 28 Inglesby TV DA Henderson JG Bartlett JAMA 287 2236 (2002) 29 FW Janssen AJ Lund and LE Anderson Science 127 26 (1958) 30 Lide DR Ed Handbook of Chemistry and Physics CRC Press 77th Ed (1996-7) 31 Billmann J G Kovacs and A Otto Surf Sci 92 153 (1980) 32 Murray CA and S Bodoff Phys Rev B 32 671 (1985) 33 Kellogg D and J Pemberton J Phys Chem 91 1120 (1987)
SPIE-2003-5269 16
Chemical agent detection by surface-enhanced Raman spectroscopy
Stuart Farquharson Alan Gift Paul Maksymiuk Frank Inscore and Wayne Smith
Real-Time Analyzers 87 Church Street East Hartford CT 06108
Kevin Morrisey and Steven D Christesen US Army SBCCOM Aberdeen Proving Ground MD 21010
ABSTRACT
In the past decade the Unites States and its allies have been challenged by a different kind of warfare exemplified by the terrorist attacks of September 11 2001 Although suicide bombings are the most often used form of terror military personnel must consider a wide range of attack scenarios Among these is the intentional poisoning of water supplies to obstruct military operations in Afghanistan and Iraq To counter such attacks the military is developing portable analyzers that can identify and quantify potential chemical agents in water supplies at microgram per liter concentrations within 10 minutes To aid this effort we have been investigating the value of a surface-enhanced Raman spectroscopy based portable analyzer In particular we have been developing silver-doped sol-gels to generate SER spectra of chemical agents and their hydrolysis products Here we present SER spectra of several chemical agents measured in a generic tap water Repeat measurements were performed to establish statistical error associated with SERS obtained using the sol-gel coated vials Keywords Chemical agents chemical agent detection SERS Raman spectroscopy
1 INTRODUCTION
In the autumn of 2001 terrorism within US borders became a sobering reality While extensive efforts are being implemented to secure the homeland US and Coalition forces in Afghanistan and Iraq are constantly faced with terrorist attacks In addition to car-bombings the military has identified several non-traditional attack scenarios including poisoning of water supplies by chemical warfare agents (CWAs) To counter this threat the Department of Defense has been investigating numerous approaches to detect such attacks under the auspices of the Joint Service Agent Water Monitor (JSAWM) program1 The JSAWM must be capable of identifying the chemical agents with no false-positives and quantifying the agents at microgram per liter concentrations within 10 minutes2 This includes the analysis of drinking water supplies distribution and storage systems Currently colorimetric paper is used to detect agents on-site while gas chromatography combined with mass spectrometry (GCMS) is used in mobile support laboratories However both methods have severe drawbacks The paper changes color in response to contact with many chemicals besides CWAs causing a high incidence of false positives3 while GCMS although very chemically specific requires up to an hour to perform and regular re-calibration456 Vibrational spectroscopy has also been investigated as a potential method of rapidly detecting CWAs7-
11 as early as 1985 in the case of infrared spectroscopy8 and 1988 in the case of Raman spectroscopy11 Again however these techniques also have limitations when it comes to measuring trace poisons in water Infrared spectra would be dominated by the very strong absorption of water which would obscure absorptions by most other chemicals present Whereas Raman spectroscopy is simply not a very sensitive technique and detection limits are typically grams per liter Surface-enhanced Raman spectroscopy (SERS) offers several advantages over conventional vibrational methods and that may provide the necessary sensitivity required for detecting trace quantities of chemical agents in water SERS employs the interaction of surface plasmon modes of metal particles with target analytes to increase scattering efficiency by as much as 1 million times12 In 1987 the potential of this approach to measure CWAs was demonstrated by measuring a series of pesticides13 Several of these organophosphonates have chemical structures similar to CWAs in particular P=O functional groups In the past few years we and others have further explored the ability of SERS to detect CWAs14-17 and even bioagents 18-21 We have been employing silver-doped sol-gels to promote the SER effect
SPIE-2003-5269 17
in these studies The porous silica network of the sol-gel matrix offers a unique environment for immobilizing and stabilizing SER-active metal particles22-25 The silver-doped sol-gels have been coated on the internal walls of standard glass vials to allow rapid SER analysis of small sample quantities without preparation We have measured over 100 chemicals with enhancements of 104 to 106 demonstrated reversible measurements in a flowing system reproducible measurements from vial-to-vial and measurements in multiple solvents including water21-25 Previously we used these vials to perform preliminary measurements of cyanide (CN) and methylphosphonic acid (MPA) Most of the nerve agents form MPA during hydrolysis while Tabun forms CN a chemical agent in its own right In another paper including in these proceedings we examined the limits of detection (LOD) for MPA by measuring a series of concentrations down to 50 mgL and estimated a limit of detection of 100 microgL26 These measurements provide encouragement in that SERS may satisfy the needs of the JSAWM To further establish the viability of SERS in particular silver-doped sol-gels here we present analysis of cyanide mustard and VX in tap water The measurements performed at the US Armyrsquos Edgewood Chemical Biological Center Aberdeen MD also included numerous repeat measurements to establish reproducibility
2 EXPERIMENTAL 2a General All chemicals including potassium cyanide 2-chloroethylethyl sulfide and those used to prepare the silver-doped and gold-doped sol-gel coated vials were obtained and used as received from Sigma-Aldrich All samples were prepared in a chemical hood using HPLC grade water unless otherwise noted (Fischer Scientific Fair Lawn NJ) for SERS measurements Once prepared the samples were transferred into the silver-doped sol-gel vials (Real-Time Analyzers Inc East Hartford CT) The vials were coated in a manner similar to that previously reported by adding ammonium hydroxide to a solution of silver nitrate tetramethyl orthosilicate (TMOS) and methanol22 The two precursor solutions were prepared mixed and transferred to 2-mL glass vials dried and heated After sol-gel formation the incorporated metal ions were reduced with dilute sodium borohydride (1mgmL) which is followed by a water wash to remove residual reducing agent After the resultant analyte solution was introduced the SER-active vial was fixed horizontally to an XY positioning stage (Conix Research Springfield OR) just inside the focal point of an f07 aspheric lens The lens focused the beam into the sample and collected the scattered radiation back along the same axis A dichroic filter (Omega Optical Brattleborough VT) was used to reflect the excitation laser to the lens and pass the Raman scattered radiation collected by the lens An f2 achromat was used to collimate the laser beam exiting a 200 microm core diameter source fiber optic while a second f2 achromat was used to focus the scattered radiation into a 365 microm fiber optic (Spectran Avon CT) A short pass filter was placed in the excitation beam path to block the silicon Raman scattering generated in the source fiber from reflecting off sampling optics and reaching the detector A long pass filter was placed in the collection beam path to block the sample Rayleigh scattering from reaching the detector A 785 nm diode laser (Process Instruments Inc model 785-600 Salt Lake City UT) was used to deliver 100 to 150 mW of power to the sample A Fourier transform Raman spectrometer (Real-Time Analyzers model IRA-785 East Hartford CT) and a silicon photo-avalanche detector (Perkin Elmer model C30902S Stamford CT) were used to acquire the SER spectra
2b Edgewood Chemical Biological Center The surface-enhanced Raman spectral measurements at the US Armyrsquos Chemical Biological center presented here were all performed on September 12 2003 To expedite measurements a plate was machined to hold up to 12 SER-active sample vials (Figure 1) The plate fit a standard XY plate reader that could be programmed Pure KCN bis-(2-chloroethyl)sulfide (distilled mustard gas HD) and ethyl S-2-diisopropylamino ethyl methylphosphonothioate (VX) were obtained on-site and used to prepare 1 gL tap water solutions in a chemical hood with appropriate safety equipment Simulated tap water was prepared by adding 10 chemicals most often found in tap water at appropriate concentrations to distilled water (Table 1) SER measurements were also performed in a chemical hood For added safety the FT-Raman instrument was placed outside the laboratory and 30 foot fiber optic and electrical cables were used to allow remote SERS measurements and plate manipulation For each experiment 1gL samples were prepared and added to 9 individual vials which were then loaded on the plate In some cases a tenth vial was included as a blank
Table 1 Generic Tap Water Composition
Compound mgL NaHCO3 100 CaSO4 27 MgSO4bull7H2O 67 NaNO3 1 Fulvic Acid 1 K2HPO4 07 KH2PO4 03 (NH4)2HSO4 001 NaCl 001 FeSO4 0001 pH 76-78
SPIE-2003-5269 18
A software program was written that allowed selecting the sequence that the vials were measured the number of positions along the length of the vials to measure (1 to 5) and the number of scans to co-add During sample analysis the program displayed the vial being analyzed the point being analyzed and the spectrum as it was being acquired Once all the data was collected a second software program was written to rapidly analyze the data The spectra collected for all the vials on a plate could be loaded at one time and then the spectra for each point could be displayed simultaneously or separately The user could then select the Raman peak to analyze in terms of peak height or area This was accomplished by selecting points on either side of the peak to define a baseline of zero The peak height or area could then be computed for all of the spectra loaded and then exported to a spreadsheet for statistical analysis
Figure 1 A) Vial Holder 6 slots to hold 2 vials each end-to-end B) Measurement Configuration Program user selects vials to measure sequence number of points per vial (1 to 5) and number of scans per point C) Spectral Acquisition Program shows spectrum being collected which vial and position D) Spectra Analysis Program user selects spectra to analyze by plate vial and point (s) as well as two wavenumbers defining the peak and the baseline to subtract The image is of 5 repeat measurements of 10 mgL KCN in generic tap water 16 sec each 100 mW of 785 nm
3 RESULTS AND DISCUSSION Raman and surface-enhanced Raman spectra were obtained for potassium cyanide bis-(2-chloroethyl)sulfide and ethyl S-2-diisopropylamino ethyl methylphosphonothioate representing three classes of chemical agents cyanides mustards and nerve agents respectively Spectra were also obtained for 2-chloroethyl ethyl sulfide (CEES) a structural analogue to HD which was included in the study to aid in assigning spectral bands KCN salt was used for cyanide experiments to avoid the increased hazards of handling HCN gas KCN completely dissolves in water forming its conjugate acid HCN according to its Ka of 615x10-1027 and at a concentration of 1 mgmL results in a pH 107 solution This is significant in that only CN- appears to interact sufficiently with silver to produce a SER spectrum and no spectral signal is observed below pH 726 Figure 2 shows the SER and normal Raman spectra for KCN The SERS spectrum of 1mgml KCN in water shows a single intense somewhat broad feature at
A B
C D
SPIE-2003-5269 19
2100 cm-1 assigned to the single CequivN stretch The band is much sharper in the normal Raman spectra of the solid KCN salt at 2074 cm-1 This band does broaden and shift to 2080 cm-1 in solution (not shown) However the observed SERS frequency is attributed to interaction with silver and low frequency mode at 135 cm-1 attributed to a Ag-CN stretch (not shown) supports this conclusion
cm-1 band to a C-S stretch but the authors concede that it is in fact more likely a C-Cl stretch28 It appears that the most intense Raman bands at 648 692 and 747 cm-1 shift to 620 660 and 720 cm-1 in the SER spectra and are tentatively assigned as above The width of these bands suggests that they overlap underlying spectral features Additional bands in the Raman spectra occur at 972 1034 1049 1263 1286 1423 1442 2865 2935 and 2960 cm-1 Corresponding bands occur in the SER spectra at 964 1015 1054 1286 1410 1447 2865 and 2935 cm-1 Most of these bands are associated with alkane modes specifically the bands at approximately 1040 cm-1 to a C-C stretch 1290 cm-1 to a CH2 in-phase twist 1440 cm-1 to a CH2 wag 2865 cm-1 to a symmetric CH2 stretch and 2965 cm-1 to an asymmetric CH2 stretch The Raman and SER spectra of sulfur mustard were measured at the Edgewood center (Figure 4) Both spectra are largely similar to CEES The C-Cl and C-S bands in the Raman spectrum of HD now occur at 640 655 700 739 and 760 cm-1 and are more resolved possibly due to the increased molecular symmetry Theoretical calculations indicate that the first three bands are due to C-Cl stretching modes and the latter two to C-S stretching modes28 Only the C-Cl bands maintain significant intensity in the SER spectra occurring at 624 and 643 cm-1 which is attributed to the
Prior to measurements of HD CEES was examined by Raman and SER spectroscopy (Figure 3) CEES also known as half-mustard is essentially identical to HD except one of the chlorine end atoms is replaced by a hydrogen atom Again although not as toxic as HD CEES is a blister agent and dilute aqueous samples were prepared using appropriate safety equipment Both the Raman and SER spectra of CEES are similar and dominated by bands between 600 and 800 cm-1 These are associated with C-Cl and C-S stretching modes which are tentatively assigned to 648 and 747 cm-1 in the Raman spectra respectively The shoulder at 630 cm-1 the overlapped band at 660 cm-1 and the strong band at 692 cm-1 could also be due to these modes or their asymmetric counterparts It is worth noting that theoretical calculations assign the 692
Figure 2 A) SER and B) NR spectra of KCN Conditions A) 1 mgml in tap water 100 mW of 785 nm at sample 1-min acquisition time B) solid 300 mW of 785 nm 5-min All spectra are 8 cm-1 resolution
A
B
Figure 3 A) SER and B) NR spectra of CEES Conditions A) 1 vv (10 mgml) in MeOH 100 mW of 785 nm 1-min acquisition time B) neat 300 mW of 785 nm 5-min
A
B
Cl-CH2-CH2-S-CH2-CH3
A
B
Figure 4 A) SER and B) NR of HD Conditions A) 1mgml in tap water B) pure both 100 mW of 785 nm 1-min
Cl-CH2-CH2-S-CH2-CH2-Cl
SPIE-2003-5269 20
expected strong interaction between chlorine and silver and adds support to the assignment of this band to a C-Cl stretch Weaker overlapping bands occur at 670 692 and 724 cm-1 the latter possibly due to C-S stretching modes Again the alkane modes are apparent in the normal Raman spectra of HD but only a broad feature at 1300 to 1450 cm-1 suggests CH2 contributions in the SER spectrum Although the observed bands in the VX spectrum have not been assigned (Figure 4) a computer generated Raman spectrum29 predicts many of the same features with surprising accuracy and are used here Two intense bands at 460 and 530 cm-1 closely match predicted bands at 463 and 546 cm-1 assigned to a CH3-P=O bend and a PO2CS wag Three highly overlapped bands occur at 694 745 and 771 cm-1 matching predicted bands at 713 730 and 760 cm-1 The first
Table 2 Measured SER peak heights for the CN stretch at 2100 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
Figure 5 A) SERS and B) NR spectra of VX Conditions A) 1 vv (10 mgml) in MeOH B) pure sample both 100 mW of 785 nm 1-min acquisition time
A
B
two have been assigned to a C-S stretch and CH2bend respectively while the latter has been attributed to either a P-C stretch or an O-C-C stretch Although the 745 cm-1 band may alternatively be assigned to a C-S stretch based on the previous measurements of CEES and HD The relatively intense bands at 890 1106 1218 1445 and 1465 cm-1 also match predicted bands at 880 1108 1216 1440 and 1464 cm-1 that are assigned to a C-C stretch CH3 rock N-C3 stretch various C-H3 bends and C-H bends respectively Both the computer generated and the measured spectra contain numerous other less intense bands One is worth mentioning A unique band appears at 370 cm-1 that is predicted at 368 cm-1 and corresponds to an O-P=O bend The surface-enhanced Raman spectrum of VX is also rich with spectral features It has the unique low frequency band at 370 cm-1 as well as a second band at 380 cm-1 that is assigned tothe S-P-O bend predicted in the normal Raman spectrum at 388 cm-1 Based on the measured and predicted normal Raman spectra the following SERS assignments are given 460 cm-1 to the CH3-P=O bend 544 cm-1 to the PO2CS wag 738 cm-1 to a C-S stretch (based on arguments above) 890 cm-1
to a C-C stretch 1101 cm-1 to a CH3 rock and 1456
cm-1 to a C-H bend The ability of SERS to measure chemical agents in water containing real-world chemical interferents was tested by using the generic tap water described in Table 1 The ability to reproduce measurements was accomplished by preparing three separate water stock solutions which were used to prepare three sample solutions each of 1 mgmL KCN in the generic tap water The samples defined as vials 1-9 were then measured at 4 points per vial in a semi-random fashion such that errors associated with stock solution preparation and errors associated with instrument drift could be identified No trends were apparent that signified such systematic errors Each spectrum collected consisted of 20 averaged scans taking 16 seconds at 8 cm-1 resolution The laser power at the sample was measured periodically during the day and it ranged from 102 to 105 mW spacccc
spot Vial 1 Vial 2 Vial 41 13754 13004 128192 13519 12692 12909
1 3 13541 12721 126394 13462 12648 12651
ave 13569 12766 12755Vial 6 Vial 3 Vial 5
1 11586 14068 10757 2 11236 14402 115122 3 11376 14546 11559
4 10894 11763 11214ave 11273 13695 11261
Vial 7 Vial8 Vial 91 11198 15804 11023
2 11243 15716 88753 3 11562 15216 11218
4 11655 15084 11077ave 11415 15455 10548
CN AVG STDEV ERR1 mgmL pk ht 12526 1560 1245
stock solution
SPIE-2003-5269 21
The CN measurements consisting of 36 data points produced an average height of 125 for the 2100 cm-1 peak with a standard deviation of 156 or 125 (Table 2) The HD and VX measurements were performed precisely the same way (Tables 3 and 4) For HD the 624 cm-1 peak was used for analysis and it had an average height of 53 with a standard deviation of 068 or 129 while for VX the 544 cm-1 peak was used for analysis and it had an average height of 1051 with a standard deviation of 308 or 293 The greater error in the VX measurements can be somewhat attributed to Vial 4 which produced lower SER signal intensities But removing this vial from the data set changes the standard deviation to 233 only a modest improvement
spot Vial 1 Vial 2 Vial 41 525 609 4682 45 675 484
1 3 527 644 5094 559 807 669
ave 51525 68375 5325Vial 6 Vial 3 Vial 5
1 521 574 575 2 536 53 4492 3 509 508 451
4 565 594 379ave 53275 5515 4635
Vial 7 Vial 8 Vial 91 457 589 413
2 583 497 4273 3 544 505 4
4 528 5 509ave 528 52275 43725
HD AVG STDEV ERR1 mgmL pk ht 530 068 1291
stock solution
spot Vial 1 Vial 2 Vial 41 1464 1034 4672 1485 99 61
1 3 1491 989 5684 1041 777 553
ave 137025 9475 5495Vial 6 Vial 3 Vial 5
1 1058 942 1293 2 697 121 965
2 3 727 1206 11124 689 1254 746
ave 79275 1153 1029Vial 7 Vial 8 Vial 9
1 1127 783 163 2 1358 812 1614
3 3 1371 908 15124 1097 875 1397
ave 123825 8445 153825
VX AVG STDEV ERR1 mgmL pk ht 1051 308 2925
stock solution
4 CONCLUSIONS In this paper we examined the ability of surface-enhanced Raman spectroscopy to reproducibly measure CN HD and VX in tap water without chemical interference Both normal and surface-enhanced Raman spectra were examined to select unique bands suitable to identify and quantify these chemical agents For SER measurements the 2100 cm-1 C-N stretch was used for CN the 624 cm-1 C-Cl stretch was used for HD and the 544 cm-1 PO2CS wag was used for VX It was determined that 1 mgmL samples of each of these chemicals measured 36 times in glass vials coated with a silver-doped sol-gel reproduced measurements with standard deviations of 125 129 and 293 It was further found that the 10 chemicals added to simulate generic tap water did not interfere with or alter the SER spectra It should be noted that the concentrations used in this study were considerably greater than those required by the JSAWM program Current work involves improving SER sensitivity and designing sampling systems with better reproducibility This includes the development of fractal silver and gold structures within the sol-gel matrix and the development of chemically selective sol-gels
ACKNOWLEDGMENTS The authors are grateful for the support of the US Army (DAAD13-02-C-0015 Joint Service Agent Water Monitor program) The authors would also like to thank Janet Jensen Ronald Crosier and Kristina Gonser for helpful discussions
Table 4 Measured SER peak heights for the VX band at 544 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
Table 3 Measured SER peak heights for the HD band at 624 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
SPIE-2003-5269 22
REFERENCES 1 Jensen JL NC Wong W R Loerop SPIE 4775-03 (2001) 2 JSAWM Requirements at wwwsbccomapgeaarmymilRDAecbcrtPRODSERJSAWMjsawmhtml 3 Erickson B ldquoThe Chemical Weapons Convention Redefines Analytical Challengerdquo Anal Chem 70 397A (1998) 4 Johnston RL Hoefler CM Fargo JC and Moberley B ldquoThe Defense Nuclear Agencyrsquos
ChemicalBiochemical Weapons Agreements Verification Technology Research Development Test and Evaluation Program and its Requirements for On-Site Analysisrdquo AT-ONSITE 5-8 (1994)
5 Black RM Clarke RJ Read RW and Reid MT ldquoApplication of GC-MS and GC-tandem MS to the analysis of chemical warfare samples found to contain residues of the nerve agent sarin sulphur mustard and their degradation productsrdquo J Chrom 662 301-321 (1994)
6 USA Army Center for Health Promotion and Preventive Medicine (USACHPPM) Technical Guideline 230A 7 Hoffland LD Piffath RJ Bouck JB rdquoSpectral signatures of chemical agents and simulantsrdquo Optical
Engineering 24 982-984 (1985) 8 Braue EHJ Pannella MGrdquoCIRCLE CELL FT-IR Analysis of Chemical Warfare Agents in Aqueous Solutionsrdquo
App Spectrosc 44 1513-1520 (1990) 9 Tseng C-H Mann CK and Vickers TJ ldquoDetermination of Organics on Metal Surfaces by Raman
Spectroscopyrdquo App Spectrosc 47 1767-1771 (1993) 10 Staff Reporter ldquoUS Army Tests Lidar to Detect Biological Toxinsrdquo Photonic Spectra pg 50 December 1998 11 Christesen SD Raman cross sections of chemical agents and simulants App Spectrosc 42 318-321 (1988) 12 Weaver MJ Farquharson S Tadayyoni MA ldquoSurface-enhancement factors for Raman scattering at silver electrodesrdquo J Chem Phys 82 4867-4874 (1985) 13 Alak AM and Vo-Dinh T ldquoSERS of Organophosphorous Chemical Agentsrdquo Anal Chem 59 2149-2153 (1987) 14 Lee Y Farquharson S ldquoRapid chemical agent identification by SERSrdquo SPIE 4378 21-26 (2001) 15 Farquharson S Maksymiuk P Ong K Christesen S ldquoChemical agent identification by surface-enhanced Raman spectroscopyrdquo SPIE 4577 166-173 (2001) 16 Spencer KM Sylvia J Clauson S and Janni J ldquoSurface Enhanced Raman as a Water Monitor for Warfare
Agents in Waterrdquo SPIE 4577 158-165 (2001) 17 Tessier P Christesen S Ong K Clemente E Lenhoff A Kaler E Velev O ldquoOn-line spectroscopic characterization of sodium cyanide with nanostructured gold surface-enhanced Raman spectroscopy substratesrdquo
App Spectrosc 56 1524-1530 (2002) 18 Farquharson S W W Smith S Elliott and J F Sperry Rapid biological agent identification by surface- enhanced Raman spectroscopy SPIE 3855110-116 (1999) 19 Farquharson S WW Smith YH Lee S Elliott and J F Sperry Detection of biological signatures A
comparison of electrolytic and metal-doped sol-gel surface-enhanced Raman media SPIE 4575 62-72 (2002) 20 Guzelian AA Sylvia J Janni J Clauson S and Spencer KM ldquoSERS of whole cell bacteria and trace levels of biological moleculesrdquo SPIE 4577 182-192 (2001) 21 Shende C Inscore F Gift A Maksymiuk P Farquharson S ldquoRapid dipicolinic acid extraction from Bacillus spores detected by surface-enhanced Raman spectroscopyrdquo App Spectrosc 58 accepted 22 Lee Y and Farquharson S ldquoSERS Sample Vials Based on Sol-Gel Process for Trace Pesticide Analysisrdquo SPIE
4206 140-146 (2000) 23 Farquharson S and Lee Y ldquoTrace Drug Analysis by SERSrdquo SPIE 4200-16 (2000) 24 Lee Y Farquharson S and Rainey P M Surface-Enhanced Raman Sensor for Trace Chemical Detection in
Water SPIE 3857 76-84 (1999) 25 Lee Y Farquharson S Kwong H and Shahriari M ldquoSol-Gel Chemical Sensor for Surface-Enhanced Raman
Spectroscopyrdquo SPIE 3537 252-260 (1998) 26 Farquharson S A Gift P Maksymiuk F Inscore and W Smith ldquopH dependence of methyl phosphonic acid
dipicolinic acid and cyanide by surface-enhanced Raman spectroscopyrdquo SPIE 5269-19 (2003) 27 Lide DR Ed Handbook of Chemistry and Physics CRC Press 77th Ed (1996-7) 28 Sosa C RJ Bartlett K KuBulat and WB Person ldquoA theoretical study of harmonic vibrational frequencies and
infrared intensities of XCH2CH2SCH2CH2X and XCH2CH2SH (= H Cl)rdquo J Phys Chem 93 577-588 (1993) 29 Hameka HF and JO Jenson ldquoComputer-generated predictions of the structure and the IR and Raman spectra of
VXrdquo ERDEC-TR-065 May 1993
- Chemagents Appendicespdf
-
- SPIE2001-4575-Bioagent-SERSapdf
-
- Detection of bioagent signatures A comparison of electrolytic and metal-doped sol-gel surface-enhanced Raman media
- Stuart Farquharson Wayne Smith and Yuan Lee
-
- Real-Time Analyzers 87 Church Street East Hartford CT 06108
-
- ABSTRACT
- 4 CONCLUSIONS
- 5 ACKNOWLEDGEMENTS
- 6 REFERENCES
-
167
Table 1 Chemical Agent Structures Hydrolysis Half-lives and JSAWM Thresholds Agent Short-Hand Chemical Structure Hydrolysis
Half-Life JSAWM
Thresholds
Sarin (GB) F-[O=P-CH3]-O-CH(CH3)2 213 hours 32 microgL Soman (GD) F-[O=P-CH3]-O-CH(CH3)-(C-(CH3)3) 23 hours 32 microgL Tabun (GA) (CH3)2-N-[O=P-CN]-O-C2H5 41 hours 32 microgL VX C2H5O-[O=P-CH3]-S-(CH2)2-N-(CH(CH2)2)2 821 hours 32 microgL EA2192 HO-[O=P-CH3]-S-(CH2)2-N-(CH(CH2)2)2 gt9 years 32 microgL Mustard (H) ClCH2CH2-S-CH2CH2Cl encapsulates 47 microgL Lewisite (L) ClCH=CH-As-Cl2 rapid 27 microgL HCN HCN rapid 20 mgL BZ C7NH12-O-[C=O]-COH(C6H5)2 23 microgL T-2 Toxin 87 microgL at pH 7 to 75 and 20 to 25 oC
demonstrated by vibrational spectroscopy7-10 Hoffland et al7 reported infrared absorbance spectra and absolute Raman cross sections for several chemical agents while Christesen measured Raman cross sections for sarin tabun mustard gas and VX (ethyl S-2-diisopropylamino ethyl methylphosphonothioate)11 Again however these techniques also have limitations Raman spectroscopy is simply not a very sensitive technique and detection limits are typically 01 (1000 ppm) While infrared spectroscopy would have limited value in analyzing poisoned water since the very strong infrared absorption of water would obscure most other chemicals present Nevertheless efforts to overcome these limitations have been demonstrated Braue and Pannella8 quantified the G-series nerve agents (tabun sarin and soman) in terms of infrared attenuated total reflectance using a circle-cell And Alak and Vo-Dinh demonstrated the possibility of surface-enhanced Raman spectroscopy (SERS) to identify CWAs by measuring several organophosphonates that simulate the nerve agents12 However quantitative measurements have not been demonstrated for the SER-active material used (silver coated on alumina particles) or other SER-active media13 Recently we developed silver-doped sol-gels to promote the SER effect14-17 The porous silica network of the sol-gel matrix offers a unique environment for stabilizing SER-active metal particles and the sol-gel provides a high surface area that effectively increases the number of molecules observed within the Raman scattering volume The silver-doped sol-gels have been coated on the internal walls of standard glass vials to allow rapid SER analysis of small sample quantities (lt 01 mL) without preparation We have used p-aminobenzoic acid (PABA) as a test chemical to demonstrate surface enhancements greater than 106 reversible measurements in a flowing system reproducible measurements from vial-to-vial and batch-to-batch and measurements in multiple solvents including water14-17 Recently we used these vials to measure Tabun (GB) and Sarin and several hydrolysis products pinacolyl methyl phosphonate (PMP from Soman) and methyl phosphonic acid (MPA from all G-agents Figure 1) Although a number of unique vibrational bands are observed (eg C-N stretch doublet and P-C stretch) the G-agents were only observed for 5 concentrations and all spectra required baseline corrections Figure 1 Surface enhanced Raman spectra of ~5 vv A) Tabun and B) Sarin C) 1 vv PMP and D) 10 ppm MPA using sol-gel sample vials 785 nm excitation 1-min scan and CCD detection Performed at Aberdeen Proving Ground
C A
Wavenumber (∆cm-1) Wavenumber (∆cm-1)
D B
790
545 C-N
2135 2190 P-C 770
760 1290
Proc SPIE Vol 4577
168
Nevertheless MPA was readily observed for a 10 ppm sample with an estimated detection limit of 04 ppm (based on a signal-to-noise ratio of 3 for the 760 cm-1 band intensity) This measurement provides encouragement in that SERS may satisfy the needs of a JSAWM Furthermore MPA is also a hydrolysis product of VX and V-gas and EA2192 (Figure 2) and may prove a valuable indicator of agent usage Figure 2 Hydrolysis of Sarin to form hydrofluoric acid (HF) methylphosphonic acid 1-methylethyl ester (MPAMME) methyl phosphonic acid (MPA) and 2-propanol With this initial albeit modest success we began analyzing chemicals with various sol-gel compositions that we have been developing Here we describe four sol-gel compositions that select for 1) polar-positive 2) polar-negative 3) weakly polar-positive and 4) weakly polar-negative chemical species The ability of these sol-gels to select and enhance Raman scattering is described for several test chemicals and MPA
2 EXPERIMENTAL The chemicals analyzed as well as all chemicals used to prepare the metal-doped sol-gels were obtained at their purest commercially available grade from Aldrich (Milwaukee WI) The sol-gel designed to select for polar-negative species was prepared from a silver amine complex tetramethyl orthosilicate (TMOS) and methanol After mixing 02 mL of the sol-gel solution was transferred into a glass vial (2 mL) dried and heated The incorporated silver ions were then reduced using dilute sodium borohydride The vials were washed and dried prior to the addition of a sample solution In a similar manner the sol-gel designed to select for polar-positive species was prepared from a gold salt TMOS and methanol The sol-gel designed to select for weakly polar-negative species was prepared from a silver amine complex tetraethyl orthosilicate (TEOS) and methanol And the last sol-gel designed to select for weakly polar-positive species was prepared from a gold salt TEOS and methanol All samples were prepared in a chemical hood and transferred into plain or SER-active vials for analysis Normal Raman spectral measurements employed 1-mL pure samples that were placed in a 1-cm3 cuvette and weighed This yielded a powder density that allowed accurate calculation of molecules in the optical collection field SERS measurements employed 1-mg sample per mL water concentrations unless otherwise stated Once prepared a 01 mL sample was placed into one of the four selective sample vials which in turn was placed into the sample compartment of a Raman spectrometer for analysis A prototype Fourier transform Raman spectrometer (RTA) was used for these measurements18 The system consisted of a NdYAG laser (Brimrose) for excitation at 1064 nm an interferometer built by On-Line Technologies (OLT East Hartford CT) for frequency separation an uncooled InGaAs detector for signal detection (RTA) and an Intel 400 MHz Pentium II based laptop computer (Dell Round Rock TX) for interferometric control data acquisition (OLT) and analysis (LabVIEW by National Instruments Austin TX) Additional components included a Notch filter (Kaiser Ann Arbor MI) and interferometer entrance and exit optics (Edmund Scientific Barrington NJ) Fiber optics were used to deliver the excitation beam to the sample and the scattered radiation to the interferometer (1 meter lengths of 200 and 365 micron core diameter respectively Spectran Avon CT) A second Notch filter (Kaiser) was used as a beam splitter to direct the excitation beam along the same axis as the collected radiation A microscope object (20x04 Newport Irvine CA) was used to focus the beam into the sample and to collect the scattered radiation back along the same axis In this co-axial backscattering arrangement the excitation beam was passed through the outside of a glass vial and focused onto the silver-doped sol-gel film (01-03 mm thickness) containing the sample
3 RESULTS AND DISCUSSION p-aminobenzoic acid (PABA) and phenyl acetylene (PA) and were used to refine the selectivity and SER-activity of the four different metal-doped sol-gels PABA is a popular chemical used to evaluate the performance of SER-active media Here the polar end groups can be used to test selectivity of the polar-negative and polar-positive sol-gels PA is essentially non-
2O+ H HF + +OH
OH3H C
OP
OF
CH
CH3
3
3H C
OCP
OOH
CH
CH3
3
3H C
O
CPHO
CH
CH
3
3
C
Sarin MPAMME MPA 2-propanol
Proc SPIE Vol 4577
169
polar but a high electron density in the cylindrical π cloud around the carbon-carbon triple bond allows testing the selectivity of the weakly polar-negative and weakly polar-positive sol-gels As Figure 3 illustrates PABA passes through the polar sol-gel and is enhanced by either the silver or gold particles At 1 mgml the concentration of neutral PABA is ca 20 times that of the ionized form (pKa = 48) For electropositive silver the PABA anion is expected to interact through the carboxylate group and the associated vibrational modes are expected to dominate the spectrum Conversely for electronegative gold either form of PABA is expected to interact through the amine group The clear differences in our spectra support this expectation Furthermore bands at 840 and 1405 cm-1 assigned to a COO- bend and stretch respectively are significantly more intense for silver than gold Additional bands at 1140 and 1195 cm-1 are assigned to CH bending modes while bands at 1450 1500 and 1605 cm-1 are assigned to ring vibrational modes A very similar SER spectrum for PABA on a silver-coated alumina substrate has previously been reported with similar assignments19 For the gold-doped sol-gel new bands appear at 690 1355 and 1585 cm-1 The first band is assigned to a ring-H bending mode the second band to a ring-N- stretching mode and the third band to a possible NH2 scissors mode or ring mode The second band is not observed in the normal Raman spectrum but infrared bands occur at this frequency for aromatic ring-secondary amine stretching modes The scissors mode occurs at this frequency in Raman spectra for several chemicals but is absent in the PABA Raman spectrum Alternatively this mode may be the1600 cm-1 ring mode that has been shifted by the gold interaction Again a very similar SER spectrum of PABA has been reported but surprisingly using silver (colloids)2021 not gold as the enhancement medium These researchers also assumed the primary interaction of PABA with silver was through the carboxylate anion and made assignments accordingly For example they assigned the 1359 cm-1 to a COO- stretch not to the amine group as we have They also favor the ring stretching mode assignment for the 1582 cm-1 band Finally it should be said that other researchers have argued that the most dominant band in the SER spectra at 1450 cm-1 a ring vibration mode suggests that PABA lies flat on the surface and the π-orbitals dominate the surface interaction22
Figure 3 SER spectra of A) PABA using polar-negative and B) polar-positive sol-gels and C) PA using weakly polar-negative and D) weakly polar-positive sol-gels PABA is 1 mgmL PA is 1 vv Spectral conditions 75 mw 1064 nm 100 scans (15 min) 8 cm-1 resolution Non-polar PA passes through the non-polar sol-gels and is also enhanced by both metals The spectra are easily understood For electropositive silver PA interacts through the cylindrical triple bond π electron cloud and a -CequivC- doublet occurs near 2000 cm-1 The interaction is reasonably strong since this band appears at 2112 cm-1 in the normal Raman spectrum For electronegative gold this interaction is unlikely and only very weak bands occur near 2000 cm-1 The remaining bands are at 1000 cm-1 1200 cm-1 doublet and 1595 cm-1 all appear in the normal Raman spectra at virtually the same frequencies and are assigned to the symmetric ring-breathing mode CH bending modes and the trigonal ring-breathing mode respectively The polarnon-polar selectivity of the polar-negative and weakly polar-negative sol-gels was tested by adding a 11 molar mixture of PABA and PA The selective enhancement is quite good (Figure 4) The spectrum obtained using the polar sol-gel represents 78 PABA and 22 PA while the spectrum obtained using the weakly polar sol-gel represents 9 PABA and 91 PA The band peak intensities at 2000 cm-1 for PA and 1450 cm-1 for PABA were used for these calculations and are expanded in Figure 4 for clarity
C A
Wavenumber (∆cm-1) Wavenumber (∆cm-1)
2NH COOH
D B
C CH
Proc SPIE Vol 4577
170
Figure 4 SERS of 11 MM of PABA and PA in A) polar-negative and C) weakly polar-negative sol-gels The lower traces compare the pure chemicals B) 1 mgml PABA in polar-negative sol-gel and D) 1 PA in weakly polar-negative sol-gel while the insets magnify the minority species for clarity (x5 in A and x10 in B) Spectral conditions as in Figure 3 Following this development of selective sol-gels that maintained SER activity we measured cyanide and MPA (Figure 5) Not surprisingly the best sensitivity for both hydrolysis products was obtained using the polar-negative sol-gel The interaction of the cyanide anion with the silver surface is sufficient to shift the CequivN stretch observed at 2080 cm-1 in the normal Raman spectrum to 2145 cm-1 in the surface-enhanced Raman spectrum Furthermore the band is substantially broadened This anion has been extensively studied by electrolytic SERS and this shift and broadening have been attributed to the formation of a tetrahedral Ag(CN)3
2- surface structure23 Figure 5 Surface-enhanced (upper traces) and normal Raman spectra (bottom traces) of A) CN- and B) MPA in silver-doped TMOS SERS conditions as in Figure 3 and 1 mgmL Note MPA yields two distinct spectra for neutral (top) and acidic pH (middle) The normal Raman spectra employed pure powders 500 scans and 900 mW of 1064 nm SER measurements of MPA with the polar-negative sol-gel yielded two unique spectral signatures that depended on solution pH (Figure 5) For more neutral solutions the P-C stretch of MPA at 762 cm-1 dominates and the CH2 stretch at 2922 cm-1 appears The SN is sufficiently high that the anti-Stokes Raman shift at -762 cm-1 is observed For deprotonated MPA an oxygen-surface mode appears at 325 cm-1 (as well as its anti-Stokes complement) suggesting a strong interaction This results in substantial enhancement of the P-O-C mode at 1051 cm-1 (upper trace) Others report that this mode dominates the infrared spectra of nerve agents measured in water8 Comparison of the two spectra suggests the following molecule-to-surface orientations The appearance of the oxygen-surface and P-O-C modes in the upper spectrum of Figure 5B indicates that the tetrahedral molecule interacts with the silver surface through the deprotonated oxygen and is oriented end-on The
A B
Wavenumber (∆cm-1)
C A
Wavenumber (∆cm-1) Wavenumber (∆cm-1)
D B
Wavenumber (∆cm-1)
Proc SPIE Vol 4577
171
dominance of the P-C and the CH2 stretches and the disappearance of the P-O-C mode in the upper spectrum suggest the molecule is oriented with the methyl-hydroxyl-hydroxyl face towards the surface However considerably more research must be performed to verify these points Both chemical forms were measured as a function of concentration to determine the onset of monolayer coverage and estimate expected detection limits (Figure 6) Below monolayer coverage the signal to concentration dependence should be linear and the SN of any spectral measurement in this range can be used to predict the detection limit In the spectra presented here the peak height was used as the signal while the noise as root-mean-squared (RMS) was measured between 4400-4600 cm-1 Since noise is distributed evenly throughout the spectrum when transformed this region was used since it does not have any contributions from signals or baseline offsets Figure 6 shows a series of spectra for MPA along with a plot of the 1050 cm-1 band intensity (the noise was identical for this series) as a function of sample concentration A clear discontinuity appears in the vicinity of 01 mgmL (19 ppm) indicating the onset of monolayer coverage A detection limit defined as a SN of 3 was calculated for the 01 and 005gmL samples at 24x10-4 and 25x10-4 gL respectively A more modest detection limit of 101x10-4 gL was obtained using the 760 cm-1 band in the second series of concentration measurements These detection limits correspond to 46 and 190 ppb for a 3-min scan and 75 mW of incident laser power
Figure 6 A) Concentration dependence of MPA SERS measured in silver-doped TMOS) B) Concentrations are 001 005 01 05 1 gL (188 94 188 94 188 ppm) I760 series (bull) and I1050 series (∆)
Enhancement factors for cyanide and methyl phosphonic acid can be determined by estimating the number of molecules contributing to the surface-enhanced and normal Raman spectra The enhancement factor EF can be defined by the following equation
EF = (ISERSINR)bull(MNRMSERS) bull(PNRPSERS) bull(TNRTSERS)12
where I is the spectral band intensity M is the sample mass P is the incident laser power and T is the measurement time (or number of scans) for the two measurements For the normal Raman spectra a cylindrical scattering volume is assumed based on the laser area (28x10-7m2 6x10-4m diameter spot) and the penetration depth (1x10-3 m)24 The density of KCN and MPA as powders were measured at 0572 and 0516 gcm3 indicating that 16x10-4 and 144x10-4 g produced the normal Raman signals in Figure 5 respectively The number of molecules contributing to the SER spectra are those on the silver particles embedded in the sol-gel The total silver surface area can be determined from the average particle size concentration and the scattering volume Previous scanning electron micrographs showed the average silver particle size to be 40 nm in diameter (335x10-23m3)17 The silver concentration is 012M based on the reactant molar concentrations and dilution factors And the scattering volume is 76x10-
11m3 again based on a cylindrical scattering volume defined by a laser area of 28x10-7m2 and a sol-gel thickness of 27x10-
4m This volume contains 123x10-6g of silver equivalent to 35x109 silver particles with a collective surface area of 18x10-
5m2 However it may reasonably be assumed that at least half of the silver surface is in contact with the sol-gel matrix and unavailable for analyte interaction If we assume monolayer coverage and that each CN molecule occupies 15x10-20m2 then approximately 62x1014 molecules or 27x10-8g of CN contribute to the SER spectrum (20x10-19m2 46x1013 molecules 74x10-9g for MPA) Accordingly the EF for cyanide equals 48x104 ((180599) bull(16x10-427x10-8) bull(90075) bull(500100)12) The EF for MPA is considerably higher at 87x106 ((60326) bull(144x10-474x10-9) bull(90075) bull(500200)12)
0
20
40
60
80
100
120
140
0 02 04 06 08 1 12[MPA] (mgmL)
I (76
0)
0
100
200
300
400
500
600
I (10
50)
Wavenumber (∆cm-1)
A B
Proc SPIE Vol 4577
172
4 CONCLUSIONS Here we present for the first time surface-enhanced Raman spectra of Tabun and Sarin obtained using silver-doped sol-gels However the inferior enhancement suggested employing SER-active sol-gels varying in composition to improve both detection limits and selectivity of the target analytes To this end we successfully demonstrated the capabilities of four sol-gels that select for 1) polar-positive 2) polar-negative 3) weakly polar-positive and 4) weakly polar-negative chemical species p-aminobenzoic acid was used to show that silver could be used to attract polar-negative chemicals or functional groups (carboxylate anion) and that gold could be used to attract polar-positive chemicals (amine) while a mixture of p-aminobenzoic acid and phenylacetylene was used to show that tetramethyl orthosilicate preferentially solvates polar chemicals and that tetraethyl orthosilicate preferentially solvates weakly polar chemicals This increased sample control was applied to cyanide and methyl phosphonic acid two hydrolysis products of chemical warfare agents Exceptional results were obtained for methyl phosphonic acid allowing measurement of 1x10-2 gL for a 3-min scan and 75 mW of incident laser power with an estimated detection limit of 245x10-4 gL and an enhancement factor of 87x106 However this detection limit is 76 times less sensitive than required for the JSAWM (32x10-6gL for the G-agents) Finally we note that the measurements performed here employed an FT-Raman spectrometer This instrumentation was chosen over dispersive Raman instrumentation for the high wavelength accuracy afforded by the HeNe reference laser (Connes Advantage25) which would allow reliable spectral subtraction matching of observed spectra to stored library spectra and confident use of chemometric approaches Such data analysis is likely to be required to identify the chemical agents as well as distinguish hydrolysis products However this instrumentation which employs 1064 nm excitation and InGaAs detection sacrifices sensitivity We believe that the measurement sensitivity can be improved by at least two orders of magnitude by using shorter laser excitation wavelengths (eg 532 nm) This would provide more efficient Raman scattering (fourth power dependence on laser excitation wavelength) more efficient generation of plasmon modes and allow using more efficient detector material (Si vs InGaAs) These modifications are underway
5 ACKNOWLEDGEMENTS The authors would like to thank Dr R Yin and J Jensen of the US Army for supporting this work (Contract Number DAAD13-01-C-0019) They would also like to thank Advanced Fuel Research for making their laboratory facilities available
6 REFERENCES 1 Tu Anthony ldquoOverview of Sarin Terrorist Incidents in Japan in 1994 and 1995rdquo 6th CBW Protection Symposium
Stockholm Sweden 10-15 May 1998 2 Jensen JL NC Wong W R Loerop SPIE 4775-03 (2001) 3 JSAWM Requirements at wwwsbccomapgeaarmymilRDAecbcrtPRODSERJSAWMjsawmhtml 4 ldquoThe Chemical Weapons Convention Redefines Analytical Challengerdquo Analytical Chemistry News amp Features June 1
397A (1998) 5 Johnston RL Hoefler CM Fargo JC and Moberley B ldquoThe Defense Nuclear Agencyrsquos ChemicalBiochemical
Weapons Agreements Verification Technology Research Development Test and Evaluation Program and its Requirements for On-Site Analysisrdquo AT-ONSITE 5-8 (1994)
6 Black RM Clarke RJ Read RW and Reid MT ldquoApplication of gas chromatography-mass spectrometry and gas chromatography-tandem mass spectrometry to the analysis of chemical warfare samples found to contain residues of the nerve agent sarin sulphur mustard and their degradation productsrdquo J Chromatography 662 301-321 (1994)
7 Hoffland LD Piffath RJ Bouck JBrdquoSpectral signatures of chemical agents and simulantsrdquo Optical Engineering 24 982-984 (1985)
8 Braue EHJ Pannella MGrdquoCIRCLE CELL FT-IR Analysis of Chemical Warfare Agents in Aqueous Solutionsrdquo Applied Spectroscopy 44 1513-1520 (1990)
9 Tseng C-H Mann CK and Vickers TJ ldquoDetermination of Organics on Metal Surfaces by Raman Spectroscopyrdquo Applied Spectroscopy 47 1767-1771 (1993)
10 Staff Reporter ldquoUS Army Tests Lidar to Detect Biological Toxinsrdquo Photonic Spectra pg 50 December 1998
Proc SPIE Vol 4577
Proc SPIE Vol 4577
173
11 Christesen SD Raman cross sections of chemical agents and simulants Applied Spectroscopy 42 318-321 (1988) 12 Alak AM and Vo-Dinh T ldquoSERS of Organophosphorous Chemical Agentsrdquo Analytical Chemistry 59 2149-2153
(1987) 13 Norrod KL Sudnik LM Rousell D and Rowlen KL ldquoQuantitative Comparison of Five SERS Substrates
Sensitivity and Detection Limitrdquo Applied Spectroscopy 51 994-1001 (1997) 14 Lee Y and Farquharson S ldquoSERS Sample Vials Based on Sol-Gel Process for Trace Pesticide Analysisrdquo SPIE 4206
140-146 (2000) 15 Farquharson S and Lee Y ldquoTrace Drug Analysis by Surface-Enhanced Raman Spectroscopyrdquo SPIE 4200-16 (2000) 16 Lee Y Farquharson S and Rainey P M Surface-Enhanced Raman Sensor for Trace Chemical Detection in Water
SPIE 3857 76-84 (1999) 17 Lee Y Farquharson S Kwong H and Shahriari M ldquoSurface-Enhanced Raman Sensor for Surface-Enhanced Raman
Spectroscopyrdquo SPIE 3537 252-260 (1998) 18 Farquharson S Smith W Carangelo R C and Brouillette C ldquoIndustrial Raman Providing Easy Immediate Cost
Effective Chemical Analysis Anywhererdquo SPIE 3859 14-23 (1999) 19 Narayanan VA JM Bello JD Stokes and T Vo-Dinh Analusis 19 307-310 (1991) 20 Laserna JJ E L Torres and JD Winefordner Analytica Chemica Acta 469-480 (1987) 21 Torres EL and JD Winefordner Analytical Chemistry 59 1626-1632 (1987) 22 Suh JS DP DiLella M Moskovits J Phys Chem 87 1540-1544 (1983) 23 Benner RE R Dornhaus R Chang and BL Laube Correlations in the Raman spectra of cyanide complexes adsorbed
at silver electrodes with voltammograms Surface Science 101 341 (1980) 24 Chase D B and JF Rabolt Fourier Transform Raman Spectroscopy Acad Press Ch1 p 131 (1994) 25 Connes J Rev Opt Theor Instrum 40 45 (1961)
APPLIED SPECTROSCOPY 351
focusing the 488 nm laser beam 2 mm deep into thebulk of the crystals to avoid contributions from potentialdepletion layers As shown in Fig 1b the dependence ofthe Raman shift on the degree of deuteration is almostperfectly linear and ts very well with D 5 22684cmR1 24526 where D is the degree of deuteration (in )and R is the spectral mean of the PO4 vibration in cm21A linear correlation coef cient of 0998 indicates an ex-cellent linear dependence of the Raman peak shift withdegree of deuteration This result shows that the shift ofthe PO 4 peak is simply caused by the linear increase inatomic mass due to isotope substitution which decreasesthe length of hydrogen-like bonds
This excellent linear dependence allows us to map thepro le of the DH exchange layer at the surface of DKDPcrystals by acquiring Raman spectra and determining theposition of the PO 4 peak for various depths This methodis preferable over other methods such as determining thestrength of the OD vibration directly (eg at 715 cm21)because the position of the most intense peak in the Ra-man spectrum can be measured more precisely than theintensity of some of the weakest peaks in the spectrumThis is demonstrated in Fig 2 where depth-dependentRaman spectra (Fig 2a) and the resulting exchange layerpro les for two DKDP crystals are shown (Fig 2b) Thespectra in Fig 2a were obtained from a depth scan of aDKDP crystal with 75 degree of deuteration in thebulk grown at 45 8C The spectra start out as DKDP with30 deuteration close to the surface and approach thebulk DKDP spectrum within a few micrometers of depthThe fact that the relative degree of deuteration does notextend to 0 D is due to the limited depth resolution ofthe Raman microprobe which averages over 4 mm indepth Figure 2b depicts the resulting DH exchange layerpro les for this and a second crystal grown at 63 8Crespectively Both crystals had the same exposure to am-bient conditions and their main difference is the temper-ature at which they were grown The different exchangelayer pro les indicate that crystals grown at differenttemperatures have differen t proton conductiv ities 13
which leads to a difference in their rate of deuteriumdepletion The parameters controlling this behavior arecurrently the objective of a detailed study the results ofwhich will be reported elsewhere
CONCLUSION
In conclusion we have shown that the shift of the to-tally symmetric PO 4 stretch mode in the Raman spectrumof DKDP crystals scales linearly with degree of deuter-ation This allows us to correlate Raman peak positionsto deuteration levels in these crystals We have presenteda new technique to determine DH diffusion pro les inDKDP frequency conversion crystals based on micro-Ra-man spectroscopy This technique is fast inexpensiveand works under various environmental conditionswhich will allow us to better understand and control deu-terium depletion in DKDP crystals
ACKNOWLEDGMENTS
We would like to thank M Runkel for rst discovering DKDP crack-ing R Floyd for providing DKDP crystals and L Chase and A Burn-ham for their support and helpful discussions This work was performedunder the auspices of the US Department of Energy by the University
of California Lawrence Livermore National Laboratory under contract W-7405-Eng-48
1 J J De Yoreo A K Burnham and P K Whitman Int Mat Rev47 113 (2002)
2 C E Barker R A Sacks B M Van Wonterghern J A Caird JR Murray J H Campbell K Kyle R B Ehrlich and N DNielsen Proc SPIE-Int Soc Opt Eng 2633 501 (1995)
3 T Suratwala paper to be published4 Cleveland Crystals Inc httpwwwclevelandcrystalscom
KDPshtmltable5 E A Popova I T Savatinova and I A Velichko Sov Phys Solid
State 12 1543 (1971)6 I P Kaminow R C C Leite and S P S Porto J Phys Chem
Solids 26 2085 (1965)7 J A Subramony B J Marquardt J W Macklin and B Kahr
Chem Mat 11 1312 (1999)8 H Tanaka M Tokunaga and I Tatsuzaki Solid State Commun
49 153 (1984)9 R J Nelmes G M Meyer and J E Tibballs J Phys C 15 59
(1982)10 M A Yakshin D W Kim Y S Kim Y Y Broslavets O E
Sidoryuk and S Goldstein Laser Physics 7 941 (1997)11 I Takenaga Y Tominaga S Endo and M Kobayashi Solid State
Commun 84 931 (1992)12 C Krenn personal communication13 M Sharon and A K Kalia J Solid State Chem 21 171 (1977)
Rapid Dipicolinic Acid Extractionfrom Bacillus Spores Detectedby Surface-EnhancedRaman Spectroscopy
STUART FARQUHARSON ALAND GIFT PAUL MAKSYMIUK andFRANK E INSCOREReal-Time Analyzers Inc East Hartford Connecticut06108
Index Headings Dipicolinic acid Bacillus spores Anthrax Surface-enhanced Raman spectroscopy
INTRODUCTION
The anxiety caused by the distribution of anthrax en-dospores through the US postal system in October 2001was exacerbated by the long time required for positiveidenti cation of the Bacillus anthracis spores and the un-known extent of their distribution Since that time manymethods capable of rapid eld analysis have been inves-tigated to augment or replace the laboratory method ofgrowing microorganisms in culture media which takesdays to perform12 Prominent among these approachesare polymerase chain reactions (PCR)3 immunoassaysand detection of calcium dipicolinate as a biochemicalsignature PCR employs primers to separate organism-
Received 10 October 2003 accepted 14 November 2003 Author to whom correspondence should be sent
352 Volume 58 Number 3 2004
speci c nucleic acid sequences (eg capsular protein en-coding gene for Bacillus anthracis)4 and polymerases toamplify the segment until it is detectable Recently am-pli cation times have been substantially reduced andcomplete analysis can now be performed in an hour orless Immunoassay methods are also being developed thatuse competitive binding of the bioagent (as an antigen)and its labeled conjugate for a limited number of anti-bodies Although analyses can be performed in under 30minutes a well-de ned anthrax antigen has not yet beenidenti ed5ndash7 and consequently the false-positive rate isunacceptably high8
A number of other methods are being developed witha focus on the detection of calcium dipicolinate (CaDPA)and its derivatives as a B anthracis signature This is sobecause only spore-forming bacteria contain CaDPA andthe most common potentially interfering spores such aspollen and mold spores do not Relatively fast methodshave been developed to chemically extract CaDPA andthen detect it directly by uorescence9 or indirectly byluminescence1011 In the latter case hot dodecylamine(DDA) has been used to extract dipicolinic acid (DPA)and terbium has been utilized to form a highly lumines-cent DPA complex11 Although measurements have beenperformed in as little as ve minutes it was found thatas many as three concentration-dependent complexes canform each with different lifetimes This coupled withthe fact that the Tb31 cation produces the same lumines-cence spectrum makes determinations of low spore con-centrations problematic
It has been long known that Raman spectra of Bacillispores are dominated by bands associated with CaDPA12
and that these spectra may provide a suitable anthrax sig-nature at the genus level13 Since that time considerableimprovements in Raman instrumentation have led to lab-oratory measurements of single Bacilli spores14 and to eld measurements of spores captured from a mail-sort-ing system15 However the single spore measurementsrequired complex instrumentation that is not ruggedwhile the eld measurements required milligrams of sam-ple Furthermore the Raman spectra of both measure-ments contained uorescence contributions that would in-crease uncertainty in quanti cation
In related research we demonstrated that nanogramquantities of DPA could be detected by uorescence-freesurface-enhanced Raman spectroscopy (SERS)16 We alsodemonstrated that microliter volumes of chemicals canbe detected by SERS using metal-doped sol-gel-packedglass capillaries17 Towards the goal of developing a rap-id eld SERS-based anthrax spore detector we havecombined our previous research and we now report thatDPA can be extracted from a 10 mg B cereus spore sam-ple using DDA in 1 minute and can be detected by SERSin an additional 1 minute
EXPERIMENTAL
Dipicolinic acid (26-pyridinedicarboxylic acid DPA)and dodecylamine (DDA) were used as received fromSigma-Aldrich (Milwaukee WI) Lyophilized B cereusspores prepared according to the literature13 were sup-plied by the University of Rhode Island and used as re-ceived Multiple particles approximately 01 mm3 each
were separated and weighed at 5 to 15 mg representing05 to 15 million spores The sample masses were con-sistent with a previous determination of spore density at0081 gmL that indicated a high degree of entrained air
All chemicals used to prepare the silver-doped sol-gelcoated capillaries were also obtained and used as receivedfrom Sigma-Aldrich According to previously publishedprocedures17 two precursor solutions were preparedmixed and then drawn into 1-mm-diameter glass capil-laries The silver amine precursor consisted of a 51 vvratio of 1 N AgNO 3 to 28 NH3OH while the alkoxideprecursor consisted of a 21 vv ratio of methanol to te-tramethyl orthosilicate The alkoxide precursors weremixed with silver amine precursor in an 81 vv ratioApproximately 015 mL was drawn into the capillarycoating a 15-mm length After sol-gel formation the in-corporated silver ions were reduced with dilute sodiumborohydride which was followed by a water wash to re-move residual reducing agent
A 100 mL drop of a 50 mM DDA solution in ethanolpre-heated to 78 8C was added to each of the B cereusparticles to digest the spore coat After 1 minute the re-sultant solution was drawn into a SER-active capillarythat was immediately xed horizontally to an XY posi-tioning stage (Conix Research Spring eld OR) just in-side the focal point of an f 07 aspheric lens The lensfocused the beam into the sample and collected the scat-tered radiation back along the same axis A dichroic lter(Omega Optical Brattleborough VT) was used to re ectthe excitation laser to the lens and pass the Raman scat-tered radiation collected by the lens An f 2 achromatwas used to collimate the laser beam exiting a 200-mm-core-diameter source ber optic while a second f 2 ach-romat was used to focus the scattered radiation into a 365mm ber optic (Spectran Avon CT) A short-pass lterwas placed in the excitation beam path to block the sil-icon Raman scattering generated in the source ber fromre ecting off sampling optics and reaching the detectorA long-pass lter was placed in the collection beam pathto block the sample Rayleigh scattering from reachingthe detector A 785 nm diode laser (Process InstrumentsInc model 785-600 Salt Lake City UT) was used todeliver 100 to 150 mW of power to the sample A Fouriertransform Raman spectrometer (Real-Time Analyzersmodel IRA-785 East Hartford CT) and a silicon photo-avalanche detector (Perkin Elmer model C30902S Stam-ford CT) were used to acquire the SER spectra
RESULTS AND DISCUSSION
As an initial experiment the SER spectrum of 1 gLof DPA in water was measured using the newly devel-oped silver-doped sol-gel-coated capillaries (Fig 1A) Atthis concentration a high signal-to-noise ratio (SN) isobtained in 1 min In fact a reasonable spectrum is ob-tained in the same time frame for 1 mgL (Fig 1B) TheSER spectra are reasonably similar to the normal Raman(NR) spectrum obtained for a saturated solution of DPAin 1 N KOH (Fig 1C) and the following band shifts areobserved (NR to SER) 647 to 657 cm21 817 to 815cm21 998 to 1008 cm21 1384 to 1382 cm21 1434 to1428 cm21 and 1569 to 1567 cm21 Many of these bandshave been previously assigned1213 such as 998 cm21 to
APPLIED SPECTROSCOPY 353
FIG 1 SERS of DPA in water using silver-doped sol-gel-coated glasscapillary for (A) 1 gL and (B) 1 mgL (C ) NR of saturated DPA in 1N KOH in a glass capillary Spectral conditions (A) and (B) 150 mWof 785 nm 1-min acquisition time (C) 450 mW of 785 nm 5-minacquisition time both 8 cm 21 resolution
FIG 2 SERS of DPA extracted from 10 mg B cereus particle using100 mL of 50 mM hot DDA acquired in (A) 1 minute and (B) 2 seconds(C ) Attempted SERS of 50 mM hot DDA in ethanol using silver-dopedsol-gel-coated glass capillary acquired in 1 min Spectral conditions150 mW of 785 nm 8 cm21 resolution
the symmetric ring stretch 1384 cm21 to the OndashCndashOsymmetric stretch 1428 cm21 to the symmetric ring CndashH bend and 1569 cm21 to the asymmetric OndashCndashOstretch
The rst B cereus samples consisted of 2 mg of sporesin 2 mL of 5 mM hot DDA The samples were main-tained at 78 8C for 40 min and while hot approximately10 mL was drawn into a SER-active capillary Since spec-tra of DPA were obtained for these initial samples small-er spore masses higher DDA concentrations and shorterheating periods were examined In due course it wasfound that 10 mg of spores could be digested by 100 mLof 50 mM hot DDA in one minute and detected (Fig2A) In fact the signal was suf ciently intense that it canbe observed in as little as two seconds (Fig 2B) Theamount of DPA that was extracted was estimated to bebetween 5 and 10 mgL by comparing the signal intensityof the 1008 cm21 band to that measured for DPA in waterThis is consistent with previous research that found thatthe majority of the DPA is extracted from spores usingDDA11 and that B cereus spores contain approximately10 DPA by weight18 The SN of 127 for the 1008 cm21
band in the 1-minute SER spectrum suggests a limit ofdetection of approximately 250 ng of B cereus sporesbased on a SN of 3 Finally it should be noted that DDAdid not produce a detectable SER spectrum as shown inFig 2C
CONCLUSION
We have demonstrated that by combining rapid extrac-tion of dipicolinic acid from Bacillus cereus spores withchemical identi cation by surface-enhanced Raman spec-troscopy as little as 10 mg of spores can be detected Infact the entire measurement from the time of adding hotdodecylamine to the spores to the time when the dipi-colinic acid SER spectrum is acquired and analyzedcould be performed in less than two minutes The abilityof this method to distinguish between spore-forming bac-teria such as Bacillus anthracis and non-DPA containingpowders could help prevent costly shutdowns associated
with the appearance of suspicious material or intentionalmailing of common substances as an anthrax hoax Thismethod could also prove useful in detecting the locationof anthrax endospores in mail distribution facilities if an-other veri ed attack should occur
Research continues to fully characterize the surface-enhanced Raman spectroscopy signal intensities as afunction of sample concentration and to explore otherextractants that do not require the use of elevated tem-perature
ACKNOWLEDGMENTS
The authors are grateful for the support of the National Science Foun-dation (DMI-0296116 and DMI-0215819) and the US Army(DAAD13-02-C-0015 Joint Service Agent Water Monitor program)The authors are indebted to Chetan Shende for preparing the sol-gelcapillaries The authors also thank James Gillespie Nicholas Fell andAugustus Fountain for providing important background informationMark Farquharson for laboratory support and Professor Jay Sperry ofthe University of Rhode Island for supplying B cereus spores
1 V A Pasechnik C C Shone and P Hambleton Bioseparations 3267 (1993)
2 P J Jackson M E Hugh-Jones D M Adair G Green K K HillC R Kuske L M Grinberg F A Abramova and P Keim ProcNatl Acad Sci USA 95 1224 (1998)
3 B R Glick and J J Pasternak Molecular Biology Principles andApplications of Recombinant DNA (ASM Press Washington DC1994)
4 C A Bell J R Uhl T L Had eld J C David R F Meyer TF Smith and F R Cockerill III J Clin Microbiol 40 2897(2002)
5 D L Gatto-Menking H Yu J G Bruno M T Goode M Millerand A W Zulich Biosens Bioelectron 10 501 (1995)
6 J J Quinlan and P M Foegeding J Rapid Methods AutomationMicrobiol 6 1 (1998)
7 A A Hindle and E A H Hall Analyst (Cambridge UK) 1241599 (1999)
8 M S Ascher US Department of Health amp Human Services(httpwwwhhsgovophppresentationsAscherdoc)
9 R Nudelman B V Bronk and S Efrima Appl Spectrosc 54 445(2000)
10 D L Rosen C Sharpless and L B McBrown Anal Chem 691082 (1997)
354 Volume 58 Number 3 2004
11 P M Pellegrino N F Fell Jr and J B Gillespie Anal ChimActa 455 167 (2002)
12 W H Woodruff T G Spiro and C Gilvarg Biochem BiophysRes Commun 58 197 (1974)
13 E Ghiamati R S Manoharan W H Nelson and J F SperryAppl Spectrosc 46 357 (1992)
14 A P Esposito C E Talley T Huser C W Hollars C M Schal-dach and S M Lane Appl Spectrosc 57 868 (2003)
15 S Farquharson L Grigely V Khitrov W W Smith J F Sperryand G Fenerty J Raman Spectrosc paper accep ted (2003)
16 S Farquharson W W Smith S Elliott and J F Sperry SPIE-IntSoc Opt Eng 3855 110 (1999)
17 S Farquharson and P Maksymiuk Appl Spectrosc 57 479(2003)
18 F W Janssen A J Lund and L E Anderson Science (Washing-ton DC) 127 26 (1958)
SPIE -2003-5269 117
pH dependence of methyl phosphonic acid dipicolinic acid and cyanide by surface-enhanced Raman spectroscopy
Stuart Farquharson Alan Gift Paul Maksymiuk Frank Inscore and Wayne Smith
Real-Time Analyzers 87 Church Street East Hartford CT 06108
ABSTRACT US and Coalition forces fighting terrorism in Afghanistan and Iraq must consider a wide range of attack scenarios in addition to car bombings Among these is the intentional poisoning of water supplies to obstruct military operations To counter such attacks the military is developing portable analyzers that can identify and quantify potential chemical agents in water supplies at microgram per liter concentrations within 10 minutes To aid this effort we have been investigating the value of a surface-enhanced Raman spectroscopy based portable analyzer In particular we have been developing silver-doped sol-gels to generate SER spectra of chemical agents and their hydrolysis products Here we present SER spectra of methyl phosphonic acid and cyanide as a function of pH an important factor affecting quantitation measurements which to our knowledge has not been examined In addition dipicolinic acid a chemical signature associated with anthrax-causing spores is also presented Keywords Chemical warfare agents agent detection agent hydrolysis SERS Raman spectroscopy homeland security
1 INTRODUCTION In the past decade the Unites States and its allies have been challenged by a different kind of warfare exemplified by the terrorist attacks of September 11 2001 Suicide bombings and the use of chemical agents are the norm and military personnel must consider a wide range of attack scenarios Among these is the intentional poisoning of water supplies to obstruct military operations in Afghanistan and Iraq Currently colorimetric paper is used to detect agents on-site while gas chromatography combined with mass spectrometry (GCMS) is used in mobile support laboratories However both methods have severe drawbacks The paper changes color in response to contact with many chemicals besides CWAs causing a high incidence of false positives1 while GCMS although very chemically specific requires hours to perform and constant re-calibration234 Military operations would be greatly aided by a portable analyzer that can identify and quantify potential chemical agents at concentrations that impact safety This includes the analysis of drinking water supplies distribution and storage systems To meet this goal the Department of Defense has been investigating numerous approaches under the auspices of the Joint Service Agent Water Monitor (JSAWM) program5 The JSAWM must be capable of identifying the chemical agents with no false-positives and quantifying the agents at microgram per liter concentrations within 10 minutes678 Recently we and others have been investigating the ability of surface-enhanced Raman spectroscopy (SERS) to measure chemical agents9-12 bioagents13-17 and their hydrolysis products in water SERS employs the interaction of surface plasmon modes of metal particles with target analytes to increase scattering efficiency by as much as 1 million times18 In 1987 the potential of this approach to measure CWAs was demonstrated by measuring a series of pesticides19 Several of these organophosphonates have chemical structures similar to CWAs in particular P=O functional groups In our studies we have been employing silver-doped sol-gels to promote the SER effect The porous silica network of the sol-gel matrix offers a unique environment for immobilizing and stabilizing SER-active metal particles20-23 The sol-gels have been coated on the internal walls of standard glass vials to allow rapid SER analysis of small sample quantities without preparation We have measured over 100 chemicals with enhancements of 104 to 106 demonstrated reversible measurements in a flowing system reproducible measurements from vial-to-vial and measurements in multiple solvents including water20-23 Previously we used these vials to perform preliminary measurements of cyanide (CN) methylphosphonic acid (MPA) and dipicolinic acid (DPA) MPA is a hydrolysis product of the nerve agents (eg sarin Reaction 1) and may be a valuable indicator of nerve agent usage particularly since the alkyl methylphosphonic acids are relatively more stable than their corresponding parent complexes24 DPA is
SPIE -2003-5269 118
a chemical signature of spore forming bacteria such as Bacillus anthracis And in light of the inability to rapidly detect the anthrax spores distributed through the US mail in October 2001 a number of methods are being developed to extract and analyze this signature Reaction 1 Stepwise hydrolysis of Sarin to form hydrofluoric acid (HF) isopropyl methylphosphonic acid (IMPA) then methyl phosphonic acid (MPA) and 2-propanol In our previous SERS investigations MPA and DPA were measured at 50 and 100 mgL respectively In both cases limits of detection (LOD) were estimated at 100 microgL providing encouragement in that SERS may satisfy the needs of the JSAWM Since it has been shown that pH can substantially influence the intensity of SER bands25 which would clearly influence quantitative analysis we undertook the present study to determine the severity of these effects for cyanide methyl phosphonic acid and dipicolinic acid Furthermore we previously observed a band at 1050 cm-1 for MPA6 possibly due to an anion formed at basic pH Here we investigate the source of this spectral anomaly
2 EXPERIMENTAL All chemicals including potassium cyanide methyl phosphonic acid dipicolinic acid and those used to prepare the silver-doped and gold-doped sol-gel coated vials were obtained and used as received from Sigma-Aldrich All samples were prepared in HPLC grade water (Fischer Scientific Fair Lawn NJ) for SERS measurements The pH of these samples was adjusted using dilute nitric acid or potassium hydroxide and verified using a pH electrode (Corning Inc Corning NY) that had been calibrated with pH 400 700 and 1000 buffered standards from Fischer Scientific Once prepared the samples were transferred into the silver-doped sol-gel vials (Real-Time Analyzers Inc East Hartford CT) The vials were coated in a manner similar to that previously reported by adding ammonium hydroxide to a solution of silver nitrate tetramethyl orthosilicate (TMOS) and methanol20 Gold-vials were coated by adding nitric acid to a solution of gold tetrachloride TMOS and methanol The two precursor solutions were prepared mixed and transferred to 2-mL glass vials dried and heated After sol-gel formation the incorporated metal ions were reduced with dilute sodium borohydride (1mgmL) which was followed by a water wash to remove residual reducing agent After the resultant analyte solution was introduced the SER-active vial was immediately fixed horizontally to an XY positioning stage (Conix Research Springfield OR) just inside the focal point of an f07 aspheric lens The lens focused the beam into the sample and collected the scattered radiation back along the same axis A dichroic filter (Omega Optical Brattleborough VT) was used to reflect the excitation laser to the lens and pass the Raman scattered radiation collected by the lens An f2 achromat was used to collimate the laser beam exiting a 200 microm core diameter source fiber optic while a second f2 achromat was used to focus the scattered radiation into a 365 microm fiber optic (Spectran Avon CT) A short pass filter was placed in the excitation beam path to block the silicon Raman scattering generated in the source fiber from reflecting off sampling optics and reaching the detector A long pass filter was placed in the collection beam path to block the sample Rayleigh scattering from reaching the detector A 785 nm diode laser (Process Instruments Inc model 785-600 Salt Lake City UT) was used to deliver 100 to 150 mW of power to the sample A Fourier transform Raman spectrometer (Real-Time Analyzers model IRA-785 East Hartford CT) and a silicon photo-avalanche detector (Perkin Elmer model C30902S Stamford CT) were used to acquire the SER spectra
3 RESULTS AND DISCUSSION In a previous study of MPA6 aimed at developing a concentration calibration curve and determining limits of detection (LOD) we observed an anomaly at 1050 cm-1 Since it was found that the band intensity changed as a function of concentration the band must be associated with a sample parameter Two possible parameters photon flux and pH are examined here The first parameter was investigated by irradiating a 1mgmL MPA sample in a SER-active vial with laser powers of 200 mW and above and monitoring spectral changes It was immediately found that the 1050 cm-1 band
2O+ H HF + +OH
OH3H C
OP
OF
CH
CH3
3
3H C
OCP
OOH
CH
CH3
3
3H C
O
CPHO
CH
CH
3
3
C
Sarin IMPA MPA 2-propanol
SPIE -2003-5269 119
appeared and grew as a function of time and that the higher the laser power the faster the growth Figure 1 shows the growth of the 1050 cm-1 band over the course of 30 minutes when using 150 mW of 785 nm excitation while Figure 2A shows that the growth can be fit with a first order exponential equation namely I1050 = 03+05e-013t Furthermore the 760 cm-1 band could be fit with a first order decay equation with an identical exponential rate constant ie I760 = 08-
08e-013t The rates represent classical first order kinetics and their correspondence allows one to conclude that MPA is being transformed one-for-one into a photo-generated product At this time the photoproduct has not been positively identified but phosphonic (phosphorous) acid and phosphonate are likely candidates since the symmetric P(OH)3 stretch occurs at ~1050cm-1 Our studies show that a reduction in laser power to 100 mW at the sample for MPA essentially eliminates this degradation process This laser power or lower was used for further measurements
Methyl phosphonic acid is a diprotic acid that stepwise dissociates into two anions MPA- and MPA= according to the following reactions26
MPA MPA- + H+ pKa1 = 212 Reaction 2
MPA- MPA= + H+ pKa2 = 729 Reaction 3 The relative concentrations of MPA MPA- and MPA= can be determined at any pH by expressing [MPA] and [MPA=] in terms of [MPA-] using Reactions 2 and 3 and summing all three to equal the total starting concentration here 2 mgmL (0021M MW = 9602) viz
[MPA] + [MPA-] + [MPA=] = 0021M Equation 1 substituting from Reactions 1 and 2
([H+][MPA-])K1a + [MPA-] + (K2a[MPA-])[H+] = 0021M Equation 2 rearranging [MPA-] = 0021M(1+[H+]K1a + K2a[H+]) Equation 3 The relative concentrations of MPA MPA- and MPA= as a function of pH are shown in Figure 3 It is worth noting that near neutral pH both MPA- and MPA= will be present To confirm that the SER signal followed this pH dependence a starting solution consisting of 20 mg of MPA in 10 mL HPLC grade water was prepared and brought to pH of 20 using dilute nitric acid From this solution 2 mL were added to a SER-active vial and the SER spectrum recorded At this pH a peak at 760 cm-1 was barely discernable The 2 mL solution was returned to the starting solution and the pH was re-measured to correct for any changes that the silver-doped sol-gel vials might cause In most cases the change was less than 02 pH units and the pH is reported as the before and after average Next the pH of the
Figure 2 A) Exponential growth of 1050 cm-1 band and B) exponential decay of 760 cm-1 band for spectral series in Figure 1
Figure 1 Growth of 1050 cm-1 band as a function of time due to exposure to 150 mW of 785 nm Spectra are 5 sec each collected every 100-sec from 0 to 30-min
B
570 770 970 1170 1370Raman Shift (cm-1)
Arbit
rary
Unit
s
MPA Photodegradation
600 800 1000 1200 1400 Wavenumber (cm-1)
A
0 10 20 30 0 10 20 30 time (min) time (min)
Ram
an In
tens
ity (r
elat
ive)
30
min
0
SPIE -2003-5269 120
starting solution was adjusted to 325 using dilute KOH Again 2 mL were added to a vial and the SER spectrum recorded At this pH a reasonably strong 760 cm-1 band was observed This process was repeated as spectra were recorded at pHs of 70 74 75 79 85 and 100 A total of 1 mL of KOH was added diluting the total concentration by 10 Next the pH of the starting solution was made acidic by adding dilute nitric acid dropwise This time spectra were recorded at pHs of 72 69 64 and 37 Figure 4 shows the SER spectra for representative pHs (spectra were left out to simplify the figure) while Figure 3 shows the 760 cm-1 peak intensities as a function of pH (The band intensities were adjusted to compensate for dilution effects caused by the addition of HNO3 and KOH then normalized to 0021 M for the most intense band observed at pH 37) It is clear from Figure 3 that the 760 cm-1 band follows the MPA- concentration as a function of pH and must be assigned to this anion No bands were observed that corresponded to MPA or MPA= The lack of an MPA SER spectrum may be due to the absence of an attraction between the neutral analyte and the electropositive silver surface The same reasoning suggests that a strong SER spectrum should be observed for MPA= but it is not and a satisfactory explanation has not been found
0000
0005
0010
0015
0020
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14pH
Con
cent
ratio
n [M
]
pK1 = 212 pK2 = 729
MPA- MPA=
MPA
Figure 4 SER spectra of 002M MPA as a function of pH Conditions 100 mW of 785 nm 36 scans (1 min) 8 cm-1 recorded 2 min after sample introduction pH 19 and 20 not apparent on this scale pH 69 and 74 near identical to 70 and 75 and not shown for clarity
00
02
04
06
08
10
12
14
16
18
0 5 10 15 20 25 30Measurement Number
Ram
an In
tens
ity (7
60 c
m-1
)
Figure 3 Concentration dependence of MPA MPA- and MPA= as a function of pH for a 002M sample Intensity of 760 cm-1
band from Figure 6 as a function of pH ( for increasing basic adjustment for increasing acidic adjustment error was measured at ~10 for pH 64)
Figure 5 SER spectra of 002M MPA at pH 64 measured around a vial at three heights (9 points per height) Conditions as in Figure 4 but 10-sec scans
Since these measurements involved the removal and replacement of the SER-active vial in the sample holder to remove and add sample variation in the intensity as a function of vial position was minimized by illuminating the exact same height along the vial wall But this does not account for variability of the SERS response of the sol-gel coating around the vial To analyze this effect a vial containing MPA at pH 64 was rotated at ~ 40o
intervals at the original height and 18rdquo above and below this value Figure 5 shows the intensity of the 760 cm-1
band for the 27 positions It was found that the average value was 137plusmn014 an RSD of 10 overall and 5 for each height An error bar is included in Figure 4 for the pH 64 measurement
SPIE -2003-5269 121
With the above analyses in mind a preliminary investigation of the SER spectral response for MPA (as MPA-) as a function of concentration was performed A single vial was used for these measurements beginning with 1 mgL followed by measurements of 10 100 and 1000 mgL In all cases the pH was ~7 and 3 positions around the vial were measured per concentration Since the 760 cm-1 band was not observed for concentrations of 1 or 10 mgL using 100 mW of 785 nm the laser power at the sample was raised to 200 mW beginning with the 10 mgL concentration Photo-degradation was largely avoided (and not observed) by exposing the sample for only 33 seconds per spectral acquisition Representative spectra for 10 100 and 1000 mgL are shown in Figure 6 while a plot of the 760 cm-1 band intensity as a function of concentration is shown in Figure 7 These values were also used to estimate limits of detection based on the signal-to-noise ratio (SN) of the 760 cm-1 band using 100 mW of 785 nm laser excitation and a 10-min acquisition time As summarized in the Figure 7 inset the lower the measured concentration the lower the predicted LOD Even if as estimated 210 microgL could be measured using the silver-doped sol-gel vials an improvement of a factor of 70 is still required to meet JSAWM goals of measuring 3 microgL in 10-minutes Similar to MPA DPA is a diprotic acid (pKa1 = 216 and pKa2 = 692) and variations in pH will effect the relative concentrations of DPA DPA- and DPA= and possibly the SER spectra and band intensities This could prove significant if an acid or base is used to denature anthrax spores with the goal of extracting and analyzing DPA The MPA pH study described above was mimicked for DPA except that the starting solution consisted of 20 mg of DPA in 20 mL HPLC grade water (60x10-3M MW = 1671) The initial solution had a pH of 245 which was made basic by dilute KOH to pHs of 355 433 487 559 1069 and 1166 SER spectra were recorded at each pH using 100 mW of 785 nm and a 44-sec acquisition time Next one drop of concentrated nitric acid was used to remake the solution acidic at a pH of 200 Again sequential pH measurements were performed at 383 510 735 and 822 The solution pH was made acidic a third time but to pH 219 171 then 135 Throughout this process no more than 20 drops of acid or base were added and therefore the concentration was diluted by no more than 10 Most of the spectral bands showed a minor decrease in intensity as a function of increasing pH values However the bands at 525 and 795 cm-1 showed the most dramatic changes which occurred at acid pH Figure 8 shows the SER spectra of DPA for the spectral region and pH range of interest The identity of the DPA species was determined by plotting the normalized peak intensities with the lowest value set to 0 and the highest to 0006 M as a function of pH and overlaying these values on a plot of the relative concentrations for DPA DPA- and DPA= as previously done for MPA (Figure 9) As can be seen the 525 cm-1 band clearly corresponds to DPA The correspondence of the 795 cm-1 band to this species is less clear as the band retains intensity until pH of 55 This can be attributed to contribution to the overlapping band at 810 cm-1 which does not change as a function of pH The fact that most bands are observed at all pHs suggest that the primary interaction with silver is through the ring nitrogen This is supported by the fact that the most intense band occurs at 1008 cm-1 attributed to a symmetric ring breathing mode and that this interaction has been characterized for pyridine in numerous papers27
0
1
2
3
4
5
6
7
0 200 400 600 800 1000 1200
MPA Concentration (mgL)
760
Ban
d In
tens
ity
Figure 6 SER spectra of MPA in water at A) 1000 B) 100 and C) 10 mgL Conditions pH of 7 silver-doped sol-gel coated vial 200 mW of 785 nm 33-sec 8 cm-1 resolution
Figure 7 Plot of SER intensity of 760 cm-1 band of MPA as a function of concentration using 200 mW of 785 nm Inset table includes average intensity LOD standard deviation and percent deviation for each concentration but for 100 mW and 10 min
conc (mgL) LOD ave stddev dev10 021 002 1021100 072 010 14301000 312 040 1280
A
B C
SPIE -2003-5269 122
The ability to detect dipicolinic acid (DPA) by SERS is demonstrated in Figure 10 Even at 1 mgL the primary bands are visible The intensity of the dominant ring breathing mode observed at 1008 cm-1 for DPA in water (pH = 25 to 55) is plotted as a function of concentration in Figure 11 These values were also used to estimate limits of detection based on the SN of the 1008 cm-1 band using 100 mW of 785 nm laser excitation and a 10-min acquisition time Again the lower the measured concentration the lower the predicted LOD (see Figure 11 inset) and detection of 160 microgL is possible Gastrointestinal anthrax requires significant more spores than inhalation anthrax28 and a limit of detection might be placed at 1 million spores in 1 liter of water or 10 microgL Since each spore contains ~10 CaDPA by weight29 a goal for DPA might be 1 microgL indicating that the present measurements must be improved by nearly two orders of magnitude Due to the increased hazards of handling HCN gas KCN salt was used for these experiments Nevertheless all sample preparations were performed in a chemical hood KCN completely dissolves in water but its conjugate acid HCN is formed and has a Ka of 615x10-1030 viz
HCN CN- + H+ pKa = 921 Reaction 4
Figure 8 SER spectra of 1 mgmL DPA as a function of pH Conditions 100 mW 785 nm 100 scans (44-sec)
Figure 9 Concentration dependence of DPA DPA- and DPA= as a function of pH for a 0006M sample Intensity of 525 () and 795 (diams) cm-1 bands from Fig 8 as a function of pH
pH
135 171 219 383
0
0001
0002
0003
0004
0005
0006
0007
0 2 4 6 8 10 12 14pH
Con
cent
ratio
n [M
]
DPADPA-DPA=795525
DPA DPA=DPA-
pK2 = 692pK1 = 216
795 525
0
05
1
15
2
25
0 200 400 600 800 1000 1200
DPA Concentration (mgL)
1008
Ban
d In
tens
ity
conc (mgL) lod-10min-100mw1 017
10 016100 103
1000 355
Figure 10 SER spectra of DPA in water at A) 1000 B) 100 C) 10 and D) 1 mgL Conditions pH of 25-55 silver-doped sol-gel coated vial 175 mW of 785 nm 1-min 8 cm-1 D) has been multiplied by x10 to make bands visible
A
B C
D
Figure 11 Plot of SER intensity of 1008 cm-1 band of DPA as a function of concentration using 175 mW of 785 nm Inset table includes LOD in mgL for each concentration but for 100 mW and 10 min
SPIE -2003-5269 123
Consequently the cyanide concentration must be determined for each initial KCN concentration Specifically the samples prepared with concentrations of 01 1 10 100 and 1000 mgL of KCN produced CN- concentrations of 63x10-3 033 69 89 and 964 mgL at pHs of 816 90 967 102 and 107 respectively The pH dependence for the HCN and CN- concentrations are shown in Figure 12 Thus as the amount of KCN added to the solution decreases so does the pH of the solution (becomes less basic) and according to Reaction 4 the relative amount of CN- to HCN also decreases For example in the preparation of a 01 mgL solution of KCN the pH is shifted from 7 for pure water to only 816 and only 63 of the starting material becomes CN- or 63x10-3 mgL In comparison for a solution of 1000 mgL the pH is shifted from 7 to 107 and 96 of the starting material becomes CN- This is significant since the cyanide ion is better able to adsorb onto the silver particles and become SERS active SER spectra of 10 100 and 1000 mgL of KCN samples are shown in Figure 13 and a corresponding plot of the 2100 cm-1 band intensity as a function of concentration is shown in Figure 14
000102030405060708091011
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14pH
Con
cent
ratio
n [m
gm
L]
pKa = 921
CN -HCN
The SER spectra of cyanide are dominated by the single CN stretch at 2100 cm-1 which occurs in normal Raman spectra of solutions at 2080 cm-1 However a low frequency mode occurs at 135 cm-1 due to a Ag-CN stretch (not shown) indicative of a strong surface interaction It is also observed that as the concentration decreases the CN stretch shifts to 2140 cm-1 This shift has been attribute to the formation of a tetrahedral Ag(CN)3
2- surface structure31 as well
Figure 14 Concentration dependence of KCN SERS measured under conditions in Fig 11 Concentrations are 1 01 and 001 mgml Intensities are measured for the CN stretch at 2100 cm-1 Inset table includes LOD in mgL for each concentration in Figs 13 and 15 but for 100 mW and 10 min
Figure 13 SER spectra of KCN in water at A) 1000 B) 100 and C) 10 mgL Conditions pHs of 107 102 and 97 silver-doped sol-gel coated vial 100 mW of 785 nm 1-min 8 cm-1 C) has been multiplied by x10 to make band visible
Figure 12 Concentration dependence of HCN and CN- as a function of pH for a 1 mgmL sample Calculated intensity of 2100 () cm-1 band for a 1 mgml sample at pHs of 816 90 967 102 and 107
A
B
C
0
20
40
60
80
100
120
140
0 200 400 600 800 1000 1200CN Concentration (mgL)
2100
Ban
d In
tens
ity
Figure 15 SER spectra of KCN in water at A) 10 B) 1 and C) 01 mgL Conditions pHs of 97 12 and 12 gold-doped sol-gel coated vial 100 mW of 785 nm 1-min 8 cm-1 C) has been multiplied by x10 to make band visible
A
B
C
conc (mgL) lod-10min-100mw Condition01 001 Au-pH 121 007 Au-pH 12
69 003 Au-pH 9769 016 Ag-pH 9789 022 Ag-pH 102946 113 Ag-pH 107
SPIE -2003-5269 124
as to CN adsorbed to two different surface sites32 Alternatively the 2140 cm-1 band could be attributed to HCN since this species dominates at lower concentrations However it is unlikely that this species would be attracted to the electropositive silver surface Further both peaks should be present at pHs between 85 and 105 but this is not observed It has also been suggested that at concentrations near and above monolayer coverage the CN- species is forced to adsorb end-on due to crowding and at lower concentrations the molecule can reorient to lie flat33 This suggests that the 2100 and 2140 cm-1 bands correspond to the end-on and flat orientations respectively As Figures 13 and 14 show the intensity of the CN stretch for the 89 mgL sample is nearly as intense as the 964 mgmL band This suggests that the Raman signal for the flat orientation is more enhanced However more extensive measurements are required to verify this point Since resent research has suggested that cyanide may be more effectively detected on gold measurements of KCN solutions were also performed using gold-doped sol-gel vials Preliminary measurements are shown in Figure 15 for samples prepared from 01 1 and 10 mgL KCN Since the pHs are 816 90 967 the resultant CN- concentrations are 63x10-3 033 and 69 mgL Initially only the highest concentration was observed and the signal intensity was significantly better than the equivalent concentration measured using silver In an effort to shift Reaction 4 to the left transforming HCN to CN- (Le Chatelierrsquos principle) KOH was added to the lower concentration samples producing solutions with pH 12 This effectively forces all of the cyanide in solution to be CN- or 01 and 10 mgL respectively More importantly the CN stretch is now observed in the SER spectra The band appears at 2125 cm-1 as has been previously reported for gold12 As calculated for MPA and DPA LODs can be estimated from this data For the three concentrations of cyanide on silver the LODs are 016 to 11 mgL for 100 mW of 785 nm laser excitation and a 10-min acquisition time For gold pH adjusted the LODs are10 to 70 microgL an improvement of more than 10 times silver Nevertheless either substrate is sufficient to meet the JSAWM goals of measuring 3 mgL in 10-minutes as the requirements form cyanide are much less stringent than the nerve agents
4 CONCLUSIONS Here we examined the surface-enhanced Raman spectral response for methyl phosphonic acid dipicolinic acid and cyanide as a function of pH It was determined that the most prominent peak at 760 cm-1 reaches a maximum intensity between pH 3 and 7 and corresponds to the MPA- species Neither the MPA nor MPA= species appear to generate a SER spectrum and consequently no spectra were observed below pH 2 or above pH 8 In this study we also found that higher laser powers could cause photodegradation of MPA signified by the exponential growth of a band at 1050 cm-1 which is tentatively assigned to phosphorous acid Unlike MPA DPA was observed at all pHs This is attributed to the dominant interaction of the pyridine functional group with silver Minor spectral changes were observed at acid pHs and were assigned to neutral DPA Like MPA SER spectra of cyanide were pH dependent No spectra were observed for the HCN species while CN- was best observed at pHs more basic than 8 Preliminary concentration studies for the three analytes allowed estimating limits of detection for MPA DPA and CN using 100 mW of 785 nm and a 10-min acquisition time of 210 165 and 70 microgL respectively Although the latter value suggests that cyanide can be measured at sufficiently low concentrations to meet JSAWM goal improvements by 100 to 200 times are required for MPA and DPA It is clear from this study that pH of the sample is important and must be taken into account when developing concentration calibration curves as well as fieldable analyzers Future work will include tailoring the sol-gel with specific functional groups to dictate sample pH and thereby optimize sensitivity
ACKNOWLEDGMENTS The authors are grateful for the support of the US Army (DAAD13-02-C-0015 Joint Service Agent Water Monitor program) The authors would also like to thank Dr Steve Christensen of the US Army SBCCOM for helpful discussions and Mr Chetan Shende of Real-Time Analyzers for assistance in development of the gold-doped sol-gels
REFERENCES 1 Erickson B Analytical Chemistry News amp Features June 1 397A (1998)
SPIE -2003-5269 125
2 Johnston RL Hoefler CM Fargo JC and Moberley B AT-ONSITE 5-8 (1994) 3 Black RM RJ Clarke RW Read and MT Reid J Chromatography 662 301-321 (1994) 4 USA Army Center for Health Promotion and Preventive Medicine (USACHPPM) Technical Guideline 230A 5 Jensen JL NC Wong W R Loerop SPIE 4775-03 (2001) 6 Hoenig SL Handbook of Chemical Warfare and Terrorism Greenwood Press Wesport CT (2002) 7 Munro NB SS Talmage GD Griffin LC Waters AP Watson JF King and V Hauschild Env Health
Persp 107 933-974 (1999) 8 Holstege CP Kirk M Sidell FR Crit Care Clin13 923-42 (1997) 9 Farquharson S P Maksymiuk K Ong and S Christesen SPIE 4577 166-173 (2001) 10 Lee Y and S Farquharson SPIE 4378 21-26 (2001) 11 Spencer KM J Sylvia S Clauson and J Janni SPIE 4577 158-165 (2001) 12 Tessier P S Christesen K Ong E Clemente A Lenhoff E Kaler and O Velev Applied Spectroscopy 56
1524-1530 (2002) 13 Farquharson S WW Smith S Elliott and JF Sperry SPIE 3533 207-214 (1998) 14 Farquharson S WW Smith S Elliott and JF Sperry SPIE 3855110-116 (1999) 15 Farquharson S WW Smith YH Lee S Elliott and JF Sperry SPIE 4575 62-72 (2002) 16 Guzelian AA J Sylvia J Janni S Clauson and KM Spencer SPIE 4577 182-192 (2001) 17 Shende C F Inscore A Gift P Maksymiuk and S Farquharson in press 18 Weaver MJ S Farquharson and MA Tadayyoni J Chem Phys 82 4867-4874 (1985) 19 Alak AM and T Vo-Dinh Analytical Chemistry 59 2149-2153 (1987) 20 Lee Y and S Farquharson SPIE 4206 140-146 (2000) 21 Farquharson S and Y Lee SPIE 4200-16 (2000) 22 Lee Y S Farquharson and P M Rainey SPIE 3857 76-84 (1999) 23 Lee Y S Farquharson H Kwong and M Shahriari SPIE 3537 252-260 (1998) 24 Wang J M Pumera G Collins and A Mulchandani Analytical Chemistry 74 6121-6125 (2002) 25 Dou X YM Jung Z-Q Cao and Y Ozaki Applied Spectroscopy 53 1440-1447 (1999) 26 Data supplied by S Christesen and K Ewing 27 Kerker M and B Thompson Eds SPIE MS 10 (1990) 28 Inglesby TV DA Henderson JG Bartlett JAMA 287 2236 (2002) 29 FW Janssen AJ Lund and LE Anderson Science 127 26 (1958) 30 Lide DR Ed Handbook of Chemistry and Physics CRC Press 77th Ed (1996-7) 31 Billmann J G Kovacs and A Otto Surf Sci 92 153 (1980) 32 Murray CA and S Bodoff Phys Rev B 32 671 (1985) 33 Kellogg D and J Pemberton J Phys Chem 91 1120 (1987)
SPIE-2003-5269 16
Chemical agent detection by surface-enhanced Raman spectroscopy
Stuart Farquharson Alan Gift Paul Maksymiuk Frank Inscore and Wayne Smith
Real-Time Analyzers 87 Church Street East Hartford CT 06108
Kevin Morrisey and Steven D Christesen US Army SBCCOM Aberdeen Proving Ground MD 21010
ABSTRACT
In the past decade the Unites States and its allies have been challenged by a different kind of warfare exemplified by the terrorist attacks of September 11 2001 Although suicide bombings are the most often used form of terror military personnel must consider a wide range of attack scenarios Among these is the intentional poisoning of water supplies to obstruct military operations in Afghanistan and Iraq To counter such attacks the military is developing portable analyzers that can identify and quantify potential chemical agents in water supplies at microgram per liter concentrations within 10 minutes To aid this effort we have been investigating the value of a surface-enhanced Raman spectroscopy based portable analyzer In particular we have been developing silver-doped sol-gels to generate SER spectra of chemical agents and their hydrolysis products Here we present SER spectra of several chemical agents measured in a generic tap water Repeat measurements were performed to establish statistical error associated with SERS obtained using the sol-gel coated vials Keywords Chemical agents chemical agent detection SERS Raman spectroscopy
1 INTRODUCTION
In the autumn of 2001 terrorism within US borders became a sobering reality While extensive efforts are being implemented to secure the homeland US and Coalition forces in Afghanistan and Iraq are constantly faced with terrorist attacks In addition to car-bombings the military has identified several non-traditional attack scenarios including poisoning of water supplies by chemical warfare agents (CWAs) To counter this threat the Department of Defense has been investigating numerous approaches to detect such attacks under the auspices of the Joint Service Agent Water Monitor (JSAWM) program1 The JSAWM must be capable of identifying the chemical agents with no false-positives and quantifying the agents at microgram per liter concentrations within 10 minutes2 This includes the analysis of drinking water supplies distribution and storage systems Currently colorimetric paper is used to detect agents on-site while gas chromatography combined with mass spectrometry (GCMS) is used in mobile support laboratories However both methods have severe drawbacks The paper changes color in response to contact with many chemicals besides CWAs causing a high incidence of false positives3 while GCMS although very chemically specific requires up to an hour to perform and regular re-calibration456 Vibrational spectroscopy has also been investigated as a potential method of rapidly detecting CWAs7-
11 as early as 1985 in the case of infrared spectroscopy8 and 1988 in the case of Raman spectroscopy11 Again however these techniques also have limitations when it comes to measuring trace poisons in water Infrared spectra would be dominated by the very strong absorption of water which would obscure absorptions by most other chemicals present Whereas Raman spectroscopy is simply not a very sensitive technique and detection limits are typically grams per liter Surface-enhanced Raman spectroscopy (SERS) offers several advantages over conventional vibrational methods and that may provide the necessary sensitivity required for detecting trace quantities of chemical agents in water SERS employs the interaction of surface plasmon modes of metal particles with target analytes to increase scattering efficiency by as much as 1 million times12 In 1987 the potential of this approach to measure CWAs was demonstrated by measuring a series of pesticides13 Several of these organophosphonates have chemical structures similar to CWAs in particular P=O functional groups In the past few years we and others have further explored the ability of SERS to detect CWAs14-17 and even bioagents 18-21 We have been employing silver-doped sol-gels to promote the SER effect
SPIE-2003-5269 17
in these studies The porous silica network of the sol-gel matrix offers a unique environment for immobilizing and stabilizing SER-active metal particles22-25 The silver-doped sol-gels have been coated on the internal walls of standard glass vials to allow rapid SER analysis of small sample quantities without preparation We have measured over 100 chemicals with enhancements of 104 to 106 demonstrated reversible measurements in a flowing system reproducible measurements from vial-to-vial and measurements in multiple solvents including water21-25 Previously we used these vials to perform preliminary measurements of cyanide (CN) and methylphosphonic acid (MPA) Most of the nerve agents form MPA during hydrolysis while Tabun forms CN a chemical agent in its own right In another paper including in these proceedings we examined the limits of detection (LOD) for MPA by measuring a series of concentrations down to 50 mgL and estimated a limit of detection of 100 microgL26 These measurements provide encouragement in that SERS may satisfy the needs of the JSAWM To further establish the viability of SERS in particular silver-doped sol-gels here we present analysis of cyanide mustard and VX in tap water The measurements performed at the US Armyrsquos Edgewood Chemical Biological Center Aberdeen MD also included numerous repeat measurements to establish reproducibility
2 EXPERIMENTAL 2a General All chemicals including potassium cyanide 2-chloroethylethyl sulfide and those used to prepare the silver-doped and gold-doped sol-gel coated vials were obtained and used as received from Sigma-Aldrich All samples were prepared in a chemical hood using HPLC grade water unless otherwise noted (Fischer Scientific Fair Lawn NJ) for SERS measurements Once prepared the samples were transferred into the silver-doped sol-gel vials (Real-Time Analyzers Inc East Hartford CT) The vials were coated in a manner similar to that previously reported by adding ammonium hydroxide to a solution of silver nitrate tetramethyl orthosilicate (TMOS) and methanol22 The two precursor solutions were prepared mixed and transferred to 2-mL glass vials dried and heated After sol-gel formation the incorporated metal ions were reduced with dilute sodium borohydride (1mgmL) which is followed by a water wash to remove residual reducing agent After the resultant analyte solution was introduced the SER-active vial was fixed horizontally to an XY positioning stage (Conix Research Springfield OR) just inside the focal point of an f07 aspheric lens The lens focused the beam into the sample and collected the scattered radiation back along the same axis A dichroic filter (Omega Optical Brattleborough VT) was used to reflect the excitation laser to the lens and pass the Raman scattered radiation collected by the lens An f2 achromat was used to collimate the laser beam exiting a 200 microm core diameter source fiber optic while a second f2 achromat was used to focus the scattered radiation into a 365 microm fiber optic (Spectran Avon CT) A short pass filter was placed in the excitation beam path to block the silicon Raman scattering generated in the source fiber from reflecting off sampling optics and reaching the detector A long pass filter was placed in the collection beam path to block the sample Rayleigh scattering from reaching the detector A 785 nm diode laser (Process Instruments Inc model 785-600 Salt Lake City UT) was used to deliver 100 to 150 mW of power to the sample A Fourier transform Raman spectrometer (Real-Time Analyzers model IRA-785 East Hartford CT) and a silicon photo-avalanche detector (Perkin Elmer model C30902S Stamford CT) were used to acquire the SER spectra
2b Edgewood Chemical Biological Center The surface-enhanced Raman spectral measurements at the US Armyrsquos Chemical Biological center presented here were all performed on September 12 2003 To expedite measurements a plate was machined to hold up to 12 SER-active sample vials (Figure 1) The plate fit a standard XY plate reader that could be programmed Pure KCN bis-(2-chloroethyl)sulfide (distilled mustard gas HD) and ethyl S-2-diisopropylamino ethyl methylphosphonothioate (VX) were obtained on-site and used to prepare 1 gL tap water solutions in a chemical hood with appropriate safety equipment Simulated tap water was prepared by adding 10 chemicals most often found in tap water at appropriate concentrations to distilled water (Table 1) SER measurements were also performed in a chemical hood For added safety the FT-Raman instrument was placed outside the laboratory and 30 foot fiber optic and electrical cables were used to allow remote SERS measurements and plate manipulation For each experiment 1gL samples were prepared and added to 9 individual vials which were then loaded on the plate In some cases a tenth vial was included as a blank
Table 1 Generic Tap Water Composition
Compound mgL NaHCO3 100 CaSO4 27 MgSO4bull7H2O 67 NaNO3 1 Fulvic Acid 1 K2HPO4 07 KH2PO4 03 (NH4)2HSO4 001 NaCl 001 FeSO4 0001 pH 76-78
SPIE-2003-5269 18
A software program was written that allowed selecting the sequence that the vials were measured the number of positions along the length of the vials to measure (1 to 5) and the number of scans to co-add During sample analysis the program displayed the vial being analyzed the point being analyzed and the spectrum as it was being acquired Once all the data was collected a second software program was written to rapidly analyze the data The spectra collected for all the vials on a plate could be loaded at one time and then the spectra for each point could be displayed simultaneously or separately The user could then select the Raman peak to analyze in terms of peak height or area This was accomplished by selecting points on either side of the peak to define a baseline of zero The peak height or area could then be computed for all of the spectra loaded and then exported to a spreadsheet for statistical analysis
Figure 1 A) Vial Holder 6 slots to hold 2 vials each end-to-end B) Measurement Configuration Program user selects vials to measure sequence number of points per vial (1 to 5) and number of scans per point C) Spectral Acquisition Program shows spectrum being collected which vial and position D) Spectra Analysis Program user selects spectra to analyze by plate vial and point (s) as well as two wavenumbers defining the peak and the baseline to subtract The image is of 5 repeat measurements of 10 mgL KCN in generic tap water 16 sec each 100 mW of 785 nm
3 RESULTS AND DISCUSSION Raman and surface-enhanced Raman spectra were obtained for potassium cyanide bis-(2-chloroethyl)sulfide and ethyl S-2-diisopropylamino ethyl methylphosphonothioate representing three classes of chemical agents cyanides mustards and nerve agents respectively Spectra were also obtained for 2-chloroethyl ethyl sulfide (CEES) a structural analogue to HD which was included in the study to aid in assigning spectral bands KCN salt was used for cyanide experiments to avoid the increased hazards of handling HCN gas KCN completely dissolves in water forming its conjugate acid HCN according to its Ka of 615x10-1027 and at a concentration of 1 mgmL results in a pH 107 solution This is significant in that only CN- appears to interact sufficiently with silver to produce a SER spectrum and no spectral signal is observed below pH 726 Figure 2 shows the SER and normal Raman spectra for KCN The SERS spectrum of 1mgml KCN in water shows a single intense somewhat broad feature at
A B
C D
SPIE-2003-5269 19
2100 cm-1 assigned to the single CequivN stretch The band is much sharper in the normal Raman spectra of the solid KCN salt at 2074 cm-1 This band does broaden and shift to 2080 cm-1 in solution (not shown) However the observed SERS frequency is attributed to interaction with silver and low frequency mode at 135 cm-1 attributed to a Ag-CN stretch (not shown) supports this conclusion
cm-1 band to a C-S stretch but the authors concede that it is in fact more likely a C-Cl stretch28 It appears that the most intense Raman bands at 648 692 and 747 cm-1 shift to 620 660 and 720 cm-1 in the SER spectra and are tentatively assigned as above The width of these bands suggests that they overlap underlying spectral features Additional bands in the Raman spectra occur at 972 1034 1049 1263 1286 1423 1442 2865 2935 and 2960 cm-1 Corresponding bands occur in the SER spectra at 964 1015 1054 1286 1410 1447 2865 and 2935 cm-1 Most of these bands are associated with alkane modes specifically the bands at approximately 1040 cm-1 to a C-C stretch 1290 cm-1 to a CH2 in-phase twist 1440 cm-1 to a CH2 wag 2865 cm-1 to a symmetric CH2 stretch and 2965 cm-1 to an asymmetric CH2 stretch The Raman and SER spectra of sulfur mustard were measured at the Edgewood center (Figure 4) Both spectra are largely similar to CEES The C-Cl and C-S bands in the Raman spectrum of HD now occur at 640 655 700 739 and 760 cm-1 and are more resolved possibly due to the increased molecular symmetry Theoretical calculations indicate that the first three bands are due to C-Cl stretching modes and the latter two to C-S stretching modes28 Only the C-Cl bands maintain significant intensity in the SER spectra occurring at 624 and 643 cm-1 which is attributed to the
Prior to measurements of HD CEES was examined by Raman and SER spectroscopy (Figure 3) CEES also known as half-mustard is essentially identical to HD except one of the chlorine end atoms is replaced by a hydrogen atom Again although not as toxic as HD CEES is a blister agent and dilute aqueous samples were prepared using appropriate safety equipment Both the Raman and SER spectra of CEES are similar and dominated by bands between 600 and 800 cm-1 These are associated with C-Cl and C-S stretching modes which are tentatively assigned to 648 and 747 cm-1 in the Raman spectra respectively The shoulder at 630 cm-1 the overlapped band at 660 cm-1 and the strong band at 692 cm-1 could also be due to these modes or their asymmetric counterparts It is worth noting that theoretical calculations assign the 692
Figure 2 A) SER and B) NR spectra of KCN Conditions A) 1 mgml in tap water 100 mW of 785 nm at sample 1-min acquisition time B) solid 300 mW of 785 nm 5-min All spectra are 8 cm-1 resolution
A
B
Figure 3 A) SER and B) NR spectra of CEES Conditions A) 1 vv (10 mgml) in MeOH 100 mW of 785 nm 1-min acquisition time B) neat 300 mW of 785 nm 5-min
A
B
Cl-CH2-CH2-S-CH2-CH3
A
B
Figure 4 A) SER and B) NR of HD Conditions A) 1mgml in tap water B) pure both 100 mW of 785 nm 1-min
Cl-CH2-CH2-S-CH2-CH2-Cl
SPIE-2003-5269 20
expected strong interaction between chlorine and silver and adds support to the assignment of this band to a C-Cl stretch Weaker overlapping bands occur at 670 692 and 724 cm-1 the latter possibly due to C-S stretching modes Again the alkane modes are apparent in the normal Raman spectra of HD but only a broad feature at 1300 to 1450 cm-1 suggests CH2 contributions in the SER spectrum Although the observed bands in the VX spectrum have not been assigned (Figure 4) a computer generated Raman spectrum29 predicts many of the same features with surprising accuracy and are used here Two intense bands at 460 and 530 cm-1 closely match predicted bands at 463 and 546 cm-1 assigned to a CH3-P=O bend and a PO2CS wag Three highly overlapped bands occur at 694 745 and 771 cm-1 matching predicted bands at 713 730 and 760 cm-1 The first
Table 2 Measured SER peak heights for the CN stretch at 2100 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
Figure 5 A) SERS and B) NR spectra of VX Conditions A) 1 vv (10 mgml) in MeOH B) pure sample both 100 mW of 785 nm 1-min acquisition time
A
B
two have been assigned to a C-S stretch and CH2bend respectively while the latter has been attributed to either a P-C stretch or an O-C-C stretch Although the 745 cm-1 band may alternatively be assigned to a C-S stretch based on the previous measurements of CEES and HD The relatively intense bands at 890 1106 1218 1445 and 1465 cm-1 also match predicted bands at 880 1108 1216 1440 and 1464 cm-1 that are assigned to a C-C stretch CH3 rock N-C3 stretch various C-H3 bends and C-H bends respectively Both the computer generated and the measured spectra contain numerous other less intense bands One is worth mentioning A unique band appears at 370 cm-1 that is predicted at 368 cm-1 and corresponds to an O-P=O bend The surface-enhanced Raman spectrum of VX is also rich with spectral features It has the unique low frequency band at 370 cm-1 as well as a second band at 380 cm-1 that is assigned tothe S-P-O bend predicted in the normal Raman spectrum at 388 cm-1 Based on the measured and predicted normal Raman spectra the following SERS assignments are given 460 cm-1 to the CH3-P=O bend 544 cm-1 to the PO2CS wag 738 cm-1 to a C-S stretch (based on arguments above) 890 cm-1
to a C-C stretch 1101 cm-1 to a CH3 rock and 1456
cm-1 to a C-H bend The ability of SERS to measure chemical agents in water containing real-world chemical interferents was tested by using the generic tap water described in Table 1 The ability to reproduce measurements was accomplished by preparing three separate water stock solutions which were used to prepare three sample solutions each of 1 mgmL KCN in the generic tap water The samples defined as vials 1-9 were then measured at 4 points per vial in a semi-random fashion such that errors associated with stock solution preparation and errors associated with instrument drift could be identified No trends were apparent that signified such systematic errors Each spectrum collected consisted of 20 averaged scans taking 16 seconds at 8 cm-1 resolution The laser power at the sample was measured periodically during the day and it ranged from 102 to 105 mW spacccc
spot Vial 1 Vial 2 Vial 41 13754 13004 128192 13519 12692 12909
1 3 13541 12721 126394 13462 12648 12651
ave 13569 12766 12755Vial 6 Vial 3 Vial 5
1 11586 14068 10757 2 11236 14402 115122 3 11376 14546 11559
4 10894 11763 11214ave 11273 13695 11261
Vial 7 Vial8 Vial 91 11198 15804 11023
2 11243 15716 88753 3 11562 15216 11218
4 11655 15084 11077ave 11415 15455 10548
CN AVG STDEV ERR1 mgmL pk ht 12526 1560 1245
stock solution
SPIE-2003-5269 21
The CN measurements consisting of 36 data points produced an average height of 125 for the 2100 cm-1 peak with a standard deviation of 156 or 125 (Table 2) The HD and VX measurements were performed precisely the same way (Tables 3 and 4) For HD the 624 cm-1 peak was used for analysis and it had an average height of 53 with a standard deviation of 068 or 129 while for VX the 544 cm-1 peak was used for analysis and it had an average height of 1051 with a standard deviation of 308 or 293 The greater error in the VX measurements can be somewhat attributed to Vial 4 which produced lower SER signal intensities But removing this vial from the data set changes the standard deviation to 233 only a modest improvement
spot Vial 1 Vial 2 Vial 41 525 609 4682 45 675 484
1 3 527 644 5094 559 807 669
ave 51525 68375 5325Vial 6 Vial 3 Vial 5
1 521 574 575 2 536 53 4492 3 509 508 451
4 565 594 379ave 53275 5515 4635
Vial 7 Vial 8 Vial 91 457 589 413
2 583 497 4273 3 544 505 4
4 528 5 509ave 528 52275 43725
HD AVG STDEV ERR1 mgmL pk ht 530 068 1291
stock solution
spot Vial 1 Vial 2 Vial 41 1464 1034 4672 1485 99 61
1 3 1491 989 5684 1041 777 553
ave 137025 9475 5495Vial 6 Vial 3 Vial 5
1 1058 942 1293 2 697 121 965
2 3 727 1206 11124 689 1254 746
ave 79275 1153 1029Vial 7 Vial 8 Vial 9
1 1127 783 163 2 1358 812 1614
3 3 1371 908 15124 1097 875 1397
ave 123825 8445 153825
VX AVG STDEV ERR1 mgmL pk ht 1051 308 2925
stock solution
4 CONCLUSIONS In this paper we examined the ability of surface-enhanced Raman spectroscopy to reproducibly measure CN HD and VX in tap water without chemical interference Both normal and surface-enhanced Raman spectra were examined to select unique bands suitable to identify and quantify these chemical agents For SER measurements the 2100 cm-1 C-N stretch was used for CN the 624 cm-1 C-Cl stretch was used for HD and the 544 cm-1 PO2CS wag was used for VX It was determined that 1 mgmL samples of each of these chemicals measured 36 times in glass vials coated with a silver-doped sol-gel reproduced measurements with standard deviations of 125 129 and 293 It was further found that the 10 chemicals added to simulate generic tap water did not interfere with or alter the SER spectra It should be noted that the concentrations used in this study were considerably greater than those required by the JSAWM program Current work involves improving SER sensitivity and designing sampling systems with better reproducibility This includes the development of fractal silver and gold structures within the sol-gel matrix and the development of chemically selective sol-gels
ACKNOWLEDGMENTS The authors are grateful for the support of the US Army (DAAD13-02-C-0015 Joint Service Agent Water Monitor program) The authors would also like to thank Janet Jensen Ronald Crosier and Kristina Gonser for helpful discussions
Table 4 Measured SER peak heights for the VX band at 544 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
Table 3 Measured SER peak heights for the HD band at 624 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
SPIE-2003-5269 22
REFERENCES 1 Jensen JL NC Wong W R Loerop SPIE 4775-03 (2001) 2 JSAWM Requirements at wwwsbccomapgeaarmymilRDAecbcrtPRODSERJSAWMjsawmhtml 3 Erickson B ldquoThe Chemical Weapons Convention Redefines Analytical Challengerdquo Anal Chem 70 397A (1998) 4 Johnston RL Hoefler CM Fargo JC and Moberley B ldquoThe Defense Nuclear Agencyrsquos
ChemicalBiochemical Weapons Agreements Verification Technology Research Development Test and Evaluation Program and its Requirements for On-Site Analysisrdquo AT-ONSITE 5-8 (1994)
5 Black RM Clarke RJ Read RW and Reid MT ldquoApplication of GC-MS and GC-tandem MS to the analysis of chemical warfare samples found to contain residues of the nerve agent sarin sulphur mustard and their degradation productsrdquo J Chrom 662 301-321 (1994)
6 USA Army Center for Health Promotion and Preventive Medicine (USACHPPM) Technical Guideline 230A 7 Hoffland LD Piffath RJ Bouck JB rdquoSpectral signatures of chemical agents and simulantsrdquo Optical
Engineering 24 982-984 (1985) 8 Braue EHJ Pannella MGrdquoCIRCLE CELL FT-IR Analysis of Chemical Warfare Agents in Aqueous Solutionsrdquo
App Spectrosc 44 1513-1520 (1990) 9 Tseng C-H Mann CK and Vickers TJ ldquoDetermination of Organics on Metal Surfaces by Raman
Spectroscopyrdquo App Spectrosc 47 1767-1771 (1993) 10 Staff Reporter ldquoUS Army Tests Lidar to Detect Biological Toxinsrdquo Photonic Spectra pg 50 December 1998 11 Christesen SD Raman cross sections of chemical agents and simulants App Spectrosc 42 318-321 (1988) 12 Weaver MJ Farquharson S Tadayyoni MA ldquoSurface-enhancement factors for Raman scattering at silver electrodesrdquo J Chem Phys 82 4867-4874 (1985) 13 Alak AM and Vo-Dinh T ldquoSERS of Organophosphorous Chemical Agentsrdquo Anal Chem 59 2149-2153 (1987) 14 Lee Y Farquharson S ldquoRapid chemical agent identification by SERSrdquo SPIE 4378 21-26 (2001) 15 Farquharson S Maksymiuk P Ong K Christesen S ldquoChemical agent identification by surface-enhanced Raman spectroscopyrdquo SPIE 4577 166-173 (2001) 16 Spencer KM Sylvia J Clauson S and Janni J ldquoSurface Enhanced Raman as a Water Monitor for Warfare
Agents in Waterrdquo SPIE 4577 158-165 (2001) 17 Tessier P Christesen S Ong K Clemente E Lenhoff A Kaler E Velev O ldquoOn-line spectroscopic characterization of sodium cyanide with nanostructured gold surface-enhanced Raman spectroscopy substratesrdquo
App Spectrosc 56 1524-1530 (2002) 18 Farquharson S W W Smith S Elliott and J F Sperry Rapid biological agent identification by surface- enhanced Raman spectroscopy SPIE 3855110-116 (1999) 19 Farquharson S WW Smith YH Lee S Elliott and J F Sperry Detection of biological signatures A
comparison of electrolytic and metal-doped sol-gel surface-enhanced Raman media SPIE 4575 62-72 (2002) 20 Guzelian AA Sylvia J Janni J Clauson S and Spencer KM ldquoSERS of whole cell bacteria and trace levels of biological moleculesrdquo SPIE 4577 182-192 (2001) 21 Shende C Inscore F Gift A Maksymiuk P Farquharson S ldquoRapid dipicolinic acid extraction from Bacillus spores detected by surface-enhanced Raman spectroscopyrdquo App Spectrosc 58 accepted 22 Lee Y and Farquharson S ldquoSERS Sample Vials Based on Sol-Gel Process for Trace Pesticide Analysisrdquo SPIE
4206 140-146 (2000) 23 Farquharson S and Lee Y ldquoTrace Drug Analysis by SERSrdquo SPIE 4200-16 (2000) 24 Lee Y Farquharson S and Rainey P M Surface-Enhanced Raman Sensor for Trace Chemical Detection in
Water SPIE 3857 76-84 (1999) 25 Lee Y Farquharson S Kwong H and Shahriari M ldquoSol-Gel Chemical Sensor for Surface-Enhanced Raman
Spectroscopyrdquo SPIE 3537 252-260 (1998) 26 Farquharson S A Gift P Maksymiuk F Inscore and W Smith ldquopH dependence of methyl phosphonic acid
dipicolinic acid and cyanide by surface-enhanced Raman spectroscopyrdquo SPIE 5269-19 (2003) 27 Lide DR Ed Handbook of Chemistry and Physics CRC Press 77th Ed (1996-7) 28 Sosa C RJ Bartlett K KuBulat and WB Person ldquoA theoretical study of harmonic vibrational frequencies and
infrared intensities of XCH2CH2SCH2CH2X and XCH2CH2SH (= H Cl)rdquo J Phys Chem 93 577-588 (1993) 29 Hameka HF and JO Jenson ldquoComputer-generated predictions of the structure and the IR and Raman spectra of
VXrdquo ERDEC-TR-065 May 1993
- Chemagents Appendicespdf
-
- SPIE2001-4575-Bioagent-SERSapdf
-
- Detection of bioagent signatures A comparison of electrolytic and metal-doped sol-gel surface-enhanced Raman media
- Stuart Farquharson Wayne Smith and Yuan Lee
-
- Real-Time Analyzers 87 Church Street East Hartford CT 06108
-
- ABSTRACT
- 4 CONCLUSIONS
- 5 ACKNOWLEDGEMENTS
- 6 REFERENCES
-
168
Nevertheless MPA was readily observed for a 10 ppm sample with an estimated detection limit of 04 ppm (based on a signal-to-noise ratio of 3 for the 760 cm-1 band intensity) This measurement provides encouragement in that SERS may satisfy the needs of a JSAWM Furthermore MPA is also a hydrolysis product of VX and V-gas and EA2192 (Figure 2) and may prove a valuable indicator of agent usage Figure 2 Hydrolysis of Sarin to form hydrofluoric acid (HF) methylphosphonic acid 1-methylethyl ester (MPAMME) methyl phosphonic acid (MPA) and 2-propanol With this initial albeit modest success we began analyzing chemicals with various sol-gel compositions that we have been developing Here we describe four sol-gel compositions that select for 1) polar-positive 2) polar-negative 3) weakly polar-positive and 4) weakly polar-negative chemical species The ability of these sol-gels to select and enhance Raman scattering is described for several test chemicals and MPA
2 EXPERIMENTAL The chemicals analyzed as well as all chemicals used to prepare the metal-doped sol-gels were obtained at their purest commercially available grade from Aldrich (Milwaukee WI) The sol-gel designed to select for polar-negative species was prepared from a silver amine complex tetramethyl orthosilicate (TMOS) and methanol After mixing 02 mL of the sol-gel solution was transferred into a glass vial (2 mL) dried and heated The incorporated silver ions were then reduced using dilute sodium borohydride The vials were washed and dried prior to the addition of a sample solution In a similar manner the sol-gel designed to select for polar-positive species was prepared from a gold salt TMOS and methanol The sol-gel designed to select for weakly polar-negative species was prepared from a silver amine complex tetraethyl orthosilicate (TEOS) and methanol And the last sol-gel designed to select for weakly polar-positive species was prepared from a gold salt TEOS and methanol All samples were prepared in a chemical hood and transferred into plain or SER-active vials for analysis Normal Raman spectral measurements employed 1-mL pure samples that were placed in a 1-cm3 cuvette and weighed This yielded a powder density that allowed accurate calculation of molecules in the optical collection field SERS measurements employed 1-mg sample per mL water concentrations unless otherwise stated Once prepared a 01 mL sample was placed into one of the four selective sample vials which in turn was placed into the sample compartment of a Raman spectrometer for analysis A prototype Fourier transform Raman spectrometer (RTA) was used for these measurements18 The system consisted of a NdYAG laser (Brimrose) for excitation at 1064 nm an interferometer built by On-Line Technologies (OLT East Hartford CT) for frequency separation an uncooled InGaAs detector for signal detection (RTA) and an Intel 400 MHz Pentium II based laptop computer (Dell Round Rock TX) for interferometric control data acquisition (OLT) and analysis (LabVIEW by National Instruments Austin TX) Additional components included a Notch filter (Kaiser Ann Arbor MI) and interferometer entrance and exit optics (Edmund Scientific Barrington NJ) Fiber optics were used to deliver the excitation beam to the sample and the scattered radiation to the interferometer (1 meter lengths of 200 and 365 micron core diameter respectively Spectran Avon CT) A second Notch filter (Kaiser) was used as a beam splitter to direct the excitation beam along the same axis as the collected radiation A microscope object (20x04 Newport Irvine CA) was used to focus the beam into the sample and to collect the scattered radiation back along the same axis In this co-axial backscattering arrangement the excitation beam was passed through the outside of a glass vial and focused onto the silver-doped sol-gel film (01-03 mm thickness) containing the sample
3 RESULTS AND DISCUSSION p-aminobenzoic acid (PABA) and phenyl acetylene (PA) and were used to refine the selectivity and SER-activity of the four different metal-doped sol-gels PABA is a popular chemical used to evaluate the performance of SER-active media Here the polar end groups can be used to test selectivity of the polar-negative and polar-positive sol-gels PA is essentially non-
2O+ H HF + +OH
OH3H C
OP
OF
CH
CH3
3
3H C
OCP
OOH
CH
CH3
3
3H C
O
CPHO
CH
CH
3
3
C
Sarin MPAMME MPA 2-propanol
Proc SPIE Vol 4577
169
polar but a high electron density in the cylindrical π cloud around the carbon-carbon triple bond allows testing the selectivity of the weakly polar-negative and weakly polar-positive sol-gels As Figure 3 illustrates PABA passes through the polar sol-gel and is enhanced by either the silver or gold particles At 1 mgml the concentration of neutral PABA is ca 20 times that of the ionized form (pKa = 48) For electropositive silver the PABA anion is expected to interact through the carboxylate group and the associated vibrational modes are expected to dominate the spectrum Conversely for electronegative gold either form of PABA is expected to interact through the amine group The clear differences in our spectra support this expectation Furthermore bands at 840 and 1405 cm-1 assigned to a COO- bend and stretch respectively are significantly more intense for silver than gold Additional bands at 1140 and 1195 cm-1 are assigned to CH bending modes while bands at 1450 1500 and 1605 cm-1 are assigned to ring vibrational modes A very similar SER spectrum for PABA on a silver-coated alumina substrate has previously been reported with similar assignments19 For the gold-doped sol-gel new bands appear at 690 1355 and 1585 cm-1 The first band is assigned to a ring-H bending mode the second band to a ring-N- stretching mode and the third band to a possible NH2 scissors mode or ring mode The second band is not observed in the normal Raman spectrum but infrared bands occur at this frequency for aromatic ring-secondary amine stretching modes The scissors mode occurs at this frequency in Raman spectra for several chemicals but is absent in the PABA Raman spectrum Alternatively this mode may be the1600 cm-1 ring mode that has been shifted by the gold interaction Again a very similar SER spectrum of PABA has been reported but surprisingly using silver (colloids)2021 not gold as the enhancement medium These researchers also assumed the primary interaction of PABA with silver was through the carboxylate anion and made assignments accordingly For example they assigned the 1359 cm-1 to a COO- stretch not to the amine group as we have They also favor the ring stretching mode assignment for the 1582 cm-1 band Finally it should be said that other researchers have argued that the most dominant band in the SER spectra at 1450 cm-1 a ring vibration mode suggests that PABA lies flat on the surface and the π-orbitals dominate the surface interaction22
Figure 3 SER spectra of A) PABA using polar-negative and B) polar-positive sol-gels and C) PA using weakly polar-negative and D) weakly polar-positive sol-gels PABA is 1 mgmL PA is 1 vv Spectral conditions 75 mw 1064 nm 100 scans (15 min) 8 cm-1 resolution Non-polar PA passes through the non-polar sol-gels and is also enhanced by both metals The spectra are easily understood For electropositive silver PA interacts through the cylindrical triple bond π electron cloud and a -CequivC- doublet occurs near 2000 cm-1 The interaction is reasonably strong since this band appears at 2112 cm-1 in the normal Raman spectrum For electronegative gold this interaction is unlikely and only very weak bands occur near 2000 cm-1 The remaining bands are at 1000 cm-1 1200 cm-1 doublet and 1595 cm-1 all appear in the normal Raman spectra at virtually the same frequencies and are assigned to the symmetric ring-breathing mode CH bending modes and the trigonal ring-breathing mode respectively The polarnon-polar selectivity of the polar-negative and weakly polar-negative sol-gels was tested by adding a 11 molar mixture of PABA and PA The selective enhancement is quite good (Figure 4) The spectrum obtained using the polar sol-gel represents 78 PABA and 22 PA while the spectrum obtained using the weakly polar sol-gel represents 9 PABA and 91 PA The band peak intensities at 2000 cm-1 for PA and 1450 cm-1 for PABA were used for these calculations and are expanded in Figure 4 for clarity
C A
Wavenumber (∆cm-1) Wavenumber (∆cm-1)
2NH COOH
D B
C CH
Proc SPIE Vol 4577
170
Figure 4 SERS of 11 MM of PABA and PA in A) polar-negative and C) weakly polar-negative sol-gels The lower traces compare the pure chemicals B) 1 mgml PABA in polar-negative sol-gel and D) 1 PA in weakly polar-negative sol-gel while the insets magnify the minority species for clarity (x5 in A and x10 in B) Spectral conditions as in Figure 3 Following this development of selective sol-gels that maintained SER activity we measured cyanide and MPA (Figure 5) Not surprisingly the best sensitivity for both hydrolysis products was obtained using the polar-negative sol-gel The interaction of the cyanide anion with the silver surface is sufficient to shift the CequivN stretch observed at 2080 cm-1 in the normal Raman spectrum to 2145 cm-1 in the surface-enhanced Raman spectrum Furthermore the band is substantially broadened This anion has been extensively studied by electrolytic SERS and this shift and broadening have been attributed to the formation of a tetrahedral Ag(CN)3
2- surface structure23 Figure 5 Surface-enhanced (upper traces) and normal Raman spectra (bottom traces) of A) CN- and B) MPA in silver-doped TMOS SERS conditions as in Figure 3 and 1 mgmL Note MPA yields two distinct spectra for neutral (top) and acidic pH (middle) The normal Raman spectra employed pure powders 500 scans and 900 mW of 1064 nm SER measurements of MPA with the polar-negative sol-gel yielded two unique spectral signatures that depended on solution pH (Figure 5) For more neutral solutions the P-C stretch of MPA at 762 cm-1 dominates and the CH2 stretch at 2922 cm-1 appears The SN is sufficiently high that the anti-Stokes Raman shift at -762 cm-1 is observed For deprotonated MPA an oxygen-surface mode appears at 325 cm-1 (as well as its anti-Stokes complement) suggesting a strong interaction This results in substantial enhancement of the P-O-C mode at 1051 cm-1 (upper trace) Others report that this mode dominates the infrared spectra of nerve agents measured in water8 Comparison of the two spectra suggests the following molecule-to-surface orientations The appearance of the oxygen-surface and P-O-C modes in the upper spectrum of Figure 5B indicates that the tetrahedral molecule interacts with the silver surface through the deprotonated oxygen and is oriented end-on The
A B
Wavenumber (∆cm-1)
C A
Wavenumber (∆cm-1) Wavenumber (∆cm-1)
D B
Wavenumber (∆cm-1)
Proc SPIE Vol 4577
171
dominance of the P-C and the CH2 stretches and the disappearance of the P-O-C mode in the upper spectrum suggest the molecule is oriented with the methyl-hydroxyl-hydroxyl face towards the surface However considerably more research must be performed to verify these points Both chemical forms were measured as a function of concentration to determine the onset of monolayer coverage and estimate expected detection limits (Figure 6) Below monolayer coverage the signal to concentration dependence should be linear and the SN of any spectral measurement in this range can be used to predict the detection limit In the spectra presented here the peak height was used as the signal while the noise as root-mean-squared (RMS) was measured between 4400-4600 cm-1 Since noise is distributed evenly throughout the spectrum when transformed this region was used since it does not have any contributions from signals or baseline offsets Figure 6 shows a series of spectra for MPA along with a plot of the 1050 cm-1 band intensity (the noise was identical for this series) as a function of sample concentration A clear discontinuity appears in the vicinity of 01 mgmL (19 ppm) indicating the onset of monolayer coverage A detection limit defined as a SN of 3 was calculated for the 01 and 005gmL samples at 24x10-4 and 25x10-4 gL respectively A more modest detection limit of 101x10-4 gL was obtained using the 760 cm-1 band in the second series of concentration measurements These detection limits correspond to 46 and 190 ppb for a 3-min scan and 75 mW of incident laser power
Figure 6 A) Concentration dependence of MPA SERS measured in silver-doped TMOS) B) Concentrations are 001 005 01 05 1 gL (188 94 188 94 188 ppm) I760 series (bull) and I1050 series (∆)
Enhancement factors for cyanide and methyl phosphonic acid can be determined by estimating the number of molecules contributing to the surface-enhanced and normal Raman spectra The enhancement factor EF can be defined by the following equation
EF = (ISERSINR)bull(MNRMSERS) bull(PNRPSERS) bull(TNRTSERS)12
where I is the spectral band intensity M is the sample mass P is the incident laser power and T is the measurement time (or number of scans) for the two measurements For the normal Raman spectra a cylindrical scattering volume is assumed based on the laser area (28x10-7m2 6x10-4m diameter spot) and the penetration depth (1x10-3 m)24 The density of KCN and MPA as powders were measured at 0572 and 0516 gcm3 indicating that 16x10-4 and 144x10-4 g produced the normal Raman signals in Figure 5 respectively The number of molecules contributing to the SER spectra are those on the silver particles embedded in the sol-gel The total silver surface area can be determined from the average particle size concentration and the scattering volume Previous scanning electron micrographs showed the average silver particle size to be 40 nm in diameter (335x10-23m3)17 The silver concentration is 012M based on the reactant molar concentrations and dilution factors And the scattering volume is 76x10-
11m3 again based on a cylindrical scattering volume defined by a laser area of 28x10-7m2 and a sol-gel thickness of 27x10-
4m This volume contains 123x10-6g of silver equivalent to 35x109 silver particles with a collective surface area of 18x10-
5m2 However it may reasonably be assumed that at least half of the silver surface is in contact with the sol-gel matrix and unavailable for analyte interaction If we assume monolayer coverage and that each CN molecule occupies 15x10-20m2 then approximately 62x1014 molecules or 27x10-8g of CN contribute to the SER spectrum (20x10-19m2 46x1013 molecules 74x10-9g for MPA) Accordingly the EF for cyanide equals 48x104 ((180599) bull(16x10-427x10-8) bull(90075) bull(500100)12) The EF for MPA is considerably higher at 87x106 ((60326) bull(144x10-474x10-9) bull(90075) bull(500200)12)
0
20
40
60
80
100
120
140
0 02 04 06 08 1 12[MPA] (mgmL)
I (76
0)
0
100
200
300
400
500
600
I (10
50)
Wavenumber (∆cm-1)
A B
Proc SPIE Vol 4577
172
4 CONCLUSIONS Here we present for the first time surface-enhanced Raman spectra of Tabun and Sarin obtained using silver-doped sol-gels However the inferior enhancement suggested employing SER-active sol-gels varying in composition to improve both detection limits and selectivity of the target analytes To this end we successfully demonstrated the capabilities of four sol-gels that select for 1) polar-positive 2) polar-negative 3) weakly polar-positive and 4) weakly polar-negative chemical species p-aminobenzoic acid was used to show that silver could be used to attract polar-negative chemicals or functional groups (carboxylate anion) and that gold could be used to attract polar-positive chemicals (amine) while a mixture of p-aminobenzoic acid and phenylacetylene was used to show that tetramethyl orthosilicate preferentially solvates polar chemicals and that tetraethyl orthosilicate preferentially solvates weakly polar chemicals This increased sample control was applied to cyanide and methyl phosphonic acid two hydrolysis products of chemical warfare agents Exceptional results were obtained for methyl phosphonic acid allowing measurement of 1x10-2 gL for a 3-min scan and 75 mW of incident laser power with an estimated detection limit of 245x10-4 gL and an enhancement factor of 87x106 However this detection limit is 76 times less sensitive than required for the JSAWM (32x10-6gL for the G-agents) Finally we note that the measurements performed here employed an FT-Raman spectrometer This instrumentation was chosen over dispersive Raman instrumentation for the high wavelength accuracy afforded by the HeNe reference laser (Connes Advantage25) which would allow reliable spectral subtraction matching of observed spectra to stored library spectra and confident use of chemometric approaches Such data analysis is likely to be required to identify the chemical agents as well as distinguish hydrolysis products However this instrumentation which employs 1064 nm excitation and InGaAs detection sacrifices sensitivity We believe that the measurement sensitivity can be improved by at least two orders of magnitude by using shorter laser excitation wavelengths (eg 532 nm) This would provide more efficient Raman scattering (fourth power dependence on laser excitation wavelength) more efficient generation of plasmon modes and allow using more efficient detector material (Si vs InGaAs) These modifications are underway
5 ACKNOWLEDGEMENTS The authors would like to thank Dr R Yin and J Jensen of the US Army for supporting this work (Contract Number DAAD13-01-C-0019) They would also like to thank Advanced Fuel Research for making their laboratory facilities available
6 REFERENCES 1 Tu Anthony ldquoOverview of Sarin Terrorist Incidents in Japan in 1994 and 1995rdquo 6th CBW Protection Symposium
Stockholm Sweden 10-15 May 1998 2 Jensen JL NC Wong W R Loerop SPIE 4775-03 (2001) 3 JSAWM Requirements at wwwsbccomapgeaarmymilRDAecbcrtPRODSERJSAWMjsawmhtml 4 ldquoThe Chemical Weapons Convention Redefines Analytical Challengerdquo Analytical Chemistry News amp Features June 1
397A (1998) 5 Johnston RL Hoefler CM Fargo JC and Moberley B ldquoThe Defense Nuclear Agencyrsquos ChemicalBiochemical
Weapons Agreements Verification Technology Research Development Test and Evaluation Program and its Requirements for On-Site Analysisrdquo AT-ONSITE 5-8 (1994)
6 Black RM Clarke RJ Read RW and Reid MT ldquoApplication of gas chromatography-mass spectrometry and gas chromatography-tandem mass spectrometry to the analysis of chemical warfare samples found to contain residues of the nerve agent sarin sulphur mustard and their degradation productsrdquo J Chromatography 662 301-321 (1994)
7 Hoffland LD Piffath RJ Bouck JBrdquoSpectral signatures of chemical agents and simulantsrdquo Optical Engineering 24 982-984 (1985)
8 Braue EHJ Pannella MGrdquoCIRCLE CELL FT-IR Analysis of Chemical Warfare Agents in Aqueous Solutionsrdquo Applied Spectroscopy 44 1513-1520 (1990)
9 Tseng C-H Mann CK and Vickers TJ ldquoDetermination of Organics on Metal Surfaces by Raman Spectroscopyrdquo Applied Spectroscopy 47 1767-1771 (1993)
10 Staff Reporter ldquoUS Army Tests Lidar to Detect Biological Toxinsrdquo Photonic Spectra pg 50 December 1998
Proc SPIE Vol 4577
Proc SPIE Vol 4577
173
11 Christesen SD Raman cross sections of chemical agents and simulants Applied Spectroscopy 42 318-321 (1988) 12 Alak AM and Vo-Dinh T ldquoSERS of Organophosphorous Chemical Agentsrdquo Analytical Chemistry 59 2149-2153
(1987) 13 Norrod KL Sudnik LM Rousell D and Rowlen KL ldquoQuantitative Comparison of Five SERS Substrates
Sensitivity and Detection Limitrdquo Applied Spectroscopy 51 994-1001 (1997) 14 Lee Y and Farquharson S ldquoSERS Sample Vials Based on Sol-Gel Process for Trace Pesticide Analysisrdquo SPIE 4206
140-146 (2000) 15 Farquharson S and Lee Y ldquoTrace Drug Analysis by Surface-Enhanced Raman Spectroscopyrdquo SPIE 4200-16 (2000) 16 Lee Y Farquharson S and Rainey P M Surface-Enhanced Raman Sensor for Trace Chemical Detection in Water
SPIE 3857 76-84 (1999) 17 Lee Y Farquharson S Kwong H and Shahriari M ldquoSurface-Enhanced Raman Sensor for Surface-Enhanced Raman
Spectroscopyrdquo SPIE 3537 252-260 (1998) 18 Farquharson S Smith W Carangelo R C and Brouillette C ldquoIndustrial Raman Providing Easy Immediate Cost
Effective Chemical Analysis Anywhererdquo SPIE 3859 14-23 (1999) 19 Narayanan VA JM Bello JD Stokes and T Vo-Dinh Analusis 19 307-310 (1991) 20 Laserna JJ E L Torres and JD Winefordner Analytica Chemica Acta 469-480 (1987) 21 Torres EL and JD Winefordner Analytical Chemistry 59 1626-1632 (1987) 22 Suh JS DP DiLella M Moskovits J Phys Chem 87 1540-1544 (1983) 23 Benner RE R Dornhaus R Chang and BL Laube Correlations in the Raman spectra of cyanide complexes adsorbed
at silver electrodes with voltammograms Surface Science 101 341 (1980) 24 Chase D B and JF Rabolt Fourier Transform Raman Spectroscopy Acad Press Ch1 p 131 (1994) 25 Connes J Rev Opt Theor Instrum 40 45 (1961)
APPLIED SPECTROSCOPY 351
focusing the 488 nm laser beam 2 mm deep into thebulk of the crystals to avoid contributions from potentialdepletion layers As shown in Fig 1b the dependence ofthe Raman shift on the degree of deuteration is almostperfectly linear and ts very well with D 5 22684cmR1 24526 where D is the degree of deuteration (in )and R is the spectral mean of the PO4 vibration in cm21A linear correlation coef cient of 0998 indicates an ex-cellent linear dependence of the Raman peak shift withdegree of deuteration This result shows that the shift ofthe PO 4 peak is simply caused by the linear increase inatomic mass due to isotope substitution which decreasesthe length of hydrogen-like bonds
This excellent linear dependence allows us to map thepro le of the DH exchange layer at the surface of DKDPcrystals by acquiring Raman spectra and determining theposition of the PO 4 peak for various depths This methodis preferable over other methods such as determining thestrength of the OD vibration directly (eg at 715 cm21)because the position of the most intense peak in the Ra-man spectrum can be measured more precisely than theintensity of some of the weakest peaks in the spectrumThis is demonstrated in Fig 2 where depth-dependentRaman spectra (Fig 2a) and the resulting exchange layerpro les for two DKDP crystals are shown (Fig 2b) Thespectra in Fig 2a were obtained from a depth scan of aDKDP crystal with 75 degree of deuteration in thebulk grown at 45 8C The spectra start out as DKDP with30 deuteration close to the surface and approach thebulk DKDP spectrum within a few micrometers of depthThe fact that the relative degree of deuteration does notextend to 0 D is due to the limited depth resolution ofthe Raman microprobe which averages over 4 mm indepth Figure 2b depicts the resulting DH exchange layerpro les for this and a second crystal grown at 63 8Crespectively Both crystals had the same exposure to am-bient conditions and their main difference is the temper-ature at which they were grown The different exchangelayer pro les indicate that crystals grown at differenttemperatures have differen t proton conductiv ities 13
which leads to a difference in their rate of deuteriumdepletion The parameters controlling this behavior arecurrently the objective of a detailed study the results ofwhich will be reported elsewhere
CONCLUSION
In conclusion we have shown that the shift of the to-tally symmetric PO 4 stretch mode in the Raman spectrumof DKDP crystals scales linearly with degree of deuter-ation This allows us to correlate Raman peak positionsto deuteration levels in these crystals We have presenteda new technique to determine DH diffusion pro les inDKDP frequency conversion crystals based on micro-Ra-man spectroscopy This technique is fast inexpensiveand works under various environmental conditionswhich will allow us to better understand and control deu-terium depletion in DKDP crystals
ACKNOWLEDGMENTS
We would like to thank M Runkel for rst discovering DKDP crack-ing R Floyd for providing DKDP crystals and L Chase and A Burn-ham for their support and helpful discussions This work was performedunder the auspices of the US Department of Energy by the University
of California Lawrence Livermore National Laboratory under contract W-7405-Eng-48
1 J J De Yoreo A K Burnham and P K Whitman Int Mat Rev47 113 (2002)
2 C E Barker R A Sacks B M Van Wonterghern J A Caird JR Murray J H Campbell K Kyle R B Ehrlich and N DNielsen Proc SPIE-Int Soc Opt Eng 2633 501 (1995)
3 T Suratwala paper to be published4 Cleveland Crystals Inc httpwwwclevelandcrystalscom
KDPshtmltable5 E A Popova I T Savatinova and I A Velichko Sov Phys Solid
State 12 1543 (1971)6 I P Kaminow R C C Leite and S P S Porto J Phys Chem
Solids 26 2085 (1965)7 J A Subramony B J Marquardt J W Macklin and B Kahr
Chem Mat 11 1312 (1999)8 H Tanaka M Tokunaga and I Tatsuzaki Solid State Commun
49 153 (1984)9 R J Nelmes G M Meyer and J E Tibballs J Phys C 15 59
(1982)10 M A Yakshin D W Kim Y S Kim Y Y Broslavets O E
Sidoryuk and S Goldstein Laser Physics 7 941 (1997)11 I Takenaga Y Tominaga S Endo and M Kobayashi Solid State
Commun 84 931 (1992)12 C Krenn personal communication13 M Sharon and A K Kalia J Solid State Chem 21 171 (1977)
Rapid Dipicolinic Acid Extractionfrom Bacillus Spores Detectedby Surface-EnhancedRaman Spectroscopy
STUART FARQUHARSON ALAND GIFT PAUL MAKSYMIUK andFRANK E INSCOREReal-Time Analyzers Inc East Hartford Connecticut06108
Index Headings Dipicolinic acid Bacillus spores Anthrax Surface-enhanced Raman spectroscopy
INTRODUCTION
The anxiety caused by the distribution of anthrax en-dospores through the US postal system in October 2001was exacerbated by the long time required for positiveidenti cation of the Bacillus anthracis spores and the un-known extent of their distribution Since that time manymethods capable of rapid eld analysis have been inves-tigated to augment or replace the laboratory method ofgrowing microorganisms in culture media which takesdays to perform12 Prominent among these approachesare polymerase chain reactions (PCR)3 immunoassaysand detection of calcium dipicolinate as a biochemicalsignature PCR employs primers to separate organism-
Received 10 October 2003 accepted 14 November 2003 Author to whom correspondence should be sent
352 Volume 58 Number 3 2004
speci c nucleic acid sequences (eg capsular protein en-coding gene for Bacillus anthracis)4 and polymerases toamplify the segment until it is detectable Recently am-pli cation times have been substantially reduced andcomplete analysis can now be performed in an hour orless Immunoassay methods are also being developed thatuse competitive binding of the bioagent (as an antigen)and its labeled conjugate for a limited number of anti-bodies Although analyses can be performed in under 30minutes a well-de ned anthrax antigen has not yet beenidenti ed5ndash7 and consequently the false-positive rate isunacceptably high8
A number of other methods are being developed witha focus on the detection of calcium dipicolinate (CaDPA)and its derivatives as a B anthracis signature This is sobecause only spore-forming bacteria contain CaDPA andthe most common potentially interfering spores such aspollen and mold spores do not Relatively fast methodshave been developed to chemically extract CaDPA andthen detect it directly by uorescence9 or indirectly byluminescence1011 In the latter case hot dodecylamine(DDA) has been used to extract dipicolinic acid (DPA)and terbium has been utilized to form a highly lumines-cent DPA complex11 Although measurements have beenperformed in as little as ve minutes it was found thatas many as three concentration-dependent complexes canform each with different lifetimes This coupled withthe fact that the Tb31 cation produces the same lumines-cence spectrum makes determinations of low spore con-centrations problematic
It has been long known that Raman spectra of Bacillispores are dominated by bands associated with CaDPA12
and that these spectra may provide a suitable anthrax sig-nature at the genus level13 Since that time considerableimprovements in Raman instrumentation have led to lab-oratory measurements of single Bacilli spores14 and to eld measurements of spores captured from a mail-sort-ing system15 However the single spore measurementsrequired complex instrumentation that is not ruggedwhile the eld measurements required milligrams of sam-ple Furthermore the Raman spectra of both measure-ments contained uorescence contributions that would in-crease uncertainty in quanti cation
In related research we demonstrated that nanogramquantities of DPA could be detected by uorescence-freesurface-enhanced Raman spectroscopy (SERS)16 We alsodemonstrated that microliter volumes of chemicals canbe detected by SERS using metal-doped sol-gel-packedglass capillaries17 Towards the goal of developing a rap-id eld SERS-based anthrax spore detector we havecombined our previous research and we now report thatDPA can be extracted from a 10 mg B cereus spore sam-ple using DDA in 1 minute and can be detected by SERSin an additional 1 minute
EXPERIMENTAL
Dipicolinic acid (26-pyridinedicarboxylic acid DPA)and dodecylamine (DDA) were used as received fromSigma-Aldrich (Milwaukee WI) Lyophilized B cereusspores prepared according to the literature13 were sup-plied by the University of Rhode Island and used as re-ceived Multiple particles approximately 01 mm3 each
were separated and weighed at 5 to 15 mg representing05 to 15 million spores The sample masses were con-sistent with a previous determination of spore density at0081 gmL that indicated a high degree of entrained air
All chemicals used to prepare the silver-doped sol-gelcoated capillaries were also obtained and used as receivedfrom Sigma-Aldrich According to previously publishedprocedures17 two precursor solutions were preparedmixed and then drawn into 1-mm-diameter glass capil-laries The silver amine precursor consisted of a 51 vvratio of 1 N AgNO 3 to 28 NH3OH while the alkoxideprecursor consisted of a 21 vv ratio of methanol to te-tramethyl orthosilicate The alkoxide precursors weremixed with silver amine precursor in an 81 vv ratioApproximately 015 mL was drawn into the capillarycoating a 15-mm length After sol-gel formation the in-corporated silver ions were reduced with dilute sodiumborohydride which was followed by a water wash to re-move residual reducing agent
A 100 mL drop of a 50 mM DDA solution in ethanolpre-heated to 78 8C was added to each of the B cereusparticles to digest the spore coat After 1 minute the re-sultant solution was drawn into a SER-active capillarythat was immediately xed horizontally to an XY posi-tioning stage (Conix Research Spring eld OR) just in-side the focal point of an f 07 aspheric lens The lensfocused the beam into the sample and collected the scat-tered radiation back along the same axis A dichroic lter(Omega Optical Brattleborough VT) was used to re ectthe excitation laser to the lens and pass the Raman scat-tered radiation collected by the lens An f 2 achromatwas used to collimate the laser beam exiting a 200-mm-core-diameter source ber optic while a second f 2 ach-romat was used to focus the scattered radiation into a 365mm ber optic (Spectran Avon CT) A short-pass lterwas placed in the excitation beam path to block the sil-icon Raman scattering generated in the source ber fromre ecting off sampling optics and reaching the detectorA long-pass lter was placed in the collection beam pathto block the sample Rayleigh scattering from reachingthe detector A 785 nm diode laser (Process InstrumentsInc model 785-600 Salt Lake City UT) was used todeliver 100 to 150 mW of power to the sample A Fouriertransform Raman spectrometer (Real-Time Analyzersmodel IRA-785 East Hartford CT) and a silicon photo-avalanche detector (Perkin Elmer model C30902S Stam-ford CT) were used to acquire the SER spectra
RESULTS AND DISCUSSION
As an initial experiment the SER spectrum of 1 gLof DPA in water was measured using the newly devel-oped silver-doped sol-gel-coated capillaries (Fig 1A) Atthis concentration a high signal-to-noise ratio (SN) isobtained in 1 min In fact a reasonable spectrum is ob-tained in the same time frame for 1 mgL (Fig 1B) TheSER spectra are reasonably similar to the normal Raman(NR) spectrum obtained for a saturated solution of DPAin 1 N KOH (Fig 1C) and the following band shifts areobserved (NR to SER) 647 to 657 cm21 817 to 815cm21 998 to 1008 cm21 1384 to 1382 cm21 1434 to1428 cm21 and 1569 to 1567 cm21 Many of these bandshave been previously assigned1213 such as 998 cm21 to
APPLIED SPECTROSCOPY 353
FIG 1 SERS of DPA in water using silver-doped sol-gel-coated glasscapillary for (A) 1 gL and (B) 1 mgL (C ) NR of saturated DPA in 1N KOH in a glass capillary Spectral conditions (A) and (B) 150 mWof 785 nm 1-min acquisition time (C) 450 mW of 785 nm 5-minacquisition time both 8 cm 21 resolution
FIG 2 SERS of DPA extracted from 10 mg B cereus particle using100 mL of 50 mM hot DDA acquired in (A) 1 minute and (B) 2 seconds(C ) Attempted SERS of 50 mM hot DDA in ethanol using silver-dopedsol-gel-coated glass capillary acquired in 1 min Spectral conditions150 mW of 785 nm 8 cm21 resolution
the symmetric ring stretch 1384 cm21 to the OndashCndashOsymmetric stretch 1428 cm21 to the symmetric ring CndashH bend and 1569 cm21 to the asymmetric OndashCndashOstretch
The rst B cereus samples consisted of 2 mg of sporesin 2 mL of 5 mM hot DDA The samples were main-tained at 78 8C for 40 min and while hot approximately10 mL was drawn into a SER-active capillary Since spec-tra of DPA were obtained for these initial samples small-er spore masses higher DDA concentrations and shorterheating periods were examined In due course it wasfound that 10 mg of spores could be digested by 100 mLof 50 mM hot DDA in one minute and detected (Fig2A) In fact the signal was suf ciently intense that it canbe observed in as little as two seconds (Fig 2B) Theamount of DPA that was extracted was estimated to bebetween 5 and 10 mgL by comparing the signal intensityof the 1008 cm21 band to that measured for DPA in waterThis is consistent with previous research that found thatthe majority of the DPA is extracted from spores usingDDA11 and that B cereus spores contain approximately10 DPA by weight18 The SN of 127 for the 1008 cm21
band in the 1-minute SER spectrum suggests a limit ofdetection of approximately 250 ng of B cereus sporesbased on a SN of 3 Finally it should be noted that DDAdid not produce a detectable SER spectrum as shown inFig 2C
CONCLUSION
We have demonstrated that by combining rapid extrac-tion of dipicolinic acid from Bacillus cereus spores withchemical identi cation by surface-enhanced Raman spec-troscopy as little as 10 mg of spores can be detected Infact the entire measurement from the time of adding hotdodecylamine to the spores to the time when the dipi-colinic acid SER spectrum is acquired and analyzedcould be performed in less than two minutes The abilityof this method to distinguish between spore-forming bac-teria such as Bacillus anthracis and non-DPA containingpowders could help prevent costly shutdowns associated
with the appearance of suspicious material or intentionalmailing of common substances as an anthrax hoax Thismethod could also prove useful in detecting the locationof anthrax endospores in mail distribution facilities if an-other veri ed attack should occur
Research continues to fully characterize the surface-enhanced Raman spectroscopy signal intensities as afunction of sample concentration and to explore otherextractants that do not require the use of elevated tem-perature
ACKNOWLEDGMENTS
The authors are grateful for the support of the National Science Foun-dation (DMI-0296116 and DMI-0215819) and the US Army(DAAD13-02-C-0015 Joint Service Agent Water Monitor program)The authors are indebted to Chetan Shende for preparing the sol-gelcapillaries The authors also thank James Gillespie Nicholas Fell andAugustus Fountain for providing important background informationMark Farquharson for laboratory support and Professor Jay Sperry ofthe University of Rhode Island for supplying B cereus spores
1 V A Pasechnik C C Shone and P Hambleton Bioseparations 3267 (1993)
2 P J Jackson M E Hugh-Jones D M Adair G Green K K HillC R Kuske L M Grinberg F A Abramova and P Keim ProcNatl Acad Sci USA 95 1224 (1998)
3 B R Glick and J J Pasternak Molecular Biology Principles andApplications of Recombinant DNA (ASM Press Washington DC1994)
4 C A Bell J R Uhl T L Had eld J C David R F Meyer TF Smith and F R Cockerill III J Clin Microbiol 40 2897(2002)
5 D L Gatto-Menking H Yu J G Bruno M T Goode M Millerand A W Zulich Biosens Bioelectron 10 501 (1995)
6 J J Quinlan and P M Foegeding J Rapid Methods AutomationMicrobiol 6 1 (1998)
7 A A Hindle and E A H Hall Analyst (Cambridge UK) 1241599 (1999)
8 M S Ascher US Department of Health amp Human Services(httpwwwhhsgovophppresentationsAscherdoc)
9 R Nudelman B V Bronk and S Efrima Appl Spectrosc 54 445(2000)
10 D L Rosen C Sharpless and L B McBrown Anal Chem 691082 (1997)
354 Volume 58 Number 3 2004
11 P M Pellegrino N F Fell Jr and J B Gillespie Anal ChimActa 455 167 (2002)
12 W H Woodruff T G Spiro and C Gilvarg Biochem BiophysRes Commun 58 197 (1974)
13 E Ghiamati R S Manoharan W H Nelson and J F SperryAppl Spectrosc 46 357 (1992)
14 A P Esposito C E Talley T Huser C W Hollars C M Schal-dach and S M Lane Appl Spectrosc 57 868 (2003)
15 S Farquharson L Grigely V Khitrov W W Smith J F Sperryand G Fenerty J Raman Spectrosc paper accep ted (2003)
16 S Farquharson W W Smith S Elliott and J F Sperry SPIE-IntSoc Opt Eng 3855 110 (1999)
17 S Farquharson and P Maksymiuk Appl Spectrosc 57 479(2003)
18 F W Janssen A J Lund and L E Anderson Science (Washing-ton DC) 127 26 (1958)
SPIE -2003-5269 117
pH dependence of methyl phosphonic acid dipicolinic acid and cyanide by surface-enhanced Raman spectroscopy
Stuart Farquharson Alan Gift Paul Maksymiuk Frank Inscore and Wayne Smith
Real-Time Analyzers 87 Church Street East Hartford CT 06108
ABSTRACT US and Coalition forces fighting terrorism in Afghanistan and Iraq must consider a wide range of attack scenarios in addition to car bombings Among these is the intentional poisoning of water supplies to obstruct military operations To counter such attacks the military is developing portable analyzers that can identify and quantify potential chemical agents in water supplies at microgram per liter concentrations within 10 minutes To aid this effort we have been investigating the value of a surface-enhanced Raman spectroscopy based portable analyzer In particular we have been developing silver-doped sol-gels to generate SER spectra of chemical agents and their hydrolysis products Here we present SER spectra of methyl phosphonic acid and cyanide as a function of pH an important factor affecting quantitation measurements which to our knowledge has not been examined In addition dipicolinic acid a chemical signature associated with anthrax-causing spores is also presented Keywords Chemical warfare agents agent detection agent hydrolysis SERS Raman spectroscopy homeland security
1 INTRODUCTION In the past decade the Unites States and its allies have been challenged by a different kind of warfare exemplified by the terrorist attacks of September 11 2001 Suicide bombings and the use of chemical agents are the norm and military personnel must consider a wide range of attack scenarios Among these is the intentional poisoning of water supplies to obstruct military operations in Afghanistan and Iraq Currently colorimetric paper is used to detect agents on-site while gas chromatography combined with mass spectrometry (GCMS) is used in mobile support laboratories However both methods have severe drawbacks The paper changes color in response to contact with many chemicals besides CWAs causing a high incidence of false positives1 while GCMS although very chemically specific requires hours to perform and constant re-calibration234 Military operations would be greatly aided by a portable analyzer that can identify and quantify potential chemical agents at concentrations that impact safety This includes the analysis of drinking water supplies distribution and storage systems To meet this goal the Department of Defense has been investigating numerous approaches under the auspices of the Joint Service Agent Water Monitor (JSAWM) program5 The JSAWM must be capable of identifying the chemical agents with no false-positives and quantifying the agents at microgram per liter concentrations within 10 minutes678 Recently we and others have been investigating the ability of surface-enhanced Raman spectroscopy (SERS) to measure chemical agents9-12 bioagents13-17 and their hydrolysis products in water SERS employs the interaction of surface plasmon modes of metal particles with target analytes to increase scattering efficiency by as much as 1 million times18 In 1987 the potential of this approach to measure CWAs was demonstrated by measuring a series of pesticides19 Several of these organophosphonates have chemical structures similar to CWAs in particular P=O functional groups In our studies we have been employing silver-doped sol-gels to promote the SER effect The porous silica network of the sol-gel matrix offers a unique environment for immobilizing and stabilizing SER-active metal particles20-23 The sol-gels have been coated on the internal walls of standard glass vials to allow rapid SER analysis of small sample quantities without preparation We have measured over 100 chemicals with enhancements of 104 to 106 demonstrated reversible measurements in a flowing system reproducible measurements from vial-to-vial and measurements in multiple solvents including water20-23 Previously we used these vials to perform preliminary measurements of cyanide (CN) methylphosphonic acid (MPA) and dipicolinic acid (DPA) MPA is a hydrolysis product of the nerve agents (eg sarin Reaction 1) and may be a valuable indicator of nerve agent usage particularly since the alkyl methylphosphonic acids are relatively more stable than their corresponding parent complexes24 DPA is
SPIE -2003-5269 118
a chemical signature of spore forming bacteria such as Bacillus anthracis And in light of the inability to rapidly detect the anthrax spores distributed through the US mail in October 2001 a number of methods are being developed to extract and analyze this signature Reaction 1 Stepwise hydrolysis of Sarin to form hydrofluoric acid (HF) isopropyl methylphosphonic acid (IMPA) then methyl phosphonic acid (MPA) and 2-propanol In our previous SERS investigations MPA and DPA were measured at 50 and 100 mgL respectively In both cases limits of detection (LOD) were estimated at 100 microgL providing encouragement in that SERS may satisfy the needs of the JSAWM Since it has been shown that pH can substantially influence the intensity of SER bands25 which would clearly influence quantitative analysis we undertook the present study to determine the severity of these effects for cyanide methyl phosphonic acid and dipicolinic acid Furthermore we previously observed a band at 1050 cm-1 for MPA6 possibly due to an anion formed at basic pH Here we investigate the source of this spectral anomaly
2 EXPERIMENTAL All chemicals including potassium cyanide methyl phosphonic acid dipicolinic acid and those used to prepare the silver-doped and gold-doped sol-gel coated vials were obtained and used as received from Sigma-Aldrich All samples were prepared in HPLC grade water (Fischer Scientific Fair Lawn NJ) for SERS measurements The pH of these samples was adjusted using dilute nitric acid or potassium hydroxide and verified using a pH electrode (Corning Inc Corning NY) that had been calibrated with pH 400 700 and 1000 buffered standards from Fischer Scientific Once prepared the samples were transferred into the silver-doped sol-gel vials (Real-Time Analyzers Inc East Hartford CT) The vials were coated in a manner similar to that previously reported by adding ammonium hydroxide to a solution of silver nitrate tetramethyl orthosilicate (TMOS) and methanol20 Gold-vials were coated by adding nitric acid to a solution of gold tetrachloride TMOS and methanol The two precursor solutions were prepared mixed and transferred to 2-mL glass vials dried and heated After sol-gel formation the incorporated metal ions were reduced with dilute sodium borohydride (1mgmL) which was followed by a water wash to remove residual reducing agent After the resultant analyte solution was introduced the SER-active vial was immediately fixed horizontally to an XY positioning stage (Conix Research Springfield OR) just inside the focal point of an f07 aspheric lens The lens focused the beam into the sample and collected the scattered radiation back along the same axis A dichroic filter (Omega Optical Brattleborough VT) was used to reflect the excitation laser to the lens and pass the Raman scattered radiation collected by the lens An f2 achromat was used to collimate the laser beam exiting a 200 microm core diameter source fiber optic while a second f2 achromat was used to focus the scattered radiation into a 365 microm fiber optic (Spectran Avon CT) A short pass filter was placed in the excitation beam path to block the silicon Raman scattering generated in the source fiber from reflecting off sampling optics and reaching the detector A long pass filter was placed in the collection beam path to block the sample Rayleigh scattering from reaching the detector A 785 nm diode laser (Process Instruments Inc model 785-600 Salt Lake City UT) was used to deliver 100 to 150 mW of power to the sample A Fourier transform Raman spectrometer (Real-Time Analyzers model IRA-785 East Hartford CT) and a silicon photo-avalanche detector (Perkin Elmer model C30902S Stamford CT) were used to acquire the SER spectra
3 RESULTS AND DISCUSSION In a previous study of MPA6 aimed at developing a concentration calibration curve and determining limits of detection (LOD) we observed an anomaly at 1050 cm-1 Since it was found that the band intensity changed as a function of concentration the band must be associated with a sample parameter Two possible parameters photon flux and pH are examined here The first parameter was investigated by irradiating a 1mgmL MPA sample in a SER-active vial with laser powers of 200 mW and above and monitoring spectral changes It was immediately found that the 1050 cm-1 band
2O+ H HF + +OH
OH3H C
OP
OF
CH
CH3
3
3H C
OCP
OOH
CH
CH3
3
3H C
O
CPHO
CH
CH
3
3
C
Sarin IMPA MPA 2-propanol
SPIE -2003-5269 119
appeared and grew as a function of time and that the higher the laser power the faster the growth Figure 1 shows the growth of the 1050 cm-1 band over the course of 30 minutes when using 150 mW of 785 nm excitation while Figure 2A shows that the growth can be fit with a first order exponential equation namely I1050 = 03+05e-013t Furthermore the 760 cm-1 band could be fit with a first order decay equation with an identical exponential rate constant ie I760 = 08-
08e-013t The rates represent classical first order kinetics and their correspondence allows one to conclude that MPA is being transformed one-for-one into a photo-generated product At this time the photoproduct has not been positively identified but phosphonic (phosphorous) acid and phosphonate are likely candidates since the symmetric P(OH)3 stretch occurs at ~1050cm-1 Our studies show that a reduction in laser power to 100 mW at the sample for MPA essentially eliminates this degradation process This laser power or lower was used for further measurements
Methyl phosphonic acid is a diprotic acid that stepwise dissociates into two anions MPA- and MPA= according to the following reactions26
MPA MPA- + H+ pKa1 = 212 Reaction 2
MPA- MPA= + H+ pKa2 = 729 Reaction 3 The relative concentrations of MPA MPA- and MPA= can be determined at any pH by expressing [MPA] and [MPA=] in terms of [MPA-] using Reactions 2 and 3 and summing all three to equal the total starting concentration here 2 mgmL (0021M MW = 9602) viz
[MPA] + [MPA-] + [MPA=] = 0021M Equation 1 substituting from Reactions 1 and 2
([H+][MPA-])K1a + [MPA-] + (K2a[MPA-])[H+] = 0021M Equation 2 rearranging [MPA-] = 0021M(1+[H+]K1a + K2a[H+]) Equation 3 The relative concentrations of MPA MPA- and MPA= as a function of pH are shown in Figure 3 It is worth noting that near neutral pH both MPA- and MPA= will be present To confirm that the SER signal followed this pH dependence a starting solution consisting of 20 mg of MPA in 10 mL HPLC grade water was prepared and brought to pH of 20 using dilute nitric acid From this solution 2 mL were added to a SER-active vial and the SER spectrum recorded At this pH a peak at 760 cm-1 was barely discernable The 2 mL solution was returned to the starting solution and the pH was re-measured to correct for any changes that the silver-doped sol-gel vials might cause In most cases the change was less than 02 pH units and the pH is reported as the before and after average Next the pH of the
Figure 2 A) Exponential growth of 1050 cm-1 band and B) exponential decay of 760 cm-1 band for spectral series in Figure 1
Figure 1 Growth of 1050 cm-1 band as a function of time due to exposure to 150 mW of 785 nm Spectra are 5 sec each collected every 100-sec from 0 to 30-min
B
570 770 970 1170 1370Raman Shift (cm-1)
Arbit
rary
Unit
s
MPA Photodegradation
600 800 1000 1200 1400 Wavenumber (cm-1)
A
0 10 20 30 0 10 20 30 time (min) time (min)
Ram
an In
tens
ity (r
elat
ive)
30
min
0
SPIE -2003-5269 120
starting solution was adjusted to 325 using dilute KOH Again 2 mL were added to a vial and the SER spectrum recorded At this pH a reasonably strong 760 cm-1 band was observed This process was repeated as spectra were recorded at pHs of 70 74 75 79 85 and 100 A total of 1 mL of KOH was added diluting the total concentration by 10 Next the pH of the starting solution was made acidic by adding dilute nitric acid dropwise This time spectra were recorded at pHs of 72 69 64 and 37 Figure 4 shows the SER spectra for representative pHs (spectra were left out to simplify the figure) while Figure 3 shows the 760 cm-1 peak intensities as a function of pH (The band intensities were adjusted to compensate for dilution effects caused by the addition of HNO3 and KOH then normalized to 0021 M for the most intense band observed at pH 37) It is clear from Figure 3 that the 760 cm-1 band follows the MPA- concentration as a function of pH and must be assigned to this anion No bands were observed that corresponded to MPA or MPA= The lack of an MPA SER spectrum may be due to the absence of an attraction between the neutral analyte and the electropositive silver surface The same reasoning suggests that a strong SER spectrum should be observed for MPA= but it is not and a satisfactory explanation has not been found
0000
0005
0010
0015
0020
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14pH
Con
cent
ratio
n [M
]
pK1 = 212 pK2 = 729
MPA- MPA=
MPA
Figure 4 SER spectra of 002M MPA as a function of pH Conditions 100 mW of 785 nm 36 scans (1 min) 8 cm-1 recorded 2 min after sample introduction pH 19 and 20 not apparent on this scale pH 69 and 74 near identical to 70 and 75 and not shown for clarity
00
02
04
06
08
10
12
14
16
18
0 5 10 15 20 25 30Measurement Number
Ram
an In
tens
ity (7
60 c
m-1
)
Figure 3 Concentration dependence of MPA MPA- and MPA= as a function of pH for a 002M sample Intensity of 760 cm-1
band from Figure 6 as a function of pH ( for increasing basic adjustment for increasing acidic adjustment error was measured at ~10 for pH 64)
Figure 5 SER spectra of 002M MPA at pH 64 measured around a vial at three heights (9 points per height) Conditions as in Figure 4 but 10-sec scans
Since these measurements involved the removal and replacement of the SER-active vial in the sample holder to remove and add sample variation in the intensity as a function of vial position was minimized by illuminating the exact same height along the vial wall But this does not account for variability of the SERS response of the sol-gel coating around the vial To analyze this effect a vial containing MPA at pH 64 was rotated at ~ 40o
intervals at the original height and 18rdquo above and below this value Figure 5 shows the intensity of the 760 cm-1
band for the 27 positions It was found that the average value was 137plusmn014 an RSD of 10 overall and 5 for each height An error bar is included in Figure 4 for the pH 64 measurement
SPIE -2003-5269 121
With the above analyses in mind a preliminary investigation of the SER spectral response for MPA (as MPA-) as a function of concentration was performed A single vial was used for these measurements beginning with 1 mgL followed by measurements of 10 100 and 1000 mgL In all cases the pH was ~7 and 3 positions around the vial were measured per concentration Since the 760 cm-1 band was not observed for concentrations of 1 or 10 mgL using 100 mW of 785 nm the laser power at the sample was raised to 200 mW beginning with the 10 mgL concentration Photo-degradation was largely avoided (and not observed) by exposing the sample for only 33 seconds per spectral acquisition Representative spectra for 10 100 and 1000 mgL are shown in Figure 6 while a plot of the 760 cm-1 band intensity as a function of concentration is shown in Figure 7 These values were also used to estimate limits of detection based on the signal-to-noise ratio (SN) of the 760 cm-1 band using 100 mW of 785 nm laser excitation and a 10-min acquisition time As summarized in the Figure 7 inset the lower the measured concentration the lower the predicted LOD Even if as estimated 210 microgL could be measured using the silver-doped sol-gel vials an improvement of a factor of 70 is still required to meet JSAWM goals of measuring 3 microgL in 10-minutes Similar to MPA DPA is a diprotic acid (pKa1 = 216 and pKa2 = 692) and variations in pH will effect the relative concentrations of DPA DPA- and DPA= and possibly the SER spectra and band intensities This could prove significant if an acid or base is used to denature anthrax spores with the goal of extracting and analyzing DPA The MPA pH study described above was mimicked for DPA except that the starting solution consisted of 20 mg of DPA in 20 mL HPLC grade water (60x10-3M MW = 1671) The initial solution had a pH of 245 which was made basic by dilute KOH to pHs of 355 433 487 559 1069 and 1166 SER spectra were recorded at each pH using 100 mW of 785 nm and a 44-sec acquisition time Next one drop of concentrated nitric acid was used to remake the solution acidic at a pH of 200 Again sequential pH measurements were performed at 383 510 735 and 822 The solution pH was made acidic a third time but to pH 219 171 then 135 Throughout this process no more than 20 drops of acid or base were added and therefore the concentration was diluted by no more than 10 Most of the spectral bands showed a minor decrease in intensity as a function of increasing pH values However the bands at 525 and 795 cm-1 showed the most dramatic changes which occurred at acid pH Figure 8 shows the SER spectra of DPA for the spectral region and pH range of interest The identity of the DPA species was determined by plotting the normalized peak intensities with the lowest value set to 0 and the highest to 0006 M as a function of pH and overlaying these values on a plot of the relative concentrations for DPA DPA- and DPA= as previously done for MPA (Figure 9) As can be seen the 525 cm-1 band clearly corresponds to DPA The correspondence of the 795 cm-1 band to this species is less clear as the band retains intensity until pH of 55 This can be attributed to contribution to the overlapping band at 810 cm-1 which does not change as a function of pH The fact that most bands are observed at all pHs suggest that the primary interaction with silver is through the ring nitrogen This is supported by the fact that the most intense band occurs at 1008 cm-1 attributed to a symmetric ring breathing mode and that this interaction has been characterized for pyridine in numerous papers27
0
1
2
3
4
5
6
7
0 200 400 600 800 1000 1200
MPA Concentration (mgL)
760
Ban
d In
tens
ity
Figure 6 SER spectra of MPA in water at A) 1000 B) 100 and C) 10 mgL Conditions pH of 7 silver-doped sol-gel coated vial 200 mW of 785 nm 33-sec 8 cm-1 resolution
Figure 7 Plot of SER intensity of 760 cm-1 band of MPA as a function of concentration using 200 mW of 785 nm Inset table includes average intensity LOD standard deviation and percent deviation for each concentration but for 100 mW and 10 min
conc (mgL) LOD ave stddev dev10 021 002 1021100 072 010 14301000 312 040 1280
A
B C
SPIE -2003-5269 122
The ability to detect dipicolinic acid (DPA) by SERS is demonstrated in Figure 10 Even at 1 mgL the primary bands are visible The intensity of the dominant ring breathing mode observed at 1008 cm-1 for DPA in water (pH = 25 to 55) is plotted as a function of concentration in Figure 11 These values were also used to estimate limits of detection based on the SN of the 1008 cm-1 band using 100 mW of 785 nm laser excitation and a 10-min acquisition time Again the lower the measured concentration the lower the predicted LOD (see Figure 11 inset) and detection of 160 microgL is possible Gastrointestinal anthrax requires significant more spores than inhalation anthrax28 and a limit of detection might be placed at 1 million spores in 1 liter of water or 10 microgL Since each spore contains ~10 CaDPA by weight29 a goal for DPA might be 1 microgL indicating that the present measurements must be improved by nearly two orders of magnitude Due to the increased hazards of handling HCN gas KCN salt was used for these experiments Nevertheless all sample preparations were performed in a chemical hood KCN completely dissolves in water but its conjugate acid HCN is formed and has a Ka of 615x10-1030 viz
HCN CN- + H+ pKa = 921 Reaction 4
Figure 8 SER spectra of 1 mgmL DPA as a function of pH Conditions 100 mW 785 nm 100 scans (44-sec)
Figure 9 Concentration dependence of DPA DPA- and DPA= as a function of pH for a 0006M sample Intensity of 525 () and 795 (diams) cm-1 bands from Fig 8 as a function of pH
pH
135 171 219 383
0
0001
0002
0003
0004
0005
0006
0007
0 2 4 6 8 10 12 14pH
Con
cent
ratio
n [M
]
DPADPA-DPA=795525
DPA DPA=DPA-
pK2 = 692pK1 = 216
795 525
0
05
1
15
2
25
0 200 400 600 800 1000 1200
DPA Concentration (mgL)
1008
Ban
d In
tens
ity
conc (mgL) lod-10min-100mw1 017
10 016100 103
1000 355
Figure 10 SER spectra of DPA in water at A) 1000 B) 100 C) 10 and D) 1 mgL Conditions pH of 25-55 silver-doped sol-gel coated vial 175 mW of 785 nm 1-min 8 cm-1 D) has been multiplied by x10 to make bands visible
A
B C
D
Figure 11 Plot of SER intensity of 1008 cm-1 band of DPA as a function of concentration using 175 mW of 785 nm Inset table includes LOD in mgL for each concentration but for 100 mW and 10 min
SPIE -2003-5269 123
Consequently the cyanide concentration must be determined for each initial KCN concentration Specifically the samples prepared with concentrations of 01 1 10 100 and 1000 mgL of KCN produced CN- concentrations of 63x10-3 033 69 89 and 964 mgL at pHs of 816 90 967 102 and 107 respectively The pH dependence for the HCN and CN- concentrations are shown in Figure 12 Thus as the amount of KCN added to the solution decreases so does the pH of the solution (becomes less basic) and according to Reaction 4 the relative amount of CN- to HCN also decreases For example in the preparation of a 01 mgL solution of KCN the pH is shifted from 7 for pure water to only 816 and only 63 of the starting material becomes CN- or 63x10-3 mgL In comparison for a solution of 1000 mgL the pH is shifted from 7 to 107 and 96 of the starting material becomes CN- This is significant since the cyanide ion is better able to adsorb onto the silver particles and become SERS active SER spectra of 10 100 and 1000 mgL of KCN samples are shown in Figure 13 and a corresponding plot of the 2100 cm-1 band intensity as a function of concentration is shown in Figure 14
000102030405060708091011
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14pH
Con
cent
ratio
n [m
gm
L]
pKa = 921
CN -HCN
The SER spectra of cyanide are dominated by the single CN stretch at 2100 cm-1 which occurs in normal Raman spectra of solutions at 2080 cm-1 However a low frequency mode occurs at 135 cm-1 due to a Ag-CN stretch (not shown) indicative of a strong surface interaction It is also observed that as the concentration decreases the CN stretch shifts to 2140 cm-1 This shift has been attribute to the formation of a tetrahedral Ag(CN)3
2- surface structure31 as well
Figure 14 Concentration dependence of KCN SERS measured under conditions in Fig 11 Concentrations are 1 01 and 001 mgml Intensities are measured for the CN stretch at 2100 cm-1 Inset table includes LOD in mgL for each concentration in Figs 13 and 15 but for 100 mW and 10 min
Figure 13 SER spectra of KCN in water at A) 1000 B) 100 and C) 10 mgL Conditions pHs of 107 102 and 97 silver-doped sol-gel coated vial 100 mW of 785 nm 1-min 8 cm-1 C) has been multiplied by x10 to make band visible
Figure 12 Concentration dependence of HCN and CN- as a function of pH for a 1 mgmL sample Calculated intensity of 2100 () cm-1 band for a 1 mgml sample at pHs of 816 90 967 102 and 107
A
B
C
0
20
40
60
80
100
120
140
0 200 400 600 800 1000 1200CN Concentration (mgL)
2100
Ban
d In
tens
ity
Figure 15 SER spectra of KCN in water at A) 10 B) 1 and C) 01 mgL Conditions pHs of 97 12 and 12 gold-doped sol-gel coated vial 100 mW of 785 nm 1-min 8 cm-1 C) has been multiplied by x10 to make band visible
A
B
C
conc (mgL) lod-10min-100mw Condition01 001 Au-pH 121 007 Au-pH 12
69 003 Au-pH 9769 016 Ag-pH 9789 022 Ag-pH 102946 113 Ag-pH 107
SPIE -2003-5269 124
as to CN adsorbed to two different surface sites32 Alternatively the 2140 cm-1 band could be attributed to HCN since this species dominates at lower concentrations However it is unlikely that this species would be attracted to the electropositive silver surface Further both peaks should be present at pHs between 85 and 105 but this is not observed It has also been suggested that at concentrations near and above monolayer coverage the CN- species is forced to adsorb end-on due to crowding and at lower concentrations the molecule can reorient to lie flat33 This suggests that the 2100 and 2140 cm-1 bands correspond to the end-on and flat orientations respectively As Figures 13 and 14 show the intensity of the CN stretch for the 89 mgL sample is nearly as intense as the 964 mgmL band This suggests that the Raman signal for the flat orientation is more enhanced However more extensive measurements are required to verify this point Since resent research has suggested that cyanide may be more effectively detected on gold measurements of KCN solutions were also performed using gold-doped sol-gel vials Preliminary measurements are shown in Figure 15 for samples prepared from 01 1 and 10 mgL KCN Since the pHs are 816 90 967 the resultant CN- concentrations are 63x10-3 033 and 69 mgL Initially only the highest concentration was observed and the signal intensity was significantly better than the equivalent concentration measured using silver In an effort to shift Reaction 4 to the left transforming HCN to CN- (Le Chatelierrsquos principle) KOH was added to the lower concentration samples producing solutions with pH 12 This effectively forces all of the cyanide in solution to be CN- or 01 and 10 mgL respectively More importantly the CN stretch is now observed in the SER spectra The band appears at 2125 cm-1 as has been previously reported for gold12 As calculated for MPA and DPA LODs can be estimated from this data For the three concentrations of cyanide on silver the LODs are 016 to 11 mgL for 100 mW of 785 nm laser excitation and a 10-min acquisition time For gold pH adjusted the LODs are10 to 70 microgL an improvement of more than 10 times silver Nevertheless either substrate is sufficient to meet the JSAWM goals of measuring 3 mgL in 10-minutes as the requirements form cyanide are much less stringent than the nerve agents
4 CONCLUSIONS Here we examined the surface-enhanced Raman spectral response for methyl phosphonic acid dipicolinic acid and cyanide as a function of pH It was determined that the most prominent peak at 760 cm-1 reaches a maximum intensity between pH 3 and 7 and corresponds to the MPA- species Neither the MPA nor MPA= species appear to generate a SER spectrum and consequently no spectra were observed below pH 2 or above pH 8 In this study we also found that higher laser powers could cause photodegradation of MPA signified by the exponential growth of a band at 1050 cm-1 which is tentatively assigned to phosphorous acid Unlike MPA DPA was observed at all pHs This is attributed to the dominant interaction of the pyridine functional group with silver Minor spectral changes were observed at acid pHs and were assigned to neutral DPA Like MPA SER spectra of cyanide were pH dependent No spectra were observed for the HCN species while CN- was best observed at pHs more basic than 8 Preliminary concentration studies for the three analytes allowed estimating limits of detection for MPA DPA and CN using 100 mW of 785 nm and a 10-min acquisition time of 210 165 and 70 microgL respectively Although the latter value suggests that cyanide can be measured at sufficiently low concentrations to meet JSAWM goal improvements by 100 to 200 times are required for MPA and DPA It is clear from this study that pH of the sample is important and must be taken into account when developing concentration calibration curves as well as fieldable analyzers Future work will include tailoring the sol-gel with specific functional groups to dictate sample pH and thereby optimize sensitivity
ACKNOWLEDGMENTS The authors are grateful for the support of the US Army (DAAD13-02-C-0015 Joint Service Agent Water Monitor program) The authors would also like to thank Dr Steve Christensen of the US Army SBCCOM for helpful discussions and Mr Chetan Shende of Real-Time Analyzers for assistance in development of the gold-doped sol-gels
REFERENCES 1 Erickson B Analytical Chemistry News amp Features June 1 397A (1998)
SPIE -2003-5269 125
2 Johnston RL Hoefler CM Fargo JC and Moberley B AT-ONSITE 5-8 (1994) 3 Black RM RJ Clarke RW Read and MT Reid J Chromatography 662 301-321 (1994) 4 USA Army Center for Health Promotion and Preventive Medicine (USACHPPM) Technical Guideline 230A 5 Jensen JL NC Wong W R Loerop SPIE 4775-03 (2001) 6 Hoenig SL Handbook of Chemical Warfare and Terrorism Greenwood Press Wesport CT (2002) 7 Munro NB SS Talmage GD Griffin LC Waters AP Watson JF King and V Hauschild Env Health
Persp 107 933-974 (1999) 8 Holstege CP Kirk M Sidell FR Crit Care Clin13 923-42 (1997) 9 Farquharson S P Maksymiuk K Ong and S Christesen SPIE 4577 166-173 (2001) 10 Lee Y and S Farquharson SPIE 4378 21-26 (2001) 11 Spencer KM J Sylvia S Clauson and J Janni SPIE 4577 158-165 (2001) 12 Tessier P S Christesen K Ong E Clemente A Lenhoff E Kaler and O Velev Applied Spectroscopy 56
1524-1530 (2002) 13 Farquharson S WW Smith S Elliott and JF Sperry SPIE 3533 207-214 (1998) 14 Farquharson S WW Smith S Elliott and JF Sperry SPIE 3855110-116 (1999) 15 Farquharson S WW Smith YH Lee S Elliott and JF Sperry SPIE 4575 62-72 (2002) 16 Guzelian AA J Sylvia J Janni S Clauson and KM Spencer SPIE 4577 182-192 (2001) 17 Shende C F Inscore A Gift P Maksymiuk and S Farquharson in press 18 Weaver MJ S Farquharson and MA Tadayyoni J Chem Phys 82 4867-4874 (1985) 19 Alak AM and T Vo-Dinh Analytical Chemistry 59 2149-2153 (1987) 20 Lee Y and S Farquharson SPIE 4206 140-146 (2000) 21 Farquharson S and Y Lee SPIE 4200-16 (2000) 22 Lee Y S Farquharson and P M Rainey SPIE 3857 76-84 (1999) 23 Lee Y S Farquharson H Kwong and M Shahriari SPIE 3537 252-260 (1998) 24 Wang J M Pumera G Collins and A Mulchandani Analytical Chemistry 74 6121-6125 (2002) 25 Dou X YM Jung Z-Q Cao and Y Ozaki Applied Spectroscopy 53 1440-1447 (1999) 26 Data supplied by S Christesen and K Ewing 27 Kerker M and B Thompson Eds SPIE MS 10 (1990) 28 Inglesby TV DA Henderson JG Bartlett JAMA 287 2236 (2002) 29 FW Janssen AJ Lund and LE Anderson Science 127 26 (1958) 30 Lide DR Ed Handbook of Chemistry and Physics CRC Press 77th Ed (1996-7) 31 Billmann J G Kovacs and A Otto Surf Sci 92 153 (1980) 32 Murray CA and S Bodoff Phys Rev B 32 671 (1985) 33 Kellogg D and J Pemberton J Phys Chem 91 1120 (1987)
SPIE-2003-5269 16
Chemical agent detection by surface-enhanced Raman spectroscopy
Stuart Farquharson Alan Gift Paul Maksymiuk Frank Inscore and Wayne Smith
Real-Time Analyzers 87 Church Street East Hartford CT 06108
Kevin Morrisey and Steven D Christesen US Army SBCCOM Aberdeen Proving Ground MD 21010
ABSTRACT
In the past decade the Unites States and its allies have been challenged by a different kind of warfare exemplified by the terrorist attacks of September 11 2001 Although suicide bombings are the most often used form of terror military personnel must consider a wide range of attack scenarios Among these is the intentional poisoning of water supplies to obstruct military operations in Afghanistan and Iraq To counter such attacks the military is developing portable analyzers that can identify and quantify potential chemical agents in water supplies at microgram per liter concentrations within 10 minutes To aid this effort we have been investigating the value of a surface-enhanced Raman spectroscopy based portable analyzer In particular we have been developing silver-doped sol-gels to generate SER spectra of chemical agents and their hydrolysis products Here we present SER spectra of several chemical agents measured in a generic tap water Repeat measurements were performed to establish statistical error associated with SERS obtained using the sol-gel coated vials Keywords Chemical agents chemical agent detection SERS Raman spectroscopy
1 INTRODUCTION
In the autumn of 2001 terrorism within US borders became a sobering reality While extensive efforts are being implemented to secure the homeland US and Coalition forces in Afghanistan and Iraq are constantly faced with terrorist attacks In addition to car-bombings the military has identified several non-traditional attack scenarios including poisoning of water supplies by chemical warfare agents (CWAs) To counter this threat the Department of Defense has been investigating numerous approaches to detect such attacks under the auspices of the Joint Service Agent Water Monitor (JSAWM) program1 The JSAWM must be capable of identifying the chemical agents with no false-positives and quantifying the agents at microgram per liter concentrations within 10 minutes2 This includes the analysis of drinking water supplies distribution and storage systems Currently colorimetric paper is used to detect agents on-site while gas chromatography combined with mass spectrometry (GCMS) is used in mobile support laboratories However both methods have severe drawbacks The paper changes color in response to contact with many chemicals besides CWAs causing a high incidence of false positives3 while GCMS although very chemically specific requires up to an hour to perform and regular re-calibration456 Vibrational spectroscopy has also been investigated as a potential method of rapidly detecting CWAs7-
11 as early as 1985 in the case of infrared spectroscopy8 and 1988 in the case of Raman spectroscopy11 Again however these techniques also have limitations when it comes to measuring trace poisons in water Infrared spectra would be dominated by the very strong absorption of water which would obscure absorptions by most other chemicals present Whereas Raman spectroscopy is simply not a very sensitive technique and detection limits are typically grams per liter Surface-enhanced Raman spectroscopy (SERS) offers several advantages over conventional vibrational methods and that may provide the necessary sensitivity required for detecting trace quantities of chemical agents in water SERS employs the interaction of surface plasmon modes of metal particles with target analytes to increase scattering efficiency by as much as 1 million times12 In 1987 the potential of this approach to measure CWAs was demonstrated by measuring a series of pesticides13 Several of these organophosphonates have chemical structures similar to CWAs in particular P=O functional groups In the past few years we and others have further explored the ability of SERS to detect CWAs14-17 and even bioagents 18-21 We have been employing silver-doped sol-gels to promote the SER effect
SPIE-2003-5269 17
in these studies The porous silica network of the sol-gel matrix offers a unique environment for immobilizing and stabilizing SER-active metal particles22-25 The silver-doped sol-gels have been coated on the internal walls of standard glass vials to allow rapid SER analysis of small sample quantities without preparation We have measured over 100 chemicals with enhancements of 104 to 106 demonstrated reversible measurements in a flowing system reproducible measurements from vial-to-vial and measurements in multiple solvents including water21-25 Previously we used these vials to perform preliminary measurements of cyanide (CN) and methylphosphonic acid (MPA) Most of the nerve agents form MPA during hydrolysis while Tabun forms CN a chemical agent in its own right In another paper including in these proceedings we examined the limits of detection (LOD) for MPA by measuring a series of concentrations down to 50 mgL and estimated a limit of detection of 100 microgL26 These measurements provide encouragement in that SERS may satisfy the needs of the JSAWM To further establish the viability of SERS in particular silver-doped sol-gels here we present analysis of cyanide mustard and VX in tap water The measurements performed at the US Armyrsquos Edgewood Chemical Biological Center Aberdeen MD also included numerous repeat measurements to establish reproducibility
2 EXPERIMENTAL 2a General All chemicals including potassium cyanide 2-chloroethylethyl sulfide and those used to prepare the silver-doped and gold-doped sol-gel coated vials were obtained and used as received from Sigma-Aldrich All samples were prepared in a chemical hood using HPLC grade water unless otherwise noted (Fischer Scientific Fair Lawn NJ) for SERS measurements Once prepared the samples were transferred into the silver-doped sol-gel vials (Real-Time Analyzers Inc East Hartford CT) The vials were coated in a manner similar to that previously reported by adding ammonium hydroxide to a solution of silver nitrate tetramethyl orthosilicate (TMOS) and methanol22 The two precursor solutions were prepared mixed and transferred to 2-mL glass vials dried and heated After sol-gel formation the incorporated metal ions were reduced with dilute sodium borohydride (1mgmL) which is followed by a water wash to remove residual reducing agent After the resultant analyte solution was introduced the SER-active vial was fixed horizontally to an XY positioning stage (Conix Research Springfield OR) just inside the focal point of an f07 aspheric lens The lens focused the beam into the sample and collected the scattered radiation back along the same axis A dichroic filter (Omega Optical Brattleborough VT) was used to reflect the excitation laser to the lens and pass the Raman scattered radiation collected by the lens An f2 achromat was used to collimate the laser beam exiting a 200 microm core diameter source fiber optic while a second f2 achromat was used to focus the scattered radiation into a 365 microm fiber optic (Spectran Avon CT) A short pass filter was placed in the excitation beam path to block the silicon Raman scattering generated in the source fiber from reflecting off sampling optics and reaching the detector A long pass filter was placed in the collection beam path to block the sample Rayleigh scattering from reaching the detector A 785 nm diode laser (Process Instruments Inc model 785-600 Salt Lake City UT) was used to deliver 100 to 150 mW of power to the sample A Fourier transform Raman spectrometer (Real-Time Analyzers model IRA-785 East Hartford CT) and a silicon photo-avalanche detector (Perkin Elmer model C30902S Stamford CT) were used to acquire the SER spectra
2b Edgewood Chemical Biological Center The surface-enhanced Raman spectral measurements at the US Armyrsquos Chemical Biological center presented here were all performed on September 12 2003 To expedite measurements a plate was machined to hold up to 12 SER-active sample vials (Figure 1) The plate fit a standard XY plate reader that could be programmed Pure KCN bis-(2-chloroethyl)sulfide (distilled mustard gas HD) and ethyl S-2-diisopropylamino ethyl methylphosphonothioate (VX) were obtained on-site and used to prepare 1 gL tap water solutions in a chemical hood with appropriate safety equipment Simulated tap water was prepared by adding 10 chemicals most often found in tap water at appropriate concentrations to distilled water (Table 1) SER measurements were also performed in a chemical hood For added safety the FT-Raman instrument was placed outside the laboratory and 30 foot fiber optic and electrical cables were used to allow remote SERS measurements and plate manipulation For each experiment 1gL samples were prepared and added to 9 individual vials which were then loaded on the plate In some cases a tenth vial was included as a blank
Table 1 Generic Tap Water Composition
Compound mgL NaHCO3 100 CaSO4 27 MgSO4bull7H2O 67 NaNO3 1 Fulvic Acid 1 K2HPO4 07 KH2PO4 03 (NH4)2HSO4 001 NaCl 001 FeSO4 0001 pH 76-78
SPIE-2003-5269 18
A software program was written that allowed selecting the sequence that the vials were measured the number of positions along the length of the vials to measure (1 to 5) and the number of scans to co-add During sample analysis the program displayed the vial being analyzed the point being analyzed and the spectrum as it was being acquired Once all the data was collected a second software program was written to rapidly analyze the data The spectra collected for all the vials on a plate could be loaded at one time and then the spectra for each point could be displayed simultaneously or separately The user could then select the Raman peak to analyze in terms of peak height or area This was accomplished by selecting points on either side of the peak to define a baseline of zero The peak height or area could then be computed for all of the spectra loaded and then exported to a spreadsheet for statistical analysis
Figure 1 A) Vial Holder 6 slots to hold 2 vials each end-to-end B) Measurement Configuration Program user selects vials to measure sequence number of points per vial (1 to 5) and number of scans per point C) Spectral Acquisition Program shows spectrum being collected which vial and position D) Spectra Analysis Program user selects spectra to analyze by plate vial and point (s) as well as two wavenumbers defining the peak and the baseline to subtract The image is of 5 repeat measurements of 10 mgL KCN in generic tap water 16 sec each 100 mW of 785 nm
3 RESULTS AND DISCUSSION Raman and surface-enhanced Raman spectra were obtained for potassium cyanide bis-(2-chloroethyl)sulfide and ethyl S-2-diisopropylamino ethyl methylphosphonothioate representing three classes of chemical agents cyanides mustards and nerve agents respectively Spectra were also obtained for 2-chloroethyl ethyl sulfide (CEES) a structural analogue to HD which was included in the study to aid in assigning spectral bands KCN salt was used for cyanide experiments to avoid the increased hazards of handling HCN gas KCN completely dissolves in water forming its conjugate acid HCN according to its Ka of 615x10-1027 and at a concentration of 1 mgmL results in a pH 107 solution This is significant in that only CN- appears to interact sufficiently with silver to produce a SER spectrum and no spectral signal is observed below pH 726 Figure 2 shows the SER and normal Raman spectra for KCN The SERS spectrum of 1mgml KCN in water shows a single intense somewhat broad feature at
A B
C D
SPIE-2003-5269 19
2100 cm-1 assigned to the single CequivN stretch The band is much sharper in the normal Raman spectra of the solid KCN salt at 2074 cm-1 This band does broaden and shift to 2080 cm-1 in solution (not shown) However the observed SERS frequency is attributed to interaction with silver and low frequency mode at 135 cm-1 attributed to a Ag-CN stretch (not shown) supports this conclusion
cm-1 band to a C-S stretch but the authors concede that it is in fact more likely a C-Cl stretch28 It appears that the most intense Raman bands at 648 692 and 747 cm-1 shift to 620 660 and 720 cm-1 in the SER spectra and are tentatively assigned as above The width of these bands suggests that they overlap underlying spectral features Additional bands in the Raman spectra occur at 972 1034 1049 1263 1286 1423 1442 2865 2935 and 2960 cm-1 Corresponding bands occur in the SER spectra at 964 1015 1054 1286 1410 1447 2865 and 2935 cm-1 Most of these bands are associated with alkane modes specifically the bands at approximately 1040 cm-1 to a C-C stretch 1290 cm-1 to a CH2 in-phase twist 1440 cm-1 to a CH2 wag 2865 cm-1 to a symmetric CH2 stretch and 2965 cm-1 to an asymmetric CH2 stretch The Raman and SER spectra of sulfur mustard were measured at the Edgewood center (Figure 4) Both spectra are largely similar to CEES The C-Cl and C-S bands in the Raman spectrum of HD now occur at 640 655 700 739 and 760 cm-1 and are more resolved possibly due to the increased molecular symmetry Theoretical calculations indicate that the first three bands are due to C-Cl stretching modes and the latter two to C-S stretching modes28 Only the C-Cl bands maintain significant intensity in the SER spectra occurring at 624 and 643 cm-1 which is attributed to the
Prior to measurements of HD CEES was examined by Raman and SER spectroscopy (Figure 3) CEES also known as half-mustard is essentially identical to HD except one of the chlorine end atoms is replaced by a hydrogen atom Again although not as toxic as HD CEES is a blister agent and dilute aqueous samples were prepared using appropriate safety equipment Both the Raman and SER spectra of CEES are similar and dominated by bands between 600 and 800 cm-1 These are associated with C-Cl and C-S stretching modes which are tentatively assigned to 648 and 747 cm-1 in the Raman spectra respectively The shoulder at 630 cm-1 the overlapped band at 660 cm-1 and the strong band at 692 cm-1 could also be due to these modes or their asymmetric counterparts It is worth noting that theoretical calculations assign the 692
Figure 2 A) SER and B) NR spectra of KCN Conditions A) 1 mgml in tap water 100 mW of 785 nm at sample 1-min acquisition time B) solid 300 mW of 785 nm 5-min All spectra are 8 cm-1 resolution
A
B
Figure 3 A) SER and B) NR spectra of CEES Conditions A) 1 vv (10 mgml) in MeOH 100 mW of 785 nm 1-min acquisition time B) neat 300 mW of 785 nm 5-min
A
B
Cl-CH2-CH2-S-CH2-CH3
A
B
Figure 4 A) SER and B) NR of HD Conditions A) 1mgml in tap water B) pure both 100 mW of 785 nm 1-min
Cl-CH2-CH2-S-CH2-CH2-Cl
SPIE-2003-5269 20
expected strong interaction between chlorine and silver and adds support to the assignment of this band to a C-Cl stretch Weaker overlapping bands occur at 670 692 and 724 cm-1 the latter possibly due to C-S stretching modes Again the alkane modes are apparent in the normal Raman spectra of HD but only a broad feature at 1300 to 1450 cm-1 suggests CH2 contributions in the SER spectrum Although the observed bands in the VX spectrum have not been assigned (Figure 4) a computer generated Raman spectrum29 predicts many of the same features with surprising accuracy and are used here Two intense bands at 460 and 530 cm-1 closely match predicted bands at 463 and 546 cm-1 assigned to a CH3-P=O bend and a PO2CS wag Three highly overlapped bands occur at 694 745 and 771 cm-1 matching predicted bands at 713 730 and 760 cm-1 The first
Table 2 Measured SER peak heights for the CN stretch at 2100 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
Figure 5 A) SERS and B) NR spectra of VX Conditions A) 1 vv (10 mgml) in MeOH B) pure sample both 100 mW of 785 nm 1-min acquisition time
A
B
two have been assigned to a C-S stretch and CH2bend respectively while the latter has been attributed to either a P-C stretch or an O-C-C stretch Although the 745 cm-1 band may alternatively be assigned to a C-S stretch based on the previous measurements of CEES and HD The relatively intense bands at 890 1106 1218 1445 and 1465 cm-1 also match predicted bands at 880 1108 1216 1440 and 1464 cm-1 that are assigned to a C-C stretch CH3 rock N-C3 stretch various C-H3 bends and C-H bends respectively Both the computer generated and the measured spectra contain numerous other less intense bands One is worth mentioning A unique band appears at 370 cm-1 that is predicted at 368 cm-1 and corresponds to an O-P=O bend The surface-enhanced Raman spectrum of VX is also rich with spectral features It has the unique low frequency band at 370 cm-1 as well as a second band at 380 cm-1 that is assigned tothe S-P-O bend predicted in the normal Raman spectrum at 388 cm-1 Based on the measured and predicted normal Raman spectra the following SERS assignments are given 460 cm-1 to the CH3-P=O bend 544 cm-1 to the PO2CS wag 738 cm-1 to a C-S stretch (based on arguments above) 890 cm-1
to a C-C stretch 1101 cm-1 to a CH3 rock and 1456
cm-1 to a C-H bend The ability of SERS to measure chemical agents in water containing real-world chemical interferents was tested by using the generic tap water described in Table 1 The ability to reproduce measurements was accomplished by preparing three separate water stock solutions which were used to prepare three sample solutions each of 1 mgmL KCN in the generic tap water The samples defined as vials 1-9 were then measured at 4 points per vial in a semi-random fashion such that errors associated with stock solution preparation and errors associated with instrument drift could be identified No trends were apparent that signified such systematic errors Each spectrum collected consisted of 20 averaged scans taking 16 seconds at 8 cm-1 resolution The laser power at the sample was measured periodically during the day and it ranged from 102 to 105 mW spacccc
spot Vial 1 Vial 2 Vial 41 13754 13004 128192 13519 12692 12909
1 3 13541 12721 126394 13462 12648 12651
ave 13569 12766 12755Vial 6 Vial 3 Vial 5
1 11586 14068 10757 2 11236 14402 115122 3 11376 14546 11559
4 10894 11763 11214ave 11273 13695 11261
Vial 7 Vial8 Vial 91 11198 15804 11023
2 11243 15716 88753 3 11562 15216 11218
4 11655 15084 11077ave 11415 15455 10548
CN AVG STDEV ERR1 mgmL pk ht 12526 1560 1245
stock solution
SPIE-2003-5269 21
The CN measurements consisting of 36 data points produced an average height of 125 for the 2100 cm-1 peak with a standard deviation of 156 or 125 (Table 2) The HD and VX measurements were performed precisely the same way (Tables 3 and 4) For HD the 624 cm-1 peak was used for analysis and it had an average height of 53 with a standard deviation of 068 or 129 while for VX the 544 cm-1 peak was used for analysis and it had an average height of 1051 with a standard deviation of 308 or 293 The greater error in the VX measurements can be somewhat attributed to Vial 4 which produced lower SER signal intensities But removing this vial from the data set changes the standard deviation to 233 only a modest improvement
spot Vial 1 Vial 2 Vial 41 525 609 4682 45 675 484
1 3 527 644 5094 559 807 669
ave 51525 68375 5325Vial 6 Vial 3 Vial 5
1 521 574 575 2 536 53 4492 3 509 508 451
4 565 594 379ave 53275 5515 4635
Vial 7 Vial 8 Vial 91 457 589 413
2 583 497 4273 3 544 505 4
4 528 5 509ave 528 52275 43725
HD AVG STDEV ERR1 mgmL pk ht 530 068 1291
stock solution
spot Vial 1 Vial 2 Vial 41 1464 1034 4672 1485 99 61
1 3 1491 989 5684 1041 777 553
ave 137025 9475 5495Vial 6 Vial 3 Vial 5
1 1058 942 1293 2 697 121 965
2 3 727 1206 11124 689 1254 746
ave 79275 1153 1029Vial 7 Vial 8 Vial 9
1 1127 783 163 2 1358 812 1614
3 3 1371 908 15124 1097 875 1397
ave 123825 8445 153825
VX AVG STDEV ERR1 mgmL pk ht 1051 308 2925
stock solution
4 CONCLUSIONS In this paper we examined the ability of surface-enhanced Raman spectroscopy to reproducibly measure CN HD and VX in tap water without chemical interference Both normal and surface-enhanced Raman spectra were examined to select unique bands suitable to identify and quantify these chemical agents For SER measurements the 2100 cm-1 C-N stretch was used for CN the 624 cm-1 C-Cl stretch was used for HD and the 544 cm-1 PO2CS wag was used for VX It was determined that 1 mgmL samples of each of these chemicals measured 36 times in glass vials coated with a silver-doped sol-gel reproduced measurements with standard deviations of 125 129 and 293 It was further found that the 10 chemicals added to simulate generic tap water did not interfere with or alter the SER spectra It should be noted that the concentrations used in this study were considerably greater than those required by the JSAWM program Current work involves improving SER sensitivity and designing sampling systems with better reproducibility This includes the development of fractal silver and gold structures within the sol-gel matrix and the development of chemically selective sol-gels
ACKNOWLEDGMENTS The authors are grateful for the support of the US Army (DAAD13-02-C-0015 Joint Service Agent Water Monitor program) The authors would also like to thank Janet Jensen Ronald Crosier and Kristina Gonser for helpful discussions
Table 4 Measured SER peak heights for the VX band at 544 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
Table 3 Measured SER peak heights for the HD band at 624 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
SPIE-2003-5269 22
REFERENCES 1 Jensen JL NC Wong W R Loerop SPIE 4775-03 (2001) 2 JSAWM Requirements at wwwsbccomapgeaarmymilRDAecbcrtPRODSERJSAWMjsawmhtml 3 Erickson B ldquoThe Chemical Weapons Convention Redefines Analytical Challengerdquo Anal Chem 70 397A (1998) 4 Johnston RL Hoefler CM Fargo JC and Moberley B ldquoThe Defense Nuclear Agencyrsquos
ChemicalBiochemical Weapons Agreements Verification Technology Research Development Test and Evaluation Program and its Requirements for On-Site Analysisrdquo AT-ONSITE 5-8 (1994)
5 Black RM Clarke RJ Read RW and Reid MT ldquoApplication of GC-MS and GC-tandem MS to the analysis of chemical warfare samples found to contain residues of the nerve agent sarin sulphur mustard and their degradation productsrdquo J Chrom 662 301-321 (1994)
6 USA Army Center for Health Promotion and Preventive Medicine (USACHPPM) Technical Guideline 230A 7 Hoffland LD Piffath RJ Bouck JB rdquoSpectral signatures of chemical agents and simulantsrdquo Optical
Engineering 24 982-984 (1985) 8 Braue EHJ Pannella MGrdquoCIRCLE CELL FT-IR Analysis of Chemical Warfare Agents in Aqueous Solutionsrdquo
App Spectrosc 44 1513-1520 (1990) 9 Tseng C-H Mann CK and Vickers TJ ldquoDetermination of Organics on Metal Surfaces by Raman
Spectroscopyrdquo App Spectrosc 47 1767-1771 (1993) 10 Staff Reporter ldquoUS Army Tests Lidar to Detect Biological Toxinsrdquo Photonic Spectra pg 50 December 1998 11 Christesen SD Raman cross sections of chemical agents and simulants App Spectrosc 42 318-321 (1988) 12 Weaver MJ Farquharson S Tadayyoni MA ldquoSurface-enhancement factors for Raman scattering at silver electrodesrdquo J Chem Phys 82 4867-4874 (1985) 13 Alak AM and Vo-Dinh T ldquoSERS of Organophosphorous Chemical Agentsrdquo Anal Chem 59 2149-2153 (1987) 14 Lee Y Farquharson S ldquoRapid chemical agent identification by SERSrdquo SPIE 4378 21-26 (2001) 15 Farquharson S Maksymiuk P Ong K Christesen S ldquoChemical agent identification by surface-enhanced Raman spectroscopyrdquo SPIE 4577 166-173 (2001) 16 Spencer KM Sylvia J Clauson S and Janni J ldquoSurface Enhanced Raman as a Water Monitor for Warfare
Agents in Waterrdquo SPIE 4577 158-165 (2001) 17 Tessier P Christesen S Ong K Clemente E Lenhoff A Kaler E Velev O ldquoOn-line spectroscopic characterization of sodium cyanide with nanostructured gold surface-enhanced Raman spectroscopy substratesrdquo
App Spectrosc 56 1524-1530 (2002) 18 Farquharson S W W Smith S Elliott and J F Sperry Rapid biological agent identification by surface- enhanced Raman spectroscopy SPIE 3855110-116 (1999) 19 Farquharson S WW Smith YH Lee S Elliott and J F Sperry Detection of biological signatures A
comparison of electrolytic and metal-doped sol-gel surface-enhanced Raman media SPIE 4575 62-72 (2002) 20 Guzelian AA Sylvia J Janni J Clauson S and Spencer KM ldquoSERS of whole cell bacteria and trace levels of biological moleculesrdquo SPIE 4577 182-192 (2001) 21 Shende C Inscore F Gift A Maksymiuk P Farquharson S ldquoRapid dipicolinic acid extraction from Bacillus spores detected by surface-enhanced Raman spectroscopyrdquo App Spectrosc 58 accepted 22 Lee Y and Farquharson S ldquoSERS Sample Vials Based on Sol-Gel Process for Trace Pesticide Analysisrdquo SPIE
4206 140-146 (2000) 23 Farquharson S and Lee Y ldquoTrace Drug Analysis by SERSrdquo SPIE 4200-16 (2000) 24 Lee Y Farquharson S and Rainey P M Surface-Enhanced Raman Sensor for Trace Chemical Detection in
Water SPIE 3857 76-84 (1999) 25 Lee Y Farquharson S Kwong H and Shahriari M ldquoSol-Gel Chemical Sensor for Surface-Enhanced Raman
Spectroscopyrdquo SPIE 3537 252-260 (1998) 26 Farquharson S A Gift P Maksymiuk F Inscore and W Smith ldquopH dependence of methyl phosphonic acid
dipicolinic acid and cyanide by surface-enhanced Raman spectroscopyrdquo SPIE 5269-19 (2003) 27 Lide DR Ed Handbook of Chemistry and Physics CRC Press 77th Ed (1996-7) 28 Sosa C RJ Bartlett K KuBulat and WB Person ldquoA theoretical study of harmonic vibrational frequencies and
infrared intensities of XCH2CH2SCH2CH2X and XCH2CH2SH (= H Cl)rdquo J Phys Chem 93 577-588 (1993) 29 Hameka HF and JO Jenson ldquoComputer-generated predictions of the structure and the IR and Raman spectra of
VXrdquo ERDEC-TR-065 May 1993
- Chemagents Appendicespdf
-
- SPIE2001-4575-Bioagent-SERSapdf
-
- Detection of bioagent signatures A comparison of electrolytic and metal-doped sol-gel surface-enhanced Raman media
- Stuart Farquharson Wayne Smith and Yuan Lee
-
- Real-Time Analyzers 87 Church Street East Hartford CT 06108
-
- ABSTRACT
- 4 CONCLUSIONS
- 5 ACKNOWLEDGEMENTS
- 6 REFERENCES
-
169
polar but a high electron density in the cylindrical π cloud around the carbon-carbon triple bond allows testing the selectivity of the weakly polar-negative and weakly polar-positive sol-gels As Figure 3 illustrates PABA passes through the polar sol-gel and is enhanced by either the silver or gold particles At 1 mgml the concentration of neutral PABA is ca 20 times that of the ionized form (pKa = 48) For electropositive silver the PABA anion is expected to interact through the carboxylate group and the associated vibrational modes are expected to dominate the spectrum Conversely for electronegative gold either form of PABA is expected to interact through the amine group The clear differences in our spectra support this expectation Furthermore bands at 840 and 1405 cm-1 assigned to a COO- bend and stretch respectively are significantly more intense for silver than gold Additional bands at 1140 and 1195 cm-1 are assigned to CH bending modes while bands at 1450 1500 and 1605 cm-1 are assigned to ring vibrational modes A very similar SER spectrum for PABA on a silver-coated alumina substrate has previously been reported with similar assignments19 For the gold-doped sol-gel new bands appear at 690 1355 and 1585 cm-1 The first band is assigned to a ring-H bending mode the second band to a ring-N- stretching mode and the third band to a possible NH2 scissors mode or ring mode The second band is not observed in the normal Raman spectrum but infrared bands occur at this frequency for aromatic ring-secondary amine stretching modes The scissors mode occurs at this frequency in Raman spectra for several chemicals but is absent in the PABA Raman spectrum Alternatively this mode may be the1600 cm-1 ring mode that has been shifted by the gold interaction Again a very similar SER spectrum of PABA has been reported but surprisingly using silver (colloids)2021 not gold as the enhancement medium These researchers also assumed the primary interaction of PABA with silver was through the carboxylate anion and made assignments accordingly For example they assigned the 1359 cm-1 to a COO- stretch not to the amine group as we have They also favor the ring stretching mode assignment for the 1582 cm-1 band Finally it should be said that other researchers have argued that the most dominant band in the SER spectra at 1450 cm-1 a ring vibration mode suggests that PABA lies flat on the surface and the π-orbitals dominate the surface interaction22
Figure 3 SER spectra of A) PABA using polar-negative and B) polar-positive sol-gels and C) PA using weakly polar-negative and D) weakly polar-positive sol-gels PABA is 1 mgmL PA is 1 vv Spectral conditions 75 mw 1064 nm 100 scans (15 min) 8 cm-1 resolution Non-polar PA passes through the non-polar sol-gels and is also enhanced by both metals The spectra are easily understood For electropositive silver PA interacts through the cylindrical triple bond π electron cloud and a -CequivC- doublet occurs near 2000 cm-1 The interaction is reasonably strong since this band appears at 2112 cm-1 in the normal Raman spectrum For electronegative gold this interaction is unlikely and only very weak bands occur near 2000 cm-1 The remaining bands are at 1000 cm-1 1200 cm-1 doublet and 1595 cm-1 all appear in the normal Raman spectra at virtually the same frequencies and are assigned to the symmetric ring-breathing mode CH bending modes and the trigonal ring-breathing mode respectively The polarnon-polar selectivity of the polar-negative and weakly polar-negative sol-gels was tested by adding a 11 molar mixture of PABA and PA The selective enhancement is quite good (Figure 4) The spectrum obtained using the polar sol-gel represents 78 PABA and 22 PA while the spectrum obtained using the weakly polar sol-gel represents 9 PABA and 91 PA The band peak intensities at 2000 cm-1 for PA and 1450 cm-1 for PABA were used for these calculations and are expanded in Figure 4 for clarity
C A
Wavenumber (∆cm-1) Wavenumber (∆cm-1)
2NH COOH
D B
C CH
Proc SPIE Vol 4577
170
Figure 4 SERS of 11 MM of PABA and PA in A) polar-negative and C) weakly polar-negative sol-gels The lower traces compare the pure chemicals B) 1 mgml PABA in polar-negative sol-gel and D) 1 PA in weakly polar-negative sol-gel while the insets magnify the minority species for clarity (x5 in A and x10 in B) Spectral conditions as in Figure 3 Following this development of selective sol-gels that maintained SER activity we measured cyanide and MPA (Figure 5) Not surprisingly the best sensitivity for both hydrolysis products was obtained using the polar-negative sol-gel The interaction of the cyanide anion with the silver surface is sufficient to shift the CequivN stretch observed at 2080 cm-1 in the normal Raman spectrum to 2145 cm-1 in the surface-enhanced Raman spectrum Furthermore the band is substantially broadened This anion has been extensively studied by electrolytic SERS and this shift and broadening have been attributed to the formation of a tetrahedral Ag(CN)3
2- surface structure23 Figure 5 Surface-enhanced (upper traces) and normal Raman spectra (bottom traces) of A) CN- and B) MPA in silver-doped TMOS SERS conditions as in Figure 3 and 1 mgmL Note MPA yields two distinct spectra for neutral (top) and acidic pH (middle) The normal Raman spectra employed pure powders 500 scans and 900 mW of 1064 nm SER measurements of MPA with the polar-negative sol-gel yielded two unique spectral signatures that depended on solution pH (Figure 5) For more neutral solutions the P-C stretch of MPA at 762 cm-1 dominates and the CH2 stretch at 2922 cm-1 appears The SN is sufficiently high that the anti-Stokes Raman shift at -762 cm-1 is observed For deprotonated MPA an oxygen-surface mode appears at 325 cm-1 (as well as its anti-Stokes complement) suggesting a strong interaction This results in substantial enhancement of the P-O-C mode at 1051 cm-1 (upper trace) Others report that this mode dominates the infrared spectra of nerve agents measured in water8 Comparison of the two spectra suggests the following molecule-to-surface orientations The appearance of the oxygen-surface and P-O-C modes in the upper spectrum of Figure 5B indicates that the tetrahedral molecule interacts with the silver surface through the deprotonated oxygen and is oriented end-on The
A B
Wavenumber (∆cm-1)
C A
Wavenumber (∆cm-1) Wavenumber (∆cm-1)
D B
Wavenumber (∆cm-1)
Proc SPIE Vol 4577
171
dominance of the P-C and the CH2 stretches and the disappearance of the P-O-C mode in the upper spectrum suggest the molecule is oriented with the methyl-hydroxyl-hydroxyl face towards the surface However considerably more research must be performed to verify these points Both chemical forms were measured as a function of concentration to determine the onset of monolayer coverage and estimate expected detection limits (Figure 6) Below monolayer coverage the signal to concentration dependence should be linear and the SN of any spectral measurement in this range can be used to predict the detection limit In the spectra presented here the peak height was used as the signal while the noise as root-mean-squared (RMS) was measured between 4400-4600 cm-1 Since noise is distributed evenly throughout the spectrum when transformed this region was used since it does not have any contributions from signals or baseline offsets Figure 6 shows a series of spectra for MPA along with a plot of the 1050 cm-1 band intensity (the noise was identical for this series) as a function of sample concentration A clear discontinuity appears in the vicinity of 01 mgmL (19 ppm) indicating the onset of monolayer coverage A detection limit defined as a SN of 3 was calculated for the 01 and 005gmL samples at 24x10-4 and 25x10-4 gL respectively A more modest detection limit of 101x10-4 gL was obtained using the 760 cm-1 band in the second series of concentration measurements These detection limits correspond to 46 and 190 ppb for a 3-min scan and 75 mW of incident laser power
Figure 6 A) Concentration dependence of MPA SERS measured in silver-doped TMOS) B) Concentrations are 001 005 01 05 1 gL (188 94 188 94 188 ppm) I760 series (bull) and I1050 series (∆)
Enhancement factors for cyanide and methyl phosphonic acid can be determined by estimating the number of molecules contributing to the surface-enhanced and normal Raman spectra The enhancement factor EF can be defined by the following equation
EF = (ISERSINR)bull(MNRMSERS) bull(PNRPSERS) bull(TNRTSERS)12
where I is the spectral band intensity M is the sample mass P is the incident laser power and T is the measurement time (or number of scans) for the two measurements For the normal Raman spectra a cylindrical scattering volume is assumed based on the laser area (28x10-7m2 6x10-4m diameter spot) and the penetration depth (1x10-3 m)24 The density of KCN and MPA as powders were measured at 0572 and 0516 gcm3 indicating that 16x10-4 and 144x10-4 g produced the normal Raman signals in Figure 5 respectively The number of molecules contributing to the SER spectra are those on the silver particles embedded in the sol-gel The total silver surface area can be determined from the average particle size concentration and the scattering volume Previous scanning electron micrographs showed the average silver particle size to be 40 nm in diameter (335x10-23m3)17 The silver concentration is 012M based on the reactant molar concentrations and dilution factors And the scattering volume is 76x10-
11m3 again based on a cylindrical scattering volume defined by a laser area of 28x10-7m2 and a sol-gel thickness of 27x10-
4m This volume contains 123x10-6g of silver equivalent to 35x109 silver particles with a collective surface area of 18x10-
5m2 However it may reasonably be assumed that at least half of the silver surface is in contact with the sol-gel matrix and unavailable for analyte interaction If we assume monolayer coverage and that each CN molecule occupies 15x10-20m2 then approximately 62x1014 molecules or 27x10-8g of CN contribute to the SER spectrum (20x10-19m2 46x1013 molecules 74x10-9g for MPA) Accordingly the EF for cyanide equals 48x104 ((180599) bull(16x10-427x10-8) bull(90075) bull(500100)12) The EF for MPA is considerably higher at 87x106 ((60326) bull(144x10-474x10-9) bull(90075) bull(500200)12)
0
20
40
60
80
100
120
140
0 02 04 06 08 1 12[MPA] (mgmL)
I (76
0)
0
100
200
300
400
500
600
I (10
50)
Wavenumber (∆cm-1)
A B
Proc SPIE Vol 4577
172
4 CONCLUSIONS Here we present for the first time surface-enhanced Raman spectra of Tabun and Sarin obtained using silver-doped sol-gels However the inferior enhancement suggested employing SER-active sol-gels varying in composition to improve both detection limits and selectivity of the target analytes To this end we successfully demonstrated the capabilities of four sol-gels that select for 1) polar-positive 2) polar-negative 3) weakly polar-positive and 4) weakly polar-negative chemical species p-aminobenzoic acid was used to show that silver could be used to attract polar-negative chemicals or functional groups (carboxylate anion) and that gold could be used to attract polar-positive chemicals (amine) while a mixture of p-aminobenzoic acid and phenylacetylene was used to show that tetramethyl orthosilicate preferentially solvates polar chemicals and that tetraethyl orthosilicate preferentially solvates weakly polar chemicals This increased sample control was applied to cyanide and methyl phosphonic acid two hydrolysis products of chemical warfare agents Exceptional results were obtained for methyl phosphonic acid allowing measurement of 1x10-2 gL for a 3-min scan and 75 mW of incident laser power with an estimated detection limit of 245x10-4 gL and an enhancement factor of 87x106 However this detection limit is 76 times less sensitive than required for the JSAWM (32x10-6gL for the G-agents) Finally we note that the measurements performed here employed an FT-Raman spectrometer This instrumentation was chosen over dispersive Raman instrumentation for the high wavelength accuracy afforded by the HeNe reference laser (Connes Advantage25) which would allow reliable spectral subtraction matching of observed spectra to stored library spectra and confident use of chemometric approaches Such data analysis is likely to be required to identify the chemical agents as well as distinguish hydrolysis products However this instrumentation which employs 1064 nm excitation and InGaAs detection sacrifices sensitivity We believe that the measurement sensitivity can be improved by at least two orders of magnitude by using shorter laser excitation wavelengths (eg 532 nm) This would provide more efficient Raman scattering (fourth power dependence on laser excitation wavelength) more efficient generation of plasmon modes and allow using more efficient detector material (Si vs InGaAs) These modifications are underway
5 ACKNOWLEDGEMENTS The authors would like to thank Dr R Yin and J Jensen of the US Army for supporting this work (Contract Number DAAD13-01-C-0019) They would also like to thank Advanced Fuel Research for making their laboratory facilities available
6 REFERENCES 1 Tu Anthony ldquoOverview of Sarin Terrorist Incidents in Japan in 1994 and 1995rdquo 6th CBW Protection Symposium
Stockholm Sweden 10-15 May 1998 2 Jensen JL NC Wong W R Loerop SPIE 4775-03 (2001) 3 JSAWM Requirements at wwwsbccomapgeaarmymilRDAecbcrtPRODSERJSAWMjsawmhtml 4 ldquoThe Chemical Weapons Convention Redefines Analytical Challengerdquo Analytical Chemistry News amp Features June 1
397A (1998) 5 Johnston RL Hoefler CM Fargo JC and Moberley B ldquoThe Defense Nuclear Agencyrsquos ChemicalBiochemical
Weapons Agreements Verification Technology Research Development Test and Evaluation Program and its Requirements for On-Site Analysisrdquo AT-ONSITE 5-8 (1994)
6 Black RM Clarke RJ Read RW and Reid MT ldquoApplication of gas chromatography-mass spectrometry and gas chromatography-tandem mass spectrometry to the analysis of chemical warfare samples found to contain residues of the nerve agent sarin sulphur mustard and their degradation productsrdquo J Chromatography 662 301-321 (1994)
7 Hoffland LD Piffath RJ Bouck JBrdquoSpectral signatures of chemical agents and simulantsrdquo Optical Engineering 24 982-984 (1985)
8 Braue EHJ Pannella MGrdquoCIRCLE CELL FT-IR Analysis of Chemical Warfare Agents in Aqueous Solutionsrdquo Applied Spectroscopy 44 1513-1520 (1990)
9 Tseng C-H Mann CK and Vickers TJ ldquoDetermination of Organics on Metal Surfaces by Raman Spectroscopyrdquo Applied Spectroscopy 47 1767-1771 (1993)
10 Staff Reporter ldquoUS Army Tests Lidar to Detect Biological Toxinsrdquo Photonic Spectra pg 50 December 1998
Proc SPIE Vol 4577
Proc SPIE Vol 4577
173
11 Christesen SD Raman cross sections of chemical agents and simulants Applied Spectroscopy 42 318-321 (1988) 12 Alak AM and Vo-Dinh T ldquoSERS of Organophosphorous Chemical Agentsrdquo Analytical Chemistry 59 2149-2153
(1987) 13 Norrod KL Sudnik LM Rousell D and Rowlen KL ldquoQuantitative Comparison of Five SERS Substrates
Sensitivity and Detection Limitrdquo Applied Spectroscopy 51 994-1001 (1997) 14 Lee Y and Farquharson S ldquoSERS Sample Vials Based on Sol-Gel Process for Trace Pesticide Analysisrdquo SPIE 4206
140-146 (2000) 15 Farquharson S and Lee Y ldquoTrace Drug Analysis by Surface-Enhanced Raman Spectroscopyrdquo SPIE 4200-16 (2000) 16 Lee Y Farquharson S and Rainey P M Surface-Enhanced Raman Sensor for Trace Chemical Detection in Water
SPIE 3857 76-84 (1999) 17 Lee Y Farquharson S Kwong H and Shahriari M ldquoSurface-Enhanced Raman Sensor for Surface-Enhanced Raman
Spectroscopyrdquo SPIE 3537 252-260 (1998) 18 Farquharson S Smith W Carangelo R C and Brouillette C ldquoIndustrial Raman Providing Easy Immediate Cost
Effective Chemical Analysis Anywhererdquo SPIE 3859 14-23 (1999) 19 Narayanan VA JM Bello JD Stokes and T Vo-Dinh Analusis 19 307-310 (1991) 20 Laserna JJ E L Torres and JD Winefordner Analytica Chemica Acta 469-480 (1987) 21 Torres EL and JD Winefordner Analytical Chemistry 59 1626-1632 (1987) 22 Suh JS DP DiLella M Moskovits J Phys Chem 87 1540-1544 (1983) 23 Benner RE R Dornhaus R Chang and BL Laube Correlations in the Raman spectra of cyanide complexes adsorbed
at silver electrodes with voltammograms Surface Science 101 341 (1980) 24 Chase D B and JF Rabolt Fourier Transform Raman Spectroscopy Acad Press Ch1 p 131 (1994) 25 Connes J Rev Opt Theor Instrum 40 45 (1961)
APPLIED SPECTROSCOPY 351
focusing the 488 nm laser beam 2 mm deep into thebulk of the crystals to avoid contributions from potentialdepletion layers As shown in Fig 1b the dependence ofthe Raman shift on the degree of deuteration is almostperfectly linear and ts very well with D 5 22684cmR1 24526 where D is the degree of deuteration (in )and R is the spectral mean of the PO4 vibration in cm21A linear correlation coef cient of 0998 indicates an ex-cellent linear dependence of the Raman peak shift withdegree of deuteration This result shows that the shift ofthe PO 4 peak is simply caused by the linear increase inatomic mass due to isotope substitution which decreasesthe length of hydrogen-like bonds
This excellent linear dependence allows us to map thepro le of the DH exchange layer at the surface of DKDPcrystals by acquiring Raman spectra and determining theposition of the PO 4 peak for various depths This methodis preferable over other methods such as determining thestrength of the OD vibration directly (eg at 715 cm21)because the position of the most intense peak in the Ra-man spectrum can be measured more precisely than theintensity of some of the weakest peaks in the spectrumThis is demonstrated in Fig 2 where depth-dependentRaman spectra (Fig 2a) and the resulting exchange layerpro les for two DKDP crystals are shown (Fig 2b) Thespectra in Fig 2a were obtained from a depth scan of aDKDP crystal with 75 degree of deuteration in thebulk grown at 45 8C The spectra start out as DKDP with30 deuteration close to the surface and approach thebulk DKDP spectrum within a few micrometers of depthThe fact that the relative degree of deuteration does notextend to 0 D is due to the limited depth resolution ofthe Raman microprobe which averages over 4 mm indepth Figure 2b depicts the resulting DH exchange layerpro les for this and a second crystal grown at 63 8Crespectively Both crystals had the same exposure to am-bient conditions and their main difference is the temper-ature at which they were grown The different exchangelayer pro les indicate that crystals grown at differenttemperatures have differen t proton conductiv ities 13
which leads to a difference in their rate of deuteriumdepletion The parameters controlling this behavior arecurrently the objective of a detailed study the results ofwhich will be reported elsewhere
CONCLUSION
In conclusion we have shown that the shift of the to-tally symmetric PO 4 stretch mode in the Raman spectrumof DKDP crystals scales linearly with degree of deuter-ation This allows us to correlate Raman peak positionsto deuteration levels in these crystals We have presenteda new technique to determine DH diffusion pro les inDKDP frequency conversion crystals based on micro-Ra-man spectroscopy This technique is fast inexpensiveand works under various environmental conditionswhich will allow us to better understand and control deu-terium depletion in DKDP crystals
ACKNOWLEDGMENTS
We would like to thank M Runkel for rst discovering DKDP crack-ing R Floyd for providing DKDP crystals and L Chase and A Burn-ham for their support and helpful discussions This work was performedunder the auspices of the US Department of Energy by the University
of California Lawrence Livermore National Laboratory under contract W-7405-Eng-48
1 J J De Yoreo A K Burnham and P K Whitman Int Mat Rev47 113 (2002)
2 C E Barker R A Sacks B M Van Wonterghern J A Caird JR Murray J H Campbell K Kyle R B Ehrlich and N DNielsen Proc SPIE-Int Soc Opt Eng 2633 501 (1995)
3 T Suratwala paper to be published4 Cleveland Crystals Inc httpwwwclevelandcrystalscom
KDPshtmltable5 E A Popova I T Savatinova and I A Velichko Sov Phys Solid
State 12 1543 (1971)6 I P Kaminow R C C Leite and S P S Porto J Phys Chem
Solids 26 2085 (1965)7 J A Subramony B J Marquardt J W Macklin and B Kahr
Chem Mat 11 1312 (1999)8 H Tanaka M Tokunaga and I Tatsuzaki Solid State Commun
49 153 (1984)9 R J Nelmes G M Meyer and J E Tibballs J Phys C 15 59
(1982)10 M A Yakshin D W Kim Y S Kim Y Y Broslavets O E
Sidoryuk and S Goldstein Laser Physics 7 941 (1997)11 I Takenaga Y Tominaga S Endo and M Kobayashi Solid State
Commun 84 931 (1992)12 C Krenn personal communication13 M Sharon and A K Kalia J Solid State Chem 21 171 (1977)
Rapid Dipicolinic Acid Extractionfrom Bacillus Spores Detectedby Surface-EnhancedRaman Spectroscopy
STUART FARQUHARSON ALAND GIFT PAUL MAKSYMIUK andFRANK E INSCOREReal-Time Analyzers Inc East Hartford Connecticut06108
Index Headings Dipicolinic acid Bacillus spores Anthrax Surface-enhanced Raman spectroscopy
INTRODUCTION
The anxiety caused by the distribution of anthrax en-dospores through the US postal system in October 2001was exacerbated by the long time required for positiveidenti cation of the Bacillus anthracis spores and the un-known extent of their distribution Since that time manymethods capable of rapid eld analysis have been inves-tigated to augment or replace the laboratory method ofgrowing microorganisms in culture media which takesdays to perform12 Prominent among these approachesare polymerase chain reactions (PCR)3 immunoassaysand detection of calcium dipicolinate as a biochemicalsignature PCR employs primers to separate organism-
Received 10 October 2003 accepted 14 November 2003 Author to whom correspondence should be sent
352 Volume 58 Number 3 2004
speci c nucleic acid sequences (eg capsular protein en-coding gene for Bacillus anthracis)4 and polymerases toamplify the segment until it is detectable Recently am-pli cation times have been substantially reduced andcomplete analysis can now be performed in an hour orless Immunoassay methods are also being developed thatuse competitive binding of the bioagent (as an antigen)and its labeled conjugate for a limited number of anti-bodies Although analyses can be performed in under 30minutes a well-de ned anthrax antigen has not yet beenidenti ed5ndash7 and consequently the false-positive rate isunacceptably high8
A number of other methods are being developed witha focus on the detection of calcium dipicolinate (CaDPA)and its derivatives as a B anthracis signature This is sobecause only spore-forming bacteria contain CaDPA andthe most common potentially interfering spores such aspollen and mold spores do not Relatively fast methodshave been developed to chemically extract CaDPA andthen detect it directly by uorescence9 or indirectly byluminescence1011 In the latter case hot dodecylamine(DDA) has been used to extract dipicolinic acid (DPA)and terbium has been utilized to form a highly lumines-cent DPA complex11 Although measurements have beenperformed in as little as ve minutes it was found thatas many as three concentration-dependent complexes canform each with different lifetimes This coupled withthe fact that the Tb31 cation produces the same lumines-cence spectrum makes determinations of low spore con-centrations problematic
It has been long known that Raman spectra of Bacillispores are dominated by bands associated with CaDPA12
and that these spectra may provide a suitable anthrax sig-nature at the genus level13 Since that time considerableimprovements in Raman instrumentation have led to lab-oratory measurements of single Bacilli spores14 and to eld measurements of spores captured from a mail-sort-ing system15 However the single spore measurementsrequired complex instrumentation that is not ruggedwhile the eld measurements required milligrams of sam-ple Furthermore the Raman spectra of both measure-ments contained uorescence contributions that would in-crease uncertainty in quanti cation
In related research we demonstrated that nanogramquantities of DPA could be detected by uorescence-freesurface-enhanced Raman spectroscopy (SERS)16 We alsodemonstrated that microliter volumes of chemicals canbe detected by SERS using metal-doped sol-gel-packedglass capillaries17 Towards the goal of developing a rap-id eld SERS-based anthrax spore detector we havecombined our previous research and we now report thatDPA can be extracted from a 10 mg B cereus spore sam-ple using DDA in 1 minute and can be detected by SERSin an additional 1 minute
EXPERIMENTAL
Dipicolinic acid (26-pyridinedicarboxylic acid DPA)and dodecylamine (DDA) were used as received fromSigma-Aldrich (Milwaukee WI) Lyophilized B cereusspores prepared according to the literature13 were sup-plied by the University of Rhode Island and used as re-ceived Multiple particles approximately 01 mm3 each
were separated and weighed at 5 to 15 mg representing05 to 15 million spores The sample masses were con-sistent with a previous determination of spore density at0081 gmL that indicated a high degree of entrained air
All chemicals used to prepare the silver-doped sol-gelcoated capillaries were also obtained and used as receivedfrom Sigma-Aldrich According to previously publishedprocedures17 two precursor solutions were preparedmixed and then drawn into 1-mm-diameter glass capil-laries The silver amine precursor consisted of a 51 vvratio of 1 N AgNO 3 to 28 NH3OH while the alkoxideprecursor consisted of a 21 vv ratio of methanol to te-tramethyl orthosilicate The alkoxide precursors weremixed with silver amine precursor in an 81 vv ratioApproximately 015 mL was drawn into the capillarycoating a 15-mm length After sol-gel formation the in-corporated silver ions were reduced with dilute sodiumborohydride which was followed by a water wash to re-move residual reducing agent
A 100 mL drop of a 50 mM DDA solution in ethanolpre-heated to 78 8C was added to each of the B cereusparticles to digest the spore coat After 1 minute the re-sultant solution was drawn into a SER-active capillarythat was immediately xed horizontally to an XY posi-tioning stage (Conix Research Spring eld OR) just in-side the focal point of an f 07 aspheric lens The lensfocused the beam into the sample and collected the scat-tered radiation back along the same axis A dichroic lter(Omega Optical Brattleborough VT) was used to re ectthe excitation laser to the lens and pass the Raman scat-tered radiation collected by the lens An f 2 achromatwas used to collimate the laser beam exiting a 200-mm-core-diameter source ber optic while a second f 2 ach-romat was used to focus the scattered radiation into a 365mm ber optic (Spectran Avon CT) A short-pass lterwas placed in the excitation beam path to block the sil-icon Raman scattering generated in the source ber fromre ecting off sampling optics and reaching the detectorA long-pass lter was placed in the collection beam pathto block the sample Rayleigh scattering from reachingthe detector A 785 nm diode laser (Process InstrumentsInc model 785-600 Salt Lake City UT) was used todeliver 100 to 150 mW of power to the sample A Fouriertransform Raman spectrometer (Real-Time Analyzersmodel IRA-785 East Hartford CT) and a silicon photo-avalanche detector (Perkin Elmer model C30902S Stam-ford CT) were used to acquire the SER spectra
RESULTS AND DISCUSSION
As an initial experiment the SER spectrum of 1 gLof DPA in water was measured using the newly devel-oped silver-doped sol-gel-coated capillaries (Fig 1A) Atthis concentration a high signal-to-noise ratio (SN) isobtained in 1 min In fact a reasonable spectrum is ob-tained in the same time frame for 1 mgL (Fig 1B) TheSER spectra are reasonably similar to the normal Raman(NR) spectrum obtained for a saturated solution of DPAin 1 N KOH (Fig 1C) and the following band shifts areobserved (NR to SER) 647 to 657 cm21 817 to 815cm21 998 to 1008 cm21 1384 to 1382 cm21 1434 to1428 cm21 and 1569 to 1567 cm21 Many of these bandshave been previously assigned1213 such as 998 cm21 to
APPLIED SPECTROSCOPY 353
FIG 1 SERS of DPA in water using silver-doped sol-gel-coated glasscapillary for (A) 1 gL and (B) 1 mgL (C ) NR of saturated DPA in 1N KOH in a glass capillary Spectral conditions (A) and (B) 150 mWof 785 nm 1-min acquisition time (C) 450 mW of 785 nm 5-minacquisition time both 8 cm 21 resolution
FIG 2 SERS of DPA extracted from 10 mg B cereus particle using100 mL of 50 mM hot DDA acquired in (A) 1 minute and (B) 2 seconds(C ) Attempted SERS of 50 mM hot DDA in ethanol using silver-dopedsol-gel-coated glass capillary acquired in 1 min Spectral conditions150 mW of 785 nm 8 cm21 resolution
the symmetric ring stretch 1384 cm21 to the OndashCndashOsymmetric stretch 1428 cm21 to the symmetric ring CndashH bend and 1569 cm21 to the asymmetric OndashCndashOstretch
The rst B cereus samples consisted of 2 mg of sporesin 2 mL of 5 mM hot DDA The samples were main-tained at 78 8C for 40 min and while hot approximately10 mL was drawn into a SER-active capillary Since spec-tra of DPA were obtained for these initial samples small-er spore masses higher DDA concentrations and shorterheating periods were examined In due course it wasfound that 10 mg of spores could be digested by 100 mLof 50 mM hot DDA in one minute and detected (Fig2A) In fact the signal was suf ciently intense that it canbe observed in as little as two seconds (Fig 2B) Theamount of DPA that was extracted was estimated to bebetween 5 and 10 mgL by comparing the signal intensityof the 1008 cm21 band to that measured for DPA in waterThis is consistent with previous research that found thatthe majority of the DPA is extracted from spores usingDDA11 and that B cereus spores contain approximately10 DPA by weight18 The SN of 127 for the 1008 cm21
band in the 1-minute SER spectrum suggests a limit ofdetection of approximately 250 ng of B cereus sporesbased on a SN of 3 Finally it should be noted that DDAdid not produce a detectable SER spectrum as shown inFig 2C
CONCLUSION
We have demonstrated that by combining rapid extrac-tion of dipicolinic acid from Bacillus cereus spores withchemical identi cation by surface-enhanced Raman spec-troscopy as little as 10 mg of spores can be detected Infact the entire measurement from the time of adding hotdodecylamine to the spores to the time when the dipi-colinic acid SER spectrum is acquired and analyzedcould be performed in less than two minutes The abilityof this method to distinguish between spore-forming bac-teria such as Bacillus anthracis and non-DPA containingpowders could help prevent costly shutdowns associated
with the appearance of suspicious material or intentionalmailing of common substances as an anthrax hoax Thismethod could also prove useful in detecting the locationof anthrax endospores in mail distribution facilities if an-other veri ed attack should occur
Research continues to fully characterize the surface-enhanced Raman spectroscopy signal intensities as afunction of sample concentration and to explore otherextractants that do not require the use of elevated tem-perature
ACKNOWLEDGMENTS
The authors are grateful for the support of the National Science Foun-dation (DMI-0296116 and DMI-0215819) and the US Army(DAAD13-02-C-0015 Joint Service Agent Water Monitor program)The authors are indebted to Chetan Shende for preparing the sol-gelcapillaries The authors also thank James Gillespie Nicholas Fell andAugustus Fountain for providing important background informationMark Farquharson for laboratory support and Professor Jay Sperry ofthe University of Rhode Island for supplying B cereus spores
1 V A Pasechnik C C Shone and P Hambleton Bioseparations 3267 (1993)
2 P J Jackson M E Hugh-Jones D M Adair G Green K K HillC R Kuske L M Grinberg F A Abramova and P Keim ProcNatl Acad Sci USA 95 1224 (1998)
3 B R Glick and J J Pasternak Molecular Biology Principles andApplications of Recombinant DNA (ASM Press Washington DC1994)
4 C A Bell J R Uhl T L Had eld J C David R F Meyer TF Smith and F R Cockerill III J Clin Microbiol 40 2897(2002)
5 D L Gatto-Menking H Yu J G Bruno M T Goode M Millerand A W Zulich Biosens Bioelectron 10 501 (1995)
6 J J Quinlan and P M Foegeding J Rapid Methods AutomationMicrobiol 6 1 (1998)
7 A A Hindle and E A H Hall Analyst (Cambridge UK) 1241599 (1999)
8 M S Ascher US Department of Health amp Human Services(httpwwwhhsgovophppresentationsAscherdoc)
9 R Nudelman B V Bronk and S Efrima Appl Spectrosc 54 445(2000)
10 D L Rosen C Sharpless and L B McBrown Anal Chem 691082 (1997)
354 Volume 58 Number 3 2004
11 P M Pellegrino N F Fell Jr and J B Gillespie Anal ChimActa 455 167 (2002)
12 W H Woodruff T G Spiro and C Gilvarg Biochem BiophysRes Commun 58 197 (1974)
13 E Ghiamati R S Manoharan W H Nelson and J F SperryAppl Spectrosc 46 357 (1992)
14 A P Esposito C E Talley T Huser C W Hollars C M Schal-dach and S M Lane Appl Spectrosc 57 868 (2003)
15 S Farquharson L Grigely V Khitrov W W Smith J F Sperryand G Fenerty J Raman Spectrosc paper accep ted (2003)
16 S Farquharson W W Smith S Elliott and J F Sperry SPIE-IntSoc Opt Eng 3855 110 (1999)
17 S Farquharson and P Maksymiuk Appl Spectrosc 57 479(2003)
18 F W Janssen A J Lund and L E Anderson Science (Washing-ton DC) 127 26 (1958)
SPIE -2003-5269 117
pH dependence of methyl phosphonic acid dipicolinic acid and cyanide by surface-enhanced Raman spectroscopy
Stuart Farquharson Alan Gift Paul Maksymiuk Frank Inscore and Wayne Smith
Real-Time Analyzers 87 Church Street East Hartford CT 06108
ABSTRACT US and Coalition forces fighting terrorism in Afghanistan and Iraq must consider a wide range of attack scenarios in addition to car bombings Among these is the intentional poisoning of water supplies to obstruct military operations To counter such attacks the military is developing portable analyzers that can identify and quantify potential chemical agents in water supplies at microgram per liter concentrations within 10 minutes To aid this effort we have been investigating the value of a surface-enhanced Raman spectroscopy based portable analyzer In particular we have been developing silver-doped sol-gels to generate SER spectra of chemical agents and their hydrolysis products Here we present SER spectra of methyl phosphonic acid and cyanide as a function of pH an important factor affecting quantitation measurements which to our knowledge has not been examined In addition dipicolinic acid a chemical signature associated with anthrax-causing spores is also presented Keywords Chemical warfare agents agent detection agent hydrolysis SERS Raman spectroscopy homeland security
1 INTRODUCTION In the past decade the Unites States and its allies have been challenged by a different kind of warfare exemplified by the terrorist attacks of September 11 2001 Suicide bombings and the use of chemical agents are the norm and military personnel must consider a wide range of attack scenarios Among these is the intentional poisoning of water supplies to obstruct military operations in Afghanistan and Iraq Currently colorimetric paper is used to detect agents on-site while gas chromatography combined with mass spectrometry (GCMS) is used in mobile support laboratories However both methods have severe drawbacks The paper changes color in response to contact with many chemicals besides CWAs causing a high incidence of false positives1 while GCMS although very chemically specific requires hours to perform and constant re-calibration234 Military operations would be greatly aided by a portable analyzer that can identify and quantify potential chemical agents at concentrations that impact safety This includes the analysis of drinking water supplies distribution and storage systems To meet this goal the Department of Defense has been investigating numerous approaches under the auspices of the Joint Service Agent Water Monitor (JSAWM) program5 The JSAWM must be capable of identifying the chemical agents with no false-positives and quantifying the agents at microgram per liter concentrations within 10 minutes678 Recently we and others have been investigating the ability of surface-enhanced Raman spectroscopy (SERS) to measure chemical agents9-12 bioagents13-17 and their hydrolysis products in water SERS employs the interaction of surface plasmon modes of metal particles with target analytes to increase scattering efficiency by as much as 1 million times18 In 1987 the potential of this approach to measure CWAs was demonstrated by measuring a series of pesticides19 Several of these organophosphonates have chemical structures similar to CWAs in particular P=O functional groups In our studies we have been employing silver-doped sol-gels to promote the SER effect The porous silica network of the sol-gel matrix offers a unique environment for immobilizing and stabilizing SER-active metal particles20-23 The sol-gels have been coated on the internal walls of standard glass vials to allow rapid SER analysis of small sample quantities without preparation We have measured over 100 chemicals with enhancements of 104 to 106 demonstrated reversible measurements in a flowing system reproducible measurements from vial-to-vial and measurements in multiple solvents including water20-23 Previously we used these vials to perform preliminary measurements of cyanide (CN) methylphosphonic acid (MPA) and dipicolinic acid (DPA) MPA is a hydrolysis product of the nerve agents (eg sarin Reaction 1) and may be a valuable indicator of nerve agent usage particularly since the alkyl methylphosphonic acids are relatively more stable than their corresponding parent complexes24 DPA is
SPIE -2003-5269 118
a chemical signature of spore forming bacteria such as Bacillus anthracis And in light of the inability to rapidly detect the anthrax spores distributed through the US mail in October 2001 a number of methods are being developed to extract and analyze this signature Reaction 1 Stepwise hydrolysis of Sarin to form hydrofluoric acid (HF) isopropyl methylphosphonic acid (IMPA) then methyl phosphonic acid (MPA) and 2-propanol In our previous SERS investigations MPA and DPA were measured at 50 and 100 mgL respectively In both cases limits of detection (LOD) were estimated at 100 microgL providing encouragement in that SERS may satisfy the needs of the JSAWM Since it has been shown that pH can substantially influence the intensity of SER bands25 which would clearly influence quantitative analysis we undertook the present study to determine the severity of these effects for cyanide methyl phosphonic acid and dipicolinic acid Furthermore we previously observed a band at 1050 cm-1 for MPA6 possibly due to an anion formed at basic pH Here we investigate the source of this spectral anomaly
2 EXPERIMENTAL All chemicals including potassium cyanide methyl phosphonic acid dipicolinic acid and those used to prepare the silver-doped and gold-doped sol-gel coated vials were obtained and used as received from Sigma-Aldrich All samples were prepared in HPLC grade water (Fischer Scientific Fair Lawn NJ) for SERS measurements The pH of these samples was adjusted using dilute nitric acid or potassium hydroxide and verified using a pH electrode (Corning Inc Corning NY) that had been calibrated with pH 400 700 and 1000 buffered standards from Fischer Scientific Once prepared the samples were transferred into the silver-doped sol-gel vials (Real-Time Analyzers Inc East Hartford CT) The vials were coated in a manner similar to that previously reported by adding ammonium hydroxide to a solution of silver nitrate tetramethyl orthosilicate (TMOS) and methanol20 Gold-vials were coated by adding nitric acid to a solution of gold tetrachloride TMOS and methanol The two precursor solutions were prepared mixed and transferred to 2-mL glass vials dried and heated After sol-gel formation the incorporated metal ions were reduced with dilute sodium borohydride (1mgmL) which was followed by a water wash to remove residual reducing agent After the resultant analyte solution was introduced the SER-active vial was immediately fixed horizontally to an XY positioning stage (Conix Research Springfield OR) just inside the focal point of an f07 aspheric lens The lens focused the beam into the sample and collected the scattered radiation back along the same axis A dichroic filter (Omega Optical Brattleborough VT) was used to reflect the excitation laser to the lens and pass the Raman scattered radiation collected by the lens An f2 achromat was used to collimate the laser beam exiting a 200 microm core diameter source fiber optic while a second f2 achromat was used to focus the scattered radiation into a 365 microm fiber optic (Spectran Avon CT) A short pass filter was placed in the excitation beam path to block the silicon Raman scattering generated in the source fiber from reflecting off sampling optics and reaching the detector A long pass filter was placed in the collection beam path to block the sample Rayleigh scattering from reaching the detector A 785 nm diode laser (Process Instruments Inc model 785-600 Salt Lake City UT) was used to deliver 100 to 150 mW of power to the sample A Fourier transform Raman spectrometer (Real-Time Analyzers model IRA-785 East Hartford CT) and a silicon photo-avalanche detector (Perkin Elmer model C30902S Stamford CT) were used to acquire the SER spectra
3 RESULTS AND DISCUSSION In a previous study of MPA6 aimed at developing a concentration calibration curve and determining limits of detection (LOD) we observed an anomaly at 1050 cm-1 Since it was found that the band intensity changed as a function of concentration the band must be associated with a sample parameter Two possible parameters photon flux and pH are examined here The first parameter was investigated by irradiating a 1mgmL MPA sample in a SER-active vial with laser powers of 200 mW and above and monitoring spectral changes It was immediately found that the 1050 cm-1 band
2O+ H HF + +OH
OH3H C
OP
OF
CH
CH3
3
3H C
OCP
OOH
CH
CH3
3
3H C
O
CPHO
CH
CH
3
3
C
Sarin IMPA MPA 2-propanol
SPIE -2003-5269 119
appeared and grew as a function of time and that the higher the laser power the faster the growth Figure 1 shows the growth of the 1050 cm-1 band over the course of 30 minutes when using 150 mW of 785 nm excitation while Figure 2A shows that the growth can be fit with a first order exponential equation namely I1050 = 03+05e-013t Furthermore the 760 cm-1 band could be fit with a first order decay equation with an identical exponential rate constant ie I760 = 08-
08e-013t The rates represent classical first order kinetics and their correspondence allows one to conclude that MPA is being transformed one-for-one into a photo-generated product At this time the photoproduct has not been positively identified but phosphonic (phosphorous) acid and phosphonate are likely candidates since the symmetric P(OH)3 stretch occurs at ~1050cm-1 Our studies show that a reduction in laser power to 100 mW at the sample for MPA essentially eliminates this degradation process This laser power or lower was used for further measurements
Methyl phosphonic acid is a diprotic acid that stepwise dissociates into two anions MPA- and MPA= according to the following reactions26
MPA MPA- + H+ pKa1 = 212 Reaction 2
MPA- MPA= + H+ pKa2 = 729 Reaction 3 The relative concentrations of MPA MPA- and MPA= can be determined at any pH by expressing [MPA] and [MPA=] in terms of [MPA-] using Reactions 2 and 3 and summing all three to equal the total starting concentration here 2 mgmL (0021M MW = 9602) viz
[MPA] + [MPA-] + [MPA=] = 0021M Equation 1 substituting from Reactions 1 and 2
([H+][MPA-])K1a + [MPA-] + (K2a[MPA-])[H+] = 0021M Equation 2 rearranging [MPA-] = 0021M(1+[H+]K1a + K2a[H+]) Equation 3 The relative concentrations of MPA MPA- and MPA= as a function of pH are shown in Figure 3 It is worth noting that near neutral pH both MPA- and MPA= will be present To confirm that the SER signal followed this pH dependence a starting solution consisting of 20 mg of MPA in 10 mL HPLC grade water was prepared and brought to pH of 20 using dilute nitric acid From this solution 2 mL were added to a SER-active vial and the SER spectrum recorded At this pH a peak at 760 cm-1 was barely discernable The 2 mL solution was returned to the starting solution and the pH was re-measured to correct for any changes that the silver-doped sol-gel vials might cause In most cases the change was less than 02 pH units and the pH is reported as the before and after average Next the pH of the
Figure 2 A) Exponential growth of 1050 cm-1 band and B) exponential decay of 760 cm-1 band for spectral series in Figure 1
Figure 1 Growth of 1050 cm-1 band as a function of time due to exposure to 150 mW of 785 nm Spectra are 5 sec each collected every 100-sec from 0 to 30-min
B
570 770 970 1170 1370Raman Shift (cm-1)
Arbit
rary
Unit
s
MPA Photodegradation
600 800 1000 1200 1400 Wavenumber (cm-1)
A
0 10 20 30 0 10 20 30 time (min) time (min)
Ram
an In
tens
ity (r
elat
ive)
30
min
0
SPIE -2003-5269 120
starting solution was adjusted to 325 using dilute KOH Again 2 mL were added to a vial and the SER spectrum recorded At this pH a reasonably strong 760 cm-1 band was observed This process was repeated as spectra were recorded at pHs of 70 74 75 79 85 and 100 A total of 1 mL of KOH was added diluting the total concentration by 10 Next the pH of the starting solution was made acidic by adding dilute nitric acid dropwise This time spectra were recorded at pHs of 72 69 64 and 37 Figure 4 shows the SER spectra for representative pHs (spectra were left out to simplify the figure) while Figure 3 shows the 760 cm-1 peak intensities as a function of pH (The band intensities were adjusted to compensate for dilution effects caused by the addition of HNO3 and KOH then normalized to 0021 M for the most intense band observed at pH 37) It is clear from Figure 3 that the 760 cm-1 band follows the MPA- concentration as a function of pH and must be assigned to this anion No bands were observed that corresponded to MPA or MPA= The lack of an MPA SER spectrum may be due to the absence of an attraction between the neutral analyte and the electropositive silver surface The same reasoning suggests that a strong SER spectrum should be observed for MPA= but it is not and a satisfactory explanation has not been found
0000
0005
0010
0015
0020
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14pH
Con
cent
ratio
n [M
]
pK1 = 212 pK2 = 729
MPA- MPA=
MPA
Figure 4 SER spectra of 002M MPA as a function of pH Conditions 100 mW of 785 nm 36 scans (1 min) 8 cm-1 recorded 2 min after sample introduction pH 19 and 20 not apparent on this scale pH 69 and 74 near identical to 70 and 75 and not shown for clarity
00
02
04
06
08
10
12
14
16
18
0 5 10 15 20 25 30Measurement Number
Ram
an In
tens
ity (7
60 c
m-1
)
Figure 3 Concentration dependence of MPA MPA- and MPA= as a function of pH for a 002M sample Intensity of 760 cm-1
band from Figure 6 as a function of pH ( for increasing basic adjustment for increasing acidic adjustment error was measured at ~10 for pH 64)
Figure 5 SER spectra of 002M MPA at pH 64 measured around a vial at three heights (9 points per height) Conditions as in Figure 4 but 10-sec scans
Since these measurements involved the removal and replacement of the SER-active vial in the sample holder to remove and add sample variation in the intensity as a function of vial position was minimized by illuminating the exact same height along the vial wall But this does not account for variability of the SERS response of the sol-gel coating around the vial To analyze this effect a vial containing MPA at pH 64 was rotated at ~ 40o
intervals at the original height and 18rdquo above and below this value Figure 5 shows the intensity of the 760 cm-1
band for the 27 positions It was found that the average value was 137plusmn014 an RSD of 10 overall and 5 for each height An error bar is included in Figure 4 for the pH 64 measurement
SPIE -2003-5269 121
With the above analyses in mind a preliminary investigation of the SER spectral response for MPA (as MPA-) as a function of concentration was performed A single vial was used for these measurements beginning with 1 mgL followed by measurements of 10 100 and 1000 mgL In all cases the pH was ~7 and 3 positions around the vial were measured per concentration Since the 760 cm-1 band was not observed for concentrations of 1 or 10 mgL using 100 mW of 785 nm the laser power at the sample was raised to 200 mW beginning with the 10 mgL concentration Photo-degradation was largely avoided (and not observed) by exposing the sample for only 33 seconds per spectral acquisition Representative spectra for 10 100 and 1000 mgL are shown in Figure 6 while a plot of the 760 cm-1 band intensity as a function of concentration is shown in Figure 7 These values were also used to estimate limits of detection based on the signal-to-noise ratio (SN) of the 760 cm-1 band using 100 mW of 785 nm laser excitation and a 10-min acquisition time As summarized in the Figure 7 inset the lower the measured concentration the lower the predicted LOD Even if as estimated 210 microgL could be measured using the silver-doped sol-gel vials an improvement of a factor of 70 is still required to meet JSAWM goals of measuring 3 microgL in 10-minutes Similar to MPA DPA is a diprotic acid (pKa1 = 216 and pKa2 = 692) and variations in pH will effect the relative concentrations of DPA DPA- and DPA= and possibly the SER spectra and band intensities This could prove significant if an acid or base is used to denature anthrax spores with the goal of extracting and analyzing DPA The MPA pH study described above was mimicked for DPA except that the starting solution consisted of 20 mg of DPA in 20 mL HPLC grade water (60x10-3M MW = 1671) The initial solution had a pH of 245 which was made basic by dilute KOH to pHs of 355 433 487 559 1069 and 1166 SER spectra were recorded at each pH using 100 mW of 785 nm and a 44-sec acquisition time Next one drop of concentrated nitric acid was used to remake the solution acidic at a pH of 200 Again sequential pH measurements were performed at 383 510 735 and 822 The solution pH was made acidic a third time but to pH 219 171 then 135 Throughout this process no more than 20 drops of acid or base were added and therefore the concentration was diluted by no more than 10 Most of the spectral bands showed a minor decrease in intensity as a function of increasing pH values However the bands at 525 and 795 cm-1 showed the most dramatic changes which occurred at acid pH Figure 8 shows the SER spectra of DPA for the spectral region and pH range of interest The identity of the DPA species was determined by plotting the normalized peak intensities with the lowest value set to 0 and the highest to 0006 M as a function of pH and overlaying these values on a plot of the relative concentrations for DPA DPA- and DPA= as previously done for MPA (Figure 9) As can be seen the 525 cm-1 band clearly corresponds to DPA The correspondence of the 795 cm-1 band to this species is less clear as the band retains intensity until pH of 55 This can be attributed to contribution to the overlapping band at 810 cm-1 which does not change as a function of pH The fact that most bands are observed at all pHs suggest that the primary interaction with silver is through the ring nitrogen This is supported by the fact that the most intense band occurs at 1008 cm-1 attributed to a symmetric ring breathing mode and that this interaction has been characterized for pyridine in numerous papers27
0
1
2
3
4
5
6
7
0 200 400 600 800 1000 1200
MPA Concentration (mgL)
760
Ban
d In
tens
ity
Figure 6 SER spectra of MPA in water at A) 1000 B) 100 and C) 10 mgL Conditions pH of 7 silver-doped sol-gel coated vial 200 mW of 785 nm 33-sec 8 cm-1 resolution
Figure 7 Plot of SER intensity of 760 cm-1 band of MPA as a function of concentration using 200 mW of 785 nm Inset table includes average intensity LOD standard deviation and percent deviation for each concentration but for 100 mW and 10 min
conc (mgL) LOD ave stddev dev10 021 002 1021100 072 010 14301000 312 040 1280
A
B C
SPIE -2003-5269 122
The ability to detect dipicolinic acid (DPA) by SERS is demonstrated in Figure 10 Even at 1 mgL the primary bands are visible The intensity of the dominant ring breathing mode observed at 1008 cm-1 for DPA in water (pH = 25 to 55) is plotted as a function of concentration in Figure 11 These values were also used to estimate limits of detection based on the SN of the 1008 cm-1 band using 100 mW of 785 nm laser excitation and a 10-min acquisition time Again the lower the measured concentration the lower the predicted LOD (see Figure 11 inset) and detection of 160 microgL is possible Gastrointestinal anthrax requires significant more spores than inhalation anthrax28 and a limit of detection might be placed at 1 million spores in 1 liter of water or 10 microgL Since each spore contains ~10 CaDPA by weight29 a goal for DPA might be 1 microgL indicating that the present measurements must be improved by nearly two orders of magnitude Due to the increased hazards of handling HCN gas KCN salt was used for these experiments Nevertheless all sample preparations were performed in a chemical hood KCN completely dissolves in water but its conjugate acid HCN is formed and has a Ka of 615x10-1030 viz
HCN CN- + H+ pKa = 921 Reaction 4
Figure 8 SER spectra of 1 mgmL DPA as a function of pH Conditions 100 mW 785 nm 100 scans (44-sec)
Figure 9 Concentration dependence of DPA DPA- and DPA= as a function of pH for a 0006M sample Intensity of 525 () and 795 (diams) cm-1 bands from Fig 8 as a function of pH
pH
135 171 219 383
0
0001
0002
0003
0004
0005
0006
0007
0 2 4 6 8 10 12 14pH
Con
cent
ratio
n [M
]
DPADPA-DPA=795525
DPA DPA=DPA-
pK2 = 692pK1 = 216
795 525
0
05
1
15
2
25
0 200 400 600 800 1000 1200
DPA Concentration (mgL)
1008
Ban
d In
tens
ity
conc (mgL) lod-10min-100mw1 017
10 016100 103
1000 355
Figure 10 SER spectra of DPA in water at A) 1000 B) 100 C) 10 and D) 1 mgL Conditions pH of 25-55 silver-doped sol-gel coated vial 175 mW of 785 nm 1-min 8 cm-1 D) has been multiplied by x10 to make bands visible
A
B C
D
Figure 11 Plot of SER intensity of 1008 cm-1 band of DPA as a function of concentration using 175 mW of 785 nm Inset table includes LOD in mgL for each concentration but for 100 mW and 10 min
SPIE -2003-5269 123
Consequently the cyanide concentration must be determined for each initial KCN concentration Specifically the samples prepared with concentrations of 01 1 10 100 and 1000 mgL of KCN produced CN- concentrations of 63x10-3 033 69 89 and 964 mgL at pHs of 816 90 967 102 and 107 respectively The pH dependence for the HCN and CN- concentrations are shown in Figure 12 Thus as the amount of KCN added to the solution decreases so does the pH of the solution (becomes less basic) and according to Reaction 4 the relative amount of CN- to HCN also decreases For example in the preparation of a 01 mgL solution of KCN the pH is shifted from 7 for pure water to only 816 and only 63 of the starting material becomes CN- or 63x10-3 mgL In comparison for a solution of 1000 mgL the pH is shifted from 7 to 107 and 96 of the starting material becomes CN- This is significant since the cyanide ion is better able to adsorb onto the silver particles and become SERS active SER spectra of 10 100 and 1000 mgL of KCN samples are shown in Figure 13 and a corresponding plot of the 2100 cm-1 band intensity as a function of concentration is shown in Figure 14
000102030405060708091011
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14pH
Con
cent
ratio
n [m
gm
L]
pKa = 921
CN -HCN
The SER spectra of cyanide are dominated by the single CN stretch at 2100 cm-1 which occurs in normal Raman spectra of solutions at 2080 cm-1 However a low frequency mode occurs at 135 cm-1 due to a Ag-CN stretch (not shown) indicative of a strong surface interaction It is also observed that as the concentration decreases the CN stretch shifts to 2140 cm-1 This shift has been attribute to the formation of a tetrahedral Ag(CN)3
2- surface structure31 as well
Figure 14 Concentration dependence of KCN SERS measured under conditions in Fig 11 Concentrations are 1 01 and 001 mgml Intensities are measured for the CN stretch at 2100 cm-1 Inset table includes LOD in mgL for each concentration in Figs 13 and 15 but for 100 mW and 10 min
Figure 13 SER spectra of KCN in water at A) 1000 B) 100 and C) 10 mgL Conditions pHs of 107 102 and 97 silver-doped sol-gel coated vial 100 mW of 785 nm 1-min 8 cm-1 C) has been multiplied by x10 to make band visible
Figure 12 Concentration dependence of HCN and CN- as a function of pH for a 1 mgmL sample Calculated intensity of 2100 () cm-1 band for a 1 mgml sample at pHs of 816 90 967 102 and 107
A
B
C
0
20
40
60
80
100
120
140
0 200 400 600 800 1000 1200CN Concentration (mgL)
2100
Ban
d In
tens
ity
Figure 15 SER spectra of KCN in water at A) 10 B) 1 and C) 01 mgL Conditions pHs of 97 12 and 12 gold-doped sol-gel coated vial 100 mW of 785 nm 1-min 8 cm-1 C) has been multiplied by x10 to make band visible
A
B
C
conc (mgL) lod-10min-100mw Condition01 001 Au-pH 121 007 Au-pH 12
69 003 Au-pH 9769 016 Ag-pH 9789 022 Ag-pH 102946 113 Ag-pH 107
SPIE -2003-5269 124
as to CN adsorbed to two different surface sites32 Alternatively the 2140 cm-1 band could be attributed to HCN since this species dominates at lower concentrations However it is unlikely that this species would be attracted to the electropositive silver surface Further both peaks should be present at pHs between 85 and 105 but this is not observed It has also been suggested that at concentrations near and above monolayer coverage the CN- species is forced to adsorb end-on due to crowding and at lower concentrations the molecule can reorient to lie flat33 This suggests that the 2100 and 2140 cm-1 bands correspond to the end-on and flat orientations respectively As Figures 13 and 14 show the intensity of the CN stretch for the 89 mgL sample is nearly as intense as the 964 mgmL band This suggests that the Raman signal for the flat orientation is more enhanced However more extensive measurements are required to verify this point Since resent research has suggested that cyanide may be more effectively detected on gold measurements of KCN solutions were also performed using gold-doped sol-gel vials Preliminary measurements are shown in Figure 15 for samples prepared from 01 1 and 10 mgL KCN Since the pHs are 816 90 967 the resultant CN- concentrations are 63x10-3 033 and 69 mgL Initially only the highest concentration was observed and the signal intensity was significantly better than the equivalent concentration measured using silver In an effort to shift Reaction 4 to the left transforming HCN to CN- (Le Chatelierrsquos principle) KOH was added to the lower concentration samples producing solutions with pH 12 This effectively forces all of the cyanide in solution to be CN- or 01 and 10 mgL respectively More importantly the CN stretch is now observed in the SER spectra The band appears at 2125 cm-1 as has been previously reported for gold12 As calculated for MPA and DPA LODs can be estimated from this data For the three concentrations of cyanide on silver the LODs are 016 to 11 mgL for 100 mW of 785 nm laser excitation and a 10-min acquisition time For gold pH adjusted the LODs are10 to 70 microgL an improvement of more than 10 times silver Nevertheless either substrate is sufficient to meet the JSAWM goals of measuring 3 mgL in 10-minutes as the requirements form cyanide are much less stringent than the nerve agents
4 CONCLUSIONS Here we examined the surface-enhanced Raman spectral response for methyl phosphonic acid dipicolinic acid and cyanide as a function of pH It was determined that the most prominent peak at 760 cm-1 reaches a maximum intensity between pH 3 and 7 and corresponds to the MPA- species Neither the MPA nor MPA= species appear to generate a SER spectrum and consequently no spectra were observed below pH 2 or above pH 8 In this study we also found that higher laser powers could cause photodegradation of MPA signified by the exponential growth of a band at 1050 cm-1 which is tentatively assigned to phosphorous acid Unlike MPA DPA was observed at all pHs This is attributed to the dominant interaction of the pyridine functional group with silver Minor spectral changes were observed at acid pHs and were assigned to neutral DPA Like MPA SER spectra of cyanide were pH dependent No spectra were observed for the HCN species while CN- was best observed at pHs more basic than 8 Preliminary concentration studies for the three analytes allowed estimating limits of detection for MPA DPA and CN using 100 mW of 785 nm and a 10-min acquisition time of 210 165 and 70 microgL respectively Although the latter value suggests that cyanide can be measured at sufficiently low concentrations to meet JSAWM goal improvements by 100 to 200 times are required for MPA and DPA It is clear from this study that pH of the sample is important and must be taken into account when developing concentration calibration curves as well as fieldable analyzers Future work will include tailoring the sol-gel with specific functional groups to dictate sample pH and thereby optimize sensitivity
ACKNOWLEDGMENTS The authors are grateful for the support of the US Army (DAAD13-02-C-0015 Joint Service Agent Water Monitor program) The authors would also like to thank Dr Steve Christensen of the US Army SBCCOM for helpful discussions and Mr Chetan Shende of Real-Time Analyzers for assistance in development of the gold-doped sol-gels
REFERENCES 1 Erickson B Analytical Chemistry News amp Features June 1 397A (1998)
SPIE -2003-5269 125
2 Johnston RL Hoefler CM Fargo JC and Moberley B AT-ONSITE 5-8 (1994) 3 Black RM RJ Clarke RW Read and MT Reid J Chromatography 662 301-321 (1994) 4 USA Army Center for Health Promotion and Preventive Medicine (USACHPPM) Technical Guideline 230A 5 Jensen JL NC Wong W R Loerop SPIE 4775-03 (2001) 6 Hoenig SL Handbook of Chemical Warfare and Terrorism Greenwood Press Wesport CT (2002) 7 Munro NB SS Talmage GD Griffin LC Waters AP Watson JF King and V Hauschild Env Health
Persp 107 933-974 (1999) 8 Holstege CP Kirk M Sidell FR Crit Care Clin13 923-42 (1997) 9 Farquharson S P Maksymiuk K Ong and S Christesen SPIE 4577 166-173 (2001) 10 Lee Y and S Farquharson SPIE 4378 21-26 (2001) 11 Spencer KM J Sylvia S Clauson and J Janni SPIE 4577 158-165 (2001) 12 Tessier P S Christesen K Ong E Clemente A Lenhoff E Kaler and O Velev Applied Spectroscopy 56
1524-1530 (2002) 13 Farquharson S WW Smith S Elliott and JF Sperry SPIE 3533 207-214 (1998) 14 Farquharson S WW Smith S Elliott and JF Sperry SPIE 3855110-116 (1999) 15 Farquharson S WW Smith YH Lee S Elliott and JF Sperry SPIE 4575 62-72 (2002) 16 Guzelian AA J Sylvia J Janni S Clauson and KM Spencer SPIE 4577 182-192 (2001) 17 Shende C F Inscore A Gift P Maksymiuk and S Farquharson in press 18 Weaver MJ S Farquharson and MA Tadayyoni J Chem Phys 82 4867-4874 (1985) 19 Alak AM and T Vo-Dinh Analytical Chemistry 59 2149-2153 (1987) 20 Lee Y and S Farquharson SPIE 4206 140-146 (2000) 21 Farquharson S and Y Lee SPIE 4200-16 (2000) 22 Lee Y S Farquharson and P M Rainey SPIE 3857 76-84 (1999) 23 Lee Y S Farquharson H Kwong and M Shahriari SPIE 3537 252-260 (1998) 24 Wang J M Pumera G Collins and A Mulchandani Analytical Chemistry 74 6121-6125 (2002) 25 Dou X YM Jung Z-Q Cao and Y Ozaki Applied Spectroscopy 53 1440-1447 (1999) 26 Data supplied by S Christesen and K Ewing 27 Kerker M and B Thompson Eds SPIE MS 10 (1990) 28 Inglesby TV DA Henderson JG Bartlett JAMA 287 2236 (2002) 29 FW Janssen AJ Lund and LE Anderson Science 127 26 (1958) 30 Lide DR Ed Handbook of Chemistry and Physics CRC Press 77th Ed (1996-7) 31 Billmann J G Kovacs and A Otto Surf Sci 92 153 (1980) 32 Murray CA and S Bodoff Phys Rev B 32 671 (1985) 33 Kellogg D and J Pemberton J Phys Chem 91 1120 (1987)
SPIE-2003-5269 16
Chemical agent detection by surface-enhanced Raman spectroscopy
Stuart Farquharson Alan Gift Paul Maksymiuk Frank Inscore and Wayne Smith
Real-Time Analyzers 87 Church Street East Hartford CT 06108
Kevin Morrisey and Steven D Christesen US Army SBCCOM Aberdeen Proving Ground MD 21010
ABSTRACT
In the past decade the Unites States and its allies have been challenged by a different kind of warfare exemplified by the terrorist attacks of September 11 2001 Although suicide bombings are the most often used form of terror military personnel must consider a wide range of attack scenarios Among these is the intentional poisoning of water supplies to obstruct military operations in Afghanistan and Iraq To counter such attacks the military is developing portable analyzers that can identify and quantify potential chemical agents in water supplies at microgram per liter concentrations within 10 minutes To aid this effort we have been investigating the value of a surface-enhanced Raman spectroscopy based portable analyzer In particular we have been developing silver-doped sol-gels to generate SER spectra of chemical agents and their hydrolysis products Here we present SER spectra of several chemical agents measured in a generic tap water Repeat measurements were performed to establish statistical error associated with SERS obtained using the sol-gel coated vials Keywords Chemical agents chemical agent detection SERS Raman spectroscopy
1 INTRODUCTION
In the autumn of 2001 terrorism within US borders became a sobering reality While extensive efforts are being implemented to secure the homeland US and Coalition forces in Afghanistan and Iraq are constantly faced with terrorist attacks In addition to car-bombings the military has identified several non-traditional attack scenarios including poisoning of water supplies by chemical warfare agents (CWAs) To counter this threat the Department of Defense has been investigating numerous approaches to detect such attacks under the auspices of the Joint Service Agent Water Monitor (JSAWM) program1 The JSAWM must be capable of identifying the chemical agents with no false-positives and quantifying the agents at microgram per liter concentrations within 10 minutes2 This includes the analysis of drinking water supplies distribution and storage systems Currently colorimetric paper is used to detect agents on-site while gas chromatography combined with mass spectrometry (GCMS) is used in mobile support laboratories However both methods have severe drawbacks The paper changes color in response to contact with many chemicals besides CWAs causing a high incidence of false positives3 while GCMS although very chemically specific requires up to an hour to perform and regular re-calibration456 Vibrational spectroscopy has also been investigated as a potential method of rapidly detecting CWAs7-
11 as early as 1985 in the case of infrared spectroscopy8 and 1988 in the case of Raman spectroscopy11 Again however these techniques also have limitations when it comes to measuring trace poisons in water Infrared spectra would be dominated by the very strong absorption of water which would obscure absorptions by most other chemicals present Whereas Raman spectroscopy is simply not a very sensitive technique and detection limits are typically grams per liter Surface-enhanced Raman spectroscopy (SERS) offers several advantages over conventional vibrational methods and that may provide the necessary sensitivity required for detecting trace quantities of chemical agents in water SERS employs the interaction of surface plasmon modes of metal particles with target analytes to increase scattering efficiency by as much as 1 million times12 In 1987 the potential of this approach to measure CWAs was demonstrated by measuring a series of pesticides13 Several of these organophosphonates have chemical structures similar to CWAs in particular P=O functional groups In the past few years we and others have further explored the ability of SERS to detect CWAs14-17 and even bioagents 18-21 We have been employing silver-doped sol-gels to promote the SER effect
SPIE-2003-5269 17
in these studies The porous silica network of the sol-gel matrix offers a unique environment for immobilizing and stabilizing SER-active metal particles22-25 The silver-doped sol-gels have been coated on the internal walls of standard glass vials to allow rapid SER analysis of small sample quantities without preparation We have measured over 100 chemicals with enhancements of 104 to 106 demonstrated reversible measurements in a flowing system reproducible measurements from vial-to-vial and measurements in multiple solvents including water21-25 Previously we used these vials to perform preliminary measurements of cyanide (CN) and methylphosphonic acid (MPA) Most of the nerve agents form MPA during hydrolysis while Tabun forms CN a chemical agent in its own right In another paper including in these proceedings we examined the limits of detection (LOD) for MPA by measuring a series of concentrations down to 50 mgL and estimated a limit of detection of 100 microgL26 These measurements provide encouragement in that SERS may satisfy the needs of the JSAWM To further establish the viability of SERS in particular silver-doped sol-gels here we present analysis of cyanide mustard and VX in tap water The measurements performed at the US Armyrsquos Edgewood Chemical Biological Center Aberdeen MD also included numerous repeat measurements to establish reproducibility
2 EXPERIMENTAL 2a General All chemicals including potassium cyanide 2-chloroethylethyl sulfide and those used to prepare the silver-doped and gold-doped sol-gel coated vials were obtained and used as received from Sigma-Aldrich All samples were prepared in a chemical hood using HPLC grade water unless otherwise noted (Fischer Scientific Fair Lawn NJ) for SERS measurements Once prepared the samples were transferred into the silver-doped sol-gel vials (Real-Time Analyzers Inc East Hartford CT) The vials were coated in a manner similar to that previously reported by adding ammonium hydroxide to a solution of silver nitrate tetramethyl orthosilicate (TMOS) and methanol22 The two precursor solutions were prepared mixed and transferred to 2-mL glass vials dried and heated After sol-gel formation the incorporated metal ions were reduced with dilute sodium borohydride (1mgmL) which is followed by a water wash to remove residual reducing agent After the resultant analyte solution was introduced the SER-active vial was fixed horizontally to an XY positioning stage (Conix Research Springfield OR) just inside the focal point of an f07 aspheric lens The lens focused the beam into the sample and collected the scattered radiation back along the same axis A dichroic filter (Omega Optical Brattleborough VT) was used to reflect the excitation laser to the lens and pass the Raman scattered radiation collected by the lens An f2 achromat was used to collimate the laser beam exiting a 200 microm core diameter source fiber optic while a second f2 achromat was used to focus the scattered radiation into a 365 microm fiber optic (Spectran Avon CT) A short pass filter was placed in the excitation beam path to block the silicon Raman scattering generated in the source fiber from reflecting off sampling optics and reaching the detector A long pass filter was placed in the collection beam path to block the sample Rayleigh scattering from reaching the detector A 785 nm diode laser (Process Instruments Inc model 785-600 Salt Lake City UT) was used to deliver 100 to 150 mW of power to the sample A Fourier transform Raman spectrometer (Real-Time Analyzers model IRA-785 East Hartford CT) and a silicon photo-avalanche detector (Perkin Elmer model C30902S Stamford CT) were used to acquire the SER spectra
2b Edgewood Chemical Biological Center The surface-enhanced Raman spectral measurements at the US Armyrsquos Chemical Biological center presented here were all performed on September 12 2003 To expedite measurements a plate was machined to hold up to 12 SER-active sample vials (Figure 1) The plate fit a standard XY plate reader that could be programmed Pure KCN bis-(2-chloroethyl)sulfide (distilled mustard gas HD) and ethyl S-2-diisopropylamino ethyl methylphosphonothioate (VX) were obtained on-site and used to prepare 1 gL tap water solutions in a chemical hood with appropriate safety equipment Simulated tap water was prepared by adding 10 chemicals most often found in tap water at appropriate concentrations to distilled water (Table 1) SER measurements were also performed in a chemical hood For added safety the FT-Raman instrument was placed outside the laboratory and 30 foot fiber optic and electrical cables were used to allow remote SERS measurements and plate manipulation For each experiment 1gL samples were prepared and added to 9 individual vials which were then loaded on the plate In some cases a tenth vial was included as a blank
Table 1 Generic Tap Water Composition
Compound mgL NaHCO3 100 CaSO4 27 MgSO4bull7H2O 67 NaNO3 1 Fulvic Acid 1 K2HPO4 07 KH2PO4 03 (NH4)2HSO4 001 NaCl 001 FeSO4 0001 pH 76-78
SPIE-2003-5269 18
A software program was written that allowed selecting the sequence that the vials were measured the number of positions along the length of the vials to measure (1 to 5) and the number of scans to co-add During sample analysis the program displayed the vial being analyzed the point being analyzed and the spectrum as it was being acquired Once all the data was collected a second software program was written to rapidly analyze the data The spectra collected for all the vials on a plate could be loaded at one time and then the spectra for each point could be displayed simultaneously or separately The user could then select the Raman peak to analyze in terms of peak height or area This was accomplished by selecting points on either side of the peak to define a baseline of zero The peak height or area could then be computed for all of the spectra loaded and then exported to a spreadsheet for statistical analysis
Figure 1 A) Vial Holder 6 slots to hold 2 vials each end-to-end B) Measurement Configuration Program user selects vials to measure sequence number of points per vial (1 to 5) and number of scans per point C) Spectral Acquisition Program shows spectrum being collected which vial and position D) Spectra Analysis Program user selects spectra to analyze by plate vial and point (s) as well as two wavenumbers defining the peak and the baseline to subtract The image is of 5 repeat measurements of 10 mgL KCN in generic tap water 16 sec each 100 mW of 785 nm
3 RESULTS AND DISCUSSION Raman and surface-enhanced Raman spectra were obtained for potassium cyanide bis-(2-chloroethyl)sulfide and ethyl S-2-diisopropylamino ethyl methylphosphonothioate representing three classes of chemical agents cyanides mustards and nerve agents respectively Spectra were also obtained for 2-chloroethyl ethyl sulfide (CEES) a structural analogue to HD which was included in the study to aid in assigning spectral bands KCN salt was used for cyanide experiments to avoid the increased hazards of handling HCN gas KCN completely dissolves in water forming its conjugate acid HCN according to its Ka of 615x10-1027 and at a concentration of 1 mgmL results in a pH 107 solution This is significant in that only CN- appears to interact sufficiently with silver to produce a SER spectrum and no spectral signal is observed below pH 726 Figure 2 shows the SER and normal Raman spectra for KCN The SERS spectrum of 1mgml KCN in water shows a single intense somewhat broad feature at
A B
C D
SPIE-2003-5269 19
2100 cm-1 assigned to the single CequivN stretch The band is much sharper in the normal Raman spectra of the solid KCN salt at 2074 cm-1 This band does broaden and shift to 2080 cm-1 in solution (not shown) However the observed SERS frequency is attributed to interaction with silver and low frequency mode at 135 cm-1 attributed to a Ag-CN stretch (not shown) supports this conclusion
cm-1 band to a C-S stretch but the authors concede that it is in fact more likely a C-Cl stretch28 It appears that the most intense Raman bands at 648 692 and 747 cm-1 shift to 620 660 and 720 cm-1 in the SER spectra and are tentatively assigned as above The width of these bands suggests that they overlap underlying spectral features Additional bands in the Raman spectra occur at 972 1034 1049 1263 1286 1423 1442 2865 2935 and 2960 cm-1 Corresponding bands occur in the SER spectra at 964 1015 1054 1286 1410 1447 2865 and 2935 cm-1 Most of these bands are associated with alkane modes specifically the bands at approximately 1040 cm-1 to a C-C stretch 1290 cm-1 to a CH2 in-phase twist 1440 cm-1 to a CH2 wag 2865 cm-1 to a symmetric CH2 stretch and 2965 cm-1 to an asymmetric CH2 stretch The Raman and SER spectra of sulfur mustard were measured at the Edgewood center (Figure 4) Both spectra are largely similar to CEES The C-Cl and C-S bands in the Raman spectrum of HD now occur at 640 655 700 739 and 760 cm-1 and are more resolved possibly due to the increased molecular symmetry Theoretical calculations indicate that the first three bands are due to C-Cl stretching modes and the latter two to C-S stretching modes28 Only the C-Cl bands maintain significant intensity in the SER spectra occurring at 624 and 643 cm-1 which is attributed to the
Prior to measurements of HD CEES was examined by Raman and SER spectroscopy (Figure 3) CEES also known as half-mustard is essentially identical to HD except one of the chlorine end atoms is replaced by a hydrogen atom Again although not as toxic as HD CEES is a blister agent and dilute aqueous samples were prepared using appropriate safety equipment Both the Raman and SER spectra of CEES are similar and dominated by bands between 600 and 800 cm-1 These are associated with C-Cl and C-S stretching modes which are tentatively assigned to 648 and 747 cm-1 in the Raman spectra respectively The shoulder at 630 cm-1 the overlapped band at 660 cm-1 and the strong band at 692 cm-1 could also be due to these modes or their asymmetric counterparts It is worth noting that theoretical calculations assign the 692
Figure 2 A) SER and B) NR spectra of KCN Conditions A) 1 mgml in tap water 100 mW of 785 nm at sample 1-min acquisition time B) solid 300 mW of 785 nm 5-min All spectra are 8 cm-1 resolution
A
B
Figure 3 A) SER and B) NR spectra of CEES Conditions A) 1 vv (10 mgml) in MeOH 100 mW of 785 nm 1-min acquisition time B) neat 300 mW of 785 nm 5-min
A
B
Cl-CH2-CH2-S-CH2-CH3
A
B
Figure 4 A) SER and B) NR of HD Conditions A) 1mgml in tap water B) pure both 100 mW of 785 nm 1-min
Cl-CH2-CH2-S-CH2-CH2-Cl
SPIE-2003-5269 20
expected strong interaction between chlorine and silver and adds support to the assignment of this band to a C-Cl stretch Weaker overlapping bands occur at 670 692 and 724 cm-1 the latter possibly due to C-S stretching modes Again the alkane modes are apparent in the normal Raman spectra of HD but only a broad feature at 1300 to 1450 cm-1 suggests CH2 contributions in the SER spectrum Although the observed bands in the VX spectrum have not been assigned (Figure 4) a computer generated Raman spectrum29 predicts many of the same features with surprising accuracy and are used here Two intense bands at 460 and 530 cm-1 closely match predicted bands at 463 and 546 cm-1 assigned to a CH3-P=O bend and a PO2CS wag Three highly overlapped bands occur at 694 745 and 771 cm-1 matching predicted bands at 713 730 and 760 cm-1 The first
Table 2 Measured SER peak heights for the CN stretch at 2100 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
Figure 5 A) SERS and B) NR spectra of VX Conditions A) 1 vv (10 mgml) in MeOH B) pure sample both 100 mW of 785 nm 1-min acquisition time
A
B
two have been assigned to a C-S stretch and CH2bend respectively while the latter has been attributed to either a P-C stretch or an O-C-C stretch Although the 745 cm-1 band may alternatively be assigned to a C-S stretch based on the previous measurements of CEES and HD The relatively intense bands at 890 1106 1218 1445 and 1465 cm-1 also match predicted bands at 880 1108 1216 1440 and 1464 cm-1 that are assigned to a C-C stretch CH3 rock N-C3 stretch various C-H3 bends and C-H bends respectively Both the computer generated and the measured spectra contain numerous other less intense bands One is worth mentioning A unique band appears at 370 cm-1 that is predicted at 368 cm-1 and corresponds to an O-P=O bend The surface-enhanced Raman spectrum of VX is also rich with spectral features It has the unique low frequency band at 370 cm-1 as well as a second band at 380 cm-1 that is assigned tothe S-P-O bend predicted in the normal Raman spectrum at 388 cm-1 Based on the measured and predicted normal Raman spectra the following SERS assignments are given 460 cm-1 to the CH3-P=O bend 544 cm-1 to the PO2CS wag 738 cm-1 to a C-S stretch (based on arguments above) 890 cm-1
to a C-C stretch 1101 cm-1 to a CH3 rock and 1456
cm-1 to a C-H bend The ability of SERS to measure chemical agents in water containing real-world chemical interferents was tested by using the generic tap water described in Table 1 The ability to reproduce measurements was accomplished by preparing three separate water stock solutions which were used to prepare three sample solutions each of 1 mgmL KCN in the generic tap water The samples defined as vials 1-9 were then measured at 4 points per vial in a semi-random fashion such that errors associated with stock solution preparation and errors associated with instrument drift could be identified No trends were apparent that signified such systematic errors Each spectrum collected consisted of 20 averaged scans taking 16 seconds at 8 cm-1 resolution The laser power at the sample was measured periodically during the day and it ranged from 102 to 105 mW spacccc
spot Vial 1 Vial 2 Vial 41 13754 13004 128192 13519 12692 12909
1 3 13541 12721 126394 13462 12648 12651
ave 13569 12766 12755Vial 6 Vial 3 Vial 5
1 11586 14068 10757 2 11236 14402 115122 3 11376 14546 11559
4 10894 11763 11214ave 11273 13695 11261
Vial 7 Vial8 Vial 91 11198 15804 11023
2 11243 15716 88753 3 11562 15216 11218
4 11655 15084 11077ave 11415 15455 10548
CN AVG STDEV ERR1 mgmL pk ht 12526 1560 1245
stock solution
SPIE-2003-5269 21
The CN measurements consisting of 36 data points produced an average height of 125 for the 2100 cm-1 peak with a standard deviation of 156 or 125 (Table 2) The HD and VX measurements were performed precisely the same way (Tables 3 and 4) For HD the 624 cm-1 peak was used for analysis and it had an average height of 53 with a standard deviation of 068 or 129 while for VX the 544 cm-1 peak was used for analysis and it had an average height of 1051 with a standard deviation of 308 or 293 The greater error in the VX measurements can be somewhat attributed to Vial 4 which produced lower SER signal intensities But removing this vial from the data set changes the standard deviation to 233 only a modest improvement
spot Vial 1 Vial 2 Vial 41 525 609 4682 45 675 484
1 3 527 644 5094 559 807 669
ave 51525 68375 5325Vial 6 Vial 3 Vial 5
1 521 574 575 2 536 53 4492 3 509 508 451
4 565 594 379ave 53275 5515 4635
Vial 7 Vial 8 Vial 91 457 589 413
2 583 497 4273 3 544 505 4
4 528 5 509ave 528 52275 43725
HD AVG STDEV ERR1 mgmL pk ht 530 068 1291
stock solution
spot Vial 1 Vial 2 Vial 41 1464 1034 4672 1485 99 61
1 3 1491 989 5684 1041 777 553
ave 137025 9475 5495Vial 6 Vial 3 Vial 5
1 1058 942 1293 2 697 121 965
2 3 727 1206 11124 689 1254 746
ave 79275 1153 1029Vial 7 Vial 8 Vial 9
1 1127 783 163 2 1358 812 1614
3 3 1371 908 15124 1097 875 1397
ave 123825 8445 153825
VX AVG STDEV ERR1 mgmL pk ht 1051 308 2925
stock solution
4 CONCLUSIONS In this paper we examined the ability of surface-enhanced Raman spectroscopy to reproducibly measure CN HD and VX in tap water without chemical interference Both normal and surface-enhanced Raman spectra were examined to select unique bands suitable to identify and quantify these chemical agents For SER measurements the 2100 cm-1 C-N stretch was used for CN the 624 cm-1 C-Cl stretch was used for HD and the 544 cm-1 PO2CS wag was used for VX It was determined that 1 mgmL samples of each of these chemicals measured 36 times in glass vials coated with a silver-doped sol-gel reproduced measurements with standard deviations of 125 129 and 293 It was further found that the 10 chemicals added to simulate generic tap water did not interfere with or alter the SER spectra It should be noted that the concentrations used in this study were considerably greater than those required by the JSAWM program Current work involves improving SER sensitivity and designing sampling systems with better reproducibility This includes the development of fractal silver and gold structures within the sol-gel matrix and the development of chemically selective sol-gels
ACKNOWLEDGMENTS The authors are grateful for the support of the US Army (DAAD13-02-C-0015 Joint Service Agent Water Monitor program) The authors would also like to thank Janet Jensen Ronald Crosier and Kristina Gonser for helpful discussions
Table 4 Measured SER peak heights for the VX band at 544 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
Table 3 Measured SER peak heights for the HD band at 624 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
SPIE-2003-5269 22
REFERENCES 1 Jensen JL NC Wong W R Loerop SPIE 4775-03 (2001) 2 JSAWM Requirements at wwwsbccomapgeaarmymilRDAecbcrtPRODSERJSAWMjsawmhtml 3 Erickson B ldquoThe Chemical Weapons Convention Redefines Analytical Challengerdquo Anal Chem 70 397A (1998) 4 Johnston RL Hoefler CM Fargo JC and Moberley B ldquoThe Defense Nuclear Agencyrsquos
ChemicalBiochemical Weapons Agreements Verification Technology Research Development Test and Evaluation Program and its Requirements for On-Site Analysisrdquo AT-ONSITE 5-8 (1994)
5 Black RM Clarke RJ Read RW and Reid MT ldquoApplication of GC-MS and GC-tandem MS to the analysis of chemical warfare samples found to contain residues of the nerve agent sarin sulphur mustard and their degradation productsrdquo J Chrom 662 301-321 (1994)
6 USA Army Center for Health Promotion and Preventive Medicine (USACHPPM) Technical Guideline 230A 7 Hoffland LD Piffath RJ Bouck JB rdquoSpectral signatures of chemical agents and simulantsrdquo Optical
Engineering 24 982-984 (1985) 8 Braue EHJ Pannella MGrdquoCIRCLE CELL FT-IR Analysis of Chemical Warfare Agents in Aqueous Solutionsrdquo
App Spectrosc 44 1513-1520 (1990) 9 Tseng C-H Mann CK and Vickers TJ ldquoDetermination of Organics on Metal Surfaces by Raman
Spectroscopyrdquo App Spectrosc 47 1767-1771 (1993) 10 Staff Reporter ldquoUS Army Tests Lidar to Detect Biological Toxinsrdquo Photonic Spectra pg 50 December 1998 11 Christesen SD Raman cross sections of chemical agents and simulants App Spectrosc 42 318-321 (1988) 12 Weaver MJ Farquharson S Tadayyoni MA ldquoSurface-enhancement factors for Raman scattering at silver electrodesrdquo J Chem Phys 82 4867-4874 (1985) 13 Alak AM and Vo-Dinh T ldquoSERS of Organophosphorous Chemical Agentsrdquo Anal Chem 59 2149-2153 (1987) 14 Lee Y Farquharson S ldquoRapid chemical agent identification by SERSrdquo SPIE 4378 21-26 (2001) 15 Farquharson S Maksymiuk P Ong K Christesen S ldquoChemical agent identification by surface-enhanced Raman spectroscopyrdquo SPIE 4577 166-173 (2001) 16 Spencer KM Sylvia J Clauson S and Janni J ldquoSurface Enhanced Raman as a Water Monitor for Warfare
Agents in Waterrdquo SPIE 4577 158-165 (2001) 17 Tessier P Christesen S Ong K Clemente E Lenhoff A Kaler E Velev O ldquoOn-line spectroscopic characterization of sodium cyanide with nanostructured gold surface-enhanced Raman spectroscopy substratesrdquo
App Spectrosc 56 1524-1530 (2002) 18 Farquharson S W W Smith S Elliott and J F Sperry Rapid biological agent identification by surface- enhanced Raman spectroscopy SPIE 3855110-116 (1999) 19 Farquharson S WW Smith YH Lee S Elliott and J F Sperry Detection of biological signatures A
comparison of electrolytic and metal-doped sol-gel surface-enhanced Raman media SPIE 4575 62-72 (2002) 20 Guzelian AA Sylvia J Janni J Clauson S and Spencer KM ldquoSERS of whole cell bacteria and trace levels of biological moleculesrdquo SPIE 4577 182-192 (2001) 21 Shende C Inscore F Gift A Maksymiuk P Farquharson S ldquoRapid dipicolinic acid extraction from Bacillus spores detected by surface-enhanced Raman spectroscopyrdquo App Spectrosc 58 accepted 22 Lee Y and Farquharson S ldquoSERS Sample Vials Based on Sol-Gel Process for Trace Pesticide Analysisrdquo SPIE
4206 140-146 (2000) 23 Farquharson S and Lee Y ldquoTrace Drug Analysis by SERSrdquo SPIE 4200-16 (2000) 24 Lee Y Farquharson S and Rainey P M Surface-Enhanced Raman Sensor for Trace Chemical Detection in
Water SPIE 3857 76-84 (1999) 25 Lee Y Farquharson S Kwong H and Shahriari M ldquoSol-Gel Chemical Sensor for Surface-Enhanced Raman
Spectroscopyrdquo SPIE 3537 252-260 (1998) 26 Farquharson S A Gift P Maksymiuk F Inscore and W Smith ldquopH dependence of methyl phosphonic acid
dipicolinic acid and cyanide by surface-enhanced Raman spectroscopyrdquo SPIE 5269-19 (2003) 27 Lide DR Ed Handbook of Chemistry and Physics CRC Press 77th Ed (1996-7) 28 Sosa C RJ Bartlett K KuBulat and WB Person ldquoA theoretical study of harmonic vibrational frequencies and
infrared intensities of XCH2CH2SCH2CH2X and XCH2CH2SH (= H Cl)rdquo J Phys Chem 93 577-588 (1993) 29 Hameka HF and JO Jenson ldquoComputer-generated predictions of the structure and the IR and Raman spectra of
VXrdquo ERDEC-TR-065 May 1993
- Chemagents Appendicespdf
-
- SPIE2001-4575-Bioagent-SERSapdf
-
- Detection of bioagent signatures A comparison of electrolytic and metal-doped sol-gel surface-enhanced Raman media
- Stuart Farquharson Wayne Smith and Yuan Lee
-
- Real-Time Analyzers 87 Church Street East Hartford CT 06108
-
- ABSTRACT
- 4 CONCLUSIONS
- 5 ACKNOWLEDGEMENTS
- 6 REFERENCES
-
170
Figure 4 SERS of 11 MM of PABA and PA in A) polar-negative and C) weakly polar-negative sol-gels The lower traces compare the pure chemicals B) 1 mgml PABA in polar-negative sol-gel and D) 1 PA in weakly polar-negative sol-gel while the insets magnify the minority species for clarity (x5 in A and x10 in B) Spectral conditions as in Figure 3 Following this development of selective sol-gels that maintained SER activity we measured cyanide and MPA (Figure 5) Not surprisingly the best sensitivity for both hydrolysis products was obtained using the polar-negative sol-gel The interaction of the cyanide anion with the silver surface is sufficient to shift the CequivN stretch observed at 2080 cm-1 in the normal Raman spectrum to 2145 cm-1 in the surface-enhanced Raman spectrum Furthermore the band is substantially broadened This anion has been extensively studied by electrolytic SERS and this shift and broadening have been attributed to the formation of a tetrahedral Ag(CN)3
2- surface structure23 Figure 5 Surface-enhanced (upper traces) and normal Raman spectra (bottom traces) of A) CN- and B) MPA in silver-doped TMOS SERS conditions as in Figure 3 and 1 mgmL Note MPA yields two distinct spectra for neutral (top) and acidic pH (middle) The normal Raman spectra employed pure powders 500 scans and 900 mW of 1064 nm SER measurements of MPA with the polar-negative sol-gel yielded two unique spectral signatures that depended on solution pH (Figure 5) For more neutral solutions the P-C stretch of MPA at 762 cm-1 dominates and the CH2 stretch at 2922 cm-1 appears The SN is sufficiently high that the anti-Stokes Raman shift at -762 cm-1 is observed For deprotonated MPA an oxygen-surface mode appears at 325 cm-1 (as well as its anti-Stokes complement) suggesting a strong interaction This results in substantial enhancement of the P-O-C mode at 1051 cm-1 (upper trace) Others report that this mode dominates the infrared spectra of nerve agents measured in water8 Comparison of the two spectra suggests the following molecule-to-surface orientations The appearance of the oxygen-surface and P-O-C modes in the upper spectrum of Figure 5B indicates that the tetrahedral molecule interacts with the silver surface through the deprotonated oxygen and is oriented end-on The
A B
Wavenumber (∆cm-1)
C A
Wavenumber (∆cm-1) Wavenumber (∆cm-1)
D B
Wavenumber (∆cm-1)
Proc SPIE Vol 4577
171
dominance of the P-C and the CH2 stretches and the disappearance of the P-O-C mode in the upper spectrum suggest the molecule is oriented with the methyl-hydroxyl-hydroxyl face towards the surface However considerably more research must be performed to verify these points Both chemical forms were measured as a function of concentration to determine the onset of monolayer coverage and estimate expected detection limits (Figure 6) Below monolayer coverage the signal to concentration dependence should be linear and the SN of any spectral measurement in this range can be used to predict the detection limit In the spectra presented here the peak height was used as the signal while the noise as root-mean-squared (RMS) was measured between 4400-4600 cm-1 Since noise is distributed evenly throughout the spectrum when transformed this region was used since it does not have any contributions from signals or baseline offsets Figure 6 shows a series of spectra for MPA along with a plot of the 1050 cm-1 band intensity (the noise was identical for this series) as a function of sample concentration A clear discontinuity appears in the vicinity of 01 mgmL (19 ppm) indicating the onset of monolayer coverage A detection limit defined as a SN of 3 was calculated for the 01 and 005gmL samples at 24x10-4 and 25x10-4 gL respectively A more modest detection limit of 101x10-4 gL was obtained using the 760 cm-1 band in the second series of concentration measurements These detection limits correspond to 46 and 190 ppb for a 3-min scan and 75 mW of incident laser power
Figure 6 A) Concentration dependence of MPA SERS measured in silver-doped TMOS) B) Concentrations are 001 005 01 05 1 gL (188 94 188 94 188 ppm) I760 series (bull) and I1050 series (∆)
Enhancement factors for cyanide and methyl phosphonic acid can be determined by estimating the number of molecules contributing to the surface-enhanced and normal Raman spectra The enhancement factor EF can be defined by the following equation
EF = (ISERSINR)bull(MNRMSERS) bull(PNRPSERS) bull(TNRTSERS)12
where I is the spectral band intensity M is the sample mass P is the incident laser power and T is the measurement time (or number of scans) for the two measurements For the normal Raman spectra a cylindrical scattering volume is assumed based on the laser area (28x10-7m2 6x10-4m diameter spot) and the penetration depth (1x10-3 m)24 The density of KCN and MPA as powders were measured at 0572 and 0516 gcm3 indicating that 16x10-4 and 144x10-4 g produced the normal Raman signals in Figure 5 respectively The number of molecules contributing to the SER spectra are those on the silver particles embedded in the sol-gel The total silver surface area can be determined from the average particle size concentration and the scattering volume Previous scanning electron micrographs showed the average silver particle size to be 40 nm in diameter (335x10-23m3)17 The silver concentration is 012M based on the reactant molar concentrations and dilution factors And the scattering volume is 76x10-
11m3 again based on a cylindrical scattering volume defined by a laser area of 28x10-7m2 and a sol-gel thickness of 27x10-
4m This volume contains 123x10-6g of silver equivalent to 35x109 silver particles with a collective surface area of 18x10-
5m2 However it may reasonably be assumed that at least half of the silver surface is in contact with the sol-gel matrix and unavailable for analyte interaction If we assume monolayer coverage and that each CN molecule occupies 15x10-20m2 then approximately 62x1014 molecules or 27x10-8g of CN contribute to the SER spectrum (20x10-19m2 46x1013 molecules 74x10-9g for MPA) Accordingly the EF for cyanide equals 48x104 ((180599) bull(16x10-427x10-8) bull(90075) bull(500100)12) The EF for MPA is considerably higher at 87x106 ((60326) bull(144x10-474x10-9) bull(90075) bull(500200)12)
0
20
40
60
80
100
120
140
0 02 04 06 08 1 12[MPA] (mgmL)
I (76
0)
0
100
200
300
400
500
600
I (10
50)
Wavenumber (∆cm-1)
A B
Proc SPIE Vol 4577
172
4 CONCLUSIONS Here we present for the first time surface-enhanced Raman spectra of Tabun and Sarin obtained using silver-doped sol-gels However the inferior enhancement suggested employing SER-active sol-gels varying in composition to improve both detection limits and selectivity of the target analytes To this end we successfully demonstrated the capabilities of four sol-gels that select for 1) polar-positive 2) polar-negative 3) weakly polar-positive and 4) weakly polar-negative chemical species p-aminobenzoic acid was used to show that silver could be used to attract polar-negative chemicals or functional groups (carboxylate anion) and that gold could be used to attract polar-positive chemicals (amine) while a mixture of p-aminobenzoic acid and phenylacetylene was used to show that tetramethyl orthosilicate preferentially solvates polar chemicals and that tetraethyl orthosilicate preferentially solvates weakly polar chemicals This increased sample control was applied to cyanide and methyl phosphonic acid two hydrolysis products of chemical warfare agents Exceptional results were obtained for methyl phosphonic acid allowing measurement of 1x10-2 gL for a 3-min scan and 75 mW of incident laser power with an estimated detection limit of 245x10-4 gL and an enhancement factor of 87x106 However this detection limit is 76 times less sensitive than required for the JSAWM (32x10-6gL for the G-agents) Finally we note that the measurements performed here employed an FT-Raman spectrometer This instrumentation was chosen over dispersive Raman instrumentation for the high wavelength accuracy afforded by the HeNe reference laser (Connes Advantage25) which would allow reliable spectral subtraction matching of observed spectra to stored library spectra and confident use of chemometric approaches Such data analysis is likely to be required to identify the chemical agents as well as distinguish hydrolysis products However this instrumentation which employs 1064 nm excitation and InGaAs detection sacrifices sensitivity We believe that the measurement sensitivity can be improved by at least two orders of magnitude by using shorter laser excitation wavelengths (eg 532 nm) This would provide more efficient Raman scattering (fourth power dependence on laser excitation wavelength) more efficient generation of plasmon modes and allow using more efficient detector material (Si vs InGaAs) These modifications are underway
5 ACKNOWLEDGEMENTS The authors would like to thank Dr R Yin and J Jensen of the US Army for supporting this work (Contract Number DAAD13-01-C-0019) They would also like to thank Advanced Fuel Research for making their laboratory facilities available
6 REFERENCES 1 Tu Anthony ldquoOverview of Sarin Terrorist Incidents in Japan in 1994 and 1995rdquo 6th CBW Protection Symposium
Stockholm Sweden 10-15 May 1998 2 Jensen JL NC Wong W R Loerop SPIE 4775-03 (2001) 3 JSAWM Requirements at wwwsbccomapgeaarmymilRDAecbcrtPRODSERJSAWMjsawmhtml 4 ldquoThe Chemical Weapons Convention Redefines Analytical Challengerdquo Analytical Chemistry News amp Features June 1
397A (1998) 5 Johnston RL Hoefler CM Fargo JC and Moberley B ldquoThe Defense Nuclear Agencyrsquos ChemicalBiochemical
Weapons Agreements Verification Technology Research Development Test and Evaluation Program and its Requirements for On-Site Analysisrdquo AT-ONSITE 5-8 (1994)
6 Black RM Clarke RJ Read RW and Reid MT ldquoApplication of gas chromatography-mass spectrometry and gas chromatography-tandem mass spectrometry to the analysis of chemical warfare samples found to contain residues of the nerve agent sarin sulphur mustard and their degradation productsrdquo J Chromatography 662 301-321 (1994)
7 Hoffland LD Piffath RJ Bouck JBrdquoSpectral signatures of chemical agents and simulantsrdquo Optical Engineering 24 982-984 (1985)
8 Braue EHJ Pannella MGrdquoCIRCLE CELL FT-IR Analysis of Chemical Warfare Agents in Aqueous Solutionsrdquo Applied Spectroscopy 44 1513-1520 (1990)
9 Tseng C-H Mann CK and Vickers TJ ldquoDetermination of Organics on Metal Surfaces by Raman Spectroscopyrdquo Applied Spectroscopy 47 1767-1771 (1993)
10 Staff Reporter ldquoUS Army Tests Lidar to Detect Biological Toxinsrdquo Photonic Spectra pg 50 December 1998
Proc SPIE Vol 4577
Proc SPIE Vol 4577
173
11 Christesen SD Raman cross sections of chemical agents and simulants Applied Spectroscopy 42 318-321 (1988) 12 Alak AM and Vo-Dinh T ldquoSERS of Organophosphorous Chemical Agentsrdquo Analytical Chemistry 59 2149-2153
(1987) 13 Norrod KL Sudnik LM Rousell D and Rowlen KL ldquoQuantitative Comparison of Five SERS Substrates
Sensitivity and Detection Limitrdquo Applied Spectroscopy 51 994-1001 (1997) 14 Lee Y and Farquharson S ldquoSERS Sample Vials Based on Sol-Gel Process for Trace Pesticide Analysisrdquo SPIE 4206
140-146 (2000) 15 Farquharson S and Lee Y ldquoTrace Drug Analysis by Surface-Enhanced Raman Spectroscopyrdquo SPIE 4200-16 (2000) 16 Lee Y Farquharson S and Rainey P M Surface-Enhanced Raman Sensor for Trace Chemical Detection in Water
SPIE 3857 76-84 (1999) 17 Lee Y Farquharson S Kwong H and Shahriari M ldquoSurface-Enhanced Raman Sensor for Surface-Enhanced Raman
Spectroscopyrdquo SPIE 3537 252-260 (1998) 18 Farquharson S Smith W Carangelo R C and Brouillette C ldquoIndustrial Raman Providing Easy Immediate Cost
Effective Chemical Analysis Anywhererdquo SPIE 3859 14-23 (1999) 19 Narayanan VA JM Bello JD Stokes and T Vo-Dinh Analusis 19 307-310 (1991) 20 Laserna JJ E L Torres and JD Winefordner Analytica Chemica Acta 469-480 (1987) 21 Torres EL and JD Winefordner Analytical Chemistry 59 1626-1632 (1987) 22 Suh JS DP DiLella M Moskovits J Phys Chem 87 1540-1544 (1983) 23 Benner RE R Dornhaus R Chang and BL Laube Correlations in the Raman spectra of cyanide complexes adsorbed
at silver electrodes with voltammograms Surface Science 101 341 (1980) 24 Chase D B and JF Rabolt Fourier Transform Raman Spectroscopy Acad Press Ch1 p 131 (1994) 25 Connes J Rev Opt Theor Instrum 40 45 (1961)
APPLIED SPECTROSCOPY 351
focusing the 488 nm laser beam 2 mm deep into thebulk of the crystals to avoid contributions from potentialdepletion layers As shown in Fig 1b the dependence ofthe Raman shift on the degree of deuteration is almostperfectly linear and ts very well with D 5 22684cmR1 24526 where D is the degree of deuteration (in )and R is the spectral mean of the PO4 vibration in cm21A linear correlation coef cient of 0998 indicates an ex-cellent linear dependence of the Raman peak shift withdegree of deuteration This result shows that the shift ofthe PO 4 peak is simply caused by the linear increase inatomic mass due to isotope substitution which decreasesthe length of hydrogen-like bonds
This excellent linear dependence allows us to map thepro le of the DH exchange layer at the surface of DKDPcrystals by acquiring Raman spectra and determining theposition of the PO 4 peak for various depths This methodis preferable over other methods such as determining thestrength of the OD vibration directly (eg at 715 cm21)because the position of the most intense peak in the Ra-man spectrum can be measured more precisely than theintensity of some of the weakest peaks in the spectrumThis is demonstrated in Fig 2 where depth-dependentRaman spectra (Fig 2a) and the resulting exchange layerpro les for two DKDP crystals are shown (Fig 2b) Thespectra in Fig 2a were obtained from a depth scan of aDKDP crystal with 75 degree of deuteration in thebulk grown at 45 8C The spectra start out as DKDP with30 deuteration close to the surface and approach thebulk DKDP spectrum within a few micrometers of depthThe fact that the relative degree of deuteration does notextend to 0 D is due to the limited depth resolution ofthe Raman microprobe which averages over 4 mm indepth Figure 2b depicts the resulting DH exchange layerpro les for this and a second crystal grown at 63 8Crespectively Both crystals had the same exposure to am-bient conditions and their main difference is the temper-ature at which they were grown The different exchangelayer pro les indicate that crystals grown at differenttemperatures have differen t proton conductiv ities 13
which leads to a difference in their rate of deuteriumdepletion The parameters controlling this behavior arecurrently the objective of a detailed study the results ofwhich will be reported elsewhere
CONCLUSION
In conclusion we have shown that the shift of the to-tally symmetric PO 4 stretch mode in the Raman spectrumof DKDP crystals scales linearly with degree of deuter-ation This allows us to correlate Raman peak positionsto deuteration levels in these crystals We have presenteda new technique to determine DH diffusion pro les inDKDP frequency conversion crystals based on micro-Ra-man spectroscopy This technique is fast inexpensiveand works under various environmental conditionswhich will allow us to better understand and control deu-terium depletion in DKDP crystals
ACKNOWLEDGMENTS
We would like to thank M Runkel for rst discovering DKDP crack-ing R Floyd for providing DKDP crystals and L Chase and A Burn-ham for their support and helpful discussions This work was performedunder the auspices of the US Department of Energy by the University
of California Lawrence Livermore National Laboratory under contract W-7405-Eng-48
1 J J De Yoreo A K Burnham and P K Whitman Int Mat Rev47 113 (2002)
2 C E Barker R A Sacks B M Van Wonterghern J A Caird JR Murray J H Campbell K Kyle R B Ehrlich and N DNielsen Proc SPIE-Int Soc Opt Eng 2633 501 (1995)
3 T Suratwala paper to be published4 Cleveland Crystals Inc httpwwwclevelandcrystalscom
KDPshtmltable5 E A Popova I T Savatinova and I A Velichko Sov Phys Solid
State 12 1543 (1971)6 I P Kaminow R C C Leite and S P S Porto J Phys Chem
Solids 26 2085 (1965)7 J A Subramony B J Marquardt J W Macklin and B Kahr
Chem Mat 11 1312 (1999)8 H Tanaka M Tokunaga and I Tatsuzaki Solid State Commun
49 153 (1984)9 R J Nelmes G M Meyer and J E Tibballs J Phys C 15 59
(1982)10 M A Yakshin D W Kim Y S Kim Y Y Broslavets O E
Sidoryuk and S Goldstein Laser Physics 7 941 (1997)11 I Takenaga Y Tominaga S Endo and M Kobayashi Solid State
Commun 84 931 (1992)12 C Krenn personal communication13 M Sharon and A K Kalia J Solid State Chem 21 171 (1977)
Rapid Dipicolinic Acid Extractionfrom Bacillus Spores Detectedby Surface-EnhancedRaman Spectroscopy
STUART FARQUHARSON ALAND GIFT PAUL MAKSYMIUK andFRANK E INSCOREReal-Time Analyzers Inc East Hartford Connecticut06108
Index Headings Dipicolinic acid Bacillus spores Anthrax Surface-enhanced Raman spectroscopy
INTRODUCTION
The anxiety caused by the distribution of anthrax en-dospores through the US postal system in October 2001was exacerbated by the long time required for positiveidenti cation of the Bacillus anthracis spores and the un-known extent of their distribution Since that time manymethods capable of rapid eld analysis have been inves-tigated to augment or replace the laboratory method ofgrowing microorganisms in culture media which takesdays to perform12 Prominent among these approachesare polymerase chain reactions (PCR)3 immunoassaysand detection of calcium dipicolinate as a biochemicalsignature PCR employs primers to separate organism-
Received 10 October 2003 accepted 14 November 2003 Author to whom correspondence should be sent
352 Volume 58 Number 3 2004
speci c nucleic acid sequences (eg capsular protein en-coding gene for Bacillus anthracis)4 and polymerases toamplify the segment until it is detectable Recently am-pli cation times have been substantially reduced andcomplete analysis can now be performed in an hour orless Immunoassay methods are also being developed thatuse competitive binding of the bioagent (as an antigen)and its labeled conjugate for a limited number of anti-bodies Although analyses can be performed in under 30minutes a well-de ned anthrax antigen has not yet beenidenti ed5ndash7 and consequently the false-positive rate isunacceptably high8
A number of other methods are being developed witha focus on the detection of calcium dipicolinate (CaDPA)and its derivatives as a B anthracis signature This is sobecause only spore-forming bacteria contain CaDPA andthe most common potentially interfering spores such aspollen and mold spores do not Relatively fast methodshave been developed to chemically extract CaDPA andthen detect it directly by uorescence9 or indirectly byluminescence1011 In the latter case hot dodecylamine(DDA) has been used to extract dipicolinic acid (DPA)and terbium has been utilized to form a highly lumines-cent DPA complex11 Although measurements have beenperformed in as little as ve minutes it was found thatas many as three concentration-dependent complexes canform each with different lifetimes This coupled withthe fact that the Tb31 cation produces the same lumines-cence spectrum makes determinations of low spore con-centrations problematic
It has been long known that Raman spectra of Bacillispores are dominated by bands associated with CaDPA12
and that these spectra may provide a suitable anthrax sig-nature at the genus level13 Since that time considerableimprovements in Raman instrumentation have led to lab-oratory measurements of single Bacilli spores14 and to eld measurements of spores captured from a mail-sort-ing system15 However the single spore measurementsrequired complex instrumentation that is not ruggedwhile the eld measurements required milligrams of sam-ple Furthermore the Raman spectra of both measure-ments contained uorescence contributions that would in-crease uncertainty in quanti cation
In related research we demonstrated that nanogramquantities of DPA could be detected by uorescence-freesurface-enhanced Raman spectroscopy (SERS)16 We alsodemonstrated that microliter volumes of chemicals canbe detected by SERS using metal-doped sol-gel-packedglass capillaries17 Towards the goal of developing a rap-id eld SERS-based anthrax spore detector we havecombined our previous research and we now report thatDPA can be extracted from a 10 mg B cereus spore sam-ple using DDA in 1 minute and can be detected by SERSin an additional 1 minute
EXPERIMENTAL
Dipicolinic acid (26-pyridinedicarboxylic acid DPA)and dodecylamine (DDA) were used as received fromSigma-Aldrich (Milwaukee WI) Lyophilized B cereusspores prepared according to the literature13 were sup-plied by the University of Rhode Island and used as re-ceived Multiple particles approximately 01 mm3 each
were separated and weighed at 5 to 15 mg representing05 to 15 million spores The sample masses were con-sistent with a previous determination of spore density at0081 gmL that indicated a high degree of entrained air
All chemicals used to prepare the silver-doped sol-gelcoated capillaries were also obtained and used as receivedfrom Sigma-Aldrich According to previously publishedprocedures17 two precursor solutions were preparedmixed and then drawn into 1-mm-diameter glass capil-laries The silver amine precursor consisted of a 51 vvratio of 1 N AgNO 3 to 28 NH3OH while the alkoxideprecursor consisted of a 21 vv ratio of methanol to te-tramethyl orthosilicate The alkoxide precursors weremixed with silver amine precursor in an 81 vv ratioApproximately 015 mL was drawn into the capillarycoating a 15-mm length After sol-gel formation the in-corporated silver ions were reduced with dilute sodiumborohydride which was followed by a water wash to re-move residual reducing agent
A 100 mL drop of a 50 mM DDA solution in ethanolpre-heated to 78 8C was added to each of the B cereusparticles to digest the spore coat After 1 minute the re-sultant solution was drawn into a SER-active capillarythat was immediately xed horizontally to an XY posi-tioning stage (Conix Research Spring eld OR) just in-side the focal point of an f 07 aspheric lens The lensfocused the beam into the sample and collected the scat-tered radiation back along the same axis A dichroic lter(Omega Optical Brattleborough VT) was used to re ectthe excitation laser to the lens and pass the Raman scat-tered radiation collected by the lens An f 2 achromatwas used to collimate the laser beam exiting a 200-mm-core-diameter source ber optic while a second f 2 ach-romat was used to focus the scattered radiation into a 365mm ber optic (Spectran Avon CT) A short-pass lterwas placed in the excitation beam path to block the sil-icon Raman scattering generated in the source ber fromre ecting off sampling optics and reaching the detectorA long-pass lter was placed in the collection beam pathto block the sample Rayleigh scattering from reachingthe detector A 785 nm diode laser (Process InstrumentsInc model 785-600 Salt Lake City UT) was used todeliver 100 to 150 mW of power to the sample A Fouriertransform Raman spectrometer (Real-Time Analyzersmodel IRA-785 East Hartford CT) and a silicon photo-avalanche detector (Perkin Elmer model C30902S Stam-ford CT) were used to acquire the SER spectra
RESULTS AND DISCUSSION
As an initial experiment the SER spectrum of 1 gLof DPA in water was measured using the newly devel-oped silver-doped sol-gel-coated capillaries (Fig 1A) Atthis concentration a high signal-to-noise ratio (SN) isobtained in 1 min In fact a reasonable spectrum is ob-tained in the same time frame for 1 mgL (Fig 1B) TheSER spectra are reasonably similar to the normal Raman(NR) spectrum obtained for a saturated solution of DPAin 1 N KOH (Fig 1C) and the following band shifts areobserved (NR to SER) 647 to 657 cm21 817 to 815cm21 998 to 1008 cm21 1384 to 1382 cm21 1434 to1428 cm21 and 1569 to 1567 cm21 Many of these bandshave been previously assigned1213 such as 998 cm21 to
APPLIED SPECTROSCOPY 353
FIG 1 SERS of DPA in water using silver-doped sol-gel-coated glasscapillary for (A) 1 gL and (B) 1 mgL (C ) NR of saturated DPA in 1N KOH in a glass capillary Spectral conditions (A) and (B) 150 mWof 785 nm 1-min acquisition time (C) 450 mW of 785 nm 5-minacquisition time both 8 cm 21 resolution
FIG 2 SERS of DPA extracted from 10 mg B cereus particle using100 mL of 50 mM hot DDA acquired in (A) 1 minute and (B) 2 seconds(C ) Attempted SERS of 50 mM hot DDA in ethanol using silver-dopedsol-gel-coated glass capillary acquired in 1 min Spectral conditions150 mW of 785 nm 8 cm21 resolution
the symmetric ring stretch 1384 cm21 to the OndashCndashOsymmetric stretch 1428 cm21 to the symmetric ring CndashH bend and 1569 cm21 to the asymmetric OndashCndashOstretch
The rst B cereus samples consisted of 2 mg of sporesin 2 mL of 5 mM hot DDA The samples were main-tained at 78 8C for 40 min and while hot approximately10 mL was drawn into a SER-active capillary Since spec-tra of DPA were obtained for these initial samples small-er spore masses higher DDA concentrations and shorterheating periods were examined In due course it wasfound that 10 mg of spores could be digested by 100 mLof 50 mM hot DDA in one minute and detected (Fig2A) In fact the signal was suf ciently intense that it canbe observed in as little as two seconds (Fig 2B) Theamount of DPA that was extracted was estimated to bebetween 5 and 10 mgL by comparing the signal intensityof the 1008 cm21 band to that measured for DPA in waterThis is consistent with previous research that found thatthe majority of the DPA is extracted from spores usingDDA11 and that B cereus spores contain approximately10 DPA by weight18 The SN of 127 for the 1008 cm21
band in the 1-minute SER spectrum suggests a limit ofdetection of approximately 250 ng of B cereus sporesbased on a SN of 3 Finally it should be noted that DDAdid not produce a detectable SER spectrum as shown inFig 2C
CONCLUSION
We have demonstrated that by combining rapid extrac-tion of dipicolinic acid from Bacillus cereus spores withchemical identi cation by surface-enhanced Raman spec-troscopy as little as 10 mg of spores can be detected Infact the entire measurement from the time of adding hotdodecylamine to the spores to the time when the dipi-colinic acid SER spectrum is acquired and analyzedcould be performed in less than two minutes The abilityof this method to distinguish between spore-forming bac-teria such as Bacillus anthracis and non-DPA containingpowders could help prevent costly shutdowns associated
with the appearance of suspicious material or intentionalmailing of common substances as an anthrax hoax Thismethod could also prove useful in detecting the locationof anthrax endospores in mail distribution facilities if an-other veri ed attack should occur
Research continues to fully characterize the surface-enhanced Raman spectroscopy signal intensities as afunction of sample concentration and to explore otherextractants that do not require the use of elevated tem-perature
ACKNOWLEDGMENTS
The authors are grateful for the support of the National Science Foun-dation (DMI-0296116 and DMI-0215819) and the US Army(DAAD13-02-C-0015 Joint Service Agent Water Monitor program)The authors are indebted to Chetan Shende for preparing the sol-gelcapillaries The authors also thank James Gillespie Nicholas Fell andAugustus Fountain for providing important background informationMark Farquharson for laboratory support and Professor Jay Sperry ofthe University of Rhode Island for supplying B cereus spores
1 V A Pasechnik C C Shone and P Hambleton Bioseparations 3267 (1993)
2 P J Jackson M E Hugh-Jones D M Adair G Green K K HillC R Kuske L M Grinberg F A Abramova and P Keim ProcNatl Acad Sci USA 95 1224 (1998)
3 B R Glick and J J Pasternak Molecular Biology Principles andApplications of Recombinant DNA (ASM Press Washington DC1994)
4 C A Bell J R Uhl T L Had eld J C David R F Meyer TF Smith and F R Cockerill III J Clin Microbiol 40 2897(2002)
5 D L Gatto-Menking H Yu J G Bruno M T Goode M Millerand A W Zulich Biosens Bioelectron 10 501 (1995)
6 J J Quinlan and P M Foegeding J Rapid Methods AutomationMicrobiol 6 1 (1998)
7 A A Hindle and E A H Hall Analyst (Cambridge UK) 1241599 (1999)
8 M S Ascher US Department of Health amp Human Services(httpwwwhhsgovophppresentationsAscherdoc)
9 R Nudelman B V Bronk and S Efrima Appl Spectrosc 54 445(2000)
10 D L Rosen C Sharpless and L B McBrown Anal Chem 691082 (1997)
354 Volume 58 Number 3 2004
11 P M Pellegrino N F Fell Jr and J B Gillespie Anal ChimActa 455 167 (2002)
12 W H Woodruff T G Spiro and C Gilvarg Biochem BiophysRes Commun 58 197 (1974)
13 E Ghiamati R S Manoharan W H Nelson and J F SperryAppl Spectrosc 46 357 (1992)
14 A P Esposito C E Talley T Huser C W Hollars C M Schal-dach and S M Lane Appl Spectrosc 57 868 (2003)
15 S Farquharson L Grigely V Khitrov W W Smith J F Sperryand G Fenerty J Raman Spectrosc paper accep ted (2003)
16 S Farquharson W W Smith S Elliott and J F Sperry SPIE-IntSoc Opt Eng 3855 110 (1999)
17 S Farquharson and P Maksymiuk Appl Spectrosc 57 479(2003)
18 F W Janssen A J Lund and L E Anderson Science (Washing-ton DC) 127 26 (1958)
SPIE -2003-5269 117
pH dependence of methyl phosphonic acid dipicolinic acid and cyanide by surface-enhanced Raman spectroscopy
Stuart Farquharson Alan Gift Paul Maksymiuk Frank Inscore and Wayne Smith
Real-Time Analyzers 87 Church Street East Hartford CT 06108
ABSTRACT US and Coalition forces fighting terrorism in Afghanistan and Iraq must consider a wide range of attack scenarios in addition to car bombings Among these is the intentional poisoning of water supplies to obstruct military operations To counter such attacks the military is developing portable analyzers that can identify and quantify potential chemical agents in water supplies at microgram per liter concentrations within 10 minutes To aid this effort we have been investigating the value of a surface-enhanced Raman spectroscopy based portable analyzer In particular we have been developing silver-doped sol-gels to generate SER spectra of chemical agents and their hydrolysis products Here we present SER spectra of methyl phosphonic acid and cyanide as a function of pH an important factor affecting quantitation measurements which to our knowledge has not been examined In addition dipicolinic acid a chemical signature associated with anthrax-causing spores is also presented Keywords Chemical warfare agents agent detection agent hydrolysis SERS Raman spectroscopy homeland security
1 INTRODUCTION In the past decade the Unites States and its allies have been challenged by a different kind of warfare exemplified by the terrorist attacks of September 11 2001 Suicide bombings and the use of chemical agents are the norm and military personnel must consider a wide range of attack scenarios Among these is the intentional poisoning of water supplies to obstruct military operations in Afghanistan and Iraq Currently colorimetric paper is used to detect agents on-site while gas chromatography combined with mass spectrometry (GCMS) is used in mobile support laboratories However both methods have severe drawbacks The paper changes color in response to contact with many chemicals besides CWAs causing a high incidence of false positives1 while GCMS although very chemically specific requires hours to perform and constant re-calibration234 Military operations would be greatly aided by a portable analyzer that can identify and quantify potential chemical agents at concentrations that impact safety This includes the analysis of drinking water supplies distribution and storage systems To meet this goal the Department of Defense has been investigating numerous approaches under the auspices of the Joint Service Agent Water Monitor (JSAWM) program5 The JSAWM must be capable of identifying the chemical agents with no false-positives and quantifying the agents at microgram per liter concentrations within 10 minutes678 Recently we and others have been investigating the ability of surface-enhanced Raman spectroscopy (SERS) to measure chemical agents9-12 bioagents13-17 and their hydrolysis products in water SERS employs the interaction of surface plasmon modes of metal particles with target analytes to increase scattering efficiency by as much as 1 million times18 In 1987 the potential of this approach to measure CWAs was demonstrated by measuring a series of pesticides19 Several of these organophosphonates have chemical structures similar to CWAs in particular P=O functional groups In our studies we have been employing silver-doped sol-gels to promote the SER effect The porous silica network of the sol-gel matrix offers a unique environment for immobilizing and stabilizing SER-active metal particles20-23 The sol-gels have been coated on the internal walls of standard glass vials to allow rapid SER analysis of small sample quantities without preparation We have measured over 100 chemicals with enhancements of 104 to 106 demonstrated reversible measurements in a flowing system reproducible measurements from vial-to-vial and measurements in multiple solvents including water20-23 Previously we used these vials to perform preliminary measurements of cyanide (CN) methylphosphonic acid (MPA) and dipicolinic acid (DPA) MPA is a hydrolysis product of the nerve agents (eg sarin Reaction 1) and may be a valuable indicator of nerve agent usage particularly since the alkyl methylphosphonic acids are relatively more stable than their corresponding parent complexes24 DPA is
SPIE -2003-5269 118
a chemical signature of spore forming bacteria such as Bacillus anthracis And in light of the inability to rapidly detect the anthrax spores distributed through the US mail in October 2001 a number of methods are being developed to extract and analyze this signature Reaction 1 Stepwise hydrolysis of Sarin to form hydrofluoric acid (HF) isopropyl methylphosphonic acid (IMPA) then methyl phosphonic acid (MPA) and 2-propanol In our previous SERS investigations MPA and DPA were measured at 50 and 100 mgL respectively In both cases limits of detection (LOD) were estimated at 100 microgL providing encouragement in that SERS may satisfy the needs of the JSAWM Since it has been shown that pH can substantially influence the intensity of SER bands25 which would clearly influence quantitative analysis we undertook the present study to determine the severity of these effects for cyanide methyl phosphonic acid and dipicolinic acid Furthermore we previously observed a band at 1050 cm-1 for MPA6 possibly due to an anion formed at basic pH Here we investigate the source of this spectral anomaly
2 EXPERIMENTAL All chemicals including potassium cyanide methyl phosphonic acid dipicolinic acid and those used to prepare the silver-doped and gold-doped sol-gel coated vials were obtained and used as received from Sigma-Aldrich All samples were prepared in HPLC grade water (Fischer Scientific Fair Lawn NJ) for SERS measurements The pH of these samples was adjusted using dilute nitric acid or potassium hydroxide and verified using a pH electrode (Corning Inc Corning NY) that had been calibrated with pH 400 700 and 1000 buffered standards from Fischer Scientific Once prepared the samples were transferred into the silver-doped sol-gel vials (Real-Time Analyzers Inc East Hartford CT) The vials were coated in a manner similar to that previously reported by adding ammonium hydroxide to a solution of silver nitrate tetramethyl orthosilicate (TMOS) and methanol20 Gold-vials were coated by adding nitric acid to a solution of gold tetrachloride TMOS and methanol The two precursor solutions were prepared mixed and transferred to 2-mL glass vials dried and heated After sol-gel formation the incorporated metal ions were reduced with dilute sodium borohydride (1mgmL) which was followed by a water wash to remove residual reducing agent After the resultant analyte solution was introduced the SER-active vial was immediately fixed horizontally to an XY positioning stage (Conix Research Springfield OR) just inside the focal point of an f07 aspheric lens The lens focused the beam into the sample and collected the scattered radiation back along the same axis A dichroic filter (Omega Optical Brattleborough VT) was used to reflect the excitation laser to the lens and pass the Raman scattered radiation collected by the lens An f2 achromat was used to collimate the laser beam exiting a 200 microm core diameter source fiber optic while a second f2 achromat was used to focus the scattered radiation into a 365 microm fiber optic (Spectran Avon CT) A short pass filter was placed in the excitation beam path to block the silicon Raman scattering generated in the source fiber from reflecting off sampling optics and reaching the detector A long pass filter was placed in the collection beam path to block the sample Rayleigh scattering from reaching the detector A 785 nm diode laser (Process Instruments Inc model 785-600 Salt Lake City UT) was used to deliver 100 to 150 mW of power to the sample A Fourier transform Raman spectrometer (Real-Time Analyzers model IRA-785 East Hartford CT) and a silicon photo-avalanche detector (Perkin Elmer model C30902S Stamford CT) were used to acquire the SER spectra
3 RESULTS AND DISCUSSION In a previous study of MPA6 aimed at developing a concentration calibration curve and determining limits of detection (LOD) we observed an anomaly at 1050 cm-1 Since it was found that the band intensity changed as a function of concentration the band must be associated with a sample parameter Two possible parameters photon flux and pH are examined here The first parameter was investigated by irradiating a 1mgmL MPA sample in a SER-active vial with laser powers of 200 mW and above and monitoring spectral changes It was immediately found that the 1050 cm-1 band
2O+ H HF + +OH
OH3H C
OP
OF
CH
CH3
3
3H C
OCP
OOH
CH
CH3
3
3H C
O
CPHO
CH
CH
3
3
C
Sarin IMPA MPA 2-propanol
SPIE -2003-5269 119
appeared and grew as a function of time and that the higher the laser power the faster the growth Figure 1 shows the growth of the 1050 cm-1 band over the course of 30 minutes when using 150 mW of 785 nm excitation while Figure 2A shows that the growth can be fit with a first order exponential equation namely I1050 = 03+05e-013t Furthermore the 760 cm-1 band could be fit with a first order decay equation with an identical exponential rate constant ie I760 = 08-
08e-013t The rates represent classical first order kinetics and their correspondence allows one to conclude that MPA is being transformed one-for-one into a photo-generated product At this time the photoproduct has not been positively identified but phosphonic (phosphorous) acid and phosphonate are likely candidates since the symmetric P(OH)3 stretch occurs at ~1050cm-1 Our studies show that a reduction in laser power to 100 mW at the sample for MPA essentially eliminates this degradation process This laser power or lower was used for further measurements
Methyl phosphonic acid is a diprotic acid that stepwise dissociates into two anions MPA- and MPA= according to the following reactions26
MPA MPA- + H+ pKa1 = 212 Reaction 2
MPA- MPA= + H+ pKa2 = 729 Reaction 3 The relative concentrations of MPA MPA- and MPA= can be determined at any pH by expressing [MPA] and [MPA=] in terms of [MPA-] using Reactions 2 and 3 and summing all three to equal the total starting concentration here 2 mgmL (0021M MW = 9602) viz
[MPA] + [MPA-] + [MPA=] = 0021M Equation 1 substituting from Reactions 1 and 2
([H+][MPA-])K1a + [MPA-] + (K2a[MPA-])[H+] = 0021M Equation 2 rearranging [MPA-] = 0021M(1+[H+]K1a + K2a[H+]) Equation 3 The relative concentrations of MPA MPA- and MPA= as a function of pH are shown in Figure 3 It is worth noting that near neutral pH both MPA- and MPA= will be present To confirm that the SER signal followed this pH dependence a starting solution consisting of 20 mg of MPA in 10 mL HPLC grade water was prepared and brought to pH of 20 using dilute nitric acid From this solution 2 mL were added to a SER-active vial and the SER spectrum recorded At this pH a peak at 760 cm-1 was barely discernable The 2 mL solution was returned to the starting solution and the pH was re-measured to correct for any changes that the silver-doped sol-gel vials might cause In most cases the change was less than 02 pH units and the pH is reported as the before and after average Next the pH of the
Figure 2 A) Exponential growth of 1050 cm-1 band and B) exponential decay of 760 cm-1 band for spectral series in Figure 1
Figure 1 Growth of 1050 cm-1 band as a function of time due to exposure to 150 mW of 785 nm Spectra are 5 sec each collected every 100-sec from 0 to 30-min
B
570 770 970 1170 1370Raman Shift (cm-1)
Arbit
rary
Unit
s
MPA Photodegradation
600 800 1000 1200 1400 Wavenumber (cm-1)
A
0 10 20 30 0 10 20 30 time (min) time (min)
Ram
an In
tens
ity (r
elat
ive)
30
min
0
SPIE -2003-5269 120
starting solution was adjusted to 325 using dilute KOH Again 2 mL were added to a vial and the SER spectrum recorded At this pH a reasonably strong 760 cm-1 band was observed This process was repeated as spectra were recorded at pHs of 70 74 75 79 85 and 100 A total of 1 mL of KOH was added diluting the total concentration by 10 Next the pH of the starting solution was made acidic by adding dilute nitric acid dropwise This time spectra were recorded at pHs of 72 69 64 and 37 Figure 4 shows the SER spectra for representative pHs (spectra were left out to simplify the figure) while Figure 3 shows the 760 cm-1 peak intensities as a function of pH (The band intensities were adjusted to compensate for dilution effects caused by the addition of HNO3 and KOH then normalized to 0021 M for the most intense band observed at pH 37) It is clear from Figure 3 that the 760 cm-1 band follows the MPA- concentration as a function of pH and must be assigned to this anion No bands were observed that corresponded to MPA or MPA= The lack of an MPA SER spectrum may be due to the absence of an attraction between the neutral analyte and the electropositive silver surface The same reasoning suggests that a strong SER spectrum should be observed for MPA= but it is not and a satisfactory explanation has not been found
0000
0005
0010
0015
0020
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14pH
Con
cent
ratio
n [M
]
pK1 = 212 pK2 = 729
MPA- MPA=
MPA
Figure 4 SER spectra of 002M MPA as a function of pH Conditions 100 mW of 785 nm 36 scans (1 min) 8 cm-1 recorded 2 min after sample introduction pH 19 and 20 not apparent on this scale pH 69 and 74 near identical to 70 and 75 and not shown for clarity
00
02
04
06
08
10
12
14
16
18
0 5 10 15 20 25 30Measurement Number
Ram
an In
tens
ity (7
60 c
m-1
)
Figure 3 Concentration dependence of MPA MPA- and MPA= as a function of pH for a 002M sample Intensity of 760 cm-1
band from Figure 6 as a function of pH ( for increasing basic adjustment for increasing acidic adjustment error was measured at ~10 for pH 64)
Figure 5 SER spectra of 002M MPA at pH 64 measured around a vial at three heights (9 points per height) Conditions as in Figure 4 but 10-sec scans
Since these measurements involved the removal and replacement of the SER-active vial in the sample holder to remove and add sample variation in the intensity as a function of vial position was minimized by illuminating the exact same height along the vial wall But this does not account for variability of the SERS response of the sol-gel coating around the vial To analyze this effect a vial containing MPA at pH 64 was rotated at ~ 40o
intervals at the original height and 18rdquo above and below this value Figure 5 shows the intensity of the 760 cm-1
band for the 27 positions It was found that the average value was 137plusmn014 an RSD of 10 overall and 5 for each height An error bar is included in Figure 4 for the pH 64 measurement
SPIE -2003-5269 121
With the above analyses in mind a preliminary investigation of the SER spectral response for MPA (as MPA-) as a function of concentration was performed A single vial was used for these measurements beginning with 1 mgL followed by measurements of 10 100 and 1000 mgL In all cases the pH was ~7 and 3 positions around the vial were measured per concentration Since the 760 cm-1 band was not observed for concentrations of 1 or 10 mgL using 100 mW of 785 nm the laser power at the sample was raised to 200 mW beginning with the 10 mgL concentration Photo-degradation was largely avoided (and not observed) by exposing the sample for only 33 seconds per spectral acquisition Representative spectra for 10 100 and 1000 mgL are shown in Figure 6 while a plot of the 760 cm-1 band intensity as a function of concentration is shown in Figure 7 These values were also used to estimate limits of detection based on the signal-to-noise ratio (SN) of the 760 cm-1 band using 100 mW of 785 nm laser excitation and a 10-min acquisition time As summarized in the Figure 7 inset the lower the measured concentration the lower the predicted LOD Even if as estimated 210 microgL could be measured using the silver-doped sol-gel vials an improvement of a factor of 70 is still required to meet JSAWM goals of measuring 3 microgL in 10-minutes Similar to MPA DPA is a diprotic acid (pKa1 = 216 and pKa2 = 692) and variations in pH will effect the relative concentrations of DPA DPA- and DPA= and possibly the SER spectra and band intensities This could prove significant if an acid or base is used to denature anthrax spores with the goal of extracting and analyzing DPA The MPA pH study described above was mimicked for DPA except that the starting solution consisted of 20 mg of DPA in 20 mL HPLC grade water (60x10-3M MW = 1671) The initial solution had a pH of 245 which was made basic by dilute KOH to pHs of 355 433 487 559 1069 and 1166 SER spectra were recorded at each pH using 100 mW of 785 nm and a 44-sec acquisition time Next one drop of concentrated nitric acid was used to remake the solution acidic at a pH of 200 Again sequential pH measurements were performed at 383 510 735 and 822 The solution pH was made acidic a third time but to pH 219 171 then 135 Throughout this process no more than 20 drops of acid or base were added and therefore the concentration was diluted by no more than 10 Most of the spectral bands showed a minor decrease in intensity as a function of increasing pH values However the bands at 525 and 795 cm-1 showed the most dramatic changes which occurred at acid pH Figure 8 shows the SER spectra of DPA for the spectral region and pH range of interest The identity of the DPA species was determined by plotting the normalized peak intensities with the lowest value set to 0 and the highest to 0006 M as a function of pH and overlaying these values on a plot of the relative concentrations for DPA DPA- and DPA= as previously done for MPA (Figure 9) As can be seen the 525 cm-1 band clearly corresponds to DPA The correspondence of the 795 cm-1 band to this species is less clear as the band retains intensity until pH of 55 This can be attributed to contribution to the overlapping band at 810 cm-1 which does not change as a function of pH The fact that most bands are observed at all pHs suggest that the primary interaction with silver is through the ring nitrogen This is supported by the fact that the most intense band occurs at 1008 cm-1 attributed to a symmetric ring breathing mode and that this interaction has been characterized for pyridine in numerous papers27
0
1
2
3
4
5
6
7
0 200 400 600 800 1000 1200
MPA Concentration (mgL)
760
Ban
d In
tens
ity
Figure 6 SER spectra of MPA in water at A) 1000 B) 100 and C) 10 mgL Conditions pH of 7 silver-doped sol-gel coated vial 200 mW of 785 nm 33-sec 8 cm-1 resolution
Figure 7 Plot of SER intensity of 760 cm-1 band of MPA as a function of concentration using 200 mW of 785 nm Inset table includes average intensity LOD standard deviation and percent deviation for each concentration but for 100 mW and 10 min
conc (mgL) LOD ave stddev dev10 021 002 1021100 072 010 14301000 312 040 1280
A
B C
SPIE -2003-5269 122
The ability to detect dipicolinic acid (DPA) by SERS is demonstrated in Figure 10 Even at 1 mgL the primary bands are visible The intensity of the dominant ring breathing mode observed at 1008 cm-1 for DPA in water (pH = 25 to 55) is plotted as a function of concentration in Figure 11 These values were also used to estimate limits of detection based on the SN of the 1008 cm-1 band using 100 mW of 785 nm laser excitation and a 10-min acquisition time Again the lower the measured concentration the lower the predicted LOD (see Figure 11 inset) and detection of 160 microgL is possible Gastrointestinal anthrax requires significant more spores than inhalation anthrax28 and a limit of detection might be placed at 1 million spores in 1 liter of water or 10 microgL Since each spore contains ~10 CaDPA by weight29 a goal for DPA might be 1 microgL indicating that the present measurements must be improved by nearly two orders of magnitude Due to the increased hazards of handling HCN gas KCN salt was used for these experiments Nevertheless all sample preparations were performed in a chemical hood KCN completely dissolves in water but its conjugate acid HCN is formed and has a Ka of 615x10-1030 viz
HCN CN- + H+ pKa = 921 Reaction 4
Figure 8 SER spectra of 1 mgmL DPA as a function of pH Conditions 100 mW 785 nm 100 scans (44-sec)
Figure 9 Concentration dependence of DPA DPA- and DPA= as a function of pH for a 0006M sample Intensity of 525 () and 795 (diams) cm-1 bands from Fig 8 as a function of pH
pH
135 171 219 383
0
0001
0002
0003
0004
0005
0006
0007
0 2 4 6 8 10 12 14pH
Con
cent
ratio
n [M
]
DPADPA-DPA=795525
DPA DPA=DPA-
pK2 = 692pK1 = 216
795 525
0
05
1
15
2
25
0 200 400 600 800 1000 1200
DPA Concentration (mgL)
1008
Ban
d In
tens
ity
conc (mgL) lod-10min-100mw1 017
10 016100 103
1000 355
Figure 10 SER spectra of DPA in water at A) 1000 B) 100 C) 10 and D) 1 mgL Conditions pH of 25-55 silver-doped sol-gel coated vial 175 mW of 785 nm 1-min 8 cm-1 D) has been multiplied by x10 to make bands visible
A
B C
D
Figure 11 Plot of SER intensity of 1008 cm-1 band of DPA as a function of concentration using 175 mW of 785 nm Inset table includes LOD in mgL for each concentration but for 100 mW and 10 min
SPIE -2003-5269 123
Consequently the cyanide concentration must be determined for each initial KCN concentration Specifically the samples prepared with concentrations of 01 1 10 100 and 1000 mgL of KCN produced CN- concentrations of 63x10-3 033 69 89 and 964 mgL at pHs of 816 90 967 102 and 107 respectively The pH dependence for the HCN and CN- concentrations are shown in Figure 12 Thus as the amount of KCN added to the solution decreases so does the pH of the solution (becomes less basic) and according to Reaction 4 the relative amount of CN- to HCN also decreases For example in the preparation of a 01 mgL solution of KCN the pH is shifted from 7 for pure water to only 816 and only 63 of the starting material becomes CN- or 63x10-3 mgL In comparison for a solution of 1000 mgL the pH is shifted from 7 to 107 and 96 of the starting material becomes CN- This is significant since the cyanide ion is better able to adsorb onto the silver particles and become SERS active SER spectra of 10 100 and 1000 mgL of KCN samples are shown in Figure 13 and a corresponding plot of the 2100 cm-1 band intensity as a function of concentration is shown in Figure 14
000102030405060708091011
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14pH
Con
cent
ratio
n [m
gm
L]
pKa = 921
CN -HCN
The SER spectra of cyanide are dominated by the single CN stretch at 2100 cm-1 which occurs in normal Raman spectra of solutions at 2080 cm-1 However a low frequency mode occurs at 135 cm-1 due to a Ag-CN stretch (not shown) indicative of a strong surface interaction It is also observed that as the concentration decreases the CN stretch shifts to 2140 cm-1 This shift has been attribute to the formation of a tetrahedral Ag(CN)3
2- surface structure31 as well
Figure 14 Concentration dependence of KCN SERS measured under conditions in Fig 11 Concentrations are 1 01 and 001 mgml Intensities are measured for the CN stretch at 2100 cm-1 Inset table includes LOD in mgL for each concentration in Figs 13 and 15 but for 100 mW and 10 min
Figure 13 SER spectra of KCN in water at A) 1000 B) 100 and C) 10 mgL Conditions pHs of 107 102 and 97 silver-doped sol-gel coated vial 100 mW of 785 nm 1-min 8 cm-1 C) has been multiplied by x10 to make band visible
Figure 12 Concentration dependence of HCN and CN- as a function of pH for a 1 mgmL sample Calculated intensity of 2100 () cm-1 band for a 1 mgml sample at pHs of 816 90 967 102 and 107
A
B
C
0
20
40
60
80
100
120
140
0 200 400 600 800 1000 1200CN Concentration (mgL)
2100
Ban
d In
tens
ity
Figure 15 SER spectra of KCN in water at A) 10 B) 1 and C) 01 mgL Conditions pHs of 97 12 and 12 gold-doped sol-gel coated vial 100 mW of 785 nm 1-min 8 cm-1 C) has been multiplied by x10 to make band visible
A
B
C
conc (mgL) lod-10min-100mw Condition01 001 Au-pH 121 007 Au-pH 12
69 003 Au-pH 9769 016 Ag-pH 9789 022 Ag-pH 102946 113 Ag-pH 107
SPIE -2003-5269 124
as to CN adsorbed to two different surface sites32 Alternatively the 2140 cm-1 band could be attributed to HCN since this species dominates at lower concentrations However it is unlikely that this species would be attracted to the electropositive silver surface Further both peaks should be present at pHs between 85 and 105 but this is not observed It has also been suggested that at concentrations near and above monolayer coverage the CN- species is forced to adsorb end-on due to crowding and at lower concentrations the molecule can reorient to lie flat33 This suggests that the 2100 and 2140 cm-1 bands correspond to the end-on and flat orientations respectively As Figures 13 and 14 show the intensity of the CN stretch for the 89 mgL sample is nearly as intense as the 964 mgmL band This suggests that the Raman signal for the flat orientation is more enhanced However more extensive measurements are required to verify this point Since resent research has suggested that cyanide may be more effectively detected on gold measurements of KCN solutions were also performed using gold-doped sol-gel vials Preliminary measurements are shown in Figure 15 for samples prepared from 01 1 and 10 mgL KCN Since the pHs are 816 90 967 the resultant CN- concentrations are 63x10-3 033 and 69 mgL Initially only the highest concentration was observed and the signal intensity was significantly better than the equivalent concentration measured using silver In an effort to shift Reaction 4 to the left transforming HCN to CN- (Le Chatelierrsquos principle) KOH was added to the lower concentration samples producing solutions with pH 12 This effectively forces all of the cyanide in solution to be CN- or 01 and 10 mgL respectively More importantly the CN stretch is now observed in the SER spectra The band appears at 2125 cm-1 as has been previously reported for gold12 As calculated for MPA and DPA LODs can be estimated from this data For the three concentrations of cyanide on silver the LODs are 016 to 11 mgL for 100 mW of 785 nm laser excitation and a 10-min acquisition time For gold pH adjusted the LODs are10 to 70 microgL an improvement of more than 10 times silver Nevertheless either substrate is sufficient to meet the JSAWM goals of measuring 3 mgL in 10-minutes as the requirements form cyanide are much less stringent than the nerve agents
4 CONCLUSIONS Here we examined the surface-enhanced Raman spectral response for methyl phosphonic acid dipicolinic acid and cyanide as a function of pH It was determined that the most prominent peak at 760 cm-1 reaches a maximum intensity between pH 3 and 7 and corresponds to the MPA- species Neither the MPA nor MPA= species appear to generate a SER spectrum and consequently no spectra were observed below pH 2 or above pH 8 In this study we also found that higher laser powers could cause photodegradation of MPA signified by the exponential growth of a band at 1050 cm-1 which is tentatively assigned to phosphorous acid Unlike MPA DPA was observed at all pHs This is attributed to the dominant interaction of the pyridine functional group with silver Minor spectral changes were observed at acid pHs and were assigned to neutral DPA Like MPA SER spectra of cyanide were pH dependent No spectra were observed for the HCN species while CN- was best observed at pHs more basic than 8 Preliminary concentration studies for the three analytes allowed estimating limits of detection for MPA DPA and CN using 100 mW of 785 nm and a 10-min acquisition time of 210 165 and 70 microgL respectively Although the latter value suggests that cyanide can be measured at sufficiently low concentrations to meet JSAWM goal improvements by 100 to 200 times are required for MPA and DPA It is clear from this study that pH of the sample is important and must be taken into account when developing concentration calibration curves as well as fieldable analyzers Future work will include tailoring the sol-gel with specific functional groups to dictate sample pH and thereby optimize sensitivity
ACKNOWLEDGMENTS The authors are grateful for the support of the US Army (DAAD13-02-C-0015 Joint Service Agent Water Monitor program) The authors would also like to thank Dr Steve Christensen of the US Army SBCCOM for helpful discussions and Mr Chetan Shende of Real-Time Analyzers for assistance in development of the gold-doped sol-gels
REFERENCES 1 Erickson B Analytical Chemistry News amp Features June 1 397A (1998)
SPIE -2003-5269 125
2 Johnston RL Hoefler CM Fargo JC and Moberley B AT-ONSITE 5-8 (1994) 3 Black RM RJ Clarke RW Read and MT Reid J Chromatography 662 301-321 (1994) 4 USA Army Center for Health Promotion and Preventive Medicine (USACHPPM) Technical Guideline 230A 5 Jensen JL NC Wong W R Loerop SPIE 4775-03 (2001) 6 Hoenig SL Handbook of Chemical Warfare and Terrorism Greenwood Press Wesport CT (2002) 7 Munro NB SS Talmage GD Griffin LC Waters AP Watson JF King and V Hauschild Env Health
Persp 107 933-974 (1999) 8 Holstege CP Kirk M Sidell FR Crit Care Clin13 923-42 (1997) 9 Farquharson S P Maksymiuk K Ong and S Christesen SPIE 4577 166-173 (2001) 10 Lee Y and S Farquharson SPIE 4378 21-26 (2001) 11 Spencer KM J Sylvia S Clauson and J Janni SPIE 4577 158-165 (2001) 12 Tessier P S Christesen K Ong E Clemente A Lenhoff E Kaler and O Velev Applied Spectroscopy 56
1524-1530 (2002) 13 Farquharson S WW Smith S Elliott and JF Sperry SPIE 3533 207-214 (1998) 14 Farquharson S WW Smith S Elliott and JF Sperry SPIE 3855110-116 (1999) 15 Farquharson S WW Smith YH Lee S Elliott and JF Sperry SPIE 4575 62-72 (2002) 16 Guzelian AA J Sylvia J Janni S Clauson and KM Spencer SPIE 4577 182-192 (2001) 17 Shende C F Inscore A Gift P Maksymiuk and S Farquharson in press 18 Weaver MJ S Farquharson and MA Tadayyoni J Chem Phys 82 4867-4874 (1985) 19 Alak AM and T Vo-Dinh Analytical Chemistry 59 2149-2153 (1987) 20 Lee Y and S Farquharson SPIE 4206 140-146 (2000) 21 Farquharson S and Y Lee SPIE 4200-16 (2000) 22 Lee Y S Farquharson and P M Rainey SPIE 3857 76-84 (1999) 23 Lee Y S Farquharson H Kwong and M Shahriari SPIE 3537 252-260 (1998) 24 Wang J M Pumera G Collins and A Mulchandani Analytical Chemistry 74 6121-6125 (2002) 25 Dou X YM Jung Z-Q Cao and Y Ozaki Applied Spectroscopy 53 1440-1447 (1999) 26 Data supplied by S Christesen and K Ewing 27 Kerker M and B Thompson Eds SPIE MS 10 (1990) 28 Inglesby TV DA Henderson JG Bartlett JAMA 287 2236 (2002) 29 FW Janssen AJ Lund and LE Anderson Science 127 26 (1958) 30 Lide DR Ed Handbook of Chemistry and Physics CRC Press 77th Ed (1996-7) 31 Billmann J G Kovacs and A Otto Surf Sci 92 153 (1980) 32 Murray CA and S Bodoff Phys Rev B 32 671 (1985) 33 Kellogg D and J Pemberton J Phys Chem 91 1120 (1987)
SPIE-2003-5269 16
Chemical agent detection by surface-enhanced Raman spectroscopy
Stuart Farquharson Alan Gift Paul Maksymiuk Frank Inscore and Wayne Smith
Real-Time Analyzers 87 Church Street East Hartford CT 06108
Kevin Morrisey and Steven D Christesen US Army SBCCOM Aberdeen Proving Ground MD 21010
ABSTRACT
In the past decade the Unites States and its allies have been challenged by a different kind of warfare exemplified by the terrorist attacks of September 11 2001 Although suicide bombings are the most often used form of terror military personnel must consider a wide range of attack scenarios Among these is the intentional poisoning of water supplies to obstruct military operations in Afghanistan and Iraq To counter such attacks the military is developing portable analyzers that can identify and quantify potential chemical agents in water supplies at microgram per liter concentrations within 10 minutes To aid this effort we have been investigating the value of a surface-enhanced Raman spectroscopy based portable analyzer In particular we have been developing silver-doped sol-gels to generate SER spectra of chemical agents and their hydrolysis products Here we present SER spectra of several chemical agents measured in a generic tap water Repeat measurements were performed to establish statistical error associated with SERS obtained using the sol-gel coated vials Keywords Chemical agents chemical agent detection SERS Raman spectroscopy
1 INTRODUCTION
In the autumn of 2001 terrorism within US borders became a sobering reality While extensive efforts are being implemented to secure the homeland US and Coalition forces in Afghanistan and Iraq are constantly faced with terrorist attacks In addition to car-bombings the military has identified several non-traditional attack scenarios including poisoning of water supplies by chemical warfare agents (CWAs) To counter this threat the Department of Defense has been investigating numerous approaches to detect such attacks under the auspices of the Joint Service Agent Water Monitor (JSAWM) program1 The JSAWM must be capable of identifying the chemical agents with no false-positives and quantifying the agents at microgram per liter concentrations within 10 minutes2 This includes the analysis of drinking water supplies distribution and storage systems Currently colorimetric paper is used to detect agents on-site while gas chromatography combined with mass spectrometry (GCMS) is used in mobile support laboratories However both methods have severe drawbacks The paper changes color in response to contact with many chemicals besides CWAs causing a high incidence of false positives3 while GCMS although very chemically specific requires up to an hour to perform and regular re-calibration456 Vibrational spectroscopy has also been investigated as a potential method of rapidly detecting CWAs7-
11 as early as 1985 in the case of infrared spectroscopy8 and 1988 in the case of Raman spectroscopy11 Again however these techniques also have limitations when it comes to measuring trace poisons in water Infrared spectra would be dominated by the very strong absorption of water which would obscure absorptions by most other chemicals present Whereas Raman spectroscopy is simply not a very sensitive technique and detection limits are typically grams per liter Surface-enhanced Raman spectroscopy (SERS) offers several advantages over conventional vibrational methods and that may provide the necessary sensitivity required for detecting trace quantities of chemical agents in water SERS employs the interaction of surface plasmon modes of metal particles with target analytes to increase scattering efficiency by as much as 1 million times12 In 1987 the potential of this approach to measure CWAs was demonstrated by measuring a series of pesticides13 Several of these organophosphonates have chemical structures similar to CWAs in particular P=O functional groups In the past few years we and others have further explored the ability of SERS to detect CWAs14-17 and even bioagents 18-21 We have been employing silver-doped sol-gels to promote the SER effect
SPIE-2003-5269 17
in these studies The porous silica network of the sol-gel matrix offers a unique environment for immobilizing and stabilizing SER-active metal particles22-25 The silver-doped sol-gels have been coated on the internal walls of standard glass vials to allow rapid SER analysis of small sample quantities without preparation We have measured over 100 chemicals with enhancements of 104 to 106 demonstrated reversible measurements in a flowing system reproducible measurements from vial-to-vial and measurements in multiple solvents including water21-25 Previously we used these vials to perform preliminary measurements of cyanide (CN) and methylphosphonic acid (MPA) Most of the nerve agents form MPA during hydrolysis while Tabun forms CN a chemical agent in its own right In another paper including in these proceedings we examined the limits of detection (LOD) for MPA by measuring a series of concentrations down to 50 mgL and estimated a limit of detection of 100 microgL26 These measurements provide encouragement in that SERS may satisfy the needs of the JSAWM To further establish the viability of SERS in particular silver-doped sol-gels here we present analysis of cyanide mustard and VX in tap water The measurements performed at the US Armyrsquos Edgewood Chemical Biological Center Aberdeen MD also included numerous repeat measurements to establish reproducibility
2 EXPERIMENTAL 2a General All chemicals including potassium cyanide 2-chloroethylethyl sulfide and those used to prepare the silver-doped and gold-doped sol-gel coated vials were obtained and used as received from Sigma-Aldrich All samples were prepared in a chemical hood using HPLC grade water unless otherwise noted (Fischer Scientific Fair Lawn NJ) for SERS measurements Once prepared the samples were transferred into the silver-doped sol-gel vials (Real-Time Analyzers Inc East Hartford CT) The vials were coated in a manner similar to that previously reported by adding ammonium hydroxide to a solution of silver nitrate tetramethyl orthosilicate (TMOS) and methanol22 The two precursor solutions were prepared mixed and transferred to 2-mL glass vials dried and heated After sol-gel formation the incorporated metal ions were reduced with dilute sodium borohydride (1mgmL) which is followed by a water wash to remove residual reducing agent After the resultant analyte solution was introduced the SER-active vial was fixed horizontally to an XY positioning stage (Conix Research Springfield OR) just inside the focal point of an f07 aspheric lens The lens focused the beam into the sample and collected the scattered radiation back along the same axis A dichroic filter (Omega Optical Brattleborough VT) was used to reflect the excitation laser to the lens and pass the Raman scattered radiation collected by the lens An f2 achromat was used to collimate the laser beam exiting a 200 microm core diameter source fiber optic while a second f2 achromat was used to focus the scattered radiation into a 365 microm fiber optic (Spectran Avon CT) A short pass filter was placed in the excitation beam path to block the silicon Raman scattering generated in the source fiber from reflecting off sampling optics and reaching the detector A long pass filter was placed in the collection beam path to block the sample Rayleigh scattering from reaching the detector A 785 nm diode laser (Process Instruments Inc model 785-600 Salt Lake City UT) was used to deliver 100 to 150 mW of power to the sample A Fourier transform Raman spectrometer (Real-Time Analyzers model IRA-785 East Hartford CT) and a silicon photo-avalanche detector (Perkin Elmer model C30902S Stamford CT) were used to acquire the SER spectra
2b Edgewood Chemical Biological Center The surface-enhanced Raman spectral measurements at the US Armyrsquos Chemical Biological center presented here were all performed on September 12 2003 To expedite measurements a plate was machined to hold up to 12 SER-active sample vials (Figure 1) The plate fit a standard XY plate reader that could be programmed Pure KCN bis-(2-chloroethyl)sulfide (distilled mustard gas HD) and ethyl S-2-diisopropylamino ethyl methylphosphonothioate (VX) were obtained on-site and used to prepare 1 gL tap water solutions in a chemical hood with appropriate safety equipment Simulated tap water was prepared by adding 10 chemicals most often found in tap water at appropriate concentrations to distilled water (Table 1) SER measurements were also performed in a chemical hood For added safety the FT-Raman instrument was placed outside the laboratory and 30 foot fiber optic and electrical cables were used to allow remote SERS measurements and plate manipulation For each experiment 1gL samples were prepared and added to 9 individual vials which were then loaded on the plate In some cases a tenth vial was included as a blank
Table 1 Generic Tap Water Composition
Compound mgL NaHCO3 100 CaSO4 27 MgSO4bull7H2O 67 NaNO3 1 Fulvic Acid 1 K2HPO4 07 KH2PO4 03 (NH4)2HSO4 001 NaCl 001 FeSO4 0001 pH 76-78
SPIE-2003-5269 18
A software program was written that allowed selecting the sequence that the vials were measured the number of positions along the length of the vials to measure (1 to 5) and the number of scans to co-add During sample analysis the program displayed the vial being analyzed the point being analyzed and the spectrum as it was being acquired Once all the data was collected a second software program was written to rapidly analyze the data The spectra collected for all the vials on a plate could be loaded at one time and then the spectra for each point could be displayed simultaneously or separately The user could then select the Raman peak to analyze in terms of peak height or area This was accomplished by selecting points on either side of the peak to define a baseline of zero The peak height or area could then be computed for all of the spectra loaded and then exported to a spreadsheet for statistical analysis
Figure 1 A) Vial Holder 6 slots to hold 2 vials each end-to-end B) Measurement Configuration Program user selects vials to measure sequence number of points per vial (1 to 5) and number of scans per point C) Spectral Acquisition Program shows spectrum being collected which vial and position D) Spectra Analysis Program user selects spectra to analyze by plate vial and point (s) as well as two wavenumbers defining the peak and the baseline to subtract The image is of 5 repeat measurements of 10 mgL KCN in generic tap water 16 sec each 100 mW of 785 nm
3 RESULTS AND DISCUSSION Raman and surface-enhanced Raman spectra were obtained for potassium cyanide bis-(2-chloroethyl)sulfide and ethyl S-2-diisopropylamino ethyl methylphosphonothioate representing three classes of chemical agents cyanides mustards and nerve agents respectively Spectra were also obtained for 2-chloroethyl ethyl sulfide (CEES) a structural analogue to HD which was included in the study to aid in assigning spectral bands KCN salt was used for cyanide experiments to avoid the increased hazards of handling HCN gas KCN completely dissolves in water forming its conjugate acid HCN according to its Ka of 615x10-1027 and at a concentration of 1 mgmL results in a pH 107 solution This is significant in that only CN- appears to interact sufficiently with silver to produce a SER spectrum and no spectral signal is observed below pH 726 Figure 2 shows the SER and normal Raman spectra for KCN The SERS spectrum of 1mgml KCN in water shows a single intense somewhat broad feature at
A B
C D
SPIE-2003-5269 19
2100 cm-1 assigned to the single CequivN stretch The band is much sharper in the normal Raman spectra of the solid KCN salt at 2074 cm-1 This band does broaden and shift to 2080 cm-1 in solution (not shown) However the observed SERS frequency is attributed to interaction with silver and low frequency mode at 135 cm-1 attributed to a Ag-CN stretch (not shown) supports this conclusion
cm-1 band to a C-S stretch but the authors concede that it is in fact more likely a C-Cl stretch28 It appears that the most intense Raman bands at 648 692 and 747 cm-1 shift to 620 660 and 720 cm-1 in the SER spectra and are tentatively assigned as above The width of these bands suggests that they overlap underlying spectral features Additional bands in the Raman spectra occur at 972 1034 1049 1263 1286 1423 1442 2865 2935 and 2960 cm-1 Corresponding bands occur in the SER spectra at 964 1015 1054 1286 1410 1447 2865 and 2935 cm-1 Most of these bands are associated with alkane modes specifically the bands at approximately 1040 cm-1 to a C-C stretch 1290 cm-1 to a CH2 in-phase twist 1440 cm-1 to a CH2 wag 2865 cm-1 to a symmetric CH2 stretch and 2965 cm-1 to an asymmetric CH2 stretch The Raman and SER spectra of sulfur mustard were measured at the Edgewood center (Figure 4) Both spectra are largely similar to CEES The C-Cl and C-S bands in the Raman spectrum of HD now occur at 640 655 700 739 and 760 cm-1 and are more resolved possibly due to the increased molecular symmetry Theoretical calculations indicate that the first three bands are due to C-Cl stretching modes and the latter two to C-S stretching modes28 Only the C-Cl bands maintain significant intensity in the SER spectra occurring at 624 and 643 cm-1 which is attributed to the
Prior to measurements of HD CEES was examined by Raman and SER spectroscopy (Figure 3) CEES also known as half-mustard is essentially identical to HD except one of the chlorine end atoms is replaced by a hydrogen atom Again although not as toxic as HD CEES is a blister agent and dilute aqueous samples were prepared using appropriate safety equipment Both the Raman and SER spectra of CEES are similar and dominated by bands between 600 and 800 cm-1 These are associated with C-Cl and C-S stretching modes which are tentatively assigned to 648 and 747 cm-1 in the Raman spectra respectively The shoulder at 630 cm-1 the overlapped band at 660 cm-1 and the strong band at 692 cm-1 could also be due to these modes or their asymmetric counterparts It is worth noting that theoretical calculations assign the 692
Figure 2 A) SER and B) NR spectra of KCN Conditions A) 1 mgml in tap water 100 mW of 785 nm at sample 1-min acquisition time B) solid 300 mW of 785 nm 5-min All spectra are 8 cm-1 resolution
A
B
Figure 3 A) SER and B) NR spectra of CEES Conditions A) 1 vv (10 mgml) in MeOH 100 mW of 785 nm 1-min acquisition time B) neat 300 mW of 785 nm 5-min
A
B
Cl-CH2-CH2-S-CH2-CH3
A
B
Figure 4 A) SER and B) NR of HD Conditions A) 1mgml in tap water B) pure both 100 mW of 785 nm 1-min
Cl-CH2-CH2-S-CH2-CH2-Cl
SPIE-2003-5269 20
expected strong interaction between chlorine and silver and adds support to the assignment of this band to a C-Cl stretch Weaker overlapping bands occur at 670 692 and 724 cm-1 the latter possibly due to C-S stretching modes Again the alkane modes are apparent in the normal Raman spectra of HD but only a broad feature at 1300 to 1450 cm-1 suggests CH2 contributions in the SER spectrum Although the observed bands in the VX spectrum have not been assigned (Figure 4) a computer generated Raman spectrum29 predicts many of the same features with surprising accuracy and are used here Two intense bands at 460 and 530 cm-1 closely match predicted bands at 463 and 546 cm-1 assigned to a CH3-P=O bend and a PO2CS wag Three highly overlapped bands occur at 694 745 and 771 cm-1 matching predicted bands at 713 730 and 760 cm-1 The first
Table 2 Measured SER peak heights for the CN stretch at 2100 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
Figure 5 A) SERS and B) NR spectra of VX Conditions A) 1 vv (10 mgml) in MeOH B) pure sample both 100 mW of 785 nm 1-min acquisition time
A
B
two have been assigned to a C-S stretch and CH2bend respectively while the latter has been attributed to either a P-C stretch or an O-C-C stretch Although the 745 cm-1 band may alternatively be assigned to a C-S stretch based on the previous measurements of CEES and HD The relatively intense bands at 890 1106 1218 1445 and 1465 cm-1 also match predicted bands at 880 1108 1216 1440 and 1464 cm-1 that are assigned to a C-C stretch CH3 rock N-C3 stretch various C-H3 bends and C-H bends respectively Both the computer generated and the measured spectra contain numerous other less intense bands One is worth mentioning A unique band appears at 370 cm-1 that is predicted at 368 cm-1 and corresponds to an O-P=O bend The surface-enhanced Raman spectrum of VX is also rich with spectral features It has the unique low frequency band at 370 cm-1 as well as a second band at 380 cm-1 that is assigned tothe S-P-O bend predicted in the normal Raman spectrum at 388 cm-1 Based on the measured and predicted normal Raman spectra the following SERS assignments are given 460 cm-1 to the CH3-P=O bend 544 cm-1 to the PO2CS wag 738 cm-1 to a C-S stretch (based on arguments above) 890 cm-1
to a C-C stretch 1101 cm-1 to a CH3 rock and 1456
cm-1 to a C-H bend The ability of SERS to measure chemical agents in water containing real-world chemical interferents was tested by using the generic tap water described in Table 1 The ability to reproduce measurements was accomplished by preparing three separate water stock solutions which were used to prepare three sample solutions each of 1 mgmL KCN in the generic tap water The samples defined as vials 1-9 were then measured at 4 points per vial in a semi-random fashion such that errors associated with stock solution preparation and errors associated with instrument drift could be identified No trends were apparent that signified such systematic errors Each spectrum collected consisted of 20 averaged scans taking 16 seconds at 8 cm-1 resolution The laser power at the sample was measured periodically during the day and it ranged from 102 to 105 mW spacccc
spot Vial 1 Vial 2 Vial 41 13754 13004 128192 13519 12692 12909
1 3 13541 12721 126394 13462 12648 12651
ave 13569 12766 12755Vial 6 Vial 3 Vial 5
1 11586 14068 10757 2 11236 14402 115122 3 11376 14546 11559
4 10894 11763 11214ave 11273 13695 11261
Vial 7 Vial8 Vial 91 11198 15804 11023
2 11243 15716 88753 3 11562 15216 11218
4 11655 15084 11077ave 11415 15455 10548
CN AVG STDEV ERR1 mgmL pk ht 12526 1560 1245
stock solution
SPIE-2003-5269 21
The CN measurements consisting of 36 data points produced an average height of 125 for the 2100 cm-1 peak with a standard deviation of 156 or 125 (Table 2) The HD and VX measurements were performed precisely the same way (Tables 3 and 4) For HD the 624 cm-1 peak was used for analysis and it had an average height of 53 with a standard deviation of 068 or 129 while for VX the 544 cm-1 peak was used for analysis and it had an average height of 1051 with a standard deviation of 308 or 293 The greater error in the VX measurements can be somewhat attributed to Vial 4 which produced lower SER signal intensities But removing this vial from the data set changes the standard deviation to 233 only a modest improvement
spot Vial 1 Vial 2 Vial 41 525 609 4682 45 675 484
1 3 527 644 5094 559 807 669
ave 51525 68375 5325Vial 6 Vial 3 Vial 5
1 521 574 575 2 536 53 4492 3 509 508 451
4 565 594 379ave 53275 5515 4635
Vial 7 Vial 8 Vial 91 457 589 413
2 583 497 4273 3 544 505 4
4 528 5 509ave 528 52275 43725
HD AVG STDEV ERR1 mgmL pk ht 530 068 1291
stock solution
spot Vial 1 Vial 2 Vial 41 1464 1034 4672 1485 99 61
1 3 1491 989 5684 1041 777 553
ave 137025 9475 5495Vial 6 Vial 3 Vial 5
1 1058 942 1293 2 697 121 965
2 3 727 1206 11124 689 1254 746
ave 79275 1153 1029Vial 7 Vial 8 Vial 9
1 1127 783 163 2 1358 812 1614
3 3 1371 908 15124 1097 875 1397
ave 123825 8445 153825
VX AVG STDEV ERR1 mgmL pk ht 1051 308 2925
stock solution
4 CONCLUSIONS In this paper we examined the ability of surface-enhanced Raman spectroscopy to reproducibly measure CN HD and VX in tap water without chemical interference Both normal and surface-enhanced Raman spectra were examined to select unique bands suitable to identify and quantify these chemical agents For SER measurements the 2100 cm-1 C-N stretch was used for CN the 624 cm-1 C-Cl stretch was used for HD and the 544 cm-1 PO2CS wag was used for VX It was determined that 1 mgmL samples of each of these chemicals measured 36 times in glass vials coated with a silver-doped sol-gel reproduced measurements with standard deviations of 125 129 and 293 It was further found that the 10 chemicals added to simulate generic tap water did not interfere with or alter the SER spectra It should be noted that the concentrations used in this study were considerably greater than those required by the JSAWM program Current work involves improving SER sensitivity and designing sampling systems with better reproducibility This includes the development of fractal silver and gold structures within the sol-gel matrix and the development of chemically selective sol-gels
ACKNOWLEDGMENTS The authors are grateful for the support of the US Army (DAAD13-02-C-0015 Joint Service Agent Water Monitor program) The authors would also like to thank Janet Jensen Ronald Crosier and Kristina Gonser for helpful discussions
Table 4 Measured SER peak heights for the VX band at 544 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
Table 3 Measured SER peak heights for the HD band at 624 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
SPIE-2003-5269 22
REFERENCES 1 Jensen JL NC Wong W R Loerop SPIE 4775-03 (2001) 2 JSAWM Requirements at wwwsbccomapgeaarmymilRDAecbcrtPRODSERJSAWMjsawmhtml 3 Erickson B ldquoThe Chemical Weapons Convention Redefines Analytical Challengerdquo Anal Chem 70 397A (1998) 4 Johnston RL Hoefler CM Fargo JC and Moberley B ldquoThe Defense Nuclear Agencyrsquos
ChemicalBiochemical Weapons Agreements Verification Technology Research Development Test and Evaluation Program and its Requirements for On-Site Analysisrdquo AT-ONSITE 5-8 (1994)
5 Black RM Clarke RJ Read RW and Reid MT ldquoApplication of GC-MS and GC-tandem MS to the analysis of chemical warfare samples found to contain residues of the nerve agent sarin sulphur mustard and their degradation productsrdquo J Chrom 662 301-321 (1994)
6 USA Army Center for Health Promotion and Preventive Medicine (USACHPPM) Technical Guideline 230A 7 Hoffland LD Piffath RJ Bouck JB rdquoSpectral signatures of chemical agents and simulantsrdquo Optical
Engineering 24 982-984 (1985) 8 Braue EHJ Pannella MGrdquoCIRCLE CELL FT-IR Analysis of Chemical Warfare Agents in Aqueous Solutionsrdquo
App Spectrosc 44 1513-1520 (1990) 9 Tseng C-H Mann CK and Vickers TJ ldquoDetermination of Organics on Metal Surfaces by Raman
Spectroscopyrdquo App Spectrosc 47 1767-1771 (1993) 10 Staff Reporter ldquoUS Army Tests Lidar to Detect Biological Toxinsrdquo Photonic Spectra pg 50 December 1998 11 Christesen SD Raman cross sections of chemical agents and simulants App Spectrosc 42 318-321 (1988) 12 Weaver MJ Farquharson S Tadayyoni MA ldquoSurface-enhancement factors for Raman scattering at silver electrodesrdquo J Chem Phys 82 4867-4874 (1985) 13 Alak AM and Vo-Dinh T ldquoSERS of Organophosphorous Chemical Agentsrdquo Anal Chem 59 2149-2153 (1987) 14 Lee Y Farquharson S ldquoRapid chemical agent identification by SERSrdquo SPIE 4378 21-26 (2001) 15 Farquharson S Maksymiuk P Ong K Christesen S ldquoChemical agent identification by surface-enhanced Raman spectroscopyrdquo SPIE 4577 166-173 (2001) 16 Spencer KM Sylvia J Clauson S and Janni J ldquoSurface Enhanced Raman as a Water Monitor for Warfare
Agents in Waterrdquo SPIE 4577 158-165 (2001) 17 Tessier P Christesen S Ong K Clemente E Lenhoff A Kaler E Velev O ldquoOn-line spectroscopic characterization of sodium cyanide with nanostructured gold surface-enhanced Raman spectroscopy substratesrdquo
App Spectrosc 56 1524-1530 (2002) 18 Farquharson S W W Smith S Elliott and J F Sperry Rapid biological agent identification by surface- enhanced Raman spectroscopy SPIE 3855110-116 (1999) 19 Farquharson S WW Smith YH Lee S Elliott and J F Sperry Detection of biological signatures A
comparison of electrolytic and metal-doped sol-gel surface-enhanced Raman media SPIE 4575 62-72 (2002) 20 Guzelian AA Sylvia J Janni J Clauson S and Spencer KM ldquoSERS of whole cell bacteria and trace levels of biological moleculesrdquo SPIE 4577 182-192 (2001) 21 Shende C Inscore F Gift A Maksymiuk P Farquharson S ldquoRapid dipicolinic acid extraction from Bacillus spores detected by surface-enhanced Raman spectroscopyrdquo App Spectrosc 58 accepted 22 Lee Y and Farquharson S ldquoSERS Sample Vials Based on Sol-Gel Process for Trace Pesticide Analysisrdquo SPIE
4206 140-146 (2000) 23 Farquharson S and Lee Y ldquoTrace Drug Analysis by SERSrdquo SPIE 4200-16 (2000) 24 Lee Y Farquharson S and Rainey P M Surface-Enhanced Raman Sensor for Trace Chemical Detection in
Water SPIE 3857 76-84 (1999) 25 Lee Y Farquharson S Kwong H and Shahriari M ldquoSol-Gel Chemical Sensor for Surface-Enhanced Raman
Spectroscopyrdquo SPIE 3537 252-260 (1998) 26 Farquharson S A Gift P Maksymiuk F Inscore and W Smith ldquopH dependence of methyl phosphonic acid
dipicolinic acid and cyanide by surface-enhanced Raman spectroscopyrdquo SPIE 5269-19 (2003) 27 Lide DR Ed Handbook of Chemistry and Physics CRC Press 77th Ed (1996-7) 28 Sosa C RJ Bartlett K KuBulat and WB Person ldquoA theoretical study of harmonic vibrational frequencies and
infrared intensities of XCH2CH2SCH2CH2X and XCH2CH2SH (= H Cl)rdquo J Phys Chem 93 577-588 (1993) 29 Hameka HF and JO Jenson ldquoComputer-generated predictions of the structure and the IR and Raman spectra of
VXrdquo ERDEC-TR-065 May 1993
- Chemagents Appendicespdf
-
- SPIE2001-4575-Bioagent-SERSapdf
-
- Detection of bioagent signatures A comparison of electrolytic and metal-doped sol-gel surface-enhanced Raman media
- Stuart Farquharson Wayne Smith and Yuan Lee
-
- Real-Time Analyzers 87 Church Street East Hartford CT 06108
-
- ABSTRACT
- 4 CONCLUSIONS
- 5 ACKNOWLEDGEMENTS
- 6 REFERENCES
-
171
dominance of the P-C and the CH2 stretches and the disappearance of the P-O-C mode in the upper spectrum suggest the molecule is oriented with the methyl-hydroxyl-hydroxyl face towards the surface However considerably more research must be performed to verify these points Both chemical forms were measured as a function of concentration to determine the onset of monolayer coverage and estimate expected detection limits (Figure 6) Below monolayer coverage the signal to concentration dependence should be linear and the SN of any spectral measurement in this range can be used to predict the detection limit In the spectra presented here the peak height was used as the signal while the noise as root-mean-squared (RMS) was measured between 4400-4600 cm-1 Since noise is distributed evenly throughout the spectrum when transformed this region was used since it does not have any contributions from signals or baseline offsets Figure 6 shows a series of spectra for MPA along with a plot of the 1050 cm-1 band intensity (the noise was identical for this series) as a function of sample concentration A clear discontinuity appears in the vicinity of 01 mgmL (19 ppm) indicating the onset of monolayer coverage A detection limit defined as a SN of 3 was calculated for the 01 and 005gmL samples at 24x10-4 and 25x10-4 gL respectively A more modest detection limit of 101x10-4 gL was obtained using the 760 cm-1 band in the second series of concentration measurements These detection limits correspond to 46 and 190 ppb for a 3-min scan and 75 mW of incident laser power
Figure 6 A) Concentration dependence of MPA SERS measured in silver-doped TMOS) B) Concentrations are 001 005 01 05 1 gL (188 94 188 94 188 ppm) I760 series (bull) and I1050 series (∆)
Enhancement factors for cyanide and methyl phosphonic acid can be determined by estimating the number of molecules contributing to the surface-enhanced and normal Raman spectra The enhancement factor EF can be defined by the following equation
EF = (ISERSINR)bull(MNRMSERS) bull(PNRPSERS) bull(TNRTSERS)12
where I is the spectral band intensity M is the sample mass P is the incident laser power and T is the measurement time (or number of scans) for the two measurements For the normal Raman spectra a cylindrical scattering volume is assumed based on the laser area (28x10-7m2 6x10-4m diameter spot) and the penetration depth (1x10-3 m)24 The density of KCN and MPA as powders were measured at 0572 and 0516 gcm3 indicating that 16x10-4 and 144x10-4 g produced the normal Raman signals in Figure 5 respectively The number of molecules contributing to the SER spectra are those on the silver particles embedded in the sol-gel The total silver surface area can be determined from the average particle size concentration and the scattering volume Previous scanning electron micrographs showed the average silver particle size to be 40 nm in diameter (335x10-23m3)17 The silver concentration is 012M based on the reactant molar concentrations and dilution factors And the scattering volume is 76x10-
11m3 again based on a cylindrical scattering volume defined by a laser area of 28x10-7m2 and a sol-gel thickness of 27x10-
4m This volume contains 123x10-6g of silver equivalent to 35x109 silver particles with a collective surface area of 18x10-
5m2 However it may reasonably be assumed that at least half of the silver surface is in contact with the sol-gel matrix and unavailable for analyte interaction If we assume monolayer coverage and that each CN molecule occupies 15x10-20m2 then approximately 62x1014 molecules or 27x10-8g of CN contribute to the SER spectrum (20x10-19m2 46x1013 molecules 74x10-9g for MPA) Accordingly the EF for cyanide equals 48x104 ((180599) bull(16x10-427x10-8) bull(90075) bull(500100)12) The EF for MPA is considerably higher at 87x106 ((60326) bull(144x10-474x10-9) bull(90075) bull(500200)12)
0
20
40
60
80
100
120
140
0 02 04 06 08 1 12[MPA] (mgmL)
I (76
0)
0
100
200
300
400
500
600
I (10
50)
Wavenumber (∆cm-1)
A B
Proc SPIE Vol 4577
172
4 CONCLUSIONS Here we present for the first time surface-enhanced Raman spectra of Tabun and Sarin obtained using silver-doped sol-gels However the inferior enhancement suggested employing SER-active sol-gels varying in composition to improve both detection limits and selectivity of the target analytes To this end we successfully demonstrated the capabilities of four sol-gels that select for 1) polar-positive 2) polar-negative 3) weakly polar-positive and 4) weakly polar-negative chemical species p-aminobenzoic acid was used to show that silver could be used to attract polar-negative chemicals or functional groups (carboxylate anion) and that gold could be used to attract polar-positive chemicals (amine) while a mixture of p-aminobenzoic acid and phenylacetylene was used to show that tetramethyl orthosilicate preferentially solvates polar chemicals and that tetraethyl orthosilicate preferentially solvates weakly polar chemicals This increased sample control was applied to cyanide and methyl phosphonic acid two hydrolysis products of chemical warfare agents Exceptional results were obtained for methyl phosphonic acid allowing measurement of 1x10-2 gL for a 3-min scan and 75 mW of incident laser power with an estimated detection limit of 245x10-4 gL and an enhancement factor of 87x106 However this detection limit is 76 times less sensitive than required for the JSAWM (32x10-6gL for the G-agents) Finally we note that the measurements performed here employed an FT-Raman spectrometer This instrumentation was chosen over dispersive Raman instrumentation for the high wavelength accuracy afforded by the HeNe reference laser (Connes Advantage25) which would allow reliable spectral subtraction matching of observed spectra to stored library spectra and confident use of chemometric approaches Such data analysis is likely to be required to identify the chemical agents as well as distinguish hydrolysis products However this instrumentation which employs 1064 nm excitation and InGaAs detection sacrifices sensitivity We believe that the measurement sensitivity can be improved by at least two orders of magnitude by using shorter laser excitation wavelengths (eg 532 nm) This would provide more efficient Raman scattering (fourth power dependence on laser excitation wavelength) more efficient generation of plasmon modes and allow using more efficient detector material (Si vs InGaAs) These modifications are underway
5 ACKNOWLEDGEMENTS The authors would like to thank Dr R Yin and J Jensen of the US Army for supporting this work (Contract Number DAAD13-01-C-0019) They would also like to thank Advanced Fuel Research for making their laboratory facilities available
6 REFERENCES 1 Tu Anthony ldquoOverview of Sarin Terrorist Incidents in Japan in 1994 and 1995rdquo 6th CBW Protection Symposium
Stockholm Sweden 10-15 May 1998 2 Jensen JL NC Wong W R Loerop SPIE 4775-03 (2001) 3 JSAWM Requirements at wwwsbccomapgeaarmymilRDAecbcrtPRODSERJSAWMjsawmhtml 4 ldquoThe Chemical Weapons Convention Redefines Analytical Challengerdquo Analytical Chemistry News amp Features June 1
397A (1998) 5 Johnston RL Hoefler CM Fargo JC and Moberley B ldquoThe Defense Nuclear Agencyrsquos ChemicalBiochemical
Weapons Agreements Verification Technology Research Development Test and Evaluation Program and its Requirements for On-Site Analysisrdquo AT-ONSITE 5-8 (1994)
6 Black RM Clarke RJ Read RW and Reid MT ldquoApplication of gas chromatography-mass spectrometry and gas chromatography-tandem mass spectrometry to the analysis of chemical warfare samples found to contain residues of the nerve agent sarin sulphur mustard and their degradation productsrdquo J Chromatography 662 301-321 (1994)
7 Hoffland LD Piffath RJ Bouck JBrdquoSpectral signatures of chemical agents and simulantsrdquo Optical Engineering 24 982-984 (1985)
8 Braue EHJ Pannella MGrdquoCIRCLE CELL FT-IR Analysis of Chemical Warfare Agents in Aqueous Solutionsrdquo Applied Spectroscopy 44 1513-1520 (1990)
9 Tseng C-H Mann CK and Vickers TJ ldquoDetermination of Organics on Metal Surfaces by Raman Spectroscopyrdquo Applied Spectroscopy 47 1767-1771 (1993)
10 Staff Reporter ldquoUS Army Tests Lidar to Detect Biological Toxinsrdquo Photonic Spectra pg 50 December 1998
Proc SPIE Vol 4577
Proc SPIE Vol 4577
173
11 Christesen SD Raman cross sections of chemical agents and simulants Applied Spectroscopy 42 318-321 (1988) 12 Alak AM and Vo-Dinh T ldquoSERS of Organophosphorous Chemical Agentsrdquo Analytical Chemistry 59 2149-2153
(1987) 13 Norrod KL Sudnik LM Rousell D and Rowlen KL ldquoQuantitative Comparison of Five SERS Substrates
Sensitivity and Detection Limitrdquo Applied Spectroscopy 51 994-1001 (1997) 14 Lee Y and Farquharson S ldquoSERS Sample Vials Based on Sol-Gel Process for Trace Pesticide Analysisrdquo SPIE 4206
140-146 (2000) 15 Farquharson S and Lee Y ldquoTrace Drug Analysis by Surface-Enhanced Raman Spectroscopyrdquo SPIE 4200-16 (2000) 16 Lee Y Farquharson S and Rainey P M Surface-Enhanced Raman Sensor for Trace Chemical Detection in Water
SPIE 3857 76-84 (1999) 17 Lee Y Farquharson S Kwong H and Shahriari M ldquoSurface-Enhanced Raman Sensor for Surface-Enhanced Raman
Spectroscopyrdquo SPIE 3537 252-260 (1998) 18 Farquharson S Smith W Carangelo R C and Brouillette C ldquoIndustrial Raman Providing Easy Immediate Cost
Effective Chemical Analysis Anywhererdquo SPIE 3859 14-23 (1999) 19 Narayanan VA JM Bello JD Stokes and T Vo-Dinh Analusis 19 307-310 (1991) 20 Laserna JJ E L Torres and JD Winefordner Analytica Chemica Acta 469-480 (1987) 21 Torres EL and JD Winefordner Analytical Chemistry 59 1626-1632 (1987) 22 Suh JS DP DiLella M Moskovits J Phys Chem 87 1540-1544 (1983) 23 Benner RE R Dornhaus R Chang and BL Laube Correlations in the Raman spectra of cyanide complexes adsorbed
at silver electrodes with voltammograms Surface Science 101 341 (1980) 24 Chase D B and JF Rabolt Fourier Transform Raman Spectroscopy Acad Press Ch1 p 131 (1994) 25 Connes J Rev Opt Theor Instrum 40 45 (1961)
APPLIED SPECTROSCOPY 351
focusing the 488 nm laser beam 2 mm deep into thebulk of the crystals to avoid contributions from potentialdepletion layers As shown in Fig 1b the dependence ofthe Raman shift on the degree of deuteration is almostperfectly linear and ts very well with D 5 22684cmR1 24526 where D is the degree of deuteration (in )and R is the spectral mean of the PO4 vibration in cm21A linear correlation coef cient of 0998 indicates an ex-cellent linear dependence of the Raman peak shift withdegree of deuteration This result shows that the shift ofthe PO 4 peak is simply caused by the linear increase inatomic mass due to isotope substitution which decreasesthe length of hydrogen-like bonds
This excellent linear dependence allows us to map thepro le of the DH exchange layer at the surface of DKDPcrystals by acquiring Raman spectra and determining theposition of the PO 4 peak for various depths This methodis preferable over other methods such as determining thestrength of the OD vibration directly (eg at 715 cm21)because the position of the most intense peak in the Ra-man spectrum can be measured more precisely than theintensity of some of the weakest peaks in the spectrumThis is demonstrated in Fig 2 where depth-dependentRaman spectra (Fig 2a) and the resulting exchange layerpro les for two DKDP crystals are shown (Fig 2b) Thespectra in Fig 2a were obtained from a depth scan of aDKDP crystal with 75 degree of deuteration in thebulk grown at 45 8C The spectra start out as DKDP with30 deuteration close to the surface and approach thebulk DKDP spectrum within a few micrometers of depthThe fact that the relative degree of deuteration does notextend to 0 D is due to the limited depth resolution ofthe Raman microprobe which averages over 4 mm indepth Figure 2b depicts the resulting DH exchange layerpro les for this and a second crystal grown at 63 8Crespectively Both crystals had the same exposure to am-bient conditions and their main difference is the temper-ature at which they were grown The different exchangelayer pro les indicate that crystals grown at differenttemperatures have differen t proton conductiv ities 13
which leads to a difference in their rate of deuteriumdepletion The parameters controlling this behavior arecurrently the objective of a detailed study the results ofwhich will be reported elsewhere
CONCLUSION
In conclusion we have shown that the shift of the to-tally symmetric PO 4 stretch mode in the Raman spectrumof DKDP crystals scales linearly with degree of deuter-ation This allows us to correlate Raman peak positionsto deuteration levels in these crystals We have presenteda new technique to determine DH diffusion pro les inDKDP frequency conversion crystals based on micro-Ra-man spectroscopy This technique is fast inexpensiveand works under various environmental conditionswhich will allow us to better understand and control deu-terium depletion in DKDP crystals
ACKNOWLEDGMENTS
We would like to thank M Runkel for rst discovering DKDP crack-ing R Floyd for providing DKDP crystals and L Chase and A Burn-ham for their support and helpful discussions This work was performedunder the auspices of the US Department of Energy by the University
of California Lawrence Livermore National Laboratory under contract W-7405-Eng-48
1 J J De Yoreo A K Burnham and P K Whitman Int Mat Rev47 113 (2002)
2 C E Barker R A Sacks B M Van Wonterghern J A Caird JR Murray J H Campbell K Kyle R B Ehrlich and N DNielsen Proc SPIE-Int Soc Opt Eng 2633 501 (1995)
3 T Suratwala paper to be published4 Cleveland Crystals Inc httpwwwclevelandcrystalscom
KDPshtmltable5 E A Popova I T Savatinova and I A Velichko Sov Phys Solid
State 12 1543 (1971)6 I P Kaminow R C C Leite and S P S Porto J Phys Chem
Solids 26 2085 (1965)7 J A Subramony B J Marquardt J W Macklin and B Kahr
Chem Mat 11 1312 (1999)8 H Tanaka M Tokunaga and I Tatsuzaki Solid State Commun
49 153 (1984)9 R J Nelmes G M Meyer and J E Tibballs J Phys C 15 59
(1982)10 M A Yakshin D W Kim Y S Kim Y Y Broslavets O E
Sidoryuk and S Goldstein Laser Physics 7 941 (1997)11 I Takenaga Y Tominaga S Endo and M Kobayashi Solid State
Commun 84 931 (1992)12 C Krenn personal communication13 M Sharon and A K Kalia J Solid State Chem 21 171 (1977)
Rapid Dipicolinic Acid Extractionfrom Bacillus Spores Detectedby Surface-EnhancedRaman Spectroscopy
STUART FARQUHARSON ALAND GIFT PAUL MAKSYMIUK andFRANK E INSCOREReal-Time Analyzers Inc East Hartford Connecticut06108
Index Headings Dipicolinic acid Bacillus spores Anthrax Surface-enhanced Raman spectroscopy
INTRODUCTION
The anxiety caused by the distribution of anthrax en-dospores through the US postal system in October 2001was exacerbated by the long time required for positiveidenti cation of the Bacillus anthracis spores and the un-known extent of their distribution Since that time manymethods capable of rapid eld analysis have been inves-tigated to augment or replace the laboratory method ofgrowing microorganisms in culture media which takesdays to perform12 Prominent among these approachesare polymerase chain reactions (PCR)3 immunoassaysand detection of calcium dipicolinate as a biochemicalsignature PCR employs primers to separate organism-
Received 10 October 2003 accepted 14 November 2003 Author to whom correspondence should be sent
352 Volume 58 Number 3 2004
speci c nucleic acid sequences (eg capsular protein en-coding gene for Bacillus anthracis)4 and polymerases toamplify the segment until it is detectable Recently am-pli cation times have been substantially reduced andcomplete analysis can now be performed in an hour orless Immunoassay methods are also being developed thatuse competitive binding of the bioagent (as an antigen)and its labeled conjugate for a limited number of anti-bodies Although analyses can be performed in under 30minutes a well-de ned anthrax antigen has not yet beenidenti ed5ndash7 and consequently the false-positive rate isunacceptably high8
A number of other methods are being developed witha focus on the detection of calcium dipicolinate (CaDPA)and its derivatives as a B anthracis signature This is sobecause only spore-forming bacteria contain CaDPA andthe most common potentially interfering spores such aspollen and mold spores do not Relatively fast methodshave been developed to chemically extract CaDPA andthen detect it directly by uorescence9 or indirectly byluminescence1011 In the latter case hot dodecylamine(DDA) has been used to extract dipicolinic acid (DPA)and terbium has been utilized to form a highly lumines-cent DPA complex11 Although measurements have beenperformed in as little as ve minutes it was found thatas many as three concentration-dependent complexes canform each with different lifetimes This coupled withthe fact that the Tb31 cation produces the same lumines-cence spectrum makes determinations of low spore con-centrations problematic
It has been long known that Raman spectra of Bacillispores are dominated by bands associated with CaDPA12
and that these spectra may provide a suitable anthrax sig-nature at the genus level13 Since that time considerableimprovements in Raman instrumentation have led to lab-oratory measurements of single Bacilli spores14 and to eld measurements of spores captured from a mail-sort-ing system15 However the single spore measurementsrequired complex instrumentation that is not ruggedwhile the eld measurements required milligrams of sam-ple Furthermore the Raman spectra of both measure-ments contained uorescence contributions that would in-crease uncertainty in quanti cation
In related research we demonstrated that nanogramquantities of DPA could be detected by uorescence-freesurface-enhanced Raman spectroscopy (SERS)16 We alsodemonstrated that microliter volumes of chemicals canbe detected by SERS using metal-doped sol-gel-packedglass capillaries17 Towards the goal of developing a rap-id eld SERS-based anthrax spore detector we havecombined our previous research and we now report thatDPA can be extracted from a 10 mg B cereus spore sam-ple using DDA in 1 minute and can be detected by SERSin an additional 1 minute
EXPERIMENTAL
Dipicolinic acid (26-pyridinedicarboxylic acid DPA)and dodecylamine (DDA) were used as received fromSigma-Aldrich (Milwaukee WI) Lyophilized B cereusspores prepared according to the literature13 were sup-plied by the University of Rhode Island and used as re-ceived Multiple particles approximately 01 mm3 each
were separated and weighed at 5 to 15 mg representing05 to 15 million spores The sample masses were con-sistent with a previous determination of spore density at0081 gmL that indicated a high degree of entrained air
All chemicals used to prepare the silver-doped sol-gelcoated capillaries were also obtained and used as receivedfrom Sigma-Aldrich According to previously publishedprocedures17 two precursor solutions were preparedmixed and then drawn into 1-mm-diameter glass capil-laries The silver amine precursor consisted of a 51 vvratio of 1 N AgNO 3 to 28 NH3OH while the alkoxideprecursor consisted of a 21 vv ratio of methanol to te-tramethyl orthosilicate The alkoxide precursors weremixed with silver amine precursor in an 81 vv ratioApproximately 015 mL was drawn into the capillarycoating a 15-mm length After sol-gel formation the in-corporated silver ions were reduced with dilute sodiumborohydride which was followed by a water wash to re-move residual reducing agent
A 100 mL drop of a 50 mM DDA solution in ethanolpre-heated to 78 8C was added to each of the B cereusparticles to digest the spore coat After 1 minute the re-sultant solution was drawn into a SER-active capillarythat was immediately xed horizontally to an XY posi-tioning stage (Conix Research Spring eld OR) just in-side the focal point of an f 07 aspheric lens The lensfocused the beam into the sample and collected the scat-tered radiation back along the same axis A dichroic lter(Omega Optical Brattleborough VT) was used to re ectthe excitation laser to the lens and pass the Raman scat-tered radiation collected by the lens An f 2 achromatwas used to collimate the laser beam exiting a 200-mm-core-diameter source ber optic while a second f 2 ach-romat was used to focus the scattered radiation into a 365mm ber optic (Spectran Avon CT) A short-pass lterwas placed in the excitation beam path to block the sil-icon Raman scattering generated in the source ber fromre ecting off sampling optics and reaching the detectorA long-pass lter was placed in the collection beam pathto block the sample Rayleigh scattering from reachingthe detector A 785 nm diode laser (Process InstrumentsInc model 785-600 Salt Lake City UT) was used todeliver 100 to 150 mW of power to the sample A Fouriertransform Raman spectrometer (Real-Time Analyzersmodel IRA-785 East Hartford CT) and a silicon photo-avalanche detector (Perkin Elmer model C30902S Stam-ford CT) were used to acquire the SER spectra
RESULTS AND DISCUSSION
As an initial experiment the SER spectrum of 1 gLof DPA in water was measured using the newly devel-oped silver-doped sol-gel-coated capillaries (Fig 1A) Atthis concentration a high signal-to-noise ratio (SN) isobtained in 1 min In fact a reasonable spectrum is ob-tained in the same time frame for 1 mgL (Fig 1B) TheSER spectra are reasonably similar to the normal Raman(NR) spectrum obtained for a saturated solution of DPAin 1 N KOH (Fig 1C) and the following band shifts areobserved (NR to SER) 647 to 657 cm21 817 to 815cm21 998 to 1008 cm21 1384 to 1382 cm21 1434 to1428 cm21 and 1569 to 1567 cm21 Many of these bandshave been previously assigned1213 such as 998 cm21 to
APPLIED SPECTROSCOPY 353
FIG 1 SERS of DPA in water using silver-doped sol-gel-coated glasscapillary for (A) 1 gL and (B) 1 mgL (C ) NR of saturated DPA in 1N KOH in a glass capillary Spectral conditions (A) and (B) 150 mWof 785 nm 1-min acquisition time (C) 450 mW of 785 nm 5-minacquisition time both 8 cm 21 resolution
FIG 2 SERS of DPA extracted from 10 mg B cereus particle using100 mL of 50 mM hot DDA acquired in (A) 1 minute and (B) 2 seconds(C ) Attempted SERS of 50 mM hot DDA in ethanol using silver-dopedsol-gel-coated glass capillary acquired in 1 min Spectral conditions150 mW of 785 nm 8 cm21 resolution
the symmetric ring stretch 1384 cm21 to the OndashCndashOsymmetric stretch 1428 cm21 to the symmetric ring CndashH bend and 1569 cm21 to the asymmetric OndashCndashOstretch
The rst B cereus samples consisted of 2 mg of sporesin 2 mL of 5 mM hot DDA The samples were main-tained at 78 8C for 40 min and while hot approximately10 mL was drawn into a SER-active capillary Since spec-tra of DPA were obtained for these initial samples small-er spore masses higher DDA concentrations and shorterheating periods were examined In due course it wasfound that 10 mg of spores could be digested by 100 mLof 50 mM hot DDA in one minute and detected (Fig2A) In fact the signal was suf ciently intense that it canbe observed in as little as two seconds (Fig 2B) Theamount of DPA that was extracted was estimated to bebetween 5 and 10 mgL by comparing the signal intensityof the 1008 cm21 band to that measured for DPA in waterThis is consistent with previous research that found thatthe majority of the DPA is extracted from spores usingDDA11 and that B cereus spores contain approximately10 DPA by weight18 The SN of 127 for the 1008 cm21
band in the 1-minute SER spectrum suggests a limit ofdetection of approximately 250 ng of B cereus sporesbased on a SN of 3 Finally it should be noted that DDAdid not produce a detectable SER spectrum as shown inFig 2C
CONCLUSION
We have demonstrated that by combining rapid extrac-tion of dipicolinic acid from Bacillus cereus spores withchemical identi cation by surface-enhanced Raman spec-troscopy as little as 10 mg of spores can be detected Infact the entire measurement from the time of adding hotdodecylamine to the spores to the time when the dipi-colinic acid SER spectrum is acquired and analyzedcould be performed in less than two minutes The abilityof this method to distinguish between spore-forming bac-teria such as Bacillus anthracis and non-DPA containingpowders could help prevent costly shutdowns associated
with the appearance of suspicious material or intentionalmailing of common substances as an anthrax hoax Thismethod could also prove useful in detecting the locationof anthrax endospores in mail distribution facilities if an-other veri ed attack should occur
Research continues to fully characterize the surface-enhanced Raman spectroscopy signal intensities as afunction of sample concentration and to explore otherextractants that do not require the use of elevated tem-perature
ACKNOWLEDGMENTS
The authors are grateful for the support of the National Science Foun-dation (DMI-0296116 and DMI-0215819) and the US Army(DAAD13-02-C-0015 Joint Service Agent Water Monitor program)The authors are indebted to Chetan Shende for preparing the sol-gelcapillaries The authors also thank James Gillespie Nicholas Fell andAugustus Fountain for providing important background informationMark Farquharson for laboratory support and Professor Jay Sperry ofthe University of Rhode Island for supplying B cereus spores
1 V A Pasechnik C C Shone and P Hambleton Bioseparations 3267 (1993)
2 P J Jackson M E Hugh-Jones D M Adair G Green K K HillC R Kuske L M Grinberg F A Abramova and P Keim ProcNatl Acad Sci USA 95 1224 (1998)
3 B R Glick and J J Pasternak Molecular Biology Principles andApplications of Recombinant DNA (ASM Press Washington DC1994)
4 C A Bell J R Uhl T L Had eld J C David R F Meyer TF Smith and F R Cockerill III J Clin Microbiol 40 2897(2002)
5 D L Gatto-Menking H Yu J G Bruno M T Goode M Millerand A W Zulich Biosens Bioelectron 10 501 (1995)
6 J J Quinlan and P M Foegeding J Rapid Methods AutomationMicrobiol 6 1 (1998)
7 A A Hindle and E A H Hall Analyst (Cambridge UK) 1241599 (1999)
8 M S Ascher US Department of Health amp Human Services(httpwwwhhsgovophppresentationsAscherdoc)
9 R Nudelman B V Bronk and S Efrima Appl Spectrosc 54 445(2000)
10 D L Rosen C Sharpless and L B McBrown Anal Chem 691082 (1997)
354 Volume 58 Number 3 2004
11 P M Pellegrino N F Fell Jr and J B Gillespie Anal ChimActa 455 167 (2002)
12 W H Woodruff T G Spiro and C Gilvarg Biochem BiophysRes Commun 58 197 (1974)
13 E Ghiamati R S Manoharan W H Nelson and J F SperryAppl Spectrosc 46 357 (1992)
14 A P Esposito C E Talley T Huser C W Hollars C M Schal-dach and S M Lane Appl Spectrosc 57 868 (2003)
15 S Farquharson L Grigely V Khitrov W W Smith J F Sperryand G Fenerty J Raman Spectrosc paper accep ted (2003)
16 S Farquharson W W Smith S Elliott and J F Sperry SPIE-IntSoc Opt Eng 3855 110 (1999)
17 S Farquharson and P Maksymiuk Appl Spectrosc 57 479(2003)
18 F W Janssen A J Lund and L E Anderson Science (Washing-ton DC) 127 26 (1958)
SPIE -2003-5269 117
pH dependence of methyl phosphonic acid dipicolinic acid and cyanide by surface-enhanced Raman spectroscopy
Stuart Farquharson Alan Gift Paul Maksymiuk Frank Inscore and Wayne Smith
Real-Time Analyzers 87 Church Street East Hartford CT 06108
ABSTRACT US and Coalition forces fighting terrorism in Afghanistan and Iraq must consider a wide range of attack scenarios in addition to car bombings Among these is the intentional poisoning of water supplies to obstruct military operations To counter such attacks the military is developing portable analyzers that can identify and quantify potential chemical agents in water supplies at microgram per liter concentrations within 10 minutes To aid this effort we have been investigating the value of a surface-enhanced Raman spectroscopy based portable analyzer In particular we have been developing silver-doped sol-gels to generate SER spectra of chemical agents and their hydrolysis products Here we present SER spectra of methyl phosphonic acid and cyanide as a function of pH an important factor affecting quantitation measurements which to our knowledge has not been examined In addition dipicolinic acid a chemical signature associated with anthrax-causing spores is also presented Keywords Chemical warfare agents agent detection agent hydrolysis SERS Raman spectroscopy homeland security
1 INTRODUCTION In the past decade the Unites States and its allies have been challenged by a different kind of warfare exemplified by the terrorist attacks of September 11 2001 Suicide bombings and the use of chemical agents are the norm and military personnel must consider a wide range of attack scenarios Among these is the intentional poisoning of water supplies to obstruct military operations in Afghanistan and Iraq Currently colorimetric paper is used to detect agents on-site while gas chromatography combined with mass spectrometry (GCMS) is used in mobile support laboratories However both methods have severe drawbacks The paper changes color in response to contact with many chemicals besides CWAs causing a high incidence of false positives1 while GCMS although very chemically specific requires hours to perform and constant re-calibration234 Military operations would be greatly aided by a portable analyzer that can identify and quantify potential chemical agents at concentrations that impact safety This includes the analysis of drinking water supplies distribution and storage systems To meet this goal the Department of Defense has been investigating numerous approaches under the auspices of the Joint Service Agent Water Monitor (JSAWM) program5 The JSAWM must be capable of identifying the chemical agents with no false-positives and quantifying the agents at microgram per liter concentrations within 10 minutes678 Recently we and others have been investigating the ability of surface-enhanced Raman spectroscopy (SERS) to measure chemical agents9-12 bioagents13-17 and their hydrolysis products in water SERS employs the interaction of surface plasmon modes of metal particles with target analytes to increase scattering efficiency by as much as 1 million times18 In 1987 the potential of this approach to measure CWAs was demonstrated by measuring a series of pesticides19 Several of these organophosphonates have chemical structures similar to CWAs in particular P=O functional groups In our studies we have been employing silver-doped sol-gels to promote the SER effect The porous silica network of the sol-gel matrix offers a unique environment for immobilizing and stabilizing SER-active metal particles20-23 The sol-gels have been coated on the internal walls of standard glass vials to allow rapid SER analysis of small sample quantities without preparation We have measured over 100 chemicals with enhancements of 104 to 106 demonstrated reversible measurements in a flowing system reproducible measurements from vial-to-vial and measurements in multiple solvents including water20-23 Previously we used these vials to perform preliminary measurements of cyanide (CN) methylphosphonic acid (MPA) and dipicolinic acid (DPA) MPA is a hydrolysis product of the nerve agents (eg sarin Reaction 1) and may be a valuable indicator of nerve agent usage particularly since the alkyl methylphosphonic acids are relatively more stable than their corresponding parent complexes24 DPA is
SPIE -2003-5269 118
a chemical signature of spore forming bacteria such as Bacillus anthracis And in light of the inability to rapidly detect the anthrax spores distributed through the US mail in October 2001 a number of methods are being developed to extract and analyze this signature Reaction 1 Stepwise hydrolysis of Sarin to form hydrofluoric acid (HF) isopropyl methylphosphonic acid (IMPA) then methyl phosphonic acid (MPA) and 2-propanol In our previous SERS investigations MPA and DPA were measured at 50 and 100 mgL respectively In both cases limits of detection (LOD) were estimated at 100 microgL providing encouragement in that SERS may satisfy the needs of the JSAWM Since it has been shown that pH can substantially influence the intensity of SER bands25 which would clearly influence quantitative analysis we undertook the present study to determine the severity of these effects for cyanide methyl phosphonic acid and dipicolinic acid Furthermore we previously observed a band at 1050 cm-1 for MPA6 possibly due to an anion formed at basic pH Here we investigate the source of this spectral anomaly
2 EXPERIMENTAL All chemicals including potassium cyanide methyl phosphonic acid dipicolinic acid and those used to prepare the silver-doped and gold-doped sol-gel coated vials were obtained and used as received from Sigma-Aldrich All samples were prepared in HPLC grade water (Fischer Scientific Fair Lawn NJ) for SERS measurements The pH of these samples was adjusted using dilute nitric acid or potassium hydroxide and verified using a pH electrode (Corning Inc Corning NY) that had been calibrated with pH 400 700 and 1000 buffered standards from Fischer Scientific Once prepared the samples were transferred into the silver-doped sol-gel vials (Real-Time Analyzers Inc East Hartford CT) The vials were coated in a manner similar to that previously reported by adding ammonium hydroxide to a solution of silver nitrate tetramethyl orthosilicate (TMOS) and methanol20 Gold-vials were coated by adding nitric acid to a solution of gold tetrachloride TMOS and methanol The two precursor solutions were prepared mixed and transferred to 2-mL glass vials dried and heated After sol-gel formation the incorporated metal ions were reduced with dilute sodium borohydride (1mgmL) which was followed by a water wash to remove residual reducing agent After the resultant analyte solution was introduced the SER-active vial was immediately fixed horizontally to an XY positioning stage (Conix Research Springfield OR) just inside the focal point of an f07 aspheric lens The lens focused the beam into the sample and collected the scattered radiation back along the same axis A dichroic filter (Omega Optical Brattleborough VT) was used to reflect the excitation laser to the lens and pass the Raman scattered radiation collected by the lens An f2 achromat was used to collimate the laser beam exiting a 200 microm core diameter source fiber optic while a second f2 achromat was used to focus the scattered radiation into a 365 microm fiber optic (Spectran Avon CT) A short pass filter was placed in the excitation beam path to block the silicon Raman scattering generated in the source fiber from reflecting off sampling optics and reaching the detector A long pass filter was placed in the collection beam path to block the sample Rayleigh scattering from reaching the detector A 785 nm diode laser (Process Instruments Inc model 785-600 Salt Lake City UT) was used to deliver 100 to 150 mW of power to the sample A Fourier transform Raman spectrometer (Real-Time Analyzers model IRA-785 East Hartford CT) and a silicon photo-avalanche detector (Perkin Elmer model C30902S Stamford CT) were used to acquire the SER spectra
3 RESULTS AND DISCUSSION In a previous study of MPA6 aimed at developing a concentration calibration curve and determining limits of detection (LOD) we observed an anomaly at 1050 cm-1 Since it was found that the band intensity changed as a function of concentration the band must be associated with a sample parameter Two possible parameters photon flux and pH are examined here The first parameter was investigated by irradiating a 1mgmL MPA sample in a SER-active vial with laser powers of 200 mW and above and monitoring spectral changes It was immediately found that the 1050 cm-1 band
2O+ H HF + +OH
OH3H C
OP
OF
CH
CH3
3
3H C
OCP
OOH
CH
CH3
3
3H C
O
CPHO
CH
CH
3
3
C
Sarin IMPA MPA 2-propanol
SPIE -2003-5269 119
appeared and grew as a function of time and that the higher the laser power the faster the growth Figure 1 shows the growth of the 1050 cm-1 band over the course of 30 minutes when using 150 mW of 785 nm excitation while Figure 2A shows that the growth can be fit with a first order exponential equation namely I1050 = 03+05e-013t Furthermore the 760 cm-1 band could be fit with a first order decay equation with an identical exponential rate constant ie I760 = 08-
08e-013t The rates represent classical first order kinetics and their correspondence allows one to conclude that MPA is being transformed one-for-one into a photo-generated product At this time the photoproduct has not been positively identified but phosphonic (phosphorous) acid and phosphonate are likely candidates since the symmetric P(OH)3 stretch occurs at ~1050cm-1 Our studies show that a reduction in laser power to 100 mW at the sample for MPA essentially eliminates this degradation process This laser power or lower was used for further measurements
Methyl phosphonic acid is a diprotic acid that stepwise dissociates into two anions MPA- and MPA= according to the following reactions26
MPA MPA- + H+ pKa1 = 212 Reaction 2
MPA- MPA= + H+ pKa2 = 729 Reaction 3 The relative concentrations of MPA MPA- and MPA= can be determined at any pH by expressing [MPA] and [MPA=] in terms of [MPA-] using Reactions 2 and 3 and summing all three to equal the total starting concentration here 2 mgmL (0021M MW = 9602) viz
[MPA] + [MPA-] + [MPA=] = 0021M Equation 1 substituting from Reactions 1 and 2
([H+][MPA-])K1a + [MPA-] + (K2a[MPA-])[H+] = 0021M Equation 2 rearranging [MPA-] = 0021M(1+[H+]K1a + K2a[H+]) Equation 3 The relative concentrations of MPA MPA- and MPA= as a function of pH are shown in Figure 3 It is worth noting that near neutral pH both MPA- and MPA= will be present To confirm that the SER signal followed this pH dependence a starting solution consisting of 20 mg of MPA in 10 mL HPLC grade water was prepared and brought to pH of 20 using dilute nitric acid From this solution 2 mL were added to a SER-active vial and the SER spectrum recorded At this pH a peak at 760 cm-1 was barely discernable The 2 mL solution was returned to the starting solution and the pH was re-measured to correct for any changes that the silver-doped sol-gel vials might cause In most cases the change was less than 02 pH units and the pH is reported as the before and after average Next the pH of the
Figure 2 A) Exponential growth of 1050 cm-1 band and B) exponential decay of 760 cm-1 band for spectral series in Figure 1
Figure 1 Growth of 1050 cm-1 band as a function of time due to exposure to 150 mW of 785 nm Spectra are 5 sec each collected every 100-sec from 0 to 30-min
B
570 770 970 1170 1370Raman Shift (cm-1)
Arbit
rary
Unit
s
MPA Photodegradation
600 800 1000 1200 1400 Wavenumber (cm-1)
A
0 10 20 30 0 10 20 30 time (min) time (min)
Ram
an In
tens
ity (r
elat
ive)
30
min
0
SPIE -2003-5269 120
starting solution was adjusted to 325 using dilute KOH Again 2 mL were added to a vial and the SER spectrum recorded At this pH a reasonably strong 760 cm-1 band was observed This process was repeated as spectra were recorded at pHs of 70 74 75 79 85 and 100 A total of 1 mL of KOH was added diluting the total concentration by 10 Next the pH of the starting solution was made acidic by adding dilute nitric acid dropwise This time spectra were recorded at pHs of 72 69 64 and 37 Figure 4 shows the SER spectra for representative pHs (spectra were left out to simplify the figure) while Figure 3 shows the 760 cm-1 peak intensities as a function of pH (The band intensities were adjusted to compensate for dilution effects caused by the addition of HNO3 and KOH then normalized to 0021 M for the most intense band observed at pH 37) It is clear from Figure 3 that the 760 cm-1 band follows the MPA- concentration as a function of pH and must be assigned to this anion No bands were observed that corresponded to MPA or MPA= The lack of an MPA SER spectrum may be due to the absence of an attraction between the neutral analyte and the electropositive silver surface The same reasoning suggests that a strong SER spectrum should be observed for MPA= but it is not and a satisfactory explanation has not been found
0000
0005
0010
0015
0020
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14pH
Con
cent
ratio
n [M
]
pK1 = 212 pK2 = 729
MPA- MPA=
MPA
Figure 4 SER spectra of 002M MPA as a function of pH Conditions 100 mW of 785 nm 36 scans (1 min) 8 cm-1 recorded 2 min after sample introduction pH 19 and 20 not apparent on this scale pH 69 and 74 near identical to 70 and 75 and not shown for clarity
00
02
04
06
08
10
12
14
16
18
0 5 10 15 20 25 30Measurement Number
Ram
an In
tens
ity (7
60 c
m-1
)
Figure 3 Concentration dependence of MPA MPA- and MPA= as a function of pH for a 002M sample Intensity of 760 cm-1
band from Figure 6 as a function of pH ( for increasing basic adjustment for increasing acidic adjustment error was measured at ~10 for pH 64)
Figure 5 SER spectra of 002M MPA at pH 64 measured around a vial at three heights (9 points per height) Conditions as in Figure 4 but 10-sec scans
Since these measurements involved the removal and replacement of the SER-active vial in the sample holder to remove and add sample variation in the intensity as a function of vial position was minimized by illuminating the exact same height along the vial wall But this does not account for variability of the SERS response of the sol-gel coating around the vial To analyze this effect a vial containing MPA at pH 64 was rotated at ~ 40o
intervals at the original height and 18rdquo above and below this value Figure 5 shows the intensity of the 760 cm-1
band for the 27 positions It was found that the average value was 137plusmn014 an RSD of 10 overall and 5 for each height An error bar is included in Figure 4 for the pH 64 measurement
SPIE -2003-5269 121
With the above analyses in mind a preliminary investigation of the SER spectral response for MPA (as MPA-) as a function of concentration was performed A single vial was used for these measurements beginning with 1 mgL followed by measurements of 10 100 and 1000 mgL In all cases the pH was ~7 and 3 positions around the vial were measured per concentration Since the 760 cm-1 band was not observed for concentrations of 1 or 10 mgL using 100 mW of 785 nm the laser power at the sample was raised to 200 mW beginning with the 10 mgL concentration Photo-degradation was largely avoided (and not observed) by exposing the sample for only 33 seconds per spectral acquisition Representative spectra for 10 100 and 1000 mgL are shown in Figure 6 while a plot of the 760 cm-1 band intensity as a function of concentration is shown in Figure 7 These values were also used to estimate limits of detection based on the signal-to-noise ratio (SN) of the 760 cm-1 band using 100 mW of 785 nm laser excitation and a 10-min acquisition time As summarized in the Figure 7 inset the lower the measured concentration the lower the predicted LOD Even if as estimated 210 microgL could be measured using the silver-doped sol-gel vials an improvement of a factor of 70 is still required to meet JSAWM goals of measuring 3 microgL in 10-minutes Similar to MPA DPA is a diprotic acid (pKa1 = 216 and pKa2 = 692) and variations in pH will effect the relative concentrations of DPA DPA- and DPA= and possibly the SER spectra and band intensities This could prove significant if an acid or base is used to denature anthrax spores with the goal of extracting and analyzing DPA The MPA pH study described above was mimicked for DPA except that the starting solution consisted of 20 mg of DPA in 20 mL HPLC grade water (60x10-3M MW = 1671) The initial solution had a pH of 245 which was made basic by dilute KOH to pHs of 355 433 487 559 1069 and 1166 SER spectra were recorded at each pH using 100 mW of 785 nm and a 44-sec acquisition time Next one drop of concentrated nitric acid was used to remake the solution acidic at a pH of 200 Again sequential pH measurements were performed at 383 510 735 and 822 The solution pH was made acidic a third time but to pH 219 171 then 135 Throughout this process no more than 20 drops of acid or base were added and therefore the concentration was diluted by no more than 10 Most of the spectral bands showed a minor decrease in intensity as a function of increasing pH values However the bands at 525 and 795 cm-1 showed the most dramatic changes which occurred at acid pH Figure 8 shows the SER spectra of DPA for the spectral region and pH range of interest The identity of the DPA species was determined by plotting the normalized peak intensities with the lowest value set to 0 and the highest to 0006 M as a function of pH and overlaying these values on a plot of the relative concentrations for DPA DPA- and DPA= as previously done for MPA (Figure 9) As can be seen the 525 cm-1 band clearly corresponds to DPA The correspondence of the 795 cm-1 band to this species is less clear as the band retains intensity until pH of 55 This can be attributed to contribution to the overlapping band at 810 cm-1 which does not change as a function of pH The fact that most bands are observed at all pHs suggest that the primary interaction with silver is through the ring nitrogen This is supported by the fact that the most intense band occurs at 1008 cm-1 attributed to a symmetric ring breathing mode and that this interaction has been characterized for pyridine in numerous papers27
0
1
2
3
4
5
6
7
0 200 400 600 800 1000 1200
MPA Concentration (mgL)
760
Ban
d In
tens
ity
Figure 6 SER spectra of MPA in water at A) 1000 B) 100 and C) 10 mgL Conditions pH of 7 silver-doped sol-gel coated vial 200 mW of 785 nm 33-sec 8 cm-1 resolution
Figure 7 Plot of SER intensity of 760 cm-1 band of MPA as a function of concentration using 200 mW of 785 nm Inset table includes average intensity LOD standard deviation and percent deviation for each concentration but for 100 mW and 10 min
conc (mgL) LOD ave stddev dev10 021 002 1021100 072 010 14301000 312 040 1280
A
B C
SPIE -2003-5269 122
The ability to detect dipicolinic acid (DPA) by SERS is demonstrated in Figure 10 Even at 1 mgL the primary bands are visible The intensity of the dominant ring breathing mode observed at 1008 cm-1 for DPA in water (pH = 25 to 55) is plotted as a function of concentration in Figure 11 These values were also used to estimate limits of detection based on the SN of the 1008 cm-1 band using 100 mW of 785 nm laser excitation and a 10-min acquisition time Again the lower the measured concentration the lower the predicted LOD (see Figure 11 inset) and detection of 160 microgL is possible Gastrointestinal anthrax requires significant more spores than inhalation anthrax28 and a limit of detection might be placed at 1 million spores in 1 liter of water or 10 microgL Since each spore contains ~10 CaDPA by weight29 a goal for DPA might be 1 microgL indicating that the present measurements must be improved by nearly two orders of magnitude Due to the increased hazards of handling HCN gas KCN salt was used for these experiments Nevertheless all sample preparations were performed in a chemical hood KCN completely dissolves in water but its conjugate acid HCN is formed and has a Ka of 615x10-1030 viz
HCN CN- + H+ pKa = 921 Reaction 4
Figure 8 SER spectra of 1 mgmL DPA as a function of pH Conditions 100 mW 785 nm 100 scans (44-sec)
Figure 9 Concentration dependence of DPA DPA- and DPA= as a function of pH for a 0006M sample Intensity of 525 () and 795 (diams) cm-1 bands from Fig 8 as a function of pH
pH
135 171 219 383
0
0001
0002
0003
0004
0005
0006
0007
0 2 4 6 8 10 12 14pH
Con
cent
ratio
n [M
]
DPADPA-DPA=795525
DPA DPA=DPA-
pK2 = 692pK1 = 216
795 525
0
05
1
15
2
25
0 200 400 600 800 1000 1200
DPA Concentration (mgL)
1008
Ban
d In
tens
ity
conc (mgL) lod-10min-100mw1 017
10 016100 103
1000 355
Figure 10 SER spectra of DPA in water at A) 1000 B) 100 C) 10 and D) 1 mgL Conditions pH of 25-55 silver-doped sol-gel coated vial 175 mW of 785 nm 1-min 8 cm-1 D) has been multiplied by x10 to make bands visible
A
B C
D
Figure 11 Plot of SER intensity of 1008 cm-1 band of DPA as a function of concentration using 175 mW of 785 nm Inset table includes LOD in mgL for each concentration but for 100 mW and 10 min
SPIE -2003-5269 123
Consequently the cyanide concentration must be determined for each initial KCN concentration Specifically the samples prepared with concentrations of 01 1 10 100 and 1000 mgL of KCN produced CN- concentrations of 63x10-3 033 69 89 and 964 mgL at pHs of 816 90 967 102 and 107 respectively The pH dependence for the HCN and CN- concentrations are shown in Figure 12 Thus as the amount of KCN added to the solution decreases so does the pH of the solution (becomes less basic) and according to Reaction 4 the relative amount of CN- to HCN also decreases For example in the preparation of a 01 mgL solution of KCN the pH is shifted from 7 for pure water to only 816 and only 63 of the starting material becomes CN- or 63x10-3 mgL In comparison for a solution of 1000 mgL the pH is shifted from 7 to 107 and 96 of the starting material becomes CN- This is significant since the cyanide ion is better able to adsorb onto the silver particles and become SERS active SER spectra of 10 100 and 1000 mgL of KCN samples are shown in Figure 13 and a corresponding plot of the 2100 cm-1 band intensity as a function of concentration is shown in Figure 14
000102030405060708091011
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14pH
Con
cent
ratio
n [m
gm
L]
pKa = 921
CN -HCN
The SER spectra of cyanide are dominated by the single CN stretch at 2100 cm-1 which occurs in normal Raman spectra of solutions at 2080 cm-1 However a low frequency mode occurs at 135 cm-1 due to a Ag-CN stretch (not shown) indicative of a strong surface interaction It is also observed that as the concentration decreases the CN stretch shifts to 2140 cm-1 This shift has been attribute to the formation of a tetrahedral Ag(CN)3
2- surface structure31 as well
Figure 14 Concentration dependence of KCN SERS measured under conditions in Fig 11 Concentrations are 1 01 and 001 mgml Intensities are measured for the CN stretch at 2100 cm-1 Inset table includes LOD in mgL for each concentration in Figs 13 and 15 but for 100 mW and 10 min
Figure 13 SER spectra of KCN in water at A) 1000 B) 100 and C) 10 mgL Conditions pHs of 107 102 and 97 silver-doped sol-gel coated vial 100 mW of 785 nm 1-min 8 cm-1 C) has been multiplied by x10 to make band visible
Figure 12 Concentration dependence of HCN and CN- as a function of pH for a 1 mgmL sample Calculated intensity of 2100 () cm-1 band for a 1 mgml sample at pHs of 816 90 967 102 and 107
A
B
C
0
20
40
60
80
100
120
140
0 200 400 600 800 1000 1200CN Concentration (mgL)
2100
Ban
d In
tens
ity
Figure 15 SER spectra of KCN in water at A) 10 B) 1 and C) 01 mgL Conditions pHs of 97 12 and 12 gold-doped sol-gel coated vial 100 mW of 785 nm 1-min 8 cm-1 C) has been multiplied by x10 to make band visible
A
B
C
conc (mgL) lod-10min-100mw Condition01 001 Au-pH 121 007 Au-pH 12
69 003 Au-pH 9769 016 Ag-pH 9789 022 Ag-pH 102946 113 Ag-pH 107
SPIE -2003-5269 124
as to CN adsorbed to two different surface sites32 Alternatively the 2140 cm-1 band could be attributed to HCN since this species dominates at lower concentrations However it is unlikely that this species would be attracted to the electropositive silver surface Further both peaks should be present at pHs between 85 and 105 but this is not observed It has also been suggested that at concentrations near and above monolayer coverage the CN- species is forced to adsorb end-on due to crowding and at lower concentrations the molecule can reorient to lie flat33 This suggests that the 2100 and 2140 cm-1 bands correspond to the end-on and flat orientations respectively As Figures 13 and 14 show the intensity of the CN stretch for the 89 mgL sample is nearly as intense as the 964 mgmL band This suggests that the Raman signal for the flat orientation is more enhanced However more extensive measurements are required to verify this point Since resent research has suggested that cyanide may be more effectively detected on gold measurements of KCN solutions were also performed using gold-doped sol-gel vials Preliminary measurements are shown in Figure 15 for samples prepared from 01 1 and 10 mgL KCN Since the pHs are 816 90 967 the resultant CN- concentrations are 63x10-3 033 and 69 mgL Initially only the highest concentration was observed and the signal intensity was significantly better than the equivalent concentration measured using silver In an effort to shift Reaction 4 to the left transforming HCN to CN- (Le Chatelierrsquos principle) KOH was added to the lower concentration samples producing solutions with pH 12 This effectively forces all of the cyanide in solution to be CN- or 01 and 10 mgL respectively More importantly the CN stretch is now observed in the SER spectra The band appears at 2125 cm-1 as has been previously reported for gold12 As calculated for MPA and DPA LODs can be estimated from this data For the three concentrations of cyanide on silver the LODs are 016 to 11 mgL for 100 mW of 785 nm laser excitation and a 10-min acquisition time For gold pH adjusted the LODs are10 to 70 microgL an improvement of more than 10 times silver Nevertheless either substrate is sufficient to meet the JSAWM goals of measuring 3 mgL in 10-minutes as the requirements form cyanide are much less stringent than the nerve agents
4 CONCLUSIONS Here we examined the surface-enhanced Raman spectral response for methyl phosphonic acid dipicolinic acid and cyanide as a function of pH It was determined that the most prominent peak at 760 cm-1 reaches a maximum intensity between pH 3 and 7 and corresponds to the MPA- species Neither the MPA nor MPA= species appear to generate a SER spectrum and consequently no spectra were observed below pH 2 or above pH 8 In this study we also found that higher laser powers could cause photodegradation of MPA signified by the exponential growth of a band at 1050 cm-1 which is tentatively assigned to phosphorous acid Unlike MPA DPA was observed at all pHs This is attributed to the dominant interaction of the pyridine functional group with silver Minor spectral changes were observed at acid pHs and were assigned to neutral DPA Like MPA SER spectra of cyanide were pH dependent No spectra were observed for the HCN species while CN- was best observed at pHs more basic than 8 Preliminary concentration studies for the three analytes allowed estimating limits of detection for MPA DPA and CN using 100 mW of 785 nm and a 10-min acquisition time of 210 165 and 70 microgL respectively Although the latter value suggests that cyanide can be measured at sufficiently low concentrations to meet JSAWM goal improvements by 100 to 200 times are required for MPA and DPA It is clear from this study that pH of the sample is important and must be taken into account when developing concentration calibration curves as well as fieldable analyzers Future work will include tailoring the sol-gel with specific functional groups to dictate sample pH and thereby optimize sensitivity
ACKNOWLEDGMENTS The authors are grateful for the support of the US Army (DAAD13-02-C-0015 Joint Service Agent Water Monitor program) The authors would also like to thank Dr Steve Christensen of the US Army SBCCOM for helpful discussions and Mr Chetan Shende of Real-Time Analyzers for assistance in development of the gold-doped sol-gels
REFERENCES 1 Erickson B Analytical Chemistry News amp Features June 1 397A (1998)
SPIE -2003-5269 125
2 Johnston RL Hoefler CM Fargo JC and Moberley B AT-ONSITE 5-8 (1994) 3 Black RM RJ Clarke RW Read and MT Reid J Chromatography 662 301-321 (1994) 4 USA Army Center for Health Promotion and Preventive Medicine (USACHPPM) Technical Guideline 230A 5 Jensen JL NC Wong W R Loerop SPIE 4775-03 (2001) 6 Hoenig SL Handbook of Chemical Warfare and Terrorism Greenwood Press Wesport CT (2002) 7 Munro NB SS Talmage GD Griffin LC Waters AP Watson JF King and V Hauschild Env Health
Persp 107 933-974 (1999) 8 Holstege CP Kirk M Sidell FR Crit Care Clin13 923-42 (1997) 9 Farquharson S P Maksymiuk K Ong and S Christesen SPIE 4577 166-173 (2001) 10 Lee Y and S Farquharson SPIE 4378 21-26 (2001) 11 Spencer KM J Sylvia S Clauson and J Janni SPIE 4577 158-165 (2001) 12 Tessier P S Christesen K Ong E Clemente A Lenhoff E Kaler and O Velev Applied Spectroscopy 56
1524-1530 (2002) 13 Farquharson S WW Smith S Elliott and JF Sperry SPIE 3533 207-214 (1998) 14 Farquharson S WW Smith S Elliott and JF Sperry SPIE 3855110-116 (1999) 15 Farquharson S WW Smith YH Lee S Elliott and JF Sperry SPIE 4575 62-72 (2002) 16 Guzelian AA J Sylvia J Janni S Clauson and KM Spencer SPIE 4577 182-192 (2001) 17 Shende C F Inscore A Gift P Maksymiuk and S Farquharson in press 18 Weaver MJ S Farquharson and MA Tadayyoni J Chem Phys 82 4867-4874 (1985) 19 Alak AM and T Vo-Dinh Analytical Chemistry 59 2149-2153 (1987) 20 Lee Y and S Farquharson SPIE 4206 140-146 (2000) 21 Farquharson S and Y Lee SPIE 4200-16 (2000) 22 Lee Y S Farquharson and P M Rainey SPIE 3857 76-84 (1999) 23 Lee Y S Farquharson H Kwong and M Shahriari SPIE 3537 252-260 (1998) 24 Wang J M Pumera G Collins and A Mulchandani Analytical Chemistry 74 6121-6125 (2002) 25 Dou X YM Jung Z-Q Cao and Y Ozaki Applied Spectroscopy 53 1440-1447 (1999) 26 Data supplied by S Christesen and K Ewing 27 Kerker M and B Thompson Eds SPIE MS 10 (1990) 28 Inglesby TV DA Henderson JG Bartlett JAMA 287 2236 (2002) 29 FW Janssen AJ Lund and LE Anderson Science 127 26 (1958) 30 Lide DR Ed Handbook of Chemistry and Physics CRC Press 77th Ed (1996-7) 31 Billmann J G Kovacs and A Otto Surf Sci 92 153 (1980) 32 Murray CA and S Bodoff Phys Rev B 32 671 (1985) 33 Kellogg D and J Pemberton J Phys Chem 91 1120 (1987)
SPIE-2003-5269 16
Chemical agent detection by surface-enhanced Raman spectroscopy
Stuart Farquharson Alan Gift Paul Maksymiuk Frank Inscore and Wayne Smith
Real-Time Analyzers 87 Church Street East Hartford CT 06108
Kevin Morrisey and Steven D Christesen US Army SBCCOM Aberdeen Proving Ground MD 21010
ABSTRACT
In the past decade the Unites States and its allies have been challenged by a different kind of warfare exemplified by the terrorist attacks of September 11 2001 Although suicide bombings are the most often used form of terror military personnel must consider a wide range of attack scenarios Among these is the intentional poisoning of water supplies to obstruct military operations in Afghanistan and Iraq To counter such attacks the military is developing portable analyzers that can identify and quantify potential chemical agents in water supplies at microgram per liter concentrations within 10 minutes To aid this effort we have been investigating the value of a surface-enhanced Raman spectroscopy based portable analyzer In particular we have been developing silver-doped sol-gels to generate SER spectra of chemical agents and their hydrolysis products Here we present SER spectra of several chemical agents measured in a generic tap water Repeat measurements were performed to establish statistical error associated with SERS obtained using the sol-gel coated vials Keywords Chemical agents chemical agent detection SERS Raman spectroscopy
1 INTRODUCTION
In the autumn of 2001 terrorism within US borders became a sobering reality While extensive efforts are being implemented to secure the homeland US and Coalition forces in Afghanistan and Iraq are constantly faced with terrorist attacks In addition to car-bombings the military has identified several non-traditional attack scenarios including poisoning of water supplies by chemical warfare agents (CWAs) To counter this threat the Department of Defense has been investigating numerous approaches to detect such attacks under the auspices of the Joint Service Agent Water Monitor (JSAWM) program1 The JSAWM must be capable of identifying the chemical agents with no false-positives and quantifying the agents at microgram per liter concentrations within 10 minutes2 This includes the analysis of drinking water supplies distribution and storage systems Currently colorimetric paper is used to detect agents on-site while gas chromatography combined with mass spectrometry (GCMS) is used in mobile support laboratories However both methods have severe drawbacks The paper changes color in response to contact with many chemicals besides CWAs causing a high incidence of false positives3 while GCMS although very chemically specific requires up to an hour to perform and regular re-calibration456 Vibrational spectroscopy has also been investigated as a potential method of rapidly detecting CWAs7-
11 as early as 1985 in the case of infrared spectroscopy8 and 1988 in the case of Raman spectroscopy11 Again however these techniques also have limitations when it comes to measuring trace poisons in water Infrared spectra would be dominated by the very strong absorption of water which would obscure absorptions by most other chemicals present Whereas Raman spectroscopy is simply not a very sensitive technique and detection limits are typically grams per liter Surface-enhanced Raman spectroscopy (SERS) offers several advantages over conventional vibrational methods and that may provide the necessary sensitivity required for detecting trace quantities of chemical agents in water SERS employs the interaction of surface plasmon modes of metal particles with target analytes to increase scattering efficiency by as much as 1 million times12 In 1987 the potential of this approach to measure CWAs was demonstrated by measuring a series of pesticides13 Several of these organophosphonates have chemical structures similar to CWAs in particular P=O functional groups In the past few years we and others have further explored the ability of SERS to detect CWAs14-17 and even bioagents 18-21 We have been employing silver-doped sol-gels to promote the SER effect
SPIE-2003-5269 17
in these studies The porous silica network of the sol-gel matrix offers a unique environment for immobilizing and stabilizing SER-active metal particles22-25 The silver-doped sol-gels have been coated on the internal walls of standard glass vials to allow rapid SER analysis of small sample quantities without preparation We have measured over 100 chemicals with enhancements of 104 to 106 demonstrated reversible measurements in a flowing system reproducible measurements from vial-to-vial and measurements in multiple solvents including water21-25 Previously we used these vials to perform preliminary measurements of cyanide (CN) and methylphosphonic acid (MPA) Most of the nerve agents form MPA during hydrolysis while Tabun forms CN a chemical agent in its own right In another paper including in these proceedings we examined the limits of detection (LOD) for MPA by measuring a series of concentrations down to 50 mgL and estimated a limit of detection of 100 microgL26 These measurements provide encouragement in that SERS may satisfy the needs of the JSAWM To further establish the viability of SERS in particular silver-doped sol-gels here we present analysis of cyanide mustard and VX in tap water The measurements performed at the US Armyrsquos Edgewood Chemical Biological Center Aberdeen MD also included numerous repeat measurements to establish reproducibility
2 EXPERIMENTAL 2a General All chemicals including potassium cyanide 2-chloroethylethyl sulfide and those used to prepare the silver-doped and gold-doped sol-gel coated vials were obtained and used as received from Sigma-Aldrich All samples were prepared in a chemical hood using HPLC grade water unless otherwise noted (Fischer Scientific Fair Lawn NJ) for SERS measurements Once prepared the samples were transferred into the silver-doped sol-gel vials (Real-Time Analyzers Inc East Hartford CT) The vials were coated in a manner similar to that previously reported by adding ammonium hydroxide to a solution of silver nitrate tetramethyl orthosilicate (TMOS) and methanol22 The two precursor solutions were prepared mixed and transferred to 2-mL glass vials dried and heated After sol-gel formation the incorporated metal ions were reduced with dilute sodium borohydride (1mgmL) which is followed by a water wash to remove residual reducing agent After the resultant analyte solution was introduced the SER-active vial was fixed horizontally to an XY positioning stage (Conix Research Springfield OR) just inside the focal point of an f07 aspheric lens The lens focused the beam into the sample and collected the scattered radiation back along the same axis A dichroic filter (Omega Optical Brattleborough VT) was used to reflect the excitation laser to the lens and pass the Raman scattered radiation collected by the lens An f2 achromat was used to collimate the laser beam exiting a 200 microm core diameter source fiber optic while a second f2 achromat was used to focus the scattered radiation into a 365 microm fiber optic (Spectran Avon CT) A short pass filter was placed in the excitation beam path to block the silicon Raman scattering generated in the source fiber from reflecting off sampling optics and reaching the detector A long pass filter was placed in the collection beam path to block the sample Rayleigh scattering from reaching the detector A 785 nm diode laser (Process Instruments Inc model 785-600 Salt Lake City UT) was used to deliver 100 to 150 mW of power to the sample A Fourier transform Raman spectrometer (Real-Time Analyzers model IRA-785 East Hartford CT) and a silicon photo-avalanche detector (Perkin Elmer model C30902S Stamford CT) were used to acquire the SER spectra
2b Edgewood Chemical Biological Center The surface-enhanced Raman spectral measurements at the US Armyrsquos Chemical Biological center presented here were all performed on September 12 2003 To expedite measurements a plate was machined to hold up to 12 SER-active sample vials (Figure 1) The plate fit a standard XY plate reader that could be programmed Pure KCN bis-(2-chloroethyl)sulfide (distilled mustard gas HD) and ethyl S-2-diisopropylamino ethyl methylphosphonothioate (VX) were obtained on-site and used to prepare 1 gL tap water solutions in a chemical hood with appropriate safety equipment Simulated tap water was prepared by adding 10 chemicals most often found in tap water at appropriate concentrations to distilled water (Table 1) SER measurements were also performed in a chemical hood For added safety the FT-Raman instrument was placed outside the laboratory and 30 foot fiber optic and electrical cables were used to allow remote SERS measurements and plate manipulation For each experiment 1gL samples were prepared and added to 9 individual vials which were then loaded on the plate In some cases a tenth vial was included as a blank
Table 1 Generic Tap Water Composition
Compound mgL NaHCO3 100 CaSO4 27 MgSO4bull7H2O 67 NaNO3 1 Fulvic Acid 1 K2HPO4 07 KH2PO4 03 (NH4)2HSO4 001 NaCl 001 FeSO4 0001 pH 76-78
SPIE-2003-5269 18
A software program was written that allowed selecting the sequence that the vials were measured the number of positions along the length of the vials to measure (1 to 5) and the number of scans to co-add During sample analysis the program displayed the vial being analyzed the point being analyzed and the spectrum as it was being acquired Once all the data was collected a second software program was written to rapidly analyze the data The spectra collected for all the vials on a plate could be loaded at one time and then the spectra for each point could be displayed simultaneously or separately The user could then select the Raman peak to analyze in terms of peak height or area This was accomplished by selecting points on either side of the peak to define a baseline of zero The peak height or area could then be computed for all of the spectra loaded and then exported to a spreadsheet for statistical analysis
Figure 1 A) Vial Holder 6 slots to hold 2 vials each end-to-end B) Measurement Configuration Program user selects vials to measure sequence number of points per vial (1 to 5) and number of scans per point C) Spectral Acquisition Program shows spectrum being collected which vial and position D) Spectra Analysis Program user selects spectra to analyze by plate vial and point (s) as well as two wavenumbers defining the peak and the baseline to subtract The image is of 5 repeat measurements of 10 mgL KCN in generic tap water 16 sec each 100 mW of 785 nm
3 RESULTS AND DISCUSSION Raman and surface-enhanced Raman spectra were obtained for potassium cyanide bis-(2-chloroethyl)sulfide and ethyl S-2-diisopropylamino ethyl methylphosphonothioate representing three classes of chemical agents cyanides mustards and nerve agents respectively Spectra were also obtained for 2-chloroethyl ethyl sulfide (CEES) a structural analogue to HD which was included in the study to aid in assigning spectral bands KCN salt was used for cyanide experiments to avoid the increased hazards of handling HCN gas KCN completely dissolves in water forming its conjugate acid HCN according to its Ka of 615x10-1027 and at a concentration of 1 mgmL results in a pH 107 solution This is significant in that only CN- appears to interact sufficiently with silver to produce a SER spectrum and no spectral signal is observed below pH 726 Figure 2 shows the SER and normal Raman spectra for KCN The SERS spectrum of 1mgml KCN in water shows a single intense somewhat broad feature at
A B
C D
SPIE-2003-5269 19
2100 cm-1 assigned to the single CequivN stretch The band is much sharper in the normal Raman spectra of the solid KCN salt at 2074 cm-1 This band does broaden and shift to 2080 cm-1 in solution (not shown) However the observed SERS frequency is attributed to interaction with silver and low frequency mode at 135 cm-1 attributed to a Ag-CN stretch (not shown) supports this conclusion
cm-1 band to a C-S stretch but the authors concede that it is in fact more likely a C-Cl stretch28 It appears that the most intense Raman bands at 648 692 and 747 cm-1 shift to 620 660 and 720 cm-1 in the SER spectra and are tentatively assigned as above The width of these bands suggests that they overlap underlying spectral features Additional bands in the Raman spectra occur at 972 1034 1049 1263 1286 1423 1442 2865 2935 and 2960 cm-1 Corresponding bands occur in the SER spectra at 964 1015 1054 1286 1410 1447 2865 and 2935 cm-1 Most of these bands are associated with alkane modes specifically the bands at approximately 1040 cm-1 to a C-C stretch 1290 cm-1 to a CH2 in-phase twist 1440 cm-1 to a CH2 wag 2865 cm-1 to a symmetric CH2 stretch and 2965 cm-1 to an asymmetric CH2 stretch The Raman and SER spectra of sulfur mustard were measured at the Edgewood center (Figure 4) Both spectra are largely similar to CEES The C-Cl and C-S bands in the Raman spectrum of HD now occur at 640 655 700 739 and 760 cm-1 and are more resolved possibly due to the increased molecular symmetry Theoretical calculations indicate that the first three bands are due to C-Cl stretching modes and the latter two to C-S stretching modes28 Only the C-Cl bands maintain significant intensity in the SER spectra occurring at 624 and 643 cm-1 which is attributed to the
Prior to measurements of HD CEES was examined by Raman and SER spectroscopy (Figure 3) CEES also known as half-mustard is essentially identical to HD except one of the chlorine end atoms is replaced by a hydrogen atom Again although not as toxic as HD CEES is a blister agent and dilute aqueous samples were prepared using appropriate safety equipment Both the Raman and SER spectra of CEES are similar and dominated by bands between 600 and 800 cm-1 These are associated with C-Cl and C-S stretching modes which are tentatively assigned to 648 and 747 cm-1 in the Raman spectra respectively The shoulder at 630 cm-1 the overlapped band at 660 cm-1 and the strong band at 692 cm-1 could also be due to these modes or their asymmetric counterparts It is worth noting that theoretical calculations assign the 692
Figure 2 A) SER and B) NR spectra of KCN Conditions A) 1 mgml in tap water 100 mW of 785 nm at sample 1-min acquisition time B) solid 300 mW of 785 nm 5-min All spectra are 8 cm-1 resolution
A
B
Figure 3 A) SER and B) NR spectra of CEES Conditions A) 1 vv (10 mgml) in MeOH 100 mW of 785 nm 1-min acquisition time B) neat 300 mW of 785 nm 5-min
A
B
Cl-CH2-CH2-S-CH2-CH3
A
B
Figure 4 A) SER and B) NR of HD Conditions A) 1mgml in tap water B) pure both 100 mW of 785 nm 1-min
Cl-CH2-CH2-S-CH2-CH2-Cl
SPIE-2003-5269 20
expected strong interaction between chlorine and silver and adds support to the assignment of this band to a C-Cl stretch Weaker overlapping bands occur at 670 692 and 724 cm-1 the latter possibly due to C-S stretching modes Again the alkane modes are apparent in the normal Raman spectra of HD but only a broad feature at 1300 to 1450 cm-1 suggests CH2 contributions in the SER spectrum Although the observed bands in the VX spectrum have not been assigned (Figure 4) a computer generated Raman spectrum29 predicts many of the same features with surprising accuracy and are used here Two intense bands at 460 and 530 cm-1 closely match predicted bands at 463 and 546 cm-1 assigned to a CH3-P=O bend and a PO2CS wag Three highly overlapped bands occur at 694 745 and 771 cm-1 matching predicted bands at 713 730 and 760 cm-1 The first
Table 2 Measured SER peak heights for the CN stretch at 2100 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
Figure 5 A) SERS and B) NR spectra of VX Conditions A) 1 vv (10 mgml) in MeOH B) pure sample both 100 mW of 785 nm 1-min acquisition time
A
B
two have been assigned to a C-S stretch and CH2bend respectively while the latter has been attributed to either a P-C stretch or an O-C-C stretch Although the 745 cm-1 band may alternatively be assigned to a C-S stretch based on the previous measurements of CEES and HD The relatively intense bands at 890 1106 1218 1445 and 1465 cm-1 also match predicted bands at 880 1108 1216 1440 and 1464 cm-1 that are assigned to a C-C stretch CH3 rock N-C3 stretch various C-H3 bends and C-H bends respectively Both the computer generated and the measured spectra contain numerous other less intense bands One is worth mentioning A unique band appears at 370 cm-1 that is predicted at 368 cm-1 and corresponds to an O-P=O bend The surface-enhanced Raman spectrum of VX is also rich with spectral features It has the unique low frequency band at 370 cm-1 as well as a second band at 380 cm-1 that is assigned tothe S-P-O bend predicted in the normal Raman spectrum at 388 cm-1 Based on the measured and predicted normal Raman spectra the following SERS assignments are given 460 cm-1 to the CH3-P=O bend 544 cm-1 to the PO2CS wag 738 cm-1 to a C-S stretch (based on arguments above) 890 cm-1
to a C-C stretch 1101 cm-1 to a CH3 rock and 1456
cm-1 to a C-H bend The ability of SERS to measure chemical agents in water containing real-world chemical interferents was tested by using the generic tap water described in Table 1 The ability to reproduce measurements was accomplished by preparing three separate water stock solutions which were used to prepare three sample solutions each of 1 mgmL KCN in the generic tap water The samples defined as vials 1-9 were then measured at 4 points per vial in a semi-random fashion such that errors associated with stock solution preparation and errors associated with instrument drift could be identified No trends were apparent that signified such systematic errors Each spectrum collected consisted of 20 averaged scans taking 16 seconds at 8 cm-1 resolution The laser power at the sample was measured periodically during the day and it ranged from 102 to 105 mW spacccc
spot Vial 1 Vial 2 Vial 41 13754 13004 128192 13519 12692 12909
1 3 13541 12721 126394 13462 12648 12651
ave 13569 12766 12755Vial 6 Vial 3 Vial 5
1 11586 14068 10757 2 11236 14402 115122 3 11376 14546 11559
4 10894 11763 11214ave 11273 13695 11261
Vial 7 Vial8 Vial 91 11198 15804 11023
2 11243 15716 88753 3 11562 15216 11218
4 11655 15084 11077ave 11415 15455 10548
CN AVG STDEV ERR1 mgmL pk ht 12526 1560 1245
stock solution
SPIE-2003-5269 21
The CN measurements consisting of 36 data points produced an average height of 125 for the 2100 cm-1 peak with a standard deviation of 156 or 125 (Table 2) The HD and VX measurements were performed precisely the same way (Tables 3 and 4) For HD the 624 cm-1 peak was used for analysis and it had an average height of 53 with a standard deviation of 068 or 129 while for VX the 544 cm-1 peak was used for analysis and it had an average height of 1051 with a standard deviation of 308 or 293 The greater error in the VX measurements can be somewhat attributed to Vial 4 which produced lower SER signal intensities But removing this vial from the data set changes the standard deviation to 233 only a modest improvement
spot Vial 1 Vial 2 Vial 41 525 609 4682 45 675 484
1 3 527 644 5094 559 807 669
ave 51525 68375 5325Vial 6 Vial 3 Vial 5
1 521 574 575 2 536 53 4492 3 509 508 451
4 565 594 379ave 53275 5515 4635
Vial 7 Vial 8 Vial 91 457 589 413
2 583 497 4273 3 544 505 4
4 528 5 509ave 528 52275 43725
HD AVG STDEV ERR1 mgmL pk ht 530 068 1291
stock solution
spot Vial 1 Vial 2 Vial 41 1464 1034 4672 1485 99 61
1 3 1491 989 5684 1041 777 553
ave 137025 9475 5495Vial 6 Vial 3 Vial 5
1 1058 942 1293 2 697 121 965
2 3 727 1206 11124 689 1254 746
ave 79275 1153 1029Vial 7 Vial 8 Vial 9
1 1127 783 163 2 1358 812 1614
3 3 1371 908 15124 1097 875 1397
ave 123825 8445 153825
VX AVG STDEV ERR1 mgmL pk ht 1051 308 2925
stock solution
4 CONCLUSIONS In this paper we examined the ability of surface-enhanced Raman spectroscopy to reproducibly measure CN HD and VX in tap water without chemical interference Both normal and surface-enhanced Raman spectra were examined to select unique bands suitable to identify and quantify these chemical agents For SER measurements the 2100 cm-1 C-N stretch was used for CN the 624 cm-1 C-Cl stretch was used for HD and the 544 cm-1 PO2CS wag was used for VX It was determined that 1 mgmL samples of each of these chemicals measured 36 times in glass vials coated with a silver-doped sol-gel reproduced measurements with standard deviations of 125 129 and 293 It was further found that the 10 chemicals added to simulate generic tap water did not interfere with or alter the SER spectra It should be noted that the concentrations used in this study were considerably greater than those required by the JSAWM program Current work involves improving SER sensitivity and designing sampling systems with better reproducibility This includes the development of fractal silver and gold structures within the sol-gel matrix and the development of chemically selective sol-gels
ACKNOWLEDGMENTS The authors are grateful for the support of the US Army (DAAD13-02-C-0015 Joint Service Agent Water Monitor program) The authors would also like to thank Janet Jensen Ronald Crosier and Kristina Gonser for helpful discussions
Table 4 Measured SER peak heights for the VX band at 544 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
Table 3 Measured SER peak heights for the HD band at 624 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
SPIE-2003-5269 22
REFERENCES 1 Jensen JL NC Wong W R Loerop SPIE 4775-03 (2001) 2 JSAWM Requirements at wwwsbccomapgeaarmymilRDAecbcrtPRODSERJSAWMjsawmhtml 3 Erickson B ldquoThe Chemical Weapons Convention Redefines Analytical Challengerdquo Anal Chem 70 397A (1998) 4 Johnston RL Hoefler CM Fargo JC and Moberley B ldquoThe Defense Nuclear Agencyrsquos
ChemicalBiochemical Weapons Agreements Verification Technology Research Development Test and Evaluation Program and its Requirements for On-Site Analysisrdquo AT-ONSITE 5-8 (1994)
5 Black RM Clarke RJ Read RW and Reid MT ldquoApplication of GC-MS and GC-tandem MS to the analysis of chemical warfare samples found to contain residues of the nerve agent sarin sulphur mustard and their degradation productsrdquo J Chrom 662 301-321 (1994)
6 USA Army Center for Health Promotion and Preventive Medicine (USACHPPM) Technical Guideline 230A 7 Hoffland LD Piffath RJ Bouck JB rdquoSpectral signatures of chemical agents and simulantsrdquo Optical
Engineering 24 982-984 (1985) 8 Braue EHJ Pannella MGrdquoCIRCLE CELL FT-IR Analysis of Chemical Warfare Agents in Aqueous Solutionsrdquo
App Spectrosc 44 1513-1520 (1990) 9 Tseng C-H Mann CK and Vickers TJ ldquoDetermination of Organics on Metal Surfaces by Raman
Spectroscopyrdquo App Spectrosc 47 1767-1771 (1993) 10 Staff Reporter ldquoUS Army Tests Lidar to Detect Biological Toxinsrdquo Photonic Spectra pg 50 December 1998 11 Christesen SD Raman cross sections of chemical agents and simulants App Spectrosc 42 318-321 (1988) 12 Weaver MJ Farquharson S Tadayyoni MA ldquoSurface-enhancement factors for Raman scattering at silver electrodesrdquo J Chem Phys 82 4867-4874 (1985) 13 Alak AM and Vo-Dinh T ldquoSERS of Organophosphorous Chemical Agentsrdquo Anal Chem 59 2149-2153 (1987) 14 Lee Y Farquharson S ldquoRapid chemical agent identification by SERSrdquo SPIE 4378 21-26 (2001) 15 Farquharson S Maksymiuk P Ong K Christesen S ldquoChemical agent identification by surface-enhanced Raman spectroscopyrdquo SPIE 4577 166-173 (2001) 16 Spencer KM Sylvia J Clauson S and Janni J ldquoSurface Enhanced Raman as a Water Monitor for Warfare
Agents in Waterrdquo SPIE 4577 158-165 (2001) 17 Tessier P Christesen S Ong K Clemente E Lenhoff A Kaler E Velev O ldquoOn-line spectroscopic characterization of sodium cyanide with nanostructured gold surface-enhanced Raman spectroscopy substratesrdquo
App Spectrosc 56 1524-1530 (2002) 18 Farquharson S W W Smith S Elliott and J F Sperry Rapid biological agent identification by surface- enhanced Raman spectroscopy SPIE 3855110-116 (1999) 19 Farquharson S WW Smith YH Lee S Elliott and J F Sperry Detection of biological signatures A
comparison of electrolytic and metal-doped sol-gel surface-enhanced Raman media SPIE 4575 62-72 (2002) 20 Guzelian AA Sylvia J Janni J Clauson S and Spencer KM ldquoSERS of whole cell bacteria and trace levels of biological moleculesrdquo SPIE 4577 182-192 (2001) 21 Shende C Inscore F Gift A Maksymiuk P Farquharson S ldquoRapid dipicolinic acid extraction from Bacillus spores detected by surface-enhanced Raman spectroscopyrdquo App Spectrosc 58 accepted 22 Lee Y and Farquharson S ldquoSERS Sample Vials Based on Sol-Gel Process for Trace Pesticide Analysisrdquo SPIE
4206 140-146 (2000) 23 Farquharson S and Lee Y ldquoTrace Drug Analysis by SERSrdquo SPIE 4200-16 (2000) 24 Lee Y Farquharson S and Rainey P M Surface-Enhanced Raman Sensor for Trace Chemical Detection in
Water SPIE 3857 76-84 (1999) 25 Lee Y Farquharson S Kwong H and Shahriari M ldquoSol-Gel Chemical Sensor for Surface-Enhanced Raman
Spectroscopyrdquo SPIE 3537 252-260 (1998) 26 Farquharson S A Gift P Maksymiuk F Inscore and W Smith ldquopH dependence of methyl phosphonic acid
dipicolinic acid and cyanide by surface-enhanced Raman spectroscopyrdquo SPIE 5269-19 (2003) 27 Lide DR Ed Handbook of Chemistry and Physics CRC Press 77th Ed (1996-7) 28 Sosa C RJ Bartlett K KuBulat and WB Person ldquoA theoretical study of harmonic vibrational frequencies and
infrared intensities of XCH2CH2SCH2CH2X and XCH2CH2SH (= H Cl)rdquo J Phys Chem 93 577-588 (1993) 29 Hameka HF and JO Jenson ldquoComputer-generated predictions of the structure and the IR and Raman spectra of
VXrdquo ERDEC-TR-065 May 1993
- Chemagents Appendicespdf
-
- SPIE2001-4575-Bioagent-SERSapdf
-
- Detection of bioagent signatures A comparison of electrolytic and metal-doped sol-gel surface-enhanced Raman media
- Stuart Farquharson Wayne Smith and Yuan Lee
-
- Real-Time Analyzers 87 Church Street East Hartford CT 06108
-
- ABSTRACT
- 4 CONCLUSIONS
- 5 ACKNOWLEDGEMENTS
- 6 REFERENCES
-
172
4 CONCLUSIONS Here we present for the first time surface-enhanced Raman spectra of Tabun and Sarin obtained using silver-doped sol-gels However the inferior enhancement suggested employing SER-active sol-gels varying in composition to improve both detection limits and selectivity of the target analytes To this end we successfully demonstrated the capabilities of four sol-gels that select for 1) polar-positive 2) polar-negative 3) weakly polar-positive and 4) weakly polar-negative chemical species p-aminobenzoic acid was used to show that silver could be used to attract polar-negative chemicals or functional groups (carboxylate anion) and that gold could be used to attract polar-positive chemicals (amine) while a mixture of p-aminobenzoic acid and phenylacetylene was used to show that tetramethyl orthosilicate preferentially solvates polar chemicals and that tetraethyl orthosilicate preferentially solvates weakly polar chemicals This increased sample control was applied to cyanide and methyl phosphonic acid two hydrolysis products of chemical warfare agents Exceptional results were obtained for methyl phosphonic acid allowing measurement of 1x10-2 gL for a 3-min scan and 75 mW of incident laser power with an estimated detection limit of 245x10-4 gL and an enhancement factor of 87x106 However this detection limit is 76 times less sensitive than required for the JSAWM (32x10-6gL for the G-agents) Finally we note that the measurements performed here employed an FT-Raman spectrometer This instrumentation was chosen over dispersive Raman instrumentation for the high wavelength accuracy afforded by the HeNe reference laser (Connes Advantage25) which would allow reliable spectral subtraction matching of observed spectra to stored library spectra and confident use of chemometric approaches Such data analysis is likely to be required to identify the chemical agents as well as distinguish hydrolysis products However this instrumentation which employs 1064 nm excitation and InGaAs detection sacrifices sensitivity We believe that the measurement sensitivity can be improved by at least two orders of magnitude by using shorter laser excitation wavelengths (eg 532 nm) This would provide more efficient Raman scattering (fourth power dependence on laser excitation wavelength) more efficient generation of plasmon modes and allow using more efficient detector material (Si vs InGaAs) These modifications are underway
5 ACKNOWLEDGEMENTS The authors would like to thank Dr R Yin and J Jensen of the US Army for supporting this work (Contract Number DAAD13-01-C-0019) They would also like to thank Advanced Fuel Research for making their laboratory facilities available
6 REFERENCES 1 Tu Anthony ldquoOverview of Sarin Terrorist Incidents in Japan in 1994 and 1995rdquo 6th CBW Protection Symposium
Stockholm Sweden 10-15 May 1998 2 Jensen JL NC Wong W R Loerop SPIE 4775-03 (2001) 3 JSAWM Requirements at wwwsbccomapgeaarmymilRDAecbcrtPRODSERJSAWMjsawmhtml 4 ldquoThe Chemical Weapons Convention Redefines Analytical Challengerdquo Analytical Chemistry News amp Features June 1
397A (1998) 5 Johnston RL Hoefler CM Fargo JC and Moberley B ldquoThe Defense Nuclear Agencyrsquos ChemicalBiochemical
Weapons Agreements Verification Technology Research Development Test and Evaluation Program and its Requirements for On-Site Analysisrdquo AT-ONSITE 5-8 (1994)
6 Black RM Clarke RJ Read RW and Reid MT ldquoApplication of gas chromatography-mass spectrometry and gas chromatography-tandem mass spectrometry to the analysis of chemical warfare samples found to contain residues of the nerve agent sarin sulphur mustard and their degradation productsrdquo J Chromatography 662 301-321 (1994)
7 Hoffland LD Piffath RJ Bouck JBrdquoSpectral signatures of chemical agents and simulantsrdquo Optical Engineering 24 982-984 (1985)
8 Braue EHJ Pannella MGrdquoCIRCLE CELL FT-IR Analysis of Chemical Warfare Agents in Aqueous Solutionsrdquo Applied Spectroscopy 44 1513-1520 (1990)
9 Tseng C-H Mann CK and Vickers TJ ldquoDetermination of Organics on Metal Surfaces by Raman Spectroscopyrdquo Applied Spectroscopy 47 1767-1771 (1993)
10 Staff Reporter ldquoUS Army Tests Lidar to Detect Biological Toxinsrdquo Photonic Spectra pg 50 December 1998
Proc SPIE Vol 4577
Proc SPIE Vol 4577
173
11 Christesen SD Raman cross sections of chemical agents and simulants Applied Spectroscopy 42 318-321 (1988) 12 Alak AM and Vo-Dinh T ldquoSERS of Organophosphorous Chemical Agentsrdquo Analytical Chemistry 59 2149-2153
(1987) 13 Norrod KL Sudnik LM Rousell D and Rowlen KL ldquoQuantitative Comparison of Five SERS Substrates
Sensitivity and Detection Limitrdquo Applied Spectroscopy 51 994-1001 (1997) 14 Lee Y and Farquharson S ldquoSERS Sample Vials Based on Sol-Gel Process for Trace Pesticide Analysisrdquo SPIE 4206
140-146 (2000) 15 Farquharson S and Lee Y ldquoTrace Drug Analysis by Surface-Enhanced Raman Spectroscopyrdquo SPIE 4200-16 (2000) 16 Lee Y Farquharson S and Rainey P M Surface-Enhanced Raman Sensor for Trace Chemical Detection in Water
SPIE 3857 76-84 (1999) 17 Lee Y Farquharson S Kwong H and Shahriari M ldquoSurface-Enhanced Raman Sensor for Surface-Enhanced Raman
Spectroscopyrdquo SPIE 3537 252-260 (1998) 18 Farquharson S Smith W Carangelo R C and Brouillette C ldquoIndustrial Raman Providing Easy Immediate Cost
Effective Chemical Analysis Anywhererdquo SPIE 3859 14-23 (1999) 19 Narayanan VA JM Bello JD Stokes and T Vo-Dinh Analusis 19 307-310 (1991) 20 Laserna JJ E L Torres and JD Winefordner Analytica Chemica Acta 469-480 (1987) 21 Torres EL and JD Winefordner Analytical Chemistry 59 1626-1632 (1987) 22 Suh JS DP DiLella M Moskovits J Phys Chem 87 1540-1544 (1983) 23 Benner RE R Dornhaus R Chang and BL Laube Correlations in the Raman spectra of cyanide complexes adsorbed
at silver electrodes with voltammograms Surface Science 101 341 (1980) 24 Chase D B and JF Rabolt Fourier Transform Raman Spectroscopy Acad Press Ch1 p 131 (1994) 25 Connes J Rev Opt Theor Instrum 40 45 (1961)
APPLIED SPECTROSCOPY 351
focusing the 488 nm laser beam 2 mm deep into thebulk of the crystals to avoid contributions from potentialdepletion layers As shown in Fig 1b the dependence ofthe Raman shift on the degree of deuteration is almostperfectly linear and ts very well with D 5 22684cmR1 24526 where D is the degree of deuteration (in )and R is the spectral mean of the PO4 vibration in cm21A linear correlation coef cient of 0998 indicates an ex-cellent linear dependence of the Raman peak shift withdegree of deuteration This result shows that the shift ofthe PO 4 peak is simply caused by the linear increase inatomic mass due to isotope substitution which decreasesthe length of hydrogen-like bonds
This excellent linear dependence allows us to map thepro le of the DH exchange layer at the surface of DKDPcrystals by acquiring Raman spectra and determining theposition of the PO 4 peak for various depths This methodis preferable over other methods such as determining thestrength of the OD vibration directly (eg at 715 cm21)because the position of the most intense peak in the Ra-man spectrum can be measured more precisely than theintensity of some of the weakest peaks in the spectrumThis is demonstrated in Fig 2 where depth-dependentRaman spectra (Fig 2a) and the resulting exchange layerpro les for two DKDP crystals are shown (Fig 2b) Thespectra in Fig 2a were obtained from a depth scan of aDKDP crystal with 75 degree of deuteration in thebulk grown at 45 8C The spectra start out as DKDP with30 deuteration close to the surface and approach thebulk DKDP spectrum within a few micrometers of depthThe fact that the relative degree of deuteration does notextend to 0 D is due to the limited depth resolution ofthe Raman microprobe which averages over 4 mm indepth Figure 2b depicts the resulting DH exchange layerpro les for this and a second crystal grown at 63 8Crespectively Both crystals had the same exposure to am-bient conditions and their main difference is the temper-ature at which they were grown The different exchangelayer pro les indicate that crystals grown at differenttemperatures have differen t proton conductiv ities 13
which leads to a difference in their rate of deuteriumdepletion The parameters controlling this behavior arecurrently the objective of a detailed study the results ofwhich will be reported elsewhere
CONCLUSION
In conclusion we have shown that the shift of the to-tally symmetric PO 4 stretch mode in the Raman spectrumof DKDP crystals scales linearly with degree of deuter-ation This allows us to correlate Raman peak positionsto deuteration levels in these crystals We have presenteda new technique to determine DH diffusion pro les inDKDP frequency conversion crystals based on micro-Ra-man spectroscopy This technique is fast inexpensiveand works under various environmental conditionswhich will allow us to better understand and control deu-terium depletion in DKDP crystals
ACKNOWLEDGMENTS
We would like to thank M Runkel for rst discovering DKDP crack-ing R Floyd for providing DKDP crystals and L Chase and A Burn-ham for their support and helpful discussions This work was performedunder the auspices of the US Department of Energy by the University
of California Lawrence Livermore National Laboratory under contract W-7405-Eng-48
1 J J De Yoreo A K Burnham and P K Whitman Int Mat Rev47 113 (2002)
2 C E Barker R A Sacks B M Van Wonterghern J A Caird JR Murray J H Campbell K Kyle R B Ehrlich and N DNielsen Proc SPIE-Int Soc Opt Eng 2633 501 (1995)
3 T Suratwala paper to be published4 Cleveland Crystals Inc httpwwwclevelandcrystalscom
KDPshtmltable5 E A Popova I T Savatinova and I A Velichko Sov Phys Solid
State 12 1543 (1971)6 I P Kaminow R C C Leite and S P S Porto J Phys Chem
Solids 26 2085 (1965)7 J A Subramony B J Marquardt J W Macklin and B Kahr
Chem Mat 11 1312 (1999)8 H Tanaka M Tokunaga and I Tatsuzaki Solid State Commun
49 153 (1984)9 R J Nelmes G M Meyer and J E Tibballs J Phys C 15 59
(1982)10 M A Yakshin D W Kim Y S Kim Y Y Broslavets O E
Sidoryuk and S Goldstein Laser Physics 7 941 (1997)11 I Takenaga Y Tominaga S Endo and M Kobayashi Solid State
Commun 84 931 (1992)12 C Krenn personal communication13 M Sharon and A K Kalia J Solid State Chem 21 171 (1977)
Rapid Dipicolinic Acid Extractionfrom Bacillus Spores Detectedby Surface-EnhancedRaman Spectroscopy
STUART FARQUHARSON ALAND GIFT PAUL MAKSYMIUK andFRANK E INSCOREReal-Time Analyzers Inc East Hartford Connecticut06108
Index Headings Dipicolinic acid Bacillus spores Anthrax Surface-enhanced Raman spectroscopy
INTRODUCTION
The anxiety caused by the distribution of anthrax en-dospores through the US postal system in October 2001was exacerbated by the long time required for positiveidenti cation of the Bacillus anthracis spores and the un-known extent of their distribution Since that time manymethods capable of rapid eld analysis have been inves-tigated to augment or replace the laboratory method ofgrowing microorganisms in culture media which takesdays to perform12 Prominent among these approachesare polymerase chain reactions (PCR)3 immunoassaysand detection of calcium dipicolinate as a biochemicalsignature PCR employs primers to separate organism-
Received 10 October 2003 accepted 14 November 2003 Author to whom correspondence should be sent
352 Volume 58 Number 3 2004
speci c nucleic acid sequences (eg capsular protein en-coding gene for Bacillus anthracis)4 and polymerases toamplify the segment until it is detectable Recently am-pli cation times have been substantially reduced andcomplete analysis can now be performed in an hour orless Immunoassay methods are also being developed thatuse competitive binding of the bioagent (as an antigen)and its labeled conjugate for a limited number of anti-bodies Although analyses can be performed in under 30minutes a well-de ned anthrax antigen has not yet beenidenti ed5ndash7 and consequently the false-positive rate isunacceptably high8
A number of other methods are being developed witha focus on the detection of calcium dipicolinate (CaDPA)and its derivatives as a B anthracis signature This is sobecause only spore-forming bacteria contain CaDPA andthe most common potentially interfering spores such aspollen and mold spores do not Relatively fast methodshave been developed to chemically extract CaDPA andthen detect it directly by uorescence9 or indirectly byluminescence1011 In the latter case hot dodecylamine(DDA) has been used to extract dipicolinic acid (DPA)and terbium has been utilized to form a highly lumines-cent DPA complex11 Although measurements have beenperformed in as little as ve minutes it was found thatas many as three concentration-dependent complexes canform each with different lifetimes This coupled withthe fact that the Tb31 cation produces the same lumines-cence spectrum makes determinations of low spore con-centrations problematic
It has been long known that Raman spectra of Bacillispores are dominated by bands associated with CaDPA12
and that these spectra may provide a suitable anthrax sig-nature at the genus level13 Since that time considerableimprovements in Raman instrumentation have led to lab-oratory measurements of single Bacilli spores14 and to eld measurements of spores captured from a mail-sort-ing system15 However the single spore measurementsrequired complex instrumentation that is not ruggedwhile the eld measurements required milligrams of sam-ple Furthermore the Raman spectra of both measure-ments contained uorescence contributions that would in-crease uncertainty in quanti cation
In related research we demonstrated that nanogramquantities of DPA could be detected by uorescence-freesurface-enhanced Raman spectroscopy (SERS)16 We alsodemonstrated that microliter volumes of chemicals canbe detected by SERS using metal-doped sol-gel-packedglass capillaries17 Towards the goal of developing a rap-id eld SERS-based anthrax spore detector we havecombined our previous research and we now report thatDPA can be extracted from a 10 mg B cereus spore sam-ple using DDA in 1 minute and can be detected by SERSin an additional 1 minute
EXPERIMENTAL
Dipicolinic acid (26-pyridinedicarboxylic acid DPA)and dodecylamine (DDA) were used as received fromSigma-Aldrich (Milwaukee WI) Lyophilized B cereusspores prepared according to the literature13 were sup-plied by the University of Rhode Island and used as re-ceived Multiple particles approximately 01 mm3 each
were separated and weighed at 5 to 15 mg representing05 to 15 million spores The sample masses were con-sistent with a previous determination of spore density at0081 gmL that indicated a high degree of entrained air
All chemicals used to prepare the silver-doped sol-gelcoated capillaries were also obtained and used as receivedfrom Sigma-Aldrich According to previously publishedprocedures17 two precursor solutions were preparedmixed and then drawn into 1-mm-diameter glass capil-laries The silver amine precursor consisted of a 51 vvratio of 1 N AgNO 3 to 28 NH3OH while the alkoxideprecursor consisted of a 21 vv ratio of methanol to te-tramethyl orthosilicate The alkoxide precursors weremixed with silver amine precursor in an 81 vv ratioApproximately 015 mL was drawn into the capillarycoating a 15-mm length After sol-gel formation the in-corporated silver ions were reduced with dilute sodiumborohydride which was followed by a water wash to re-move residual reducing agent
A 100 mL drop of a 50 mM DDA solution in ethanolpre-heated to 78 8C was added to each of the B cereusparticles to digest the spore coat After 1 minute the re-sultant solution was drawn into a SER-active capillarythat was immediately xed horizontally to an XY posi-tioning stage (Conix Research Spring eld OR) just in-side the focal point of an f 07 aspheric lens The lensfocused the beam into the sample and collected the scat-tered radiation back along the same axis A dichroic lter(Omega Optical Brattleborough VT) was used to re ectthe excitation laser to the lens and pass the Raman scat-tered radiation collected by the lens An f 2 achromatwas used to collimate the laser beam exiting a 200-mm-core-diameter source ber optic while a second f 2 ach-romat was used to focus the scattered radiation into a 365mm ber optic (Spectran Avon CT) A short-pass lterwas placed in the excitation beam path to block the sil-icon Raman scattering generated in the source ber fromre ecting off sampling optics and reaching the detectorA long-pass lter was placed in the collection beam pathto block the sample Rayleigh scattering from reachingthe detector A 785 nm diode laser (Process InstrumentsInc model 785-600 Salt Lake City UT) was used todeliver 100 to 150 mW of power to the sample A Fouriertransform Raman spectrometer (Real-Time Analyzersmodel IRA-785 East Hartford CT) and a silicon photo-avalanche detector (Perkin Elmer model C30902S Stam-ford CT) were used to acquire the SER spectra
RESULTS AND DISCUSSION
As an initial experiment the SER spectrum of 1 gLof DPA in water was measured using the newly devel-oped silver-doped sol-gel-coated capillaries (Fig 1A) Atthis concentration a high signal-to-noise ratio (SN) isobtained in 1 min In fact a reasonable spectrum is ob-tained in the same time frame for 1 mgL (Fig 1B) TheSER spectra are reasonably similar to the normal Raman(NR) spectrum obtained for a saturated solution of DPAin 1 N KOH (Fig 1C) and the following band shifts areobserved (NR to SER) 647 to 657 cm21 817 to 815cm21 998 to 1008 cm21 1384 to 1382 cm21 1434 to1428 cm21 and 1569 to 1567 cm21 Many of these bandshave been previously assigned1213 such as 998 cm21 to
APPLIED SPECTROSCOPY 353
FIG 1 SERS of DPA in water using silver-doped sol-gel-coated glasscapillary for (A) 1 gL and (B) 1 mgL (C ) NR of saturated DPA in 1N KOH in a glass capillary Spectral conditions (A) and (B) 150 mWof 785 nm 1-min acquisition time (C) 450 mW of 785 nm 5-minacquisition time both 8 cm 21 resolution
FIG 2 SERS of DPA extracted from 10 mg B cereus particle using100 mL of 50 mM hot DDA acquired in (A) 1 minute and (B) 2 seconds(C ) Attempted SERS of 50 mM hot DDA in ethanol using silver-dopedsol-gel-coated glass capillary acquired in 1 min Spectral conditions150 mW of 785 nm 8 cm21 resolution
the symmetric ring stretch 1384 cm21 to the OndashCndashOsymmetric stretch 1428 cm21 to the symmetric ring CndashH bend and 1569 cm21 to the asymmetric OndashCndashOstretch
The rst B cereus samples consisted of 2 mg of sporesin 2 mL of 5 mM hot DDA The samples were main-tained at 78 8C for 40 min and while hot approximately10 mL was drawn into a SER-active capillary Since spec-tra of DPA were obtained for these initial samples small-er spore masses higher DDA concentrations and shorterheating periods were examined In due course it wasfound that 10 mg of spores could be digested by 100 mLof 50 mM hot DDA in one minute and detected (Fig2A) In fact the signal was suf ciently intense that it canbe observed in as little as two seconds (Fig 2B) Theamount of DPA that was extracted was estimated to bebetween 5 and 10 mgL by comparing the signal intensityof the 1008 cm21 band to that measured for DPA in waterThis is consistent with previous research that found thatthe majority of the DPA is extracted from spores usingDDA11 and that B cereus spores contain approximately10 DPA by weight18 The SN of 127 for the 1008 cm21
band in the 1-minute SER spectrum suggests a limit ofdetection of approximately 250 ng of B cereus sporesbased on a SN of 3 Finally it should be noted that DDAdid not produce a detectable SER spectrum as shown inFig 2C
CONCLUSION
We have demonstrated that by combining rapid extrac-tion of dipicolinic acid from Bacillus cereus spores withchemical identi cation by surface-enhanced Raman spec-troscopy as little as 10 mg of spores can be detected Infact the entire measurement from the time of adding hotdodecylamine to the spores to the time when the dipi-colinic acid SER spectrum is acquired and analyzedcould be performed in less than two minutes The abilityof this method to distinguish between spore-forming bac-teria such as Bacillus anthracis and non-DPA containingpowders could help prevent costly shutdowns associated
with the appearance of suspicious material or intentionalmailing of common substances as an anthrax hoax Thismethod could also prove useful in detecting the locationof anthrax endospores in mail distribution facilities if an-other veri ed attack should occur
Research continues to fully characterize the surface-enhanced Raman spectroscopy signal intensities as afunction of sample concentration and to explore otherextractants that do not require the use of elevated tem-perature
ACKNOWLEDGMENTS
The authors are grateful for the support of the National Science Foun-dation (DMI-0296116 and DMI-0215819) and the US Army(DAAD13-02-C-0015 Joint Service Agent Water Monitor program)The authors are indebted to Chetan Shende for preparing the sol-gelcapillaries The authors also thank James Gillespie Nicholas Fell andAugustus Fountain for providing important background informationMark Farquharson for laboratory support and Professor Jay Sperry ofthe University of Rhode Island for supplying B cereus spores
1 V A Pasechnik C C Shone and P Hambleton Bioseparations 3267 (1993)
2 P J Jackson M E Hugh-Jones D M Adair G Green K K HillC R Kuske L M Grinberg F A Abramova and P Keim ProcNatl Acad Sci USA 95 1224 (1998)
3 B R Glick and J J Pasternak Molecular Biology Principles andApplications of Recombinant DNA (ASM Press Washington DC1994)
4 C A Bell J R Uhl T L Had eld J C David R F Meyer TF Smith and F R Cockerill III J Clin Microbiol 40 2897(2002)
5 D L Gatto-Menking H Yu J G Bruno M T Goode M Millerand A W Zulich Biosens Bioelectron 10 501 (1995)
6 J J Quinlan and P M Foegeding J Rapid Methods AutomationMicrobiol 6 1 (1998)
7 A A Hindle and E A H Hall Analyst (Cambridge UK) 1241599 (1999)
8 M S Ascher US Department of Health amp Human Services(httpwwwhhsgovophppresentationsAscherdoc)
9 R Nudelman B V Bronk and S Efrima Appl Spectrosc 54 445(2000)
10 D L Rosen C Sharpless and L B McBrown Anal Chem 691082 (1997)
354 Volume 58 Number 3 2004
11 P M Pellegrino N F Fell Jr and J B Gillespie Anal ChimActa 455 167 (2002)
12 W H Woodruff T G Spiro and C Gilvarg Biochem BiophysRes Commun 58 197 (1974)
13 E Ghiamati R S Manoharan W H Nelson and J F SperryAppl Spectrosc 46 357 (1992)
14 A P Esposito C E Talley T Huser C W Hollars C M Schal-dach and S M Lane Appl Spectrosc 57 868 (2003)
15 S Farquharson L Grigely V Khitrov W W Smith J F Sperryand G Fenerty J Raman Spectrosc paper accep ted (2003)
16 S Farquharson W W Smith S Elliott and J F Sperry SPIE-IntSoc Opt Eng 3855 110 (1999)
17 S Farquharson and P Maksymiuk Appl Spectrosc 57 479(2003)
18 F W Janssen A J Lund and L E Anderson Science (Washing-ton DC) 127 26 (1958)
SPIE -2003-5269 117
pH dependence of methyl phosphonic acid dipicolinic acid and cyanide by surface-enhanced Raman spectroscopy
Stuart Farquharson Alan Gift Paul Maksymiuk Frank Inscore and Wayne Smith
Real-Time Analyzers 87 Church Street East Hartford CT 06108
ABSTRACT US and Coalition forces fighting terrorism in Afghanistan and Iraq must consider a wide range of attack scenarios in addition to car bombings Among these is the intentional poisoning of water supplies to obstruct military operations To counter such attacks the military is developing portable analyzers that can identify and quantify potential chemical agents in water supplies at microgram per liter concentrations within 10 minutes To aid this effort we have been investigating the value of a surface-enhanced Raman spectroscopy based portable analyzer In particular we have been developing silver-doped sol-gels to generate SER spectra of chemical agents and their hydrolysis products Here we present SER spectra of methyl phosphonic acid and cyanide as a function of pH an important factor affecting quantitation measurements which to our knowledge has not been examined In addition dipicolinic acid a chemical signature associated with anthrax-causing spores is also presented Keywords Chemical warfare agents agent detection agent hydrolysis SERS Raman spectroscopy homeland security
1 INTRODUCTION In the past decade the Unites States and its allies have been challenged by a different kind of warfare exemplified by the terrorist attacks of September 11 2001 Suicide bombings and the use of chemical agents are the norm and military personnel must consider a wide range of attack scenarios Among these is the intentional poisoning of water supplies to obstruct military operations in Afghanistan and Iraq Currently colorimetric paper is used to detect agents on-site while gas chromatography combined with mass spectrometry (GCMS) is used in mobile support laboratories However both methods have severe drawbacks The paper changes color in response to contact with many chemicals besides CWAs causing a high incidence of false positives1 while GCMS although very chemically specific requires hours to perform and constant re-calibration234 Military operations would be greatly aided by a portable analyzer that can identify and quantify potential chemical agents at concentrations that impact safety This includes the analysis of drinking water supplies distribution and storage systems To meet this goal the Department of Defense has been investigating numerous approaches under the auspices of the Joint Service Agent Water Monitor (JSAWM) program5 The JSAWM must be capable of identifying the chemical agents with no false-positives and quantifying the agents at microgram per liter concentrations within 10 minutes678 Recently we and others have been investigating the ability of surface-enhanced Raman spectroscopy (SERS) to measure chemical agents9-12 bioagents13-17 and their hydrolysis products in water SERS employs the interaction of surface plasmon modes of metal particles with target analytes to increase scattering efficiency by as much as 1 million times18 In 1987 the potential of this approach to measure CWAs was demonstrated by measuring a series of pesticides19 Several of these organophosphonates have chemical structures similar to CWAs in particular P=O functional groups In our studies we have been employing silver-doped sol-gels to promote the SER effect The porous silica network of the sol-gel matrix offers a unique environment for immobilizing and stabilizing SER-active metal particles20-23 The sol-gels have been coated on the internal walls of standard glass vials to allow rapid SER analysis of small sample quantities without preparation We have measured over 100 chemicals with enhancements of 104 to 106 demonstrated reversible measurements in a flowing system reproducible measurements from vial-to-vial and measurements in multiple solvents including water20-23 Previously we used these vials to perform preliminary measurements of cyanide (CN) methylphosphonic acid (MPA) and dipicolinic acid (DPA) MPA is a hydrolysis product of the nerve agents (eg sarin Reaction 1) and may be a valuable indicator of nerve agent usage particularly since the alkyl methylphosphonic acids are relatively more stable than their corresponding parent complexes24 DPA is
SPIE -2003-5269 118
a chemical signature of spore forming bacteria such as Bacillus anthracis And in light of the inability to rapidly detect the anthrax spores distributed through the US mail in October 2001 a number of methods are being developed to extract and analyze this signature Reaction 1 Stepwise hydrolysis of Sarin to form hydrofluoric acid (HF) isopropyl methylphosphonic acid (IMPA) then methyl phosphonic acid (MPA) and 2-propanol In our previous SERS investigations MPA and DPA were measured at 50 and 100 mgL respectively In both cases limits of detection (LOD) were estimated at 100 microgL providing encouragement in that SERS may satisfy the needs of the JSAWM Since it has been shown that pH can substantially influence the intensity of SER bands25 which would clearly influence quantitative analysis we undertook the present study to determine the severity of these effects for cyanide methyl phosphonic acid and dipicolinic acid Furthermore we previously observed a band at 1050 cm-1 for MPA6 possibly due to an anion formed at basic pH Here we investigate the source of this spectral anomaly
2 EXPERIMENTAL All chemicals including potassium cyanide methyl phosphonic acid dipicolinic acid and those used to prepare the silver-doped and gold-doped sol-gel coated vials were obtained and used as received from Sigma-Aldrich All samples were prepared in HPLC grade water (Fischer Scientific Fair Lawn NJ) for SERS measurements The pH of these samples was adjusted using dilute nitric acid or potassium hydroxide and verified using a pH electrode (Corning Inc Corning NY) that had been calibrated with pH 400 700 and 1000 buffered standards from Fischer Scientific Once prepared the samples were transferred into the silver-doped sol-gel vials (Real-Time Analyzers Inc East Hartford CT) The vials were coated in a manner similar to that previously reported by adding ammonium hydroxide to a solution of silver nitrate tetramethyl orthosilicate (TMOS) and methanol20 Gold-vials were coated by adding nitric acid to a solution of gold tetrachloride TMOS and methanol The two precursor solutions were prepared mixed and transferred to 2-mL glass vials dried and heated After sol-gel formation the incorporated metal ions were reduced with dilute sodium borohydride (1mgmL) which was followed by a water wash to remove residual reducing agent After the resultant analyte solution was introduced the SER-active vial was immediately fixed horizontally to an XY positioning stage (Conix Research Springfield OR) just inside the focal point of an f07 aspheric lens The lens focused the beam into the sample and collected the scattered radiation back along the same axis A dichroic filter (Omega Optical Brattleborough VT) was used to reflect the excitation laser to the lens and pass the Raman scattered radiation collected by the lens An f2 achromat was used to collimate the laser beam exiting a 200 microm core diameter source fiber optic while a second f2 achromat was used to focus the scattered radiation into a 365 microm fiber optic (Spectran Avon CT) A short pass filter was placed in the excitation beam path to block the silicon Raman scattering generated in the source fiber from reflecting off sampling optics and reaching the detector A long pass filter was placed in the collection beam path to block the sample Rayleigh scattering from reaching the detector A 785 nm diode laser (Process Instruments Inc model 785-600 Salt Lake City UT) was used to deliver 100 to 150 mW of power to the sample A Fourier transform Raman spectrometer (Real-Time Analyzers model IRA-785 East Hartford CT) and a silicon photo-avalanche detector (Perkin Elmer model C30902S Stamford CT) were used to acquire the SER spectra
3 RESULTS AND DISCUSSION In a previous study of MPA6 aimed at developing a concentration calibration curve and determining limits of detection (LOD) we observed an anomaly at 1050 cm-1 Since it was found that the band intensity changed as a function of concentration the band must be associated with a sample parameter Two possible parameters photon flux and pH are examined here The first parameter was investigated by irradiating a 1mgmL MPA sample in a SER-active vial with laser powers of 200 mW and above and monitoring spectral changes It was immediately found that the 1050 cm-1 band
2O+ H HF + +OH
OH3H C
OP
OF
CH
CH3
3
3H C
OCP
OOH
CH
CH3
3
3H C
O
CPHO
CH
CH
3
3
C
Sarin IMPA MPA 2-propanol
SPIE -2003-5269 119
appeared and grew as a function of time and that the higher the laser power the faster the growth Figure 1 shows the growth of the 1050 cm-1 band over the course of 30 minutes when using 150 mW of 785 nm excitation while Figure 2A shows that the growth can be fit with a first order exponential equation namely I1050 = 03+05e-013t Furthermore the 760 cm-1 band could be fit with a first order decay equation with an identical exponential rate constant ie I760 = 08-
08e-013t The rates represent classical first order kinetics and their correspondence allows one to conclude that MPA is being transformed one-for-one into a photo-generated product At this time the photoproduct has not been positively identified but phosphonic (phosphorous) acid and phosphonate are likely candidates since the symmetric P(OH)3 stretch occurs at ~1050cm-1 Our studies show that a reduction in laser power to 100 mW at the sample for MPA essentially eliminates this degradation process This laser power or lower was used for further measurements
Methyl phosphonic acid is a diprotic acid that stepwise dissociates into two anions MPA- and MPA= according to the following reactions26
MPA MPA- + H+ pKa1 = 212 Reaction 2
MPA- MPA= + H+ pKa2 = 729 Reaction 3 The relative concentrations of MPA MPA- and MPA= can be determined at any pH by expressing [MPA] and [MPA=] in terms of [MPA-] using Reactions 2 and 3 and summing all three to equal the total starting concentration here 2 mgmL (0021M MW = 9602) viz
[MPA] + [MPA-] + [MPA=] = 0021M Equation 1 substituting from Reactions 1 and 2
([H+][MPA-])K1a + [MPA-] + (K2a[MPA-])[H+] = 0021M Equation 2 rearranging [MPA-] = 0021M(1+[H+]K1a + K2a[H+]) Equation 3 The relative concentrations of MPA MPA- and MPA= as a function of pH are shown in Figure 3 It is worth noting that near neutral pH both MPA- and MPA= will be present To confirm that the SER signal followed this pH dependence a starting solution consisting of 20 mg of MPA in 10 mL HPLC grade water was prepared and brought to pH of 20 using dilute nitric acid From this solution 2 mL were added to a SER-active vial and the SER spectrum recorded At this pH a peak at 760 cm-1 was barely discernable The 2 mL solution was returned to the starting solution and the pH was re-measured to correct for any changes that the silver-doped sol-gel vials might cause In most cases the change was less than 02 pH units and the pH is reported as the before and after average Next the pH of the
Figure 2 A) Exponential growth of 1050 cm-1 band and B) exponential decay of 760 cm-1 band for spectral series in Figure 1
Figure 1 Growth of 1050 cm-1 band as a function of time due to exposure to 150 mW of 785 nm Spectra are 5 sec each collected every 100-sec from 0 to 30-min
B
570 770 970 1170 1370Raman Shift (cm-1)
Arbit
rary
Unit
s
MPA Photodegradation
600 800 1000 1200 1400 Wavenumber (cm-1)
A
0 10 20 30 0 10 20 30 time (min) time (min)
Ram
an In
tens
ity (r
elat
ive)
30
min
0
SPIE -2003-5269 120
starting solution was adjusted to 325 using dilute KOH Again 2 mL were added to a vial and the SER spectrum recorded At this pH a reasonably strong 760 cm-1 band was observed This process was repeated as spectra were recorded at pHs of 70 74 75 79 85 and 100 A total of 1 mL of KOH was added diluting the total concentration by 10 Next the pH of the starting solution was made acidic by adding dilute nitric acid dropwise This time spectra were recorded at pHs of 72 69 64 and 37 Figure 4 shows the SER spectra for representative pHs (spectra were left out to simplify the figure) while Figure 3 shows the 760 cm-1 peak intensities as a function of pH (The band intensities were adjusted to compensate for dilution effects caused by the addition of HNO3 and KOH then normalized to 0021 M for the most intense band observed at pH 37) It is clear from Figure 3 that the 760 cm-1 band follows the MPA- concentration as a function of pH and must be assigned to this anion No bands were observed that corresponded to MPA or MPA= The lack of an MPA SER spectrum may be due to the absence of an attraction between the neutral analyte and the electropositive silver surface The same reasoning suggests that a strong SER spectrum should be observed for MPA= but it is not and a satisfactory explanation has not been found
0000
0005
0010
0015
0020
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14pH
Con
cent
ratio
n [M
]
pK1 = 212 pK2 = 729
MPA- MPA=
MPA
Figure 4 SER spectra of 002M MPA as a function of pH Conditions 100 mW of 785 nm 36 scans (1 min) 8 cm-1 recorded 2 min after sample introduction pH 19 and 20 not apparent on this scale pH 69 and 74 near identical to 70 and 75 and not shown for clarity
00
02
04
06
08
10
12
14
16
18
0 5 10 15 20 25 30Measurement Number
Ram
an In
tens
ity (7
60 c
m-1
)
Figure 3 Concentration dependence of MPA MPA- and MPA= as a function of pH for a 002M sample Intensity of 760 cm-1
band from Figure 6 as a function of pH ( for increasing basic adjustment for increasing acidic adjustment error was measured at ~10 for pH 64)
Figure 5 SER spectra of 002M MPA at pH 64 measured around a vial at three heights (9 points per height) Conditions as in Figure 4 but 10-sec scans
Since these measurements involved the removal and replacement of the SER-active vial in the sample holder to remove and add sample variation in the intensity as a function of vial position was minimized by illuminating the exact same height along the vial wall But this does not account for variability of the SERS response of the sol-gel coating around the vial To analyze this effect a vial containing MPA at pH 64 was rotated at ~ 40o
intervals at the original height and 18rdquo above and below this value Figure 5 shows the intensity of the 760 cm-1
band for the 27 positions It was found that the average value was 137plusmn014 an RSD of 10 overall and 5 for each height An error bar is included in Figure 4 for the pH 64 measurement
SPIE -2003-5269 121
With the above analyses in mind a preliminary investigation of the SER spectral response for MPA (as MPA-) as a function of concentration was performed A single vial was used for these measurements beginning with 1 mgL followed by measurements of 10 100 and 1000 mgL In all cases the pH was ~7 and 3 positions around the vial were measured per concentration Since the 760 cm-1 band was not observed for concentrations of 1 or 10 mgL using 100 mW of 785 nm the laser power at the sample was raised to 200 mW beginning with the 10 mgL concentration Photo-degradation was largely avoided (and not observed) by exposing the sample for only 33 seconds per spectral acquisition Representative spectra for 10 100 and 1000 mgL are shown in Figure 6 while a plot of the 760 cm-1 band intensity as a function of concentration is shown in Figure 7 These values were also used to estimate limits of detection based on the signal-to-noise ratio (SN) of the 760 cm-1 band using 100 mW of 785 nm laser excitation and a 10-min acquisition time As summarized in the Figure 7 inset the lower the measured concentration the lower the predicted LOD Even if as estimated 210 microgL could be measured using the silver-doped sol-gel vials an improvement of a factor of 70 is still required to meet JSAWM goals of measuring 3 microgL in 10-minutes Similar to MPA DPA is a diprotic acid (pKa1 = 216 and pKa2 = 692) and variations in pH will effect the relative concentrations of DPA DPA- and DPA= and possibly the SER spectra and band intensities This could prove significant if an acid or base is used to denature anthrax spores with the goal of extracting and analyzing DPA The MPA pH study described above was mimicked for DPA except that the starting solution consisted of 20 mg of DPA in 20 mL HPLC grade water (60x10-3M MW = 1671) The initial solution had a pH of 245 which was made basic by dilute KOH to pHs of 355 433 487 559 1069 and 1166 SER spectra were recorded at each pH using 100 mW of 785 nm and a 44-sec acquisition time Next one drop of concentrated nitric acid was used to remake the solution acidic at a pH of 200 Again sequential pH measurements were performed at 383 510 735 and 822 The solution pH was made acidic a third time but to pH 219 171 then 135 Throughout this process no more than 20 drops of acid or base were added and therefore the concentration was diluted by no more than 10 Most of the spectral bands showed a minor decrease in intensity as a function of increasing pH values However the bands at 525 and 795 cm-1 showed the most dramatic changes which occurred at acid pH Figure 8 shows the SER spectra of DPA for the spectral region and pH range of interest The identity of the DPA species was determined by plotting the normalized peak intensities with the lowest value set to 0 and the highest to 0006 M as a function of pH and overlaying these values on a plot of the relative concentrations for DPA DPA- and DPA= as previously done for MPA (Figure 9) As can be seen the 525 cm-1 band clearly corresponds to DPA The correspondence of the 795 cm-1 band to this species is less clear as the band retains intensity until pH of 55 This can be attributed to contribution to the overlapping band at 810 cm-1 which does not change as a function of pH The fact that most bands are observed at all pHs suggest that the primary interaction with silver is through the ring nitrogen This is supported by the fact that the most intense band occurs at 1008 cm-1 attributed to a symmetric ring breathing mode and that this interaction has been characterized for pyridine in numerous papers27
0
1
2
3
4
5
6
7
0 200 400 600 800 1000 1200
MPA Concentration (mgL)
760
Ban
d In
tens
ity
Figure 6 SER spectra of MPA in water at A) 1000 B) 100 and C) 10 mgL Conditions pH of 7 silver-doped sol-gel coated vial 200 mW of 785 nm 33-sec 8 cm-1 resolution
Figure 7 Plot of SER intensity of 760 cm-1 band of MPA as a function of concentration using 200 mW of 785 nm Inset table includes average intensity LOD standard deviation and percent deviation for each concentration but for 100 mW and 10 min
conc (mgL) LOD ave stddev dev10 021 002 1021100 072 010 14301000 312 040 1280
A
B C
SPIE -2003-5269 122
The ability to detect dipicolinic acid (DPA) by SERS is demonstrated in Figure 10 Even at 1 mgL the primary bands are visible The intensity of the dominant ring breathing mode observed at 1008 cm-1 for DPA in water (pH = 25 to 55) is plotted as a function of concentration in Figure 11 These values were also used to estimate limits of detection based on the SN of the 1008 cm-1 band using 100 mW of 785 nm laser excitation and a 10-min acquisition time Again the lower the measured concentration the lower the predicted LOD (see Figure 11 inset) and detection of 160 microgL is possible Gastrointestinal anthrax requires significant more spores than inhalation anthrax28 and a limit of detection might be placed at 1 million spores in 1 liter of water or 10 microgL Since each spore contains ~10 CaDPA by weight29 a goal for DPA might be 1 microgL indicating that the present measurements must be improved by nearly two orders of magnitude Due to the increased hazards of handling HCN gas KCN salt was used for these experiments Nevertheless all sample preparations were performed in a chemical hood KCN completely dissolves in water but its conjugate acid HCN is formed and has a Ka of 615x10-1030 viz
HCN CN- + H+ pKa = 921 Reaction 4
Figure 8 SER spectra of 1 mgmL DPA as a function of pH Conditions 100 mW 785 nm 100 scans (44-sec)
Figure 9 Concentration dependence of DPA DPA- and DPA= as a function of pH for a 0006M sample Intensity of 525 () and 795 (diams) cm-1 bands from Fig 8 as a function of pH
pH
135 171 219 383
0
0001
0002
0003
0004
0005
0006
0007
0 2 4 6 8 10 12 14pH
Con
cent
ratio
n [M
]
DPADPA-DPA=795525
DPA DPA=DPA-
pK2 = 692pK1 = 216
795 525
0
05
1
15
2
25
0 200 400 600 800 1000 1200
DPA Concentration (mgL)
1008
Ban
d In
tens
ity
conc (mgL) lod-10min-100mw1 017
10 016100 103
1000 355
Figure 10 SER spectra of DPA in water at A) 1000 B) 100 C) 10 and D) 1 mgL Conditions pH of 25-55 silver-doped sol-gel coated vial 175 mW of 785 nm 1-min 8 cm-1 D) has been multiplied by x10 to make bands visible
A
B C
D
Figure 11 Plot of SER intensity of 1008 cm-1 band of DPA as a function of concentration using 175 mW of 785 nm Inset table includes LOD in mgL for each concentration but for 100 mW and 10 min
SPIE -2003-5269 123
Consequently the cyanide concentration must be determined for each initial KCN concentration Specifically the samples prepared with concentrations of 01 1 10 100 and 1000 mgL of KCN produced CN- concentrations of 63x10-3 033 69 89 and 964 mgL at pHs of 816 90 967 102 and 107 respectively The pH dependence for the HCN and CN- concentrations are shown in Figure 12 Thus as the amount of KCN added to the solution decreases so does the pH of the solution (becomes less basic) and according to Reaction 4 the relative amount of CN- to HCN also decreases For example in the preparation of a 01 mgL solution of KCN the pH is shifted from 7 for pure water to only 816 and only 63 of the starting material becomes CN- or 63x10-3 mgL In comparison for a solution of 1000 mgL the pH is shifted from 7 to 107 and 96 of the starting material becomes CN- This is significant since the cyanide ion is better able to adsorb onto the silver particles and become SERS active SER spectra of 10 100 and 1000 mgL of KCN samples are shown in Figure 13 and a corresponding plot of the 2100 cm-1 band intensity as a function of concentration is shown in Figure 14
000102030405060708091011
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14pH
Con
cent
ratio
n [m
gm
L]
pKa = 921
CN -HCN
The SER spectra of cyanide are dominated by the single CN stretch at 2100 cm-1 which occurs in normal Raman spectra of solutions at 2080 cm-1 However a low frequency mode occurs at 135 cm-1 due to a Ag-CN stretch (not shown) indicative of a strong surface interaction It is also observed that as the concentration decreases the CN stretch shifts to 2140 cm-1 This shift has been attribute to the formation of a tetrahedral Ag(CN)3
2- surface structure31 as well
Figure 14 Concentration dependence of KCN SERS measured under conditions in Fig 11 Concentrations are 1 01 and 001 mgml Intensities are measured for the CN stretch at 2100 cm-1 Inset table includes LOD in mgL for each concentration in Figs 13 and 15 but for 100 mW and 10 min
Figure 13 SER spectra of KCN in water at A) 1000 B) 100 and C) 10 mgL Conditions pHs of 107 102 and 97 silver-doped sol-gel coated vial 100 mW of 785 nm 1-min 8 cm-1 C) has been multiplied by x10 to make band visible
Figure 12 Concentration dependence of HCN and CN- as a function of pH for a 1 mgmL sample Calculated intensity of 2100 () cm-1 band for a 1 mgml sample at pHs of 816 90 967 102 and 107
A
B
C
0
20
40
60
80
100
120
140
0 200 400 600 800 1000 1200CN Concentration (mgL)
2100
Ban
d In
tens
ity
Figure 15 SER spectra of KCN in water at A) 10 B) 1 and C) 01 mgL Conditions pHs of 97 12 and 12 gold-doped sol-gel coated vial 100 mW of 785 nm 1-min 8 cm-1 C) has been multiplied by x10 to make band visible
A
B
C
conc (mgL) lod-10min-100mw Condition01 001 Au-pH 121 007 Au-pH 12
69 003 Au-pH 9769 016 Ag-pH 9789 022 Ag-pH 102946 113 Ag-pH 107
SPIE -2003-5269 124
as to CN adsorbed to two different surface sites32 Alternatively the 2140 cm-1 band could be attributed to HCN since this species dominates at lower concentrations However it is unlikely that this species would be attracted to the electropositive silver surface Further both peaks should be present at pHs between 85 and 105 but this is not observed It has also been suggested that at concentrations near and above monolayer coverage the CN- species is forced to adsorb end-on due to crowding and at lower concentrations the molecule can reorient to lie flat33 This suggests that the 2100 and 2140 cm-1 bands correspond to the end-on and flat orientations respectively As Figures 13 and 14 show the intensity of the CN stretch for the 89 mgL sample is nearly as intense as the 964 mgmL band This suggests that the Raman signal for the flat orientation is more enhanced However more extensive measurements are required to verify this point Since resent research has suggested that cyanide may be more effectively detected on gold measurements of KCN solutions were also performed using gold-doped sol-gel vials Preliminary measurements are shown in Figure 15 for samples prepared from 01 1 and 10 mgL KCN Since the pHs are 816 90 967 the resultant CN- concentrations are 63x10-3 033 and 69 mgL Initially only the highest concentration was observed and the signal intensity was significantly better than the equivalent concentration measured using silver In an effort to shift Reaction 4 to the left transforming HCN to CN- (Le Chatelierrsquos principle) KOH was added to the lower concentration samples producing solutions with pH 12 This effectively forces all of the cyanide in solution to be CN- or 01 and 10 mgL respectively More importantly the CN stretch is now observed in the SER spectra The band appears at 2125 cm-1 as has been previously reported for gold12 As calculated for MPA and DPA LODs can be estimated from this data For the three concentrations of cyanide on silver the LODs are 016 to 11 mgL for 100 mW of 785 nm laser excitation and a 10-min acquisition time For gold pH adjusted the LODs are10 to 70 microgL an improvement of more than 10 times silver Nevertheless either substrate is sufficient to meet the JSAWM goals of measuring 3 mgL in 10-minutes as the requirements form cyanide are much less stringent than the nerve agents
4 CONCLUSIONS Here we examined the surface-enhanced Raman spectral response for methyl phosphonic acid dipicolinic acid and cyanide as a function of pH It was determined that the most prominent peak at 760 cm-1 reaches a maximum intensity between pH 3 and 7 and corresponds to the MPA- species Neither the MPA nor MPA= species appear to generate a SER spectrum and consequently no spectra were observed below pH 2 or above pH 8 In this study we also found that higher laser powers could cause photodegradation of MPA signified by the exponential growth of a band at 1050 cm-1 which is tentatively assigned to phosphorous acid Unlike MPA DPA was observed at all pHs This is attributed to the dominant interaction of the pyridine functional group with silver Minor spectral changes were observed at acid pHs and were assigned to neutral DPA Like MPA SER spectra of cyanide were pH dependent No spectra were observed for the HCN species while CN- was best observed at pHs more basic than 8 Preliminary concentration studies for the three analytes allowed estimating limits of detection for MPA DPA and CN using 100 mW of 785 nm and a 10-min acquisition time of 210 165 and 70 microgL respectively Although the latter value suggests that cyanide can be measured at sufficiently low concentrations to meet JSAWM goal improvements by 100 to 200 times are required for MPA and DPA It is clear from this study that pH of the sample is important and must be taken into account when developing concentration calibration curves as well as fieldable analyzers Future work will include tailoring the sol-gel with specific functional groups to dictate sample pH and thereby optimize sensitivity
ACKNOWLEDGMENTS The authors are grateful for the support of the US Army (DAAD13-02-C-0015 Joint Service Agent Water Monitor program) The authors would also like to thank Dr Steve Christensen of the US Army SBCCOM for helpful discussions and Mr Chetan Shende of Real-Time Analyzers for assistance in development of the gold-doped sol-gels
REFERENCES 1 Erickson B Analytical Chemistry News amp Features June 1 397A (1998)
SPIE -2003-5269 125
2 Johnston RL Hoefler CM Fargo JC and Moberley B AT-ONSITE 5-8 (1994) 3 Black RM RJ Clarke RW Read and MT Reid J Chromatography 662 301-321 (1994) 4 USA Army Center for Health Promotion and Preventive Medicine (USACHPPM) Technical Guideline 230A 5 Jensen JL NC Wong W R Loerop SPIE 4775-03 (2001) 6 Hoenig SL Handbook of Chemical Warfare and Terrorism Greenwood Press Wesport CT (2002) 7 Munro NB SS Talmage GD Griffin LC Waters AP Watson JF King and V Hauschild Env Health
Persp 107 933-974 (1999) 8 Holstege CP Kirk M Sidell FR Crit Care Clin13 923-42 (1997) 9 Farquharson S P Maksymiuk K Ong and S Christesen SPIE 4577 166-173 (2001) 10 Lee Y and S Farquharson SPIE 4378 21-26 (2001) 11 Spencer KM J Sylvia S Clauson and J Janni SPIE 4577 158-165 (2001) 12 Tessier P S Christesen K Ong E Clemente A Lenhoff E Kaler and O Velev Applied Spectroscopy 56
1524-1530 (2002) 13 Farquharson S WW Smith S Elliott and JF Sperry SPIE 3533 207-214 (1998) 14 Farquharson S WW Smith S Elliott and JF Sperry SPIE 3855110-116 (1999) 15 Farquharson S WW Smith YH Lee S Elliott and JF Sperry SPIE 4575 62-72 (2002) 16 Guzelian AA J Sylvia J Janni S Clauson and KM Spencer SPIE 4577 182-192 (2001) 17 Shende C F Inscore A Gift P Maksymiuk and S Farquharson in press 18 Weaver MJ S Farquharson and MA Tadayyoni J Chem Phys 82 4867-4874 (1985) 19 Alak AM and T Vo-Dinh Analytical Chemistry 59 2149-2153 (1987) 20 Lee Y and S Farquharson SPIE 4206 140-146 (2000) 21 Farquharson S and Y Lee SPIE 4200-16 (2000) 22 Lee Y S Farquharson and P M Rainey SPIE 3857 76-84 (1999) 23 Lee Y S Farquharson H Kwong and M Shahriari SPIE 3537 252-260 (1998) 24 Wang J M Pumera G Collins and A Mulchandani Analytical Chemistry 74 6121-6125 (2002) 25 Dou X YM Jung Z-Q Cao and Y Ozaki Applied Spectroscopy 53 1440-1447 (1999) 26 Data supplied by S Christesen and K Ewing 27 Kerker M and B Thompson Eds SPIE MS 10 (1990) 28 Inglesby TV DA Henderson JG Bartlett JAMA 287 2236 (2002) 29 FW Janssen AJ Lund and LE Anderson Science 127 26 (1958) 30 Lide DR Ed Handbook of Chemistry and Physics CRC Press 77th Ed (1996-7) 31 Billmann J G Kovacs and A Otto Surf Sci 92 153 (1980) 32 Murray CA and S Bodoff Phys Rev B 32 671 (1985) 33 Kellogg D and J Pemberton J Phys Chem 91 1120 (1987)
SPIE-2003-5269 16
Chemical agent detection by surface-enhanced Raman spectroscopy
Stuart Farquharson Alan Gift Paul Maksymiuk Frank Inscore and Wayne Smith
Real-Time Analyzers 87 Church Street East Hartford CT 06108
Kevin Morrisey and Steven D Christesen US Army SBCCOM Aberdeen Proving Ground MD 21010
ABSTRACT
In the past decade the Unites States and its allies have been challenged by a different kind of warfare exemplified by the terrorist attacks of September 11 2001 Although suicide bombings are the most often used form of terror military personnel must consider a wide range of attack scenarios Among these is the intentional poisoning of water supplies to obstruct military operations in Afghanistan and Iraq To counter such attacks the military is developing portable analyzers that can identify and quantify potential chemical agents in water supplies at microgram per liter concentrations within 10 minutes To aid this effort we have been investigating the value of a surface-enhanced Raman spectroscopy based portable analyzer In particular we have been developing silver-doped sol-gels to generate SER spectra of chemical agents and their hydrolysis products Here we present SER spectra of several chemical agents measured in a generic tap water Repeat measurements were performed to establish statistical error associated with SERS obtained using the sol-gel coated vials Keywords Chemical agents chemical agent detection SERS Raman spectroscopy
1 INTRODUCTION
In the autumn of 2001 terrorism within US borders became a sobering reality While extensive efforts are being implemented to secure the homeland US and Coalition forces in Afghanistan and Iraq are constantly faced with terrorist attacks In addition to car-bombings the military has identified several non-traditional attack scenarios including poisoning of water supplies by chemical warfare agents (CWAs) To counter this threat the Department of Defense has been investigating numerous approaches to detect such attacks under the auspices of the Joint Service Agent Water Monitor (JSAWM) program1 The JSAWM must be capable of identifying the chemical agents with no false-positives and quantifying the agents at microgram per liter concentrations within 10 minutes2 This includes the analysis of drinking water supplies distribution and storage systems Currently colorimetric paper is used to detect agents on-site while gas chromatography combined with mass spectrometry (GCMS) is used in mobile support laboratories However both methods have severe drawbacks The paper changes color in response to contact with many chemicals besides CWAs causing a high incidence of false positives3 while GCMS although very chemically specific requires up to an hour to perform and regular re-calibration456 Vibrational spectroscopy has also been investigated as a potential method of rapidly detecting CWAs7-
11 as early as 1985 in the case of infrared spectroscopy8 and 1988 in the case of Raman spectroscopy11 Again however these techniques also have limitations when it comes to measuring trace poisons in water Infrared spectra would be dominated by the very strong absorption of water which would obscure absorptions by most other chemicals present Whereas Raman spectroscopy is simply not a very sensitive technique and detection limits are typically grams per liter Surface-enhanced Raman spectroscopy (SERS) offers several advantages over conventional vibrational methods and that may provide the necessary sensitivity required for detecting trace quantities of chemical agents in water SERS employs the interaction of surface plasmon modes of metal particles with target analytes to increase scattering efficiency by as much as 1 million times12 In 1987 the potential of this approach to measure CWAs was demonstrated by measuring a series of pesticides13 Several of these organophosphonates have chemical structures similar to CWAs in particular P=O functional groups In the past few years we and others have further explored the ability of SERS to detect CWAs14-17 and even bioagents 18-21 We have been employing silver-doped sol-gels to promote the SER effect
SPIE-2003-5269 17
in these studies The porous silica network of the sol-gel matrix offers a unique environment for immobilizing and stabilizing SER-active metal particles22-25 The silver-doped sol-gels have been coated on the internal walls of standard glass vials to allow rapid SER analysis of small sample quantities without preparation We have measured over 100 chemicals with enhancements of 104 to 106 demonstrated reversible measurements in a flowing system reproducible measurements from vial-to-vial and measurements in multiple solvents including water21-25 Previously we used these vials to perform preliminary measurements of cyanide (CN) and methylphosphonic acid (MPA) Most of the nerve agents form MPA during hydrolysis while Tabun forms CN a chemical agent in its own right In another paper including in these proceedings we examined the limits of detection (LOD) for MPA by measuring a series of concentrations down to 50 mgL and estimated a limit of detection of 100 microgL26 These measurements provide encouragement in that SERS may satisfy the needs of the JSAWM To further establish the viability of SERS in particular silver-doped sol-gels here we present analysis of cyanide mustard and VX in tap water The measurements performed at the US Armyrsquos Edgewood Chemical Biological Center Aberdeen MD also included numerous repeat measurements to establish reproducibility
2 EXPERIMENTAL 2a General All chemicals including potassium cyanide 2-chloroethylethyl sulfide and those used to prepare the silver-doped and gold-doped sol-gel coated vials were obtained and used as received from Sigma-Aldrich All samples were prepared in a chemical hood using HPLC grade water unless otherwise noted (Fischer Scientific Fair Lawn NJ) for SERS measurements Once prepared the samples were transferred into the silver-doped sol-gel vials (Real-Time Analyzers Inc East Hartford CT) The vials were coated in a manner similar to that previously reported by adding ammonium hydroxide to a solution of silver nitrate tetramethyl orthosilicate (TMOS) and methanol22 The two precursor solutions were prepared mixed and transferred to 2-mL glass vials dried and heated After sol-gel formation the incorporated metal ions were reduced with dilute sodium borohydride (1mgmL) which is followed by a water wash to remove residual reducing agent After the resultant analyte solution was introduced the SER-active vial was fixed horizontally to an XY positioning stage (Conix Research Springfield OR) just inside the focal point of an f07 aspheric lens The lens focused the beam into the sample and collected the scattered radiation back along the same axis A dichroic filter (Omega Optical Brattleborough VT) was used to reflect the excitation laser to the lens and pass the Raman scattered radiation collected by the lens An f2 achromat was used to collimate the laser beam exiting a 200 microm core diameter source fiber optic while a second f2 achromat was used to focus the scattered radiation into a 365 microm fiber optic (Spectran Avon CT) A short pass filter was placed in the excitation beam path to block the silicon Raman scattering generated in the source fiber from reflecting off sampling optics and reaching the detector A long pass filter was placed in the collection beam path to block the sample Rayleigh scattering from reaching the detector A 785 nm diode laser (Process Instruments Inc model 785-600 Salt Lake City UT) was used to deliver 100 to 150 mW of power to the sample A Fourier transform Raman spectrometer (Real-Time Analyzers model IRA-785 East Hartford CT) and a silicon photo-avalanche detector (Perkin Elmer model C30902S Stamford CT) were used to acquire the SER spectra
2b Edgewood Chemical Biological Center The surface-enhanced Raman spectral measurements at the US Armyrsquos Chemical Biological center presented here were all performed on September 12 2003 To expedite measurements a plate was machined to hold up to 12 SER-active sample vials (Figure 1) The plate fit a standard XY plate reader that could be programmed Pure KCN bis-(2-chloroethyl)sulfide (distilled mustard gas HD) and ethyl S-2-diisopropylamino ethyl methylphosphonothioate (VX) were obtained on-site and used to prepare 1 gL tap water solutions in a chemical hood with appropriate safety equipment Simulated tap water was prepared by adding 10 chemicals most often found in tap water at appropriate concentrations to distilled water (Table 1) SER measurements were also performed in a chemical hood For added safety the FT-Raman instrument was placed outside the laboratory and 30 foot fiber optic and electrical cables were used to allow remote SERS measurements and plate manipulation For each experiment 1gL samples were prepared and added to 9 individual vials which were then loaded on the plate In some cases a tenth vial was included as a blank
Table 1 Generic Tap Water Composition
Compound mgL NaHCO3 100 CaSO4 27 MgSO4bull7H2O 67 NaNO3 1 Fulvic Acid 1 K2HPO4 07 KH2PO4 03 (NH4)2HSO4 001 NaCl 001 FeSO4 0001 pH 76-78
SPIE-2003-5269 18
A software program was written that allowed selecting the sequence that the vials were measured the number of positions along the length of the vials to measure (1 to 5) and the number of scans to co-add During sample analysis the program displayed the vial being analyzed the point being analyzed and the spectrum as it was being acquired Once all the data was collected a second software program was written to rapidly analyze the data The spectra collected for all the vials on a plate could be loaded at one time and then the spectra for each point could be displayed simultaneously or separately The user could then select the Raman peak to analyze in terms of peak height or area This was accomplished by selecting points on either side of the peak to define a baseline of zero The peak height or area could then be computed for all of the spectra loaded and then exported to a spreadsheet for statistical analysis
Figure 1 A) Vial Holder 6 slots to hold 2 vials each end-to-end B) Measurement Configuration Program user selects vials to measure sequence number of points per vial (1 to 5) and number of scans per point C) Spectral Acquisition Program shows spectrum being collected which vial and position D) Spectra Analysis Program user selects spectra to analyze by plate vial and point (s) as well as two wavenumbers defining the peak and the baseline to subtract The image is of 5 repeat measurements of 10 mgL KCN in generic tap water 16 sec each 100 mW of 785 nm
3 RESULTS AND DISCUSSION Raman and surface-enhanced Raman spectra were obtained for potassium cyanide bis-(2-chloroethyl)sulfide and ethyl S-2-diisopropylamino ethyl methylphosphonothioate representing three classes of chemical agents cyanides mustards and nerve agents respectively Spectra were also obtained for 2-chloroethyl ethyl sulfide (CEES) a structural analogue to HD which was included in the study to aid in assigning spectral bands KCN salt was used for cyanide experiments to avoid the increased hazards of handling HCN gas KCN completely dissolves in water forming its conjugate acid HCN according to its Ka of 615x10-1027 and at a concentration of 1 mgmL results in a pH 107 solution This is significant in that only CN- appears to interact sufficiently with silver to produce a SER spectrum and no spectral signal is observed below pH 726 Figure 2 shows the SER and normal Raman spectra for KCN The SERS spectrum of 1mgml KCN in water shows a single intense somewhat broad feature at
A B
C D
SPIE-2003-5269 19
2100 cm-1 assigned to the single CequivN stretch The band is much sharper in the normal Raman spectra of the solid KCN salt at 2074 cm-1 This band does broaden and shift to 2080 cm-1 in solution (not shown) However the observed SERS frequency is attributed to interaction with silver and low frequency mode at 135 cm-1 attributed to a Ag-CN stretch (not shown) supports this conclusion
cm-1 band to a C-S stretch but the authors concede that it is in fact more likely a C-Cl stretch28 It appears that the most intense Raman bands at 648 692 and 747 cm-1 shift to 620 660 and 720 cm-1 in the SER spectra and are tentatively assigned as above The width of these bands suggests that they overlap underlying spectral features Additional bands in the Raman spectra occur at 972 1034 1049 1263 1286 1423 1442 2865 2935 and 2960 cm-1 Corresponding bands occur in the SER spectra at 964 1015 1054 1286 1410 1447 2865 and 2935 cm-1 Most of these bands are associated with alkane modes specifically the bands at approximately 1040 cm-1 to a C-C stretch 1290 cm-1 to a CH2 in-phase twist 1440 cm-1 to a CH2 wag 2865 cm-1 to a symmetric CH2 stretch and 2965 cm-1 to an asymmetric CH2 stretch The Raman and SER spectra of sulfur mustard were measured at the Edgewood center (Figure 4) Both spectra are largely similar to CEES The C-Cl and C-S bands in the Raman spectrum of HD now occur at 640 655 700 739 and 760 cm-1 and are more resolved possibly due to the increased molecular symmetry Theoretical calculations indicate that the first three bands are due to C-Cl stretching modes and the latter two to C-S stretching modes28 Only the C-Cl bands maintain significant intensity in the SER spectra occurring at 624 and 643 cm-1 which is attributed to the
Prior to measurements of HD CEES was examined by Raman and SER spectroscopy (Figure 3) CEES also known as half-mustard is essentially identical to HD except one of the chlorine end atoms is replaced by a hydrogen atom Again although not as toxic as HD CEES is a blister agent and dilute aqueous samples were prepared using appropriate safety equipment Both the Raman and SER spectra of CEES are similar and dominated by bands between 600 and 800 cm-1 These are associated with C-Cl and C-S stretching modes which are tentatively assigned to 648 and 747 cm-1 in the Raman spectra respectively The shoulder at 630 cm-1 the overlapped band at 660 cm-1 and the strong band at 692 cm-1 could also be due to these modes or their asymmetric counterparts It is worth noting that theoretical calculations assign the 692
Figure 2 A) SER and B) NR spectra of KCN Conditions A) 1 mgml in tap water 100 mW of 785 nm at sample 1-min acquisition time B) solid 300 mW of 785 nm 5-min All spectra are 8 cm-1 resolution
A
B
Figure 3 A) SER and B) NR spectra of CEES Conditions A) 1 vv (10 mgml) in MeOH 100 mW of 785 nm 1-min acquisition time B) neat 300 mW of 785 nm 5-min
A
B
Cl-CH2-CH2-S-CH2-CH3
A
B
Figure 4 A) SER and B) NR of HD Conditions A) 1mgml in tap water B) pure both 100 mW of 785 nm 1-min
Cl-CH2-CH2-S-CH2-CH2-Cl
SPIE-2003-5269 20
expected strong interaction between chlorine and silver and adds support to the assignment of this band to a C-Cl stretch Weaker overlapping bands occur at 670 692 and 724 cm-1 the latter possibly due to C-S stretching modes Again the alkane modes are apparent in the normal Raman spectra of HD but only a broad feature at 1300 to 1450 cm-1 suggests CH2 contributions in the SER spectrum Although the observed bands in the VX spectrum have not been assigned (Figure 4) a computer generated Raman spectrum29 predicts many of the same features with surprising accuracy and are used here Two intense bands at 460 and 530 cm-1 closely match predicted bands at 463 and 546 cm-1 assigned to a CH3-P=O bend and a PO2CS wag Three highly overlapped bands occur at 694 745 and 771 cm-1 matching predicted bands at 713 730 and 760 cm-1 The first
Table 2 Measured SER peak heights for the CN stretch at 2100 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
Figure 5 A) SERS and B) NR spectra of VX Conditions A) 1 vv (10 mgml) in MeOH B) pure sample both 100 mW of 785 nm 1-min acquisition time
A
B
two have been assigned to a C-S stretch and CH2bend respectively while the latter has been attributed to either a P-C stretch or an O-C-C stretch Although the 745 cm-1 band may alternatively be assigned to a C-S stretch based on the previous measurements of CEES and HD The relatively intense bands at 890 1106 1218 1445 and 1465 cm-1 also match predicted bands at 880 1108 1216 1440 and 1464 cm-1 that are assigned to a C-C stretch CH3 rock N-C3 stretch various C-H3 bends and C-H bends respectively Both the computer generated and the measured spectra contain numerous other less intense bands One is worth mentioning A unique band appears at 370 cm-1 that is predicted at 368 cm-1 and corresponds to an O-P=O bend The surface-enhanced Raman spectrum of VX is also rich with spectral features It has the unique low frequency band at 370 cm-1 as well as a second band at 380 cm-1 that is assigned tothe S-P-O bend predicted in the normal Raman spectrum at 388 cm-1 Based on the measured and predicted normal Raman spectra the following SERS assignments are given 460 cm-1 to the CH3-P=O bend 544 cm-1 to the PO2CS wag 738 cm-1 to a C-S stretch (based on arguments above) 890 cm-1
to a C-C stretch 1101 cm-1 to a CH3 rock and 1456
cm-1 to a C-H bend The ability of SERS to measure chemical agents in water containing real-world chemical interferents was tested by using the generic tap water described in Table 1 The ability to reproduce measurements was accomplished by preparing three separate water stock solutions which were used to prepare three sample solutions each of 1 mgmL KCN in the generic tap water The samples defined as vials 1-9 were then measured at 4 points per vial in a semi-random fashion such that errors associated with stock solution preparation and errors associated with instrument drift could be identified No trends were apparent that signified such systematic errors Each spectrum collected consisted of 20 averaged scans taking 16 seconds at 8 cm-1 resolution The laser power at the sample was measured periodically during the day and it ranged from 102 to 105 mW spacccc
spot Vial 1 Vial 2 Vial 41 13754 13004 128192 13519 12692 12909
1 3 13541 12721 126394 13462 12648 12651
ave 13569 12766 12755Vial 6 Vial 3 Vial 5
1 11586 14068 10757 2 11236 14402 115122 3 11376 14546 11559
4 10894 11763 11214ave 11273 13695 11261
Vial 7 Vial8 Vial 91 11198 15804 11023
2 11243 15716 88753 3 11562 15216 11218
4 11655 15084 11077ave 11415 15455 10548
CN AVG STDEV ERR1 mgmL pk ht 12526 1560 1245
stock solution
SPIE-2003-5269 21
The CN measurements consisting of 36 data points produced an average height of 125 for the 2100 cm-1 peak with a standard deviation of 156 or 125 (Table 2) The HD and VX measurements were performed precisely the same way (Tables 3 and 4) For HD the 624 cm-1 peak was used for analysis and it had an average height of 53 with a standard deviation of 068 or 129 while for VX the 544 cm-1 peak was used for analysis and it had an average height of 1051 with a standard deviation of 308 or 293 The greater error in the VX measurements can be somewhat attributed to Vial 4 which produced lower SER signal intensities But removing this vial from the data set changes the standard deviation to 233 only a modest improvement
spot Vial 1 Vial 2 Vial 41 525 609 4682 45 675 484
1 3 527 644 5094 559 807 669
ave 51525 68375 5325Vial 6 Vial 3 Vial 5
1 521 574 575 2 536 53 4492 3 509 508 451
4 565 594 379ave 53275 5515 4635
Vial 7 Vial 8 Vial 91 457 589 413
2 583 497 4273 3 544 505 4
4 528 5 509ave 528 52275 43725
HD AVG STDEV ERR1 mgmL pk ht 530 068 1291
stock solution
spot Vial 1 Vial 2 Vial 41 1464 1034 4672 1485 99 61
1 3 1491 989 5684 1041 777 553
ave 137025 9475 5495Vial 6 Vial 3 Vial 5
1 1058 942 1293 2 697 121 965
2 3 727 1206 11124 689 1254 746
ave 79275 1153 1029Vial 7 Vial 8 Vial 9
1 1127 783 163 2 1358 812 1614
3 3 1371 908 15124 1097 875 1397
ave 123825 8445 153825
VX AVG STDEV ERR1 mgmL pk ht 1051 308 2925
stock solution
4 CONCLUSIONS In this paper we examined the ability of surface-enhanced Raman spectroscopy to reproducibly measure CN HD and VX in tap water without chemical interference Both normal and surface-enhanced Raman spectra were examined to select unique bands suitable to identify and quantify these chemical agents For SER measurements the 2100 cm-1 C-N stretch was used for CN the 624 cm-1 C-Cl stretch was used for HD and the 544 cm-1 PO2CS wag was used for VX It was determined that 1 mgmL samples of each of these chemicals measured 36 times in glass vials coated with a silver-doped sol-gel reproduced measurements with standard deviations of 125 129 and 293 It was further found that the 10 chemicals added to simulate generic tap water did not interfere with or alter the SER spectra It should be noted that the concentrations used in this study were considerably greater than those required by the JSAWM program Current work involves improving SER sensitivity and designing sampling systems with better reproducibility This includes the development of fractal silver and gold structures within the sol-gel matrix and the development of chemically selective sol-gels
ACKNOWLEDGMENTS The authors are grateful for the support of the US Army (DAAD13-02-C-0015 Joint Service Agent Water Monitor program) The authors would also like to thank Janet Jensen Ronald Crosier and Kristina Gonser for helpful discussions
Table 4 Measured SER peak heights for the VX band at 544 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
Table 3 Measured SER peak heights for the HD band at 624 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
SPIE-2003-5269 22
REFERENCES 1 Jensen JL NC Wong W R Loerop SPIE 4775-03 (2001) 2 JSAWM Requirements at wwwsbccomapgeaarmymilRDAecbcrtPRODSERJSAWMjsawmhtml 3 Erickson B ldquoThe Chemical Weapons Convention Redefines Analytical Challengerdquo Anal Chem 70 397A (1998) 4 Johnston RL Hoefler CM Fargo JC and Moberley B ldquoThe Defense Nuclear Agencyrsquos
ChemicalBiochemical Weapons Agreements Verification Technology Research Development Test and Evaluation Program and its Requirements for On-Site Analysisrdquo AT-ONSITE 5-8 (1994)
5 Black RM Clarke RJ Read RW and Reid MT ldquoApplication of GC-MS and GC-tandem MS to the analysis of chemical warfare samples found to contain residues of the nerve agent sarin sulphur mustard and their degradation productsrdquo J Chrom 662 301-321 (1994)
6 USA Army Center for Health Promotion and Preventive Medicine (USACHPPM) Technical Guideline 230A 7 Hoffland LD Piffath RJ Bouck JB rdquoSpectral signatures of chemical agents and simulantsrdquo Optical
Engineering 24 982-984 (1985) 8 Braue EHJ Pannella MGrdquoCIRCLE CELL FT-IR Analysis of Chemical Warfare Agents in Aqueous Solutionsrdquo
App Spectrosc 44 1513-1520 (1990) 9 Tseng C-H Mann CK and Vickers TJ ldquoDetermination of Organics on Metal Surfaces by Raman
Spectroscopyrdquo App Spectrosc 47 1767-1771 (1993) 10 Staff Reporter ldquoUS Army Tests Lidar to Detect Biological Toxinsrdquo Photonic Spectra pg 50 December 1998 11 Christesen SD Raman cross sections of chemical agents and simulants App Spectrosc 42 318-321 (1988) 12 Weaver MJ Farquharson S Tadayyoni MA ldquoSurface-enhancement factors for Raman scattering at silver electrodesrdquo J Chem Phys 82 4867-4874 (1985) 13 Alak AM and Vo-Dinh T ldquoSERS of Organophosphorous Chemical Agentsrdquo Anal Chem 59 2149-2153 (1987) 14 Lee Y Farquharson S ldquoRapid chemical agent identification by SERSrdquo SPIE 4378 21-26 (2001) 15 Farquharson S Maksymiuk P Ong K Christesen S ldquoChemical agent identification by surface-enhanced Raman spectroscopyrdquo SPIE 4577 166-173 (2001) 16 Spencer KM Sylvia J Clauson S and Janni J ldquoSurface Enhanced Raman as a Water Monitor for Warfare
Agents in Waterrdquo SPIE 4577 158-165 (2001) 17 Tessier P Christesen S Ong K Clemente E Lenhoff A Kaler E Velev O ldquoOn-line spectroscopic characterization of sodium cyanide with nanostructured gold surface-enhanced Raman spectroscopy substratesrdquo
App Spectrosc 56 1524-1530 (2002) 18 Farquharson S W W Smith S Elliott and J F Sperry Rapid biological agent identification by surface- enhanced Raman spectroscopy SPIE 3855110-116 (1999) 19 Farquharson S WW Smith YH Lee S Elliott and J F Sperry Detection of biological signatures A
comparison of electrolytic and metal-doped sol-gel surface-enhanced Raman media SPIE 4575 62-72 (2002) 20 Guzelian AA Sylvia J Janni J Clauson S and Spencer KM ldquoSERS of whole cell bacteria and trace levels of biological moleculesrdquo SPIE 4577 182-192 (2001) 21 Shende C Inscore F Gift A Maksymiuk P Farquharson S ldquoRapid dipicolinic acid extraction from Bacillus spores detected by surface-enhanced Raman spectroscopyrdquo App Spectrosc 58 accepted 22 Lee Y and Farquharson S ldquoSERS Sample Vials Based on Sol-Gel Process for Trace Pesticide Analysisrdquo SPIE
4206 140-146 (2000) 23 Farquharson S and Lee Y ldquoTrace Drug Analysis by SERSrdquo SPIE 4200-16 (2000) 24 Lee Y Farquharson S and Rainey P M Surface-Enhanced Raman Sensor for Trace Chemical Detection in
Water SPIE 3857 76-84 (1999) 25 Lee Y Farquharson S Kwong H and Shahriari M ldquoSol-Gel Chemical Sensor for Surface-Enhanced Raman
Spectroscopyrdquo SPIE 3537 252-260 (1998) 26 Farquharson S A Gift P Maksymiuk F Inscore and W Smith ldquopH dependence of methyl phosphonic acid
dipicolinic acid and cyanide by surface-enhanced Raman spectroscopyrdquo SPIE 5269-19 (2003) 27 Lide DR Ed Handbook of Chemistry and Physics CRC Press 77th Ed (1996-7) 28 Sosa C RJ Bartlett K KuBulat and WB Person ldquoA theoretical study of harmonic vibrational frequencies and
infrared intensities of XCH2CH2SCH2CH2X and XCH2CH2SH (= H Cl)rdquo J Phys Chem 93 577-588 (1993) 29 Hameka HF and JO Jenson ldquoComputer-generated predictions of the structure and the IR and Raman spectra of
VXrdquo ERDEC-TR-065 May 1993
- Chemagents Appendicespdf
-
- SPIE2001-4575-Bioagent-SERSapdf
-
- Detection of bioagent signatures A comparison of electrolytic and metal-doped sol-gel surface-enhanced Raman media
- Stuart Farquharson Wayne Smith and Yuan Lee
-
- Real-Time Analyzers 87 Church Street East Hartford CT 06108
-
- ABSTRACT
- 4 CONCLUSIONS
- 5 ACKNOWLEDGEMENTS
- 6 REFERENCES
-
173
11 Christesen SD Raman cross sections of chemical agents and simulants Applied Spectroscopy 42 318-321 (1988) 12 Alak AM and Vo-Dinh T ldquoSERS of Organophosphorous Chemical Agentsrdquo Analytical Chemistry 59 2149-2153
(1987) 13 Norrod KL Sudnik LM Rousell D and Rowlen KL ldquoQuantitative Comparison of Five SERS Substrates
Sensitivity and Detection Limitrdquo Applied Spectroscopy 51 994-1001 (1997) 14 Lee Y and Farquharson S ldquoSERS Sample Vials Based on Sol-Gel Process for Trace Pesticide Analysisrdquo SPIE 4206
140-146 (2000) 15 Farquharson S and Lee Y ldquoTrace Drug Analysis by Surface-Enhanced Raman Spectroscopyrdquo SPIE 4200-16 (2000) 16 Lee Y Farquharson S and Rainey P M Surface-Enhanced Raman Sensor for Trace Chemical Detection in Water
SPIE 3857 76-84 (1999) 17 Lee Y Farquharson S Kwong H and Shahriari M ldquoSurface-Enhanced Raman Sensor for Surface-Enhanced Raman
Spectroscopyrdquo SPIE 3537 252-260 (1998) 18 Farquharson S Smith W Carangelo R C and Brouillette C ldquoIndustrial Raman Providing Easy Immediate Cost
Effective Chemical Analysis Anywhererdquo SPIE 3859 14-23 (1999) 19 Narayanan VA JM Bello JD Stokes and T Vo-Dinh Analusis 19 307-310 (1991) 20 Laserna JJ E L Torres and JD Winefordner Analytica Chemica Acta 469-480 (1987) 21 Torres EL and JD Winefordner Analytical Chemistry 59 1626-1632 (1987) 22 Suh JS DP DiLella M Moskovits J Phys Chem 87 1540-1544 (1983) 23 Benner RE R Dornhaus R Chang and BL Laube Correlations in the Raman spectra of cyanide complexes adsorbed
at silver electrodes with voltammograms Surface Science 101 341 (1980) 24 Chase D B and JF Rabolt Fourier Transform Raman Spectroscopy Acad Press Ch1 p 131 (1994) 25 Connes J Rev Opt Theor Instrum 40 45 (1961)
APPLIED SPECTROSCOPY 351
focusing the 488 nm laser beam 2 mm deep into thebulk of the crystals to avoid contributions from potentialdepletion layers As shown in Fig 1b the dependence ofthe Raman shift on the degree of deuteration is almostperfectly linear and ts very well with D 5 22684cmR1 24526 where D is the degree of deuteration (in )and R is the spectral mean of the PO4 vibration in cm21A linear correlation coef cient of 0998 indicates an ex-cellent linear dependence of the Raman peak shift withdegree of deuteration This result shows that the shift ofthe PO 4 peak is simply caused by the linear increase inatomic mass due to isotope substitution which decreasesthe length of hydrogen-like bonds
This excellent linear dependence allows us to map thepro le of the DH exchange layer at the surface of DKDPcrystals by acquiring Raman spectra and determining theposition of the PO 4 peak for various depths This methodis preferable over other methods such as determining thestrength of the OD vibration directly (eg at 715 cm21)because the position of the most intense peak in the Ra-man spectrum can be measured more precisely than theintensity of some of the weakest peaks in the spectrumThis is demonstrated in Fig 2 where depth-dependentRaman spectra (Fig 2a) and the resulting exchange layerpro les for two DKDP crystals are shown (Fig 2b) Thespectra in Fig 2a were obtained from a depth scan of aDKDP crystal with 75 degree of deuteration in thebulk grown at 45 8C The spectra start out as DKDP with30 deuteration close to the surface and approach thebulk DKDP spectrum within a few micrometers of depthThe fact that the relative degree of deuteration does notextend to 0 D is due to the limited depth resolution ofthe Raman microprobe which averages over 4 mm indepth Figure 2b depicts the resulting DH exchange layerpro les for this and a second crystal grown at 63 8Crespectively Both crystals had the same exposure to am-bient conditions and their main difference is the temper-ature at which they were grown The different exchangelayer pro les indicate that crystals grown at differenttemperatures have differen t proton conductiv ities 13
which leads to a difference in their rate of deuteriumdepletion The parameters controlling this behavior arecurrently the objective of a detailed study the results ofwhich will be reported elsewhere
CONCLUSION
In conclusion we have shown that the shift of the to-tally symmetric PO 4 stretch mode in the Raman spectrumof DKDP crystals scales linearly with degree of deuter-ation This allows us to correlate Raman peak positionsto deuteration levels in these crystals We have presenteda new technique to determine DH diffusion pro les inDKDP frequency conversion crystals based on micro-Ra-man spectroscopy This technique is fast inexpensiveand works under various environmental conditionswhich will allow us to better understand and control deu-terium depletion in DKDP crystals
ACKNOWLEDGMENTS
We would like to thank M Runkel for rst discovering DKDP crack-ing R Floyd for providing DKDP crystals and L Chase and A Burn-ham for their support and helpful discussions This work was performedunder the auspices of the US Department of Energy by the University
of California Lawrence Livermore National Laboratory under contract W-7405-Eng-48
1 J J De Yoreo A K Burnham and P K Whitman Int Mat Rev47 113 (2002)
2 C E Barker R A Sacks B M Van Wonterghern J A Caird JR Murray J H Campbell K Kyle R B Ehrlich and N DNielsen Proc SPIE-Int Soc Opt Eng 2633 501 (1995)
3 T Suratwala paper to be published4 Cleveland Crystals Inc httpwwwclevelandcrystalscom
KDPshtmltable5 E A Popova I T Savatinova and I A Velichko Sov Phys Solid
State 12 1543 (1971)6 I P Kaminow R C C Leite and S P S Porto J Phys Chem
Solids 26 2085 (1965)7 J A Subramony B J Marquardt J W Macklin and B Kahr
Chem Mat 11 1312 (1999)8 H Tanaka M Tokunaga and I Tatsuzaki Solid State Commun
49 153 (1984)9 R J Nelmes G M Meyer and J E Tibballs J Phys C 15 59
(1982)10 M A Yakshin D W Kim Y S Kim Y Y Broslavets O E
Sidoryuk and S Goldstein Laser Physics 7 941 (1997)11 I Takenaga Y Tominaga S Endo and M Kobayashi Solid State
Commun 84 931 (1992)12 C Krenn personal communication13 M Sharon and A K Kalia J Solid State Chem 21 171 (1977)
Rapid Dipicolinic Acid Extractionfrom Bacillus Spores Detectedby Surface-EnhancedRaman Spectroscopy
STUART FARQUHARSON ALAND GIFT PAUL MAKSYMIUK andFRANK E INSCOREReal-Time Analyzers Inc East Hartford Connecticut06108
Index Headings Dipicolinic acid Bacillus spores Anthrax Surface-enhanced Raman spectroscopy
INTRODUCTION
The anxiety caused by the distribution of anthrax en-dospores through the US postal system in October 2001was exacerbated by the long time required for positiveidenti cation of the Bacillus anthracis spores and the un-known extent of their distribution Since that time manymethods capable of rapid eld analysis have been inves-tigated to augment or replace the laboratory method ofgrowing microorganisms in culture media which takesdays to perform12 Prominent among these approachesare polymerase chain reactions (PCR)3 immunoassaysand detection of calcium dipicolinate as a biochemicalsignature PCR employs primers to separate organism-
Received 10 October 2003 accepted 14 November 2003 Author to whom correspondence should be sent
352 Volume 58 Number 3 2004
speci c nucleic acid sequences (eg capsular protein en-coding gene for Bacillus anthracis)4 and polymerases toamplify the segment until it is detectable Recently am-pli cation times have been substantially reduced andcomplete analysis can now be performed in an hour orless Immunoassay methods are also being developed thatuse competitive binding of the bioagent (as an antigen)and its labeled conjugate for a limited number of anti-bodies Although analyses can be performed in under 30minutes a well-de ned anthrax antigen has not yet beenidenti ed5ndash7 and consequently the false-positive rate isunacceptably high8
A number of other methods are being developed witha focus on the detection of calcium dipicolinate (CaDPA)and its derivatives as a B anthracis signature This is sobecause only spore-forming bacteria contain CaDPA andthe most common potentially interfering spores such aspollen and mold spores do not Relatively fast methodshave been developed to chemically extract CaDPA andthen detect it directly by uorescence9 or indirectly byluminescence1011 In the latter case hot dodecylamine(DDA) has been used to extract dipicolinic acid (DPA)and terbium has been utilized to form a highly lumines-cent DPA complex11 Although measurements have beenperformed in as little as ve minutes it was found thatas many as three concentration-dependent complexes canform each with different lifetimes This coupled withthe fact that the Tb31 cation produces the same lumines-cence spectrum makes determinations of low spore con-centrations problematic
It has been long known that Raman spectra of Bacillispores are dominated by bands associated with CaDPA12
and that these spectra may provide a suitable anthrax sig-nature at the genus level13 Since that time considerableimprovements in Raman instrumentation have led to lab-oratory measurements of single Bacilli spores14 and to eld measurements of spores captured from a mail-sort-ing system15 However the single spore measurementsrequired complex instrumentation that is not ruggedwhile the eld measurements required milligrams of sam-ple Furthermore the Raman spectra of both measure-ments contained uorescence contributions that would in-crease uncertainty in quanti cation
In related research we demonstrated that nanogramquantities of DPA could be detected by uorescence-freesurface-enhanced Raman spectroscopy (SERS)16 We alsodemonstrated that microliter volumes of chemicals canbe detected by SERS using metal-doped sol-gel-packedglass capillaries17 Towards the goal of developing a rap-id eld SERS-based anthrax spore detector we havecombined our previous research and we now report thatDPA can be extracted from a 10 mg B cereus spore sam-ple using DDA in 1 minute and can be detected by SERSin an additional 1 minute
EXPERIMENTAL
Dipicolinic acid (26-pyridinedicarboxylic acid DPA)and dodecylamine (DDA) were used as received fromSigma-Aldrich (Milwaukee WI) Lyophilized B cereusspores prepared according to the literature13 were sup-plied by the University of Rhode Island and used as re-ceived Multiple particles approximately 01 mm3 each
were separated and weighed at 5 to 15 mg representing05 to 15 million spores The sample masses were con-sistent with a previous determination of spore density at0081 gmL that indicated a high degree of entrained air
All chemicals used to prepare the silver-doped sol-gelcoated capillaries were also obtained and used as receivedfrom Sigma-Aldrich According to previously publishedprocedures17 two precursor solutions were preparedmixed and then drawn into 1-mm-diameter glass capil-laries The silver amine precursor consisted of a 51 vvratio of 1 N AgNO 3 to 28 NH3OH while the alkoxideprecursor consisted of a 21 vv ratio of methanol to te-tramethyl orthosilicate The alkoxide precursors weremixed with silver amine precursor in an 81 vv ratioApproximately 015 mL was drawn into the capillarycoating a 15-mm length After sol-gel formation the in-corporated silver ions were reduced with dilute sodiumborohydride which was followed by a water wash to re-move residual reducing agent
A 100 mL drop of a 50 mM DDA solution in ethanolpre-heated to 78 8C was added to each of the B cereusparticles to digest the spore coat After 1 minute the re-sultant solution was drawn into a SER-active capillarythat was immediately xed horizontally to an XY posi-tioning stage (Conix Research Spring eld OR) just in-side the focal point of an f 07 aspheric lens The lensfocused the beam into the sample and collected the scat-tered radiation back along the same axis A dichroic lter(Omega Optical Brattleborough VT) was used to re ectthe excitation laser to the lens and pass the Raman scat-tered radiation collected by the lens An f 2 achromatwas used to collimate the laser beam exiting a 200-mm-core-diameter source ber optic while a second f 2 ach-romat was used to focus the scattered radiation into a 365mm ber optic (Spectran Avon CT) A short-pass lterwas placed in the excitation beam path to block the sil-icon Raman scattering generated in the source ber fromre ecting off sampling optics and reaching the detectorA long-pass lter was placed in the collection beam pathto block the sample Rayleigh scattering from reachingthe detector A 785 nm diode laser (Process InstrumentsInc model 785-600 Salt Lake City UT) was used todeliver 100 to 150 mW of power to the sample A Fouriertransform Raman spectrometer (Real-Time Analyzersmodel IRA-785 East Hartford CT) and a silicon photo-avalanche detector (Perkin Elmer model C30902S Stam-ford CT) were used to acquire the SER spectra
RESULTS AND DISCUSSION
As an initial experiment the SER spectrum of 1 gLof DPA in water was measured using the newly devel-oped silver-doped sol-gel-coated capillaries (Fig 1A) Atthis concentration a high signal-to-noise ratio (SN) isobtained in 1 min In fact a reasonable spectrum is ob-tained in the same time frame for 1 mgL (Fig 1B) TheSER spectra are reasonably similar to the normal Raman(NR) spectrum obtained for a saturated solution of DPAin 1 N KOH (Fig 1C) and the following band shifts areobserved (NR to SER) 647 to 657 cm21 817 to 815cm21 998 to 1008 cm21 1384 to 1382 cm21 1434 to1428 cm21 and 1569 to 1567 cm21 Many of these bandshave been previously assigned1213 such as 998 cm21 to
APPLIED SPECTROSCOPY 353
FIG 1 SERS of DPA in water using silver-doped sol-gel-coated glasscapillary for (A) 1 gL and (B) 1 mgL (C ) NR of saturated DPA in 1N KOH in a glass capillary Spectral conditions (A) and (B) 150 mWof 785 nm 1-min acquisition time (C) 450 mW of 785 nm 5-minacquisition time both 8 cm 21 resolution
FIG 2 SERS of DPA extracted from 10 mg B cereus particle using100 mL of 50 mM hot DDA acquired in (A) 1 minute and (B) 2 seconds(C ) Attempted SERS of 50 mM hot DDA in ethanol using silver-dopedsol-gel-coated glass capillary acquired in 1 min Spectral conditions150 mW of 785 nm 8 cm21 resolution
the symmetric ring stretch 1384 cm21 to the OndashCndashOsymmetric stretch 1428 cm21 to the symmetric ring CndashH bend and 1569 cm21 to the asymmetric OndashCndashOstretch
The rst B cereus samples consisted of 2 mg of sporesin 2 mL of 5 mM hot DDA The samples were main-tained at 78 8C for 40 min and while hot approximately10 mL was drawn into a SER-active capillary Since spec-tra of DPA were obtained for these initial samples small-er spore masses higher DDA concentrations and shorterheating periods were examined In due course it wasfound that 10 mg of spores could be digested by 100 mLof 50 mM hot DDA in one minute and detected (Fig2A) In fact the signal was suf ciently intense that it canbe observed in as little as two seconds (Fig 2B) Theamount of DPA that was extracted was estimated to bebetween 5 and 10 mgL by comparing the signal intensityof the 1008 cm21 band to that measured for DPA in waterThis is consistent with previous research that found thatthe majority of the DPA is extracted from spores usingDDA11 and that B cereus spores contain approximately10 DPA by weight18 The SN of 127 for the 1008 cm21
band in the 1-minute SER spectrum suggests a limit ofdetection of approximately 250 ng of B cereus sporesbased on a SN of 3 Finally it should be noted that DDAdid not produce a detectable SER spectrum as shown inFig 2C
CONCLUSION
We have demonstrated that by combining rapid extrac-tion of dipicolinic acid from Bacillus cereus spores withchemical identi cation by surface-enhanced Raman spec-troscopy as little as 10 mg of spores can be detected Infact the entire measurement from the time of adding hotdodecylamine to the spores to the time when the dipi-colinic acid SER spectrum is acquired and analyzedcould be performed in less than two minutes The abilityof this method to distinguish between spore-forming bac-teria such as Bacillus anthracis and non-DPA containingpowders could help prevent costly shutdowns associated
with the appearance of suspicious material or intentionalmailing of common substances as an anthrax hoax Thismethod could also prove useful in detecting the locationof anthrax endospores in mail distribution facilities if an-other veri ed attack should occur
Research continues to fully characterize the surface-enhanced Raman spectroscopy signal intensities as afunction of sample concentration and to explore otherextractants that do not require the use of elevated tem-perature
ACKNOWLEDGMENTS
The authors are grateful for the support of the National Science Foun-dation (DMI-0296116 and DMI-0215819) and the US Army(DAAD13-02-C-0015 Joint Service Agent Water Monitor program)The authors are indebted to Chetan Shende for preparing the sol-gelcapillaries The authors also thank James Gillespie Nicholas Fell andAugustus Fountain for providing important background informationMark Farquharson for laboratory support and Professor Jay Sperry ofthe University of Rhode Island for supplying B cereus spores
1 V A Pasechnik C C Shone and P Hambleton Bioseparations 3267 (1993)
2 P J Jackson M E Hugh-Jones D M Adair G Green K K HillC R Kuske L M Grinberg F A Abramova and P Keim ProcNatl Acad Sci USA 95 1224 (1998)
3 B R Glick and J J Pasternak Molecular Biology Principles andApplications of Recombinant DNA (ASM Press Washington DC1994)
4 C A Bell J R Uhl T L Had eld J C David R F Meyer TF Smith and F R Cockerill III J Clin Microbiol 40 2897(2002)
5 D L Gatto-Menking H Yu J G Bruno M T Goode M Millerand A W Zulich Biosens Bioelectron 10 501 (1995)
6 J J Quinlan and P M Foegeding J Rapid Methods AutomationMicrobiol 6 1 (1998)
7 A A Hindle and E A H Hall Analyst (Cambridge UK) 1241599 (1999)
8 M S Ascher US Department of Health amp Human Services(httpwwwhhsgovophppresentationsAscherdoc)
9 R Nudelman B V Bronk and S Efrima Appl Spectrosc 54 445(2000)
10 D L Rosen C Sharpless and L B McBrown Anal Chem 691082 (1997)
354 Volume 58 Number 3 2004
11 P M Pellegrino N F Fell Jr and J B Gillespie Anal ChimActa 455 167 (2002)
12 W H Woodruff T G Spiro and C Gilvarg Biochem BiophysRes Commun 58 197 (1974)
13 E Ghiamati R S Manoharan W H Nelson and J F SperryAppl Spectrosc 46 357 (1992)
14 A P Esposito C E Talley T Huser C W Hollars C M Schal-dach and S M Lane Appl Spectrosc 57 868 (2003)
15 S Farquharson L Grigely V Khitrov W W Smith J F Sperryand G Fenerty J Raman Spectrosc paper accep ted (2003)
16 S Farquharson W W Smith S Elliott and J F Sperry SPIE-IntSoc Opt Eng 3855 110 (1999)
17 S Farquharson and P Maksymiuk Appl Spectrosc 57 479(2003)
18 F W Janssen A J Lund and L E Anderson Science (Washing-ton DC) 127 26 (1958)
SPIE -2003-5269 117
pH dependence of methyl phosphonic acid dipicolinic acid and cyanide by surface-enhanced Raman spectroscopy
Stuart Farquharson Alan Gift Paul Maksymiuk Frank Inscore and Wayne Smith
Real-Time Analyzers 87 Church Street East Hartford CT 06108
ABSTRACT US and Coalition forces fighting terrorism in Afghanistan and Iraq must consider a wide range of attack scenarios in addition to car bombings Among these is the intentional poisoning of water supplies to obstruct military operations To counter such attacks the military is developing portable analyzers that can identify and quantify potential chemical agents in water supplies at microgram per liter concentrations within 10 minutes To aid this effort we have been investigating the value of a surface-enhanced Raman spectroscopy based portable analyzer In particular we have been developing silver-doped sol-gels to generate SER spectra of chemical agents and their hydrolysis products Here we present SER spectra of methyl phosphonic acid and cyanide as a function of pH an important factor affecting quantitation measurements which to our knowledge has not been examined In addition dipicolinic acid a chemical signature associated with anthrax-causing spores is also presented Keywords Chemical warfare agents agent detection agent hydrolysis SERS Raman spectroscopy homeland security
1 INTRODUCTION In the past decade the Unites States and its allies have been challenged by a different kind of warfare exemplified by the terrorist attacks of September 11 2001 Suicide bombings and the use of chemical agents are the norm and military personnel must consider a wide range of attack scenarios Among these is the intentional poisoning of water supplies to obstruct military operations in Afghanistan and Iraq Currently colorimetric paper is used to detect agents on-site while gas chromatography combined with mass spectrometry (GCMS) is used in mobile support laboratories However both methods have severe drawbacks The paper changes color in response to contact with many chemicals besides CWAs causing a high incidence of false positives1 while GCMS although very chemically specific requires hours to perform and constant re-calibration234 Military operations would be greatly aided by a portable analyzer that can identify and quantify potential chemical agents at concentrations that impact safety This includes the analysis of drinking water supplies distribution and storage systems To meet this goal the Department of Defense has been investigating numerous approaches under the auspices of the Joint Service Agent Water Monitor (JSAWM) program5 The JSAWM must be capable of identifying the chemical agents with no false-positives and quantifying the agents at microgram per liter concentrations within 10 minutes678 Recently we and others have been investigating the ability of surface-enhanced Raman spectroscopy (SERS) to measure chemical agents9-12 bioagents13-17 and their hydrolysis products in water SERS employs the interaction of surface plasmon modes of metal particles with target analytes to increase scattering efficiency by as much as 1 million times18 In 1987 the potential of this approach to measure CWAs was demonstrated by measuring a series of pesticides19 Several of these organophosphonates have chemical structures similar to CWAs in particular P=O functional groups In our studies we have been employing silver-doped sol-gels to promote the SER effect The porous silica network of the sol-gel matrix offers a unique environment for immobilizing and stabilizing SER-active metal particles20-23 The sol-gels have been coated on the internal walls of standard glass vials to allow rapid SER analysis of small sample quantities without preparation We have measured over 100 chemicals with enhancements of 104 to 106 demonstrated reversible measurements in a flowing system reproducible measurements from vial-to-vial and measurements in multiple solvents including water20-23 Previously we used these vials to perform preliminary measurements of cyanide (CN) methylphosphonic acid (MPA) and dipicolinic acid (DPA) MPA is a hydrolysis product of the nerve agents (eg sarin Reaction 1) and may be a valuable indicator of nerve agent usage particularly since the alkyl methylphosphonic acids are relatively more stable than their corresponding parent complexes24 DPA is
SPIE -2003-5269 118
a chemical signature of spore forming bacteria such as Bacillus anthracis And in light of the inability to rapidly detect the anthrax spores distributed through the US mail in October 2001 a number of methods are being developed to extract and analyze this signature Reaction 1 Stepwise hydrolysis of Sarin to form hydrofluoric acid (HF) isopropyl methylphosphonic acid (IMPA) then methyl phosphonic acid (MPA) and 2-propanol In our previous SERS investigations MPA and DPA were measured at 50 and 100 mgL respectively In both cases limits of detection (LOD) were estimated at 100 microgL providing encouragement in that SERS may satisfy the needs of the JSAWM Since it has been shown that pH can substantially influence the intensity of SER bands25 which would clearly influence quantitative analysis we undertook the present study to determine the severity of these effects for cyanide methyl phosphonic acid and dipicolinic acid Furthermore we previously observed a band at 1050 cm-1 for MPA6 possibly due to an anion formed at basic pH Here we investigate the source of this spectral anomaly
2 EXPERIMENTAL All chemicals including potassium cyanide methyl phosphonic acid dipicolinic acid and those used to prepare the silver-doped and gold-doped sol-gel coated vials were obtained and used as received from Sigma-Aldrich All samples were prepared in HPLC grade water (Fischer Scientific Fair Lawn NJ) for SERS measurements The pH of these samples was adjusted using dilute nitric acid or potassium hydroxide and verified using a pH electrode (Corning Inc Corning NY) that had been calibrated with pH 400 700 and 1000 buffered standards from Fischer Scientific Once prepared the samples were transferred into the silver-doped sol-gel vials (Real-Time Analyzers Inc East Hartford CT) The vials were coated in a manner similar to that previously reported by adding ammonium hydroxide to a solution of silver nitrate tetramethyl orthosilicate (TMOS) and methanol20 Gold-vials were coated by adding nitric acid to a solution of gold tetrachloride TMOS and methanol The two precursor solutions were prepared mixed and transferred to 2-mL glass vials dried and heated After sol-gel formation the incorporated metal ions were reduced with dilute sodium borohydride (1mgmL) which was followed by a water wash to remove residual reducing agent After the resultant analyte solution was introduced the SER-active vial was immediately fixed horizontally to an XY positioning stage (Conix Research Springfield OR) just inside the focal point of an f07 aspheric lens The lens focused the beam into the sample and collected the scattered radiation back along the same axis A dichroic filter (Omega Optical Brattleborough VT) was used to reflect the excitation laser to the lens and pass the Raman scattered radiation collected by the lens An f2 achromat was used to collimate the laser beam exiting a 200 microm core diameter source fiber optic while a second f2 achromat was used to focus the scattered radiation into a 365 microm fiber optic (Spectran Avon CT) A short pass filter was placed in the excitation beam path to block the silicon Raman scattering generated in the source fiber from reflecting off sampling optics and reaching the detector A long pass filter was placed in the collection beam path to block the sample Rayleigh scattering from reaching the detector A 785 nm diode laser (Process Instruments Inc model 785-600 Salt Lake City UT) was used to deliver 100 to 150 mW of power to the sample A Fourier transform Raman spectrometer (Real-Time Analyzers model IRA-785 East Hartford CT) and a silicon photo-avalanche detector (Perkin Elmer model C30902S Stamford CT) were used to acquire the SER spectra
3 RESULTS AND DISCUSSION In a previous study of MPA6 aimed at developing a concentration calibration curve and determining limits of detection (LOD) we observed an anomaly at 1050 cm-1 Since it was found that the band intensity changed as a function of concentration the band must be associated with a sample parameter Two possible parameters photon flux and pH are examined here The first parameter was investigated by irradiating a 1mgmL MPA sample in a SER-active vial with laser powers of 200 mW and above and monitoring spectral changes It was immediately found that the 1050 cm-1 band
2O+ H HF + +OH
OH3H C
OP
OF
CH
CH3
3
3H C
OCP
OOH
CH
CH3
3
3H C
O
CPHO
CH
CH
3
3
C
Sarin IMPA MPA 2-propanol
SPIE -2003-5269 119
appeared and grew as a function of time and that the higher the laser power the faster the growth Figure 1 shows the growth of the 1050 cm-1 band over the course of 30 minutes when using 150 mW of 785 nm excitation while Figure 2A shows that the growth can be fit with a first order exponential equation namely I1050 = 03+05e-013t Furthermore the 760 cm-1 band could be fit with a first order decay equation with an identical exponential rate constant ie I760 = 08-
08e-013t The rates represent classical first order kinetics and their correspondence allows one to conclude that MPA is being transformed one-for-one into a photo-generated product At this time the photoproduct has not been positively identified but phosphonic (phosphorous) acid and phosphonate are likely candidates since the symmetric P(OH)3 stretch occurs at ~1050cm-1 Our studies show that a reduction in laser power to 100 mW at the sample for MPA essentially eliminates this degradation process This laser power or lower was used for further measurements
Methyl phosphonic acid is a diprotic acid that stepwise dissociates into two anions MPA- and MPA= according to the following reactions26
MPA MPA- + H+ pKa1 = 212 Reaction 2
MPA- MPA= + H+ pKa2 = 729 Reaction 3 The relative concentrations of MPA MPA- and MPA= can be determined at any pH by expressing [MPA] and [MPA=] in terms of [MPA-] using Reactions 2 and 3 and summing all three to equal the total starting concentration here 2 mgmL (0021M MW = 9602) viz
[MPA] + [MPA-] + [MPA=] = 0021M Equation 1 substituting from Reactions 1 and 2
([H+][MPA-])K1a + [MPA-] + (K2a[MPA-])[H+] = 0021M Equation 2 rearranging [MPA-] = 0021M(1+[H+]K1a + K2a[H+]) Equation 3 The relative concentrations of MPA MPA- and MPA= as a function of pH are shown in Figure 3 It is worth noting that near neutral pH both MPA- and MPA= will be present To confirm that the SER signal followed this pH dependence a starting solution consisting of 20 mg of MPA in 10 mL HPLC grade water was prepared and brought to pH of 20 using dilute nitric acid From this solution 2 mL were added to a SER-active vial and the SER spectrum recorded At this pH a peak at 760 cm-1 was barely discernable The 2 mL solution was returned to the starting solution and the pH was re-measured to correct for any changes that the silver-doped sol-gel vials might cause In most cases the change was less than 02 pH units and the pH is reported as the before and after average Next the pH of the
Figure 2 A) Exponential growth of 1050 cm-1 band and B) exponential decay of 760 cm-1 band for spectral series in Figure 1
Figure 1 Growth of 1050 cm-1 band as a function of time due to exposure to 150 mW of 785 nm Spectra are 5 sec each collected every 100-sec from 0 to 30-min
B
570 770 970 1170 1370Raman Shift (cm-1)
Arbit
rary
Unit
s
MPA Photodegradation
600 800 1000 1200 1400 Wavenumber (cm-1)
A
0 10 20 30 0 10 20 30 time (min) time (min)
Ram
an In
tens
ity (r
elat
ive)
30
min
0
SPIE -2003-5269 120
starting solution was adjusted to 325 using dilute KOH Again 2 mL were added to a vial and the SER spectrum recorded At this pH a reasonably strong 760 cm-1 band was observed This process was repeated as spectra were recorded at pHs of 70 74 75 79 85 and 100 A total of 1 mL of KOH was added diluting the total concentration by 10 Next the pH of the starting solution was made acidic by adding dilute nitric acid dropwise This time spectra were recorded at pHs of 72 69 64 and 37 Figure 4 shows the SER spectra for representative pHs (spectra were left out to simplify the figure) while Figure 3 shows the 760 cm-1 peak intensities as a function of pH (The band intensities were adjusted to compensate for dilution effects caused by the addition of HNO3 and KOH then normalized to 0021 M for the most intense band observed at pH 37) It is clear from Figure 3 that the 760 cm-1 band follows the MPA- concentration as a function of pH and must be assigned to this anion No bands were observed that corresponded to MPA or MPA= The lack of an MPA SER spectrum may be due to the absence of an attraction between the neutral analyte and the electropositive silver surface The same reasoning suggests that a strong SER spectrum should be observed for MPA= but it is not and a satisfactory explanation has not been found
0000
0005
0010
0015
0020
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14pH
Con
cent
ratio
n [M
]
pK1 = 212 pK2 = 729
MPA- MPA=
MPA
Figure 4 SER spectra of 002M MPA as a function of pH Conditions 100 mW of 785 nm 36 scans (1 min) 8 cm-1 recorded 2 min after sample introduction pH 19 and 20 not apparent on this scale pH 69 and 74 near identical to 70 and 75 and not shown for clarity
00
02
04
06
08
10
12
14
16
18
0 5 10 15 20 25 30Measurement Number
Ram
an In
tens
ity (7
60 c
m-1
)
Figure 3 Concentration dependence of MPA MPA- and MPA= as a function of pH for a 002M sample Intensity of 760 cm-1
band from Figure 6 as a function of pH ( for increasing basic adjustment for increasing acidic adjustment error was measured at ~10 for pH 64)
Figure 5 SER spectra of 002M MPA at pH 64 measured around a vial at three heights (9 points per height) Conditions as in Figure 4 but 10-sec scans
Since these measurements involved the removal and replacement of the SER-active vial in the sample holder to remove and add sample variation in the intensity as a function of vial position was minimized by illuminating the exact same height along the vial wall But this does not account for variability of the SERS response of the sol-gel coating around the vial To analyze this effect a vial containing MPA at pH 64 was rotated at ~ 40o
intervals at the original height and 18rdquo above and below this value Figure 5 shows the intensity of the 760 cm-1
band for the 27 positions It was found that the average value was 137plusmn014 an RSD of 10 overall and 5 for each height An error bar is included in Figure 4 for the pH 64 measurement
SPIE -2003-5269 121
With the above analyses in mind a preliminary investigation of the SER spectral response for MPA (as MPA-) as a function of concentration was performed A single vial was used for these measurements beginning with 1 mgL followed by measurements of 10 100 and 1000 mgL In all cases the pH was ~7 and 3 positions around the vial were measured per concentration Since the 760 cm-1 band was not observed for concentrations of 1 or 10 mgL using 100 mW of 785 nm the laser power at the sample was raised to 200 mW beginning with the 10 mgL concentration Photo-degradation was largely avoided (and not observed) by exposing the sample for only 33 seconds per spectral acquisition Representative spectra for 10 100 and 1000 mgL are shown in Figure 6 while a plot of the 760 cm-1 band intensity as a function of concentration is shown in Figure 7 These values were also used to estimate limits of detection based on the signal-to-noise ratio (SN) of the 760 cm-1 band using 100 mW of 785 nm laser excitation and a 10-min acquisition time As summarized in the Figure 7 inset the lower the measured concentration the lower the predicted LOD Even if as estimated 210 microgL could be measured using the silver-doped sol-gel vials an improvement of a factor of 70 is still required to meet JSAWM goals of measuring 3 microgL in 10-minutes Similar to MPA DPA is a diprotic acid (pKa1 = 216 and pKa2 = 692) and variations in pH will effect the relative concentrations of DPA DPA- and DPA= and possibly the SER spectra and band intensities This could prove significant if an acid or base is used to denature anthrax spores with the goal of extracting and analyzing DPA The MPA pH study described above was mimicked for DPA except that the starting solution consisted of 20 mg of DPA in 20 mL HPLC grade water (60x10-3M MW = 1671) The initial solution had a pH of 245 which was made basic by dilute KOH to pHs of 355 433 487 559 1069 and 1166 SER spectra were recorded at each pH using 100 mW of 785 nm and a 44-sec acquisition time Next one drop of concentrated nitric acid was used to remake the solution acidic at a pH of 200 Again sequential pH measurements were performed at 383 510 735 and 822 The solution pH was made acidic a third time but to pH 219 171 then 135 Throughout this process no more than 20 drops of acid or base were added and therefore the concentration was diluted by no more than 10 Most of the spectral bands showed a minor decrease in intensity as a function of increasing pH values However the bands at 525 and 795 cm-1 showed the most dramatic changes which occurred at acid pH Figure 8 shows the SER spectra of DPA for the spectral region and pH range of interest The identity of the DPA species was determined by plotting the normalized peak intensities with the lowest value set to 0 and the highest to 0006 M as a function of pH and overlaying these values on a plot of the relative concentrations for DPA DPA- and DPA= as previously done for MPA (Figure 9) As can be seen the 525 cm-1 band clearly corresponds to DPA The correspondence of the 795 cm-1 band to this species is less clear as the band retains intensity until pH of 55 This can be attributed to contribution to the overlapping band at 810 cm-1 which does not change as a function of pH The fact that most bands are observed at all pHs suggest that the primary interaction with silver is through the ring nitrogen This is supported by the fact that the most intense band occurs at 1008 cm-1 attributed to a symmetric ring breathing mode and that this interaction has been characterized for pyridine in numerous papers27
0
1
2
3
4
5
6
7
0 200 400 600 800 1000 1200
MPA Concentration (mgL)
760
Ban
d In
tens
ity
Figure 6 SER spectra of MPA in water at A) 1000 B) 100 and C) 10 mgL Conditions pH of 7 silver-doped sol-gel coated vial 200 mW of 785 nm 33-sec 8 cm-1 resolution
Figure 7 Plot of SER intensity of 760 cm-1 band of MPA as a function of concentration using 200 mW of 785 nm Inset table includes average intensity LOD standard deviation and percent deviation for each concentration but for 100 mW and 10 min
conc (mgL) LOD ave stddev dev10 021 002 1021100 072 010 14301000 312 040 1280
A
B C
SPIE -2003-5269 122
The ability to detect dipicolinic acid (DPA) by SERS is demonstrated in Figure 10 Even at 1 mgL the primary bands are visible The intensity of the dominant ring breathing mode observed at 1008 cm-1 for DPA in water (pH = 25 to 55) is plotted as a function of concentration in Figure 11 These values were also used to estimate limits of detection based on the SN of the 1008 cm-1 band using 100 mW of 785 nm laser excitation and a 10-min acquisition time Again the lower the measured concentration the lower the predicted LOD (see Figure 11 inset) and detection of 160 microgL is possible Gastrointestinal anthrax requires significant more spores than inhalation anthrax28 and a limit of detection might be placed at 1 million spores in 1 liter of water or 10 microgL Since each spore contains ~10 CaDPA by weight29 a goal for DPA might be 1 microgL indicating that the present measurements must be improved by nearly two orders of magnitude Due to the increased hazards of handling HCN gas KCN salt was used for these experiments Nevertheless all sample preparations were performed in a chemical hood KCN completely dissolves in water but its conjugate acid HCN is formed and has a Ka of 615x10-1030 viz
HCN CN- + H+ pKa = 921 Reaction 4
Figure 8 SER spectra of 1 mgmL DPA as a function of pH Conditions 100 mW 785 nm 100 scans (44-sec)
Figure 9 Concentration dependence of DPA DPA- and DPA= as a function of pH for a 0006M sample Intensity of 525 () and 795 (diams) cm-1 bands from Fig 8 as a function of pH
pH
135 171 219 383
0
0001
0002
0003
0004
0005
0006
0007
0 2 4 6 8 10 12 14pH
Con
cent
ratio
n [M
]
DPADPA-DPA=795525
DPA DPA=DPA-
pK2 = 692pK1 = 216
795 525
0
05
1
15
2
25
0 200 400 600 800 1000 1200
DPA Concentration (mgL)
1008
Ban
d In
tens
ity
conc (mgL) lod-10min-100mw1 017
10 016100 103
1000 355
Figure 10 SER spectra of DPA in water at A) 1000 B) 100 C) 10 and D) 1 mgL Conditions pH of 25-55 silver-doped sol-gel coated vial 175 mW of 785 nm 1-min 8 cm-1 D) has been multiplied by x10 to make bands visible
A
B C
D
Figure 11 Plot of SER intensity of 1008 cm-1 band of DPA as a function of concentration using 175 mW of 785 nm Inset table includes LOD in mgL for each concentration but for 100 mW and 10 min
SPIE -2003-5269 123
Consequently the cyanide concentration must be determined for each initial KCN concentration Specifically the samples prepared with concentrations of 01 1 10 100 and 1000 mgL of KCN produced CN- concentrations of 63x10-3 033 69 89 and 964 mgL at pHs of 816 90 967 102 and 107 respectively The pH dependence for the HCN and CN- concentrations are shown in Figure 12 Thus as the amount of KCN added to the solution decreases so does the pH of the solution (becomes less basic) and according to Reaction 4 the relative amount of CN- to HCN also decreases For example in the preparation of a 01 mgL solution of KCN the pH is shifted from 7 for pure water to only 816 and only 63 of the starting material becomes CN- or 63x10-3 mgL In comparison for a solution of 1000 mgL the pH is shifted from 7 to 107 and 96 of the starting material becomes CN- This is significant since the cyanide ion is better able to adsorb onto the silver particles and become SERS active SER spectra of 10 100 and 1000 mgL of KCN samples are shown in Figure 13 and a corresponding plot of the 2100 cm-1 band intensity as a function of concentration is shown in Figure 14
000102030405060708091011
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14pH
Con
cent
ratio
n [m
gm
L]
pKa = 921
CN -HCN
The SER spectra of cyanide are dominated by the single CN stretch at 2100 cm-1 which occurs in normal Raman spectra of solutions at 2080 cm-1 However a low frequency mode occurs at 135 cm-1 due to a Ag-CN stretch (not shown) indicative of a strong surface interaction It is also observed that as the concentration decreases the CN stretch shifts to 2140 cm-1 This shift has been attribute to the formation of a tetrahedral Ag(CN)3
2- surface structure31 as well
Figure 14 Concentration dependence of KCN SERS measured under conditions in Fig 11 Concentrations are 1 01 and 001 mgml Intensities are measured for the CN stretch at 2100 cm-1 Inset table includes LOD in mgL for each concentration in Figs 13 and 15 but for 100 mW and 10 min
Figure 13 SER spectra of KCN in water at A) 1000 B) 100 and C) 10 mgL Conditions pHs of 107 102 and 97 silver-doped sol-gel coated vial 100 mW of 785 nm 1-min 8 cm-1 C) has been multiplied by x10 to make band visible
Figure 12 Concentration dependence of HCN and CN- as a function of pH for a 1 mgmL sample Calculated intensity of 2100 () cm-1 band for a 1 mgml sample at pHs of 816 90 967 102 and 107
A
B
C
0
20
40
60
80
100
120
140
0 200 400 600 800 1000 1200CN Concentration (mgL)
2100
Ban
d In
tens
ity
Figure 15 SER spectra of KCN in water at A) 10 B) 1 and C) 01 mgL Conditions pHs of 97 12 and 12 gold-doped sol-gel coated vial 100 mW of 785 nm 1-min 8 cm-1 C) has been multiplied by x10 to make band visible
A
B
C
conc (mgL) lod-10min-100mw Condition01 001 Au-pH 121 007 Au-pH 12
69 003 Au-pH 9769 016 Ag-pH 9789 022 Ag-pH 102946 113 Ag-pH 107
SPIE -2003-5269 124
as to CN adsorbed to two different surface sites32 Alternatively the 2140 cm-1 band could be attributed to HCN since this species dominates at lower concentrations However it is unlikely that this species would be attracted to the electropositive silver surface Further both peaks should be present at pHs between 85 and 105 but this is not observed It has also been suggested that at concentrations near and above monolayer coverage the CN- species is forced to adsorb end-on due to crowding and at lower concentrations the molecule can reorient to lie flat33 This suggests that the 2100 and 2140 cm-1 bands correspond to the end-on and flat orientations respectively As Figures 13 and 14 show the intensity of the CN stretch for the 89 mgL sample is nearly as intense as the 964 mgmL band This suggests that the Raman signal for the flat orientation is more enhanced However more extensive measurements are required to verify this point Since resent research has suggested that cyanide may be more effectively detected on gold measurements of KCN solutions were also performed using gold-doped sol-gel vials Preliminary measurements are shown in Figure 15 for samples prepared from 01 1 and 10 mgL KCN Since the pHs are 816 90 967 the resultant CN- concentrations are 63x10-3 033 and 69 mgL Initially only the highest concentration was observed and the signal intensity was significantly better than the equivalent concentration measured using silver In an effort to shift Reaction 4 to the left transforming HCN to CN- (Le Chatelierrsquos principle) KOH was added to the lower concentration samples producing solutions with pH 12 This effectively forces all of the cyanide in solution to be CN- or 01 and 10 mgL respectively More importantly the CN stretch is now observed in the SER spectra The band appears at 2125 cm-1 as has been previously reported for gold12 As calculated for MPA and DPA LODs can be estimated from this data For the three concentrations of cyanide on silver the LODs are 016 to 11 mgL for 100 mW of 785 nm laser excitation and a 10-min acquisition time For gold pH adjusted the LODs are10 to 70 microgL an improvement of more than 10 times silver Nevertheless either substrate is sufficient to meet the JSAWM goals of measuring 3 mgL in 10-minutes as the requirements form cyanide are much less stringent than the nerve agents
4 CONCLUSIONS Here we examined the surface-enhanced Raman spectral response for methyl phosphonic acid dipicolinic acid and cyanide as a function of pH It was determined that the most prominent peak at 760 cm-1 reaches a maximum intensity between pH 3 and 7 and corresponds to the MPA- species Neither the MPA nor MPA= species appear to generate a SER spectrum and consequently no spectra were observed below pH 2 or above pH 8 In this study we also found that higher laser powers could cause photodegradation of MPA signified by the exponential growth of a band at 1050 cm-1 which is tentatively assigned to phosphorous acid Unlike MPA DPA was observed at all pHs This is attributed to the dominant interaction of the pyridine functional group with silver Minor spectral changes were observed at acid pHs and were assigned to neutral DPA Like MPA SER spectra of cyanide were pH dependent No spectra were observed for the HCN species while CN- was best observed at pHs more basic than 8 Preliminary concentration studies for the three analytes allowed estimating limits of detection for MPA DPA and CN using 100 mW of 785 nm and a 10-min acquisition time of 210 165 and 70 microgL respectively Although the latter value suggests that cyanide can be measured at sufficiently low concentrations to meet JSAWM goal improvements by 100 to 200 times are required for MPA and DPA It is clear from this study that pH of the sample is important and must be taken into account when developing concentration calibration curves as well as fieldable analyzers Future work will include tailoring the sol-gel with specific functional groups to dictate sample pH and thereby optimize sensitivity
ACKNOWLEDGMENTS The authors are grateful for the support of the US Army (DAAD13-02-C-0015 Joint Service Agent Water Monitor program) The authors would also like to thank Dr Steve Christensen of the US Army SBCCOM for helpful discussions and Mr Chetan Shende of Real-Time Analyzers for assistance in development of the gold-doped sol-gels
REFERENCES 1 Erickson B Analytical Chemistry News amp Features June 1 397A (1998)
SPIE -2003-5269 125
2 Johnston RL Hoefler CM Fargo JC and Moberley B AT-ONSITE 5-8 (1994) 3 Black RM RJ Clarke RW Read and MT Reid J Chromatography 662 301-321 (1994) 4 USA Army Center for Health Promotion and Preventive Medicine (USACHPPM) Technical Guideline 230A 5 Jensen JL NC Wong W R Loerop SPIE 4775-03 (2001) 6 Hoenig SL Handbook of Chemical Warfare and Terrorism Greenwood Press Wesport CT (2002) 7 Munro NB SS Talmage GD Griffin LC Waters AP Watson JF King and V Hauschild Env Health
Persp 107 933-974 (1999) 8 Holstege CP Kirk M Sidell FR Crit Care Clin13 923-42 (1997) 9 Farquharson S P Maksymiuk K Ong and S Christesen SPIE 4577 166-173 (2001) 10 Lee Y and S Farquharson SPIE 4378 21-26 (2001) 11 Spencer KM J Sylvia S Clauson and J Janni SPIE 4577 158-165 (2001) 12 Tessier P S Christesen K Ong E Clemente A Lenhoff E Kaler and O Velev Applied Spectroscopy 56
1524-1530 (2002) 13 Farquharson S WW Smith S Elliott and JF Sperry SPIE 3533 207-214 (1998) 14 Farquharson S WW Smith S Elliott and JF Sperry SPIE 3855110-116 (1999) 15 Farquharson S WW Smith YH Lee S Elliott and JF Sperry SPIE 4575 62-72 (2002) 16 Guzelian AA J Sylvia J Janni S Clauson and KM Spencer SPIE 4577 182-192 (2001) 17 Shende C F Inscore A Gift P Maksymiuk and S Farquharson in press 18 Weaver MJ S Farquharson and MA Tadayyoni J Chem Phys 82 4867-4874 (1985) 19 Alak AM and T Vo-Dinh Analytical Chemistry 59 2149-2153 (1987) 20 Lee Y and S Farquharson SPIE 4206 140-146 (2000) 21 Farquharson S and Y Lee SPIE 4200-16 (2000) 22 Lee Y S Farquharson and P M Rainey SPIE 3857 76-84 (1999) 23 Lee Y S Farquharson H Kwong and M Shahriari SPIE 3537 252-260 (1998) 24 Wang J M Pumera G Collins and A Mulchandani Analytical Chemistry 74 6121-6125 (2002) 25 Dou X YM Jung Z-Q Cao and Y Ozaki Applied Spectroscopy 53 1440-1447 (1999) 26 Data supplied by S Christesen and K Ewing 27 Kerker M and B Thompson Eds SPIE MS 10 (1990) 28 Inglesby TV DA Henderson JG Bartlett JAMA 287 2236 (2002) 29 FW Janssen AJ Lund and LE Anderson Science 127 26 (1958) 30 Lide DR Ed Handbook of Chemistry and Physics CRC Press 77th Ed (1996-7) 31 Billmann J G Kovacs and A Otto Surf Sci 92 153 (1980) 32 Murray CA and S Bodoff Phys Rev B 32 671 (1985) 33 Kellogg D and J Pemberton J Phys Chem 91 1120 (1987)
SPIE-2003-5269 16
Chemical agent detection by surface-enhanced Raman spectroscopy
Stuart Farquharson Alan Gift Paul Maksymiuk Frank Inscore and Wayne Smith
Real-Time Analyzers 87 Church Street East Hartford CT 06108
Kevin Morrisey and Steven D Christesen US Army SBCCOM Aberdeen Proving Ground MD 21010
ABSTRACT
In the past decade the Unites States and its allies have been challenged by a different kind of warfare exemplified by the terrorist attacks of September 11 2001 Although suicide bombings are the most often used form of terror military personnel must consider a wide range of attack scenarios Among these is the intentional poisoning of water supplies to obstruct military operations in Afghanistan and Iraq To counter such attacks the military is developing portable analyzers that can identify and quantify potential chemical agents in water supplies at microgram per liter concentrations within 10 minutes To aid this effort we have been investigating the value of a surface-enhanced Raman spectroscopy based portable analyzer In particular we have been developing silver-doped sol-gels to generate SER spectra of chemical agents and their hydrolysis products Here we present SER spectra of several chemical agents measured in a generic tap water Repeat measurements were performed to establish statistical error associated with SERS obtained using the sol-gel coated vials Keywords Chemical agents chemical agent detection SERS Raman spectroscopy
1 INTRODUCTION
In the autumn of 2001 terrorism within US borders became a sobering reality While extensive efforts are being implemented to secure the homeland US and Coalition forces in Afghanistan and Iraq are constantly faced with terrorist attacks In addition to car-bombings the military has identified several non-traditional attack scenarios including poisoning of water supplies by chemical warfare agents (CWAs) To counter this threat the Department of Defense has been investigating numerous approaches to detect such attacks under the auspices of the Joint Service Agent Water Monitor (JSAWM) program1 The JSAWM must be capable of identifying the chemical agents with no false-positives and quantifying the agents at microgram per liter concentrations within 10 minutes2 This includes the analysis of drinking water supplies distribution and storage systems Currently colorimetric paper is used to detect agents on-site while gas chromatography combined with mass spectrometry (GCMS) is used in mobile support laboratories However both methods have severe drawbacks The paper changes color in response to contact with many chemicals besides CWAs causing a high incidence of false positives3 while GCMS although very chemically specific requires up to an hour to perform and regular re-calibration456 Vibrational spectroscopy has also been investigated as a potential method of rapidly detecting CWAs7-
11 as early as 1985 in the case of infrared spectroscopy8 and 1988 in the case of Raman spectroscopy11 Again however these techniques also have limitations when it comes to measuring trace poisons in water Infrared spectra would be dominated by the very strong absorption of water which would obscure absorptions by most other chemicals present Whereas Raman spectroscopy is simply not a very sensitive technique and detection limits are typically grams per liter Surface-enhanced Raman spectroscopy (SERS) offers several advantages over conventional vibrational methods and that may provide the necessary sensitivity required for detecting trace quantities of chemical agents in water SERS employs the interaction of surface plasmon modes of metal particles with target analytes to increase scattering efficiency by as much as 1 million times12 In 1987 the potential of this approach to measure CWAs was demonstrated by measuring a series of pesticides13 Several of these organophosphonates have chemical structures similar to CWAs in particular P=O functional groups In the past few years we and others have further explored the ability of SERS to detect CWAs14-17 and even bioagents 18-21 We have been employing silver-doped sol-gels to promote the SER effect
SPIE-2003-5269 17
in these studies The porous silica network of the sol-gel matrix offers a unique environment for immobilizing and stabilizing SER-active metal particles22-25 The silver-doped sol-gels have been coated on the internal walls of standard glass vials to allow rapid SER analysis of small sample quantities without preparation We have measured over 100 chemicals with enhancements of 104 to 106 demonstrated reversible measurements in a flowing system reproducible measurements from vial-to-vial and measurements in multiple solvents including water21-25 Previously we used these vials to perform preliminary measurements of cyanide (CN) and methylphosphonic acid (MPA) Most of the nerve agents form MPA during hydrolysis while Tabun forms CN a chemical agent in its own right In another paper including in these proceedings we examined the limits of detection (LOD) for MPA by measuring a series of concentrations down to 50 mgL and estimated a limit of detection of 100 microgL26 These measurements provide encouragement in that SERS may satisfy the needs of the JSAWM To further establish the viability of SERS in particular silver-doped sol-gels here we present analysis of cyanide mustard and VX in tap water The measurements performed at the US Armyrsquos Edgewood Chemical Biological Center Aberdeen MD also included numerous repeat measurements to establish reproducibility
2 EXPERIMENTAL 2a General All chemicals including potassium cyanide 2-chloroethylethyl sulfide and those used to prepare the silver-doped and gold-doped sol-gel coated vials were obtained and used as received from Sigma-Aldrich All samples were prepared in a chemical hood using HPLC grade water unless otherwise noted (Fischer Scientific Fair Lawn NJ) for SERS measurements Once prepared the samples were transferred into the silver-doped sol-gel vials (Real-Time Analyzers Inc East Hartford CT) The vials were coated in a manner similar to that previously reported by adding ammonium hydroxide to a solution of silver nitrate tetramethyl orthosilicate (TMOS) and methanol22 The two precursor solutions were prepared mixed and transferred to 2-mL glass vials dried and heated After sol-gel formation the incorporated metal ions were reduced with dilute sodium borohydride (1mgmL) which is followed by a water wash to remove residual reducing agent After the resultant analyte solution was introduced the SER-active vial was fixed horizontally to an XY positioning stage (Conix Research Springfield OR) just inside the focal point of an f07 aspheric lens The lens focused the beam into the sample and collected the scattered radiation back along the same axis A dichroic filter (Omega Optical Brattleborough VT) was used to reflect the excitation laser to the lens and pass the Raman scattered radiation collected by the lens An f2 achromat was used to collimate the laser beam exiting a 200 microm core diameter source fiber optic while a second f2 achromat was used to focus the scattered radiation into a 365 microm fiber optic (Spectran Avon CT) A short pass filter was placed in the excitation beam path to block the silicon Raman scattering generated in the source fiber from reflecting off sampling optics and reaching the detector A long pass filter was placed in the collection beam path to block the sample Rayleigh scattering from reaching the detector A 785 nm diode laser (Process Instruments Inc model 785-600 Salt Lake City UT) was used to deliver 100 to 150 mW of power to the sample A Fourier transform Raman spectrometer (Real-Time Analyzers model IRA-785 East Hartford CT) and a silicon photo-avalanche detector (Perkin Elmer model C30902S Stamford CT) were used to acquire the SER spectra
2b Edgewood Chemical Biological Center The surface-enhanced Raman spectral measurements at the US Armyrsquos Chemical Biological center presented here were all performed on September 12 2003 To expedite measurements a plate was machined to hold up to 12 SER-active sample vials (Figure 1) The plate fit a standard XY plate reader that could be programmed Pure KCN bis-(2-chloroethyl)sulfide (distilled mustard gas HD) and ethyl S-2-diisopropylamino ethyl methylphosphonothioate (VX) were obtained on-site and used to prepare 1 gL tap water solutions in a chemical hood with appropriate safety equipment Simulated tap water was prepared by adding 10 chemicals most often found in tap water at appropriate concentrations to distilled water (Table 1) SER measurements were also performed in a chemical hood For added safety the FT-Raman instrument was placed outside the laboratory and 30 foot fiber optic and electrical cables were used to allow remote SERS measurements and plate manipulation For each experiment 1gL samples were prepared and added to 9 individual vials which were then loaded on the plate In some cases a tenth vial was included as a blank
Table 1 Generic Tap Water Composition
Compound mgL NaHCO3 100 CaSO4 27 MgSO4bull7H2O 67 NaNO3 1 Fulvic Acid 1 K2HPO4 07 KH2PO4 03 (NH4)2HSO4 001 NaCl 001 FeSO4 0001 pH 76-78
SPIE-2003-5269 18
A software program was written that allowed selecting the sequence that the vials were measured the number of positions along the length of the vials to measure (1 to 5) and the number of scans to co-add During sample analysis the program displayed the vial being analyzed the point being analyzed and the spectrum as it was being acquired Once all the data was collected a second software program was written to rapidly analyze the data The spectra collected for all the vials on a plate could be loaded at one time and then the spectra for each point could be displayed simultaneously or separately The user could then select the Raman peak to analyze in terms of peak height or area This was accomplished by selecting points on either side of the peak to define a baseline of zero The peak height or area could then be computed for all of the spectra loaded and then exported to a spreadsheet for statistical analysis
Figure 1 A) Vial Holder 6 slots to hold 2 vials each end-to-end B) Measurement Configuration Program user selects vials to measure sequence number of points per vial (1 to 5) and number of scans per point C) Spectral Acquisition Program shows spectrum being collected which vial and position D) Spectra Analysis Program user selects spectra to analyze by plate vial and point (s) as well as two wavenumbers defining the peak and the baseline to subtract The image is of 5 repeat measurements of 10 mgL KCN in generic tap water 16 sec each 100 mW of 785 nm
3 RESULTS AND DISCUSSION Raman and surface-enhanced Raman spectra were obtained for potassium cyanide bis-(2-chloroethyl)sulfide and ethyl S-2-diisopropylamino ethyl methylphosphonothioate representing three classes of chemical agents cyanides mustards and nerve agents respectively Spectra were also obtained for 2-chloroethyl ethyl sulfide (CEES) a structural analogue to HD which was included in the study to aid in assigning spectral bands KCN salt was used for cyanide experiments to avoid the increased hazards of handling HCN gas KCN completely dissolves in water forming its conjugate acid HCN according to its Ka of 615x10-1027 and at a concentration of 1 mgmL results in a pH 107 solution This is significant in that only CN- appears to interact sufficiently with silver to produce a SER spectrum and no spectral signal is observed below pH 726 Figure 2 shows the SER and normal Raman spectra for KCN The SERS spectrum of 1mgml KCN in water shows a single intense somewhat broad feature at
A B
C D
SPIE-2003-5269 19
2100 cm-1 assigned to the single CequivN stretch The band is much sharper in the normal Raman spectra of the solid KCN salt at 2074 cm-1 This band does broaden and shift to 2080 cm-1 in solution (not shown) However the observed SERS frequency is attributed to interaction with silver and low frequency mode at 135 cm-1 attributed to a Ag-CN stretch (not shown) supports this conclusion
cm-1 band to a C-S stretch but the authors concede that it is in fact more likely a C-Cl stretch28 It appears that the most intense Raman bands at 648 692 and 747 cm-1 shift to 620 660 and 720 cm-1 in the SER spectra and are tentatively assigned as above The width of these bands suggests that they overlap underlying spectral features Additional bands in the Raman spectra occur at 972 1034 1049 1263 1286 1423 1442 2865 2935 and 2960 cm-1 Corresponding bands occur in the SER spectra at 964 1015 1054 1286 1410 1447 2865 and 2935 cm-1 Most of these bands are associated with alkane modes specifically the bands at approximately 1040 cm-1 to a C-C stretch 1290 cm-1 to a CH2 in-phase twist 1440 cm-1 to a CH2 wag 2865 cm-1 to a symmetric CH2 stretch and 2965 cm-1 to an asymmetric CH2 stretch The Raman and SER spectra of sulfur mustard were measured at the Edgewood center (Figure 4) Both spectra are largely similar to CEES The C-Cl and C-S bands in the Raman spectrum of HD now occur at 640 655 700 739 and 760 cm-1 and are more resolved possibly due to the increased molecular symmetry Theoretical calculations indicate that the first three bands are due to C-Cl stretching modes and the latter two to C-S stretching modes28 Only the C-Cl bands maintain significant intensity in the SER spectra occurring at 624 and 643 cm-1 which is attributed to the
Prior to measurements of HD CEES was examined by Raman and SER spectroscopy (Figure 3) CEES also known as half-mustard is essentially identical to HD except one of the chlorine end atoms is replaced by a hydrogen atom Again although not as toxic as HD CEES is a blister agent and dilute aqueous samples were prepared using appropriate safety equipment Both the Raman and SER spectra of CEES are similar and dominated by bands between 600 and 800 cm-1 These are associated with C-Cl and C-S stretching modes which are tentatively assigned to 648 and 747 cm-1 in the Raman spectra respectively The shoulder at 630 cm-1 the overlapped band at 660 cm-1 and the strong band at 692 cm-1 could also be due to these modes or their asymmetric counterparts It is worth noting that theoretical calculations assign the 692
Figure 2 A) SER and B) NR spectra of KCN Conditions A) 1 mgml in tap water 100 mW of 785 nm at sample 1-min acquisition time B) solid 300 mW of 785 nm 5-min All spectra are 8 cm-1 resolution
A
B
Figure 3 A) SER and B) NR spectra of CEES Conditions A) 1 vv (10 mgml) in MeOH 100 mW of 785 nm 1-min acquisition time B) neat 300 mW of 785 nm 5-min
A
B
Cl-CH2-CH2-S-CH2-CH3
A
B
Figure 4 A) SER and B) NR of HD Conditions A) 1mgml in tap water B) pure both 100 mW of 785 nm 1-min
Cl-CH2-CH2-S-CH2-CH2-Cl
SPIE-2003-5269 20
expected strong interaction between chlorine and silver and adds support to the assignment of this band to a C-Cl stretch Weaker overlapping bands occur at 670 692 and 724 cm-1 the latter possibly due to C-S stretching modes Again the alkane modes are apparent in the normal Raman spectra of HD but only a broad feature at 1300 to 1450 cm-1 suggests CH2 contributions in the SER spectrum Although the observed bands in the VX spectrum have not been assigned (Figure 4) a computer generated Raman spectrum29 predicts many of the same features with surprising accuracy and are used here Two intense bands at 460 and 530 cm-1 closely match predicted bands at 463 and 546 cm-1 assigned to a CH3-P=O bend and a PO2CS wag Three highly overlapped bands occur at 694 745 and 771 cm-1 matching predicted bands at 713 730 and 760 cm-1 The first
Table 2 Measured SER peak heights for the CN stretch at 2100 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
Figure 5 A) SERS and B) NR spectra of VX Conditions A) 1 vv (10 mgml) in MeOH B) pure sample both 100 mW of 785 nm 1-min acquisition time
A
B
two have been assigned to a C-S stretch and CH2bend respectively while the latter has been attributed to either a P-C stretch or an O-C-C stretch Although the 745 cm-1 band may alternatively be assigned to a C-S stretch based on the previous measurements of CEES and HD The relatively intense bands at 890 1106 1218 1445 and 1465 cm-1 also match predicted bands at 880 1108 1216 1440 and 1464 cm-1 that are assigned to a C-C stretch CH3 rock N-C3 stretch various C-H3 bends and C-H bends respectively Both the computer generated and the measured spectra contain numerous other less intense bands One is worth mentioning A unique band appears at 370 cm-1 that is predicted at 368 cm-1 and corresponds to an O-P=O bend The surface-enhanced Raman spectrum of VX is also rich with spectral features It has the unique low frequency band at 370 cm-1 as well as a second band at 380 cm-1 that is assigned tothe S-P-O bend predicted in the normal Raman spectrum at 388 cm-1 Based on the measured and predicted normal Raman spectra the following SERS assignments are given 460 cm-1 to the CH3-P=O bend 544 cm-1 to the PO2CS wag 738 cm-1 to a C-S stretch (based on arguments above) 890 cm-1
to a C-C stretch 1101 cm-1 to a CH3 rock and 1456
cm-1 to a C-H bend The ability of SERS to measure chemical agents in water containing real-world chemical interferents was tested by using the generic tap water described in Table 1 The ability to reproduce measurements was accomplished by preparing three separate water stock solutions which were used to prepare three sample solutions each of 1 mgmL KCN in the generic tap water The samples defined as vials 1-9 were then measured at 4 points per vial in a semi-random fashion such that errors associated with stock solution preparation and errors associated with instrument drift could be identified No trends were apparent that signified such systematic errors Each spectrum collected consisted of 20 averaged scans taking 16 seconds at 8 cm-1 resolution The laser power at the sample was measured periodically during the day and it ranged from 102 to 105 mW spacccc
spot Vial 1 Vial 2 Vial 41 13754 13004 128192 13519 12692 12909
1 3 13541 12721 126394 13462 12648 12651
ave 13569 12766 12755Vial 6 Vial 3 Vial 5
1 11586 14068 10757 2 11236 14402 115122 3 11376 14546 11559
4 10894 11763 11214ave 11273 13695 11261
Vial 7 Vial8 Vial 91 11198 15804 11023
2 11243 15716 88753 3 11562 15216 11218
4 11655 15084 11077ave 11415 15455 10548
CN AVG STDEV ERR1 mgmL pk ht 12526 1560 1245
stock solution
SPIE-2003-5269 21
The CN measurements consisting of 36 data points produced an average height of 125 for the 2100 cm-1 peak with a standard deviation of 156 or 125 (Table 2) The HD and VX measurements were performed precisely the same way (Tables 3 and 4) For HD the 624 cm-1 peak was used for analysis and it had an average height of 53 with a standard deviation of 068 or 129 while for VX the 544 cm-1 peak was used for analysis and it had an average height of 1051 with a standard deviation of 308 or 293 The greater error in the VX measurements can be somewhat attributed to Vial 4 which produced lower SER signal intensities But removing this vial from the data set changes the standard deviation to 233 only a modest improvement
spot Vial 1 Vial 2 Vial 41 525 609 4682 45 675 484
1 3 527 644 5094 559 807 669
ave 51525 68375 5325Vial 6 Vial 3 Vial 5
1 521 574 575 2 536 53 4492 3 509 508 451
4 565 594 379ave 53275 5515 4635
Vial 7 Vial 8 Vial 91 457 589 413
2 583 497 4273 3 544 505 4
4 528 5 509ave 528 52275 43725
HD AVG STDEV ERR1 mgmL pk ht 530 068 1291
stock solution
spot Vial 1 Vial 2 Vial 41 1464 1034 4672 1485 99 61
1 3 1491 989 5684 1041 777 553
ave 137025 9475 5495Vial 6 Vial 3 Vial 5
1 1058 942 1293 2 697 121 965
2 3 727 1206 11124 689 1254 746
ave 79275 1153 1029Vial 7 Vial 8 Vial 9
1 1127 783 163 2 1358 812 1614
3 3 1371 908 15124 1097 875 1397
ave 123825 8445 153825
VX AVG STDEV ERR1 mgmL pk ht 1051 308 2925
stock solution
4 CONCLUSIONS In this paper we examined the ability of surface-enhanced Raman spectroscopy to reproducibly measure CN HD and VX in tap water without chemical interference Both normal and surface-enhanced Raman spectra were examined to select unique bands suitable to identify and quantify these chemical agents For SER measurements the 2100 cm-1 C-N stretch was used for CN the 624 cm-1 C-Cl stretch was used for HD and the 544 cm-1 PO2CS wag was used for VX It was determined that 1 mgmL samples of each of these chemicals measured 36 times in glass vials coated with a silver-doped sol-gel reproduced measurements with standard deviations of 125 129 and 293 It was further found that the 10 chemicals added to simulate generic tap water did not interfere with or alter the SER spectra It should be noted that the concentrations used in this study were considerably greater than those required by the JSAWM program Current work involves improving SER sensitivity and designing sampling systems with better reproducibility This includes the development of fractal silver and gold structures within the sol-gel matrix and the development of chemically selective sol-gels
ACKNOWLEDGMENTS The authors are grateful for the support of the US Army (DAAD13-02-C-0015 Joint Service Agent Water Monitor program) The authors would also like to thank Janet Jensen Ronald Crosier and Kristina Gonser for helpful discussions
Table 4 Measured SER peak heights for the VX band at 544 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
Table 3 Measured SER peak heights for the HD band at 624 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
SPIE-2003-5269 22
REFERENCES 1 Jensen JL NC Wong W R Loerop SPIE 4775-03 (2001) 2 JSAWM Requirements at wwwsbccomapgeaarmymilRDAecbcrtPRODSERJSAWMjsawmhtml 3 Erickson B ldquoThe Chemical Weapons Convention Redefines Analytical Challengerdquo Anal Chem 70 397A (1998) 4 Johnston RL Hoefler CM Fargo JC and Moberley B ldquoThe Defense Nuclear Agencyrsquos
ChemicalBiochemical Weapons Agreements Verification Technology Research Development Test and Evaluation Program and its Requirements for On-Site Analysisrdquo AT-ONSITE 5-8 (1994)
5 Black RM Clarke RJ Read RW and Reid MT ldquoApplication of GC-MS and GC-tandem MS to the analysis of chemical warfare samples found to contain residues of the nerve agent sarin sulphur mustard and their degradation productsrdquo J Chrom 662 301-321 (1994)
6 USA Army Center for Health Promotion and Preventive Medicine (USACHPPM) Technical Guideline 230A 7 Hoffland LD Piffath RJ Bouck JB rdquoSpectral signatures of chemical agents and simulantsrdquo Optical
Engineering 24 982-984 (1985) 8 Braue EHJ Pannella MGrdquoCIRCLE CELL FT-IR Analysis of Chemical Warfare Agents in Aqueous Solutionsrdquo
App Spectrosc 44 1513-1520 (1990) 9 Tseng C-H Mann CK and Vickers TJ ldquoDetermination of Organics on Metal Surfaces by Raman
Spectroscopyrdquo App Spectrosc 47 1767-1771 (1993) 10 Staff Reporter ldquoUS Army Tests Lidar to Detect Biological Toxinsrdquo Photonic Spectra pg 50 December 1998 11 Christesen SD Raman cross sections of chemical agents and simulants App Spectrosc 42 318-321 (1988) 12 Weaver MJ Farquharson S Tadayyoni MA ldquoSurface-enhancement factors for Raman scattering at silver electrodesrdquo J Chem Phys 82 4867-4874 (1985) 13 Alak AM and Vo-Dinh T ldquoSERS of Organophosphorous Chemical Agentsrdquo Anal Chem 59 2149-2153 (1987) 14 Lee Y Farquharson S ldquoRapid chemical agent identification by SERSrdquo SPIE 4378 21-26 (2001) 15 Farquharson S Maksymiuk P Ong K Christesen S ldquoChemical agent identification by surface-enhanced Raman spectroscopyrdquo SPIE 4577 166-173 (2001) 16 Spencer KM Sylvia J Clauson S and Janni J ldquoSurface Enhanced Raman as a Water Monitor for Warfare
Agents in Waterrdquo SPIE 4577 158-165 (2001) 17 Tessier P Christesen S Ong K Clemente E Lenhoff A Kaler E Velev O ldquoOn-line spectroscopic characterization of sodium cyanide with nanostructured gold surface-enhanced Raman spectroscopy substratesrdquo
App Spectrosc 56 1524-1530 (2002) 18 Farquharson S W W Smith S Elliott and J F Sperry Rapid biological agent identification by surface- enhanced Raman spectroscopy SPIE 3855110-116 (1999) 19 Farquharson S WW Smith YH Lee S Elliott and J F Sperry Detection of biological signatures A
comparison of electrolytic and metal-doped sol-gel surface-enhanced Raman media SPIE 4575 62-72 (2002) 20 Guzelian AA Sylvia J Janni J Clauson S and Spencer KM ldquoSERS of whole cell bacteria and trace levels of biological moleculesrdquo SPIE 4577 182-192 (2001) 21 Shende C Inscore F Gift A Maksymiuk P Farquharson S ldquoRapid dipicolinic acid extraction from Bacillus spores detected by surface-enhanced Raman spectroscopyrdquo App Spectrosc 58 accepted 22 Lee Y and Farquharson S ldquoSERS Sample Vials Based on Sol-Gel Process for Trace Pesticide Analysisrdquo SPIE
4206 140-146 (2000) 23 Farquharson S and Lee Y ldquoTrace Drug Analysis by SERSrdquo SPIE 4200-16 (2000) 24 Lee Y Farquharson S and Rainey P M Surface-Enhanced Raman Sensor for Trace Chemical Detection in
Water SPIE 3857 76-84 (1999) 25 Lee Y Farquharson S Kwong H and Shahriari M ldquoSol-Gel Chemical Sensor for Surface-Enhanced Raman
Spectroscopyrdquo SPIE 3537 252-260 (1998) 26 Farquharson S A Gift P Maksymiuk F Inscore and W Smith ldquopH dependence of methyl phosphonic acid
dipicolinic acid and cyanide by surface-enhanced Raman spectroscopyrdquo SPIE 5269-19 (2003) 27 Lide DR Ed Handbook of Chemistry and Physics CRC Press 77th Ed (1996-7) 28 Sosa C RJ Bartlett K KuBulat and WB Person ldquoA theoretical study of harmonic vibrational frequencies and
infrared intensities of XCH2CH2SCH2CH2X and XCH2CH2SH (= H Cl)rdquo J Phys Chem 93 577-588 (1993) 29 Hameka HF and JO Jenson ldquoComputer-generated predictions of the structure and the IR and Raman spectra of
VXrdquo ERDEC-TR-065 May 1993
- Chemagents Appendicespdf
-
- SPIE2001-4575-Bioagent-SERSapdf
-
- Detection of bioagent signatures A comparison of electrolytic and metal-doped sol-gel surface-enhanced Raman media
- Stuart Farquharson Wayne Smith and Yuan Lee
-
- Real-Time Analyzers 87 Church Street East Hartford CT 06108
-
- ABSTRACT
- 4 CONCLUSIONS
- 5 ACKNOWLEDGEMENTS
- 6 REFERENCES
-
APPLIED SPECTROSCOPY 351
focusing the 488 nm laser beam 2 mm deep into thebulk of the crystals to avoid contributions from potentialdepletion layers As shown in Fig 1b the dependence ofthe Raman shift on the degree of deuteration is almostperfectly linear and ts very well with D 5 22684cmR1 24526 where D is the degree of deuteration (in )and R is the spectral mean of the PO4 vibration in cm21A linear correlation coef cient of 0998 indicates an ex-cellent linear dependence of the Raman peak shift withdegree of deuteration This result shows that the shift ofthe PO 4 peak is simply caused by the linear increase inatomic mass due to isotope substitution which decreasesthe length of hydrogen-like bonds
This excellent linear dependence allows us to map thepro le of the DH exchange layer at the surface of DKDPcrystals by acquiring Raman spectra and determining theposition of the PO 4 peak for various depths This methodis preferable over other methods such as determining thestrength of the OD vibration directly (eg at 715 cm21)because the position of the most intense peak in the Ra-man spectrum can be measured more precisely than theintensity of some of the weakest peaks in the spectrumThis is demonstrated in Fig 2 where depth-dependentRaman spectra (Fig 2a) and the resulting exchange layerpro les for two DKDP crystals are shown (Fig 2b) Thespectra in Fig 2a were obtained from a depth scan of aDKDP crystal with 75 degree of deuteration in thebulk grown at 45 8C The spectra start out as DKDP with30 deuteration close to the surface and approach thebulk DKDP spectrum within a few micrometers of depthThe fact that the relative degree of deuteration does notextend to 0 D is due to the limited depth resolution ofthe Raman microprobe which averages over 4 mm indepth Figure 2b depicts the resulting DH exchange layerpro les for this and a second crystal grown at 63 8Crespectively Both crystals had the same exposure to am-bient conditions and their main difference is the temper-ature at which they were grown The different exchangelayer pro les indicate that crystals grown at differenttemperatures have differen t proton conductiv ities 13
which leads to a difference in their rate of deuteriumdepletion The parameters controlling this behavior arecurrently the objective of a detailed study the results ofwhich will be reported elsewhere
CONCLUSION
In conclusion we have shown that the shift of the to-tally symmetric PO 4 stretch mode in the Raman spectrumof DKDP crystals scales linearly with degree of deuter-ation This allows us to correlate Raman peak positionsto deuteration levels in these crystals We have presenteda new technique to determine DH diffusion pro les inDKDP frequency conversion crystals based on micro-Ra-man spectroscopy This technique is fast inexpensiveand works under various environmental conditionswhich will allow us to better understand and control deu-terium depletion in DKDP crystals
ACKNOWLEDGMENTS
We would like to thank M Runkel for rst discovering DKDP crack-ing R Floyd for providing DKDP crystals and L Chase and A Burn-ham for their support and helpful discussions This work was performedunder the auspices of the US Department of Energy by the University
of California Lawrence Livermore National Laboratory under contract W-7405-Eng-48
1 J J De Yoreo A K Burnham and P K Whitman Int Mat Rev47 113 (2002)
2 C E Barker R A Sacks B M Van Wonterghern J A Caird JR Murray J H Campbell K Kyle R B Ehrlich and N DNielsen Proc SPIE-Int Soc Opt Eng 2633 501 (1995)
3 T Suratwala paper to be published4 Cleveland Crystals Inc httpwwwclevelandcrystalscom
KDPshtmltable5 E A Popova I T Savatinova and I A Velichko Sov Phys Solid
State 12 1543 (1971)6 I P Kaminow R C C Leite and S P S Porto J Phys Chem
Solids 26 2085 (1965)7 J A Subramony B J Marquardt J W Macklin and B Kahr
Chem Mat 11 1312 (1999)8 H Tanaka M Tokunaga and I Tatsuzaki Solid State Commun
49 153 (1984)9 R J Nelmes G M Meyer and J E Tibballs J Phys C 15 59
(1982)10 M A Yakshin D W Kim Y S Kim Y Y Broslavets O E
Sidoryuk and S Goldstein Laser Physics 7 941 (1997)11 I Takenaga Y Tominaga S Endo and M Kobayashi Solid State
Commun 84 931 (1992)12 C Krenn personal communication13 M Sharon and A K Kalia J Solid State Chem 21 171 (1977)
Rapid Dipicolinic Acid Extractionfrom Bacillus Spores Detectedby Surface-EnhancedRaman Spectroscopy
STUART FARQUHARSON ALAND GIFT PAUL MAKSYMIUK andFRANK E INSCOREReal-Time Analyzers Inc East Hartford Connecticut06108
Index Headings Dipicolinic acid Bacillus spores Anthrax Surface-enhanced Raman spectroscopy
INTRODUCTION
The anxiety caused by the distribution of anthrax en-dospores through the US postal system in October 2001was exacerbated by the long time required for positiveidenti cation of the Bacillus anthracis spores and the un-known extent of their distribution Since that time manymethods capable of rapid eld analysis have been inves-tigated to augment or replace the laboratory method ofgrowing microorganisms in culture media which takesdays to perform12 Prominent among these approachesare polymerase chain reactions (PCR)3 immunoassaysand detection of calcium dipicolinate as a biochemicalsignature PCR employs primers to separate organism-
Received 10 October 2003 accepted 14 November 2003 Author to whom correspondence should be sent
352 Volume 58 Number 3 2004
speci c nucleic acid sequences (eg capsular protein en-coding gene for Bacillus anthracis)4 and polymerases toamplify the segment until it is detectable Recently am-pli cation times have been substantially reduced andcomplete analysis can now be performed in an hour orless Immunoassay methods are also being developed thatuse competitive binding of the bioagent (as an antigen)and its labeled conjugate for a limited number of anti-bodies Although analyses can be performed in under 30minutes a well-de ned anthrax antigen has not yet beenidenti ed5ndash7 and consequently the false-positive rate isunacceptably high8
A number of other methods are being developed witha focus on the detection of calcium dipicolinate (CaDPA)and its derivatives as a B anthracis signature This is sobecause only spore-forming bacteria contain CaDPA andthe most common potentially interfering spores such aspollen and mold spores do not Relatively fast methodshave been developed to chemically extract CaDPA andthen detect it directly by uorescence9 or indirectly byluminescence1011 In the latter case hot dodecylamine(DDA) has been used to extract dipicolinic acid (DPA)and terbium has been utilized to form a highly lumines-cent DPA complex11 Although measurements have beenperformed in as little as ve minutes it was found thatas many as three concentration-dependent complexes canform each with different lifetimes This coupled withthe fact that the Tb31 cation produces the same lumines-cence spectrum makes determinations of low spore con-centrations problematic
It has been long known that Raman spectra of Bacillispores are dominated by bands associated with CaDPA12
and that these spectra may provide a suitable anthrax sig-nature at the genus level13 Since that time considerableimprovements in Raman instrumentation have led to lab-oratory measurements of single Bacilli spores14 and to eld measurements of spores captured from a mail-sort-ing system15 However the single spore measurementsrequired complex instrumentation that is not ruggedwhile the eld measurements required milligrams of sam-ple Furthermore the Raman spectra of both measure-ments contained uorescence contributions that would in-crease uncertainty in quanti cation
In related research we demonstrated that nanogramquantities of DPA could be detected by uorescence-freesurface-enhanced Raman spectroscopy (SERS)16 We alsodemonstrated that microliter volumes of chemicals canbe detected by SERS using metal-doped sol-gel-packedglass capillaries17 Towards the goal of developing a rap-id eld SERS-based anthrax spore detector we havecombined our previous research and we now report thatDPA can be extracted from a 10 mg B cereus spore sam-ple using DDA in 1 minute and can be detected by SERSin an additional 1 minute
EXPERIMENTAL
Dipicolinic acid (26-pyridinedicarboxylic acid DPA)and dodecylamine (DDA) were used as received fromSigma-Aldrich (Milwaukee WI) Lyophilized B cereusspores prepared according to the literature13 were sup-plied by the University of Rhode Island and used as re-ceived Multiple particles approximately 01 mm3 each
were separated and weighed at 5 to 15 mg representing05 to 15 million spores The sample masses were con-sistent with a previous determination of spore density at0081 gmL that indicated a high degree of entrained air
All chemicals used to prepare the silver-doped sol-gelcoated capillaries were also obtained and used as receivedfrom Sigma-Aldrich According to previously publishedprocedures17 two precursor solutions were preparedmixed and then drawn into 1-mm-diameter glass capil-laries The silver amine precursor consisted of a 51 vvratio of 1 N AgNO 3 to 28 NH3OH while the alkoxideprecursor consisted of a 21 vv ratio of methanol to te-tramethyl orthosilicate The alkoxide precursors weremixed with silver amine precursor in an 81 vv ratioApproximately 015 mL was drawn into the capillarycoating a 15-mm length After sol-gel formation the in-corporated silver ions were reduced with dilute sodiumborohydride which was followed by a water wash to re-move residual reducing agent
A 100 mL drop of a 50 mM DDA solution in ethanolpre-heated to 78 8C was added to each of the B cereusparticles to digest the spore coat After 1 minute the re-sultant solution was drawn into a SER-active capillarythat was immediately xed horizontally to an XY posi-tioning stage (Conix Research Spring eld OR) just in-side the focal point of an f 07 aspheric lens The lensfocused the beam into the sample and collected the scat-tered radiation back along the same axis A dichroic lter(Omega Optical Brattleborough VT) was used to re ectthe excitation laser to the lens and pass the Raman scat-tered radiation collected by the lens An f 2 achromatwas used to collimate the laser beam exiting a 200-mm-core-diameter source ber optic while a second f 2 ach-romat was used to focus the scattered radiation into a 365mm ber optic (Spectran Avon CT) A short-pass lterwas placed in the excitation beam path to block the sil-icon Raman scattering generated in the source ber fromre ecting off sampling optics and reaching the detectorA long-pass lter was placed in the collection beam pathto block the sample Rayleigh scattering from reachingthe detector A 785 nm diode laser (Process InstrumentsInc model 785-600 Salt Lake City UT) was used todeliver 100 to 150 mW of power to the sample A Fouriertransform Raman spectrometer (Real-Time Analyzersmodel IRA-785 East Hartford CT) and a silicon photo-avalanche detector (Perkin Elmer model C30902S Stam-ford CT) were used to acquire the SER spectra
RESULTS AND DISCUSSION
As an initial experiment the SER spectrum of 1 gLof DPA in water was measured using the newly devel-oped silver-doped sol-gel-coated capillaries (Fig 1A) Atthis concentration a high signal-to-noise ratio (SN) isobtained in 1 min In fact a reasonable spectrum is ob-tained in the same time frame for 1 mgL (Fig 1B) TheSER spectra are reasonably similar to the normal Raman(NR) spectrum obtained for a saturated solution of DPAin 1 N KOH (Fig 1C) and the following band shifts areobserved (NR to SER) 647 to 657 cm21 817 to 815cm21 998 to 1008 cm21 1384 to 1382 cm21 1434 to1428 cm21 and 1569 to 1567 cm21 Many of these bandshave been previously assigned1213 such as 998 cm21 to
APPLIED SPECTROSCOPY 353
FIG 1 SERS of DPA in water using silver-doped sol-gel-coated glasscapillary for (A) 1 gL and (B) 1 mgL (C ) NR of saturated DPA in 1N KOH in a glass capillary Spectral conditions (A) and (B) 150 mWof 785 nm 1-min acquisition time (C) 450 mW of 785 nm 5-minacquisition time both 8 cm 21 resolution
FIG 2 SERS of DPA extracted from 10 mg B cereus particle using100 mL of 50 mM hot DDA acquired in (A) 1 minute and (B) 2 seconds(C ) Attempted SERS of 50 mM hot DDA in ethanol using silver-dopedsol-gel-coated glass capillary acquired in 1 min Spectral conditions150 mW of 785 nm 8 cm21 resolution
the symmetric ring stretch 1384 cm21 to the OndashCndashOsymmetric stretch 1428 cm21 to the symmetric ring CndashH bend and 1569 cm21 to the asymmetric OndashCndashOstretch
The rst B cereus samples consisted of 2 mg of sporesin 2 mL of 5 mM hot DDA The samples were main-tained at 78 8C for 40 min and while hot approximately10 mL was drawn into a SER-active capillary Since spec-tra of DPA were obtained for these initial samples small-er spore masses higher DDA concentrations and shorterheating periods were examined In due course it wasfound that 10 mg of spores could be digested by 100 mLof 50 mM hot DDA in one minute and detected (Fig2A) In fact the signal was suf ciently intense that it canbe observed in as little as two seconds (Fig 2B) Theamount of DPA that was extracted was estimated to bebetween 5 and 10 mgL by comparing the signal intensityof the 1008 cm21 band to that measured for DPA in waterThis is consistent with previous research that found thatthe majority of the DPA is extracted from spores usingDDA11 and that B cereus spores contain approximately10 DPA by weight18 The SN of 127 for the 1008 cm21
band in the 1-minute SER spectrum suggests a limit ofdetection of approximately 250 ng of B cereus sporesbased on a SN of 3 Finally it should be noted that DDAdid not produce a detectable SER spectrum as shown inFig 2C
CONCLUSION
We have demonstrated that by combining rapid extrac-tion of dipicolinic acid from Bacillus cereus spores withchemical identi cation by surface-enhanced Raman spec-troscopy as little as 10 mg of spores can be detected Infact the entire measurement from the time of adding hotdodecylamine to the spores to the time when the dipi-colinic acid SER spectrum is acquired and analyzedcould be performed in less than two minutes The abilityof this method to distinguish between spore-forming bac-teria such as Bacillus anthracis and non-DPA containingpowders could help prevent costly shutdowns associated
with the appearance of suspicious material or intentionalmailing of common substances as an anthrax hoax Thismethod could also prove useful in detecting the locationof anthrax endospores in mail distribution facilities if an-other veri ed attack should occur
Research continues to fully characterize the surface-enhanced Raman spectroscopy signal intensities as afunction of sample concentration and to explore otherextractants that do not require the use of elevated tem-perature
ACKNOWLEDGMENTS
The authors are grateful for the support of the National Science Foun-dation (DMI-0296116 and DMI-0215819) and the US Army(DAAD13-02-C-0015 Joint Service Agent Water Monitor program)The authors are indebted to Chetan Shende for preparing the sol-gelcapillaries The authors also thank James Gillespie Nicholas Fell andAugustus Fountain for providing important background informationMark Farquharson for laboratory support and Professor Jay Sperry ofthe University of Rhode Island for supplying B cereus spores
1 V A Pasechnik C C Shone and P Hambleton Bioseparations 3267 (1993)
2 P J Jackson M E Hugh-Jones D M Adair G Green K K HillC R Kuske L M Grinberg F A Abramova and P Keim ProcNatl Acad Sci USA 95 1224 (1998)
3 B R Glick and J J Pasternak Molecular Biology Principles andApplications of Recombinant DNA (ASM Press Washington DC1994)
4 C A Bell J R Uhl T L Had eld J C David R F Meyer TF Smith and F R Cockerill III J Clin Microbiol 40 2897(2002)
5 D L Gatto-Menking H Yu J G Bruno M T Goode M Millerand A W Zulich Biosens Bioelectron 10 501 (1995)
6 J J Quinlan and P M Foegeding J Rapid Methods AutomationMicrobiol 6 1 (1998)
7 A A Hindle and E A H Hall Analyst (Cambridge UK) 1241599 (1999)
8 M S Ascher US Department of Health amp Human Services(httpwwwhhsgovophppresentationsAscherdoc)
9 R Nudelman B V Bronk and S Efrima Appl Spectrosc 54 445(2000)
10 D L Rosen C Sharpless and L B McBrown Anal Chem 691082 (1997)
354 Volume 58 Number 3 2004
11 P M Pellegrino N F Fell Jr and J B Gillespie Anal ChimActa 455 167 (2002)
12 W H Woodruff T G Spiro and C Gilvarg Biochem BiophysRes Commun 58 197 (1974)
13 E Ghiamati R S Manoharan W H Nelson and J F SperryAppl Spectrosc 46 357 (1992)
14 A P Esposito C E Talley T Huser C W Hollars C M Schal-dach and S M Lane Appl Spectrosc 57 868 (2003)
15 S Farquharson L Grigely V Khitrov W W Smith J F Sperryand G Fenerty J Raman Spectrosc paper accep ted (2003)
16 S Farquharson W W Smith S Elliott and J F Sperry SPIE-IntSoc Opt Eng 3855 110 (1999)
17 S Farquharson and P Maksymiuk Appl Spectrosc 57 479(2003)
18 F W Janssen A J Lund and L E Anderson Science (Washing-ton DC) 127 26 (1958)
SPIE -2003-5269 117
pH dependence of methyl phosphonic acid dipicolinic acid and cyanide by surface-enhanced Raman spectroscopy
Stuart Farquharson Alan Gift Paul Maksymiuk Frank Inscore and Wayne Smith
Real-Time Analyzers 87 Church Street East Hartford CT 06108
ABSTRACT US and Coalition forces fighting terrorism in Afghanistan and Iraq must consider a wide range of attack scenarios in addition to car bombings Among these is the intentional poisoning of water supplies to obstruct military operations To counter such attacks the military is developing portable analyzers that can identify and quantify potential chemical agents in water supplies at microgram per liter concentrations within 10 minutes To aid this effort we have been investigating the value of a surface-enhanced Raman spectroscopy based portable analyzer In particular we have been developing silver-doped sol-gels to generate SER spectra of chemical agents and their hydrolysis products Here we present SER spectra of methyl phosphonic acid and cyanide as a function of pH an important factor affecting quantitation measurements which to our knowledge has not been examined In addition dipicolinic acid a chemical signature associated with anthrax-causing spores is also presented Keywords Chemical warfare agents agent detection agent hydrolysis SERS Raman spectroscopy homeland security
1 INTRODUCTION In the past decade the Unites States and its allies have been challenged by a different kind of warfare exemplified by the terrorist attacks of September 11 2001 Suicide bombings and the use of chemical agents are the norm and military personnel must consider a wide range of attack scenarios Among these is the intentional poisoning of water supplies to obstruct military operations in Afghanistan and Iraq Currently colorimetric paper is used to detect agents on-site while gas chromatography combined with mass spectrometry (GCMS) is used in mobile support laboratories However both methods have severe drawbacks The paper changes color in response to contact with many chemicals besides CWAs causing a high incidence of false positives1 while GCMS although very chemically specific requires hours to perform and constant re-calibration234 Military operations would be greatly aided by a portable analyzer that can identify and quantify potential chemical agents at concentrations that impact safety This includes the analysis of drinking water supplies distribution and storage systems To meet this goal the Department of Defense has been investigating numerous approaches under the auspices of the Joint Service Agent Water Monitor (JSAWM) program5 The JSAWM must be capable of identifying the chemical agents with no false-positives and quantifying the agents at microgram per liter concentrations within 10 minutes678 Recently we and others have been investigating the ability of surface-enhanced Raman spectroscopy (SERS) to measure chemical agents9-12 bioagents13-17 and their hydrolysis products in water SERS employs the interaction of surface plasmon modes of metal particles with target analytes to increase scattering efficiency by as much as 1 million times18 In 1987 the potential of this approach to measure CWAs was demonstrated by measuring a series of pesticides19 Several of these organophosphonates have chemical structures similar to CWAs in particular P=O functional groups In our studies we have been employing silver-doped sol-gels to promote the SER effect The porous silica network of the sol-gel matrix offers a unique environment for immobilizing and stabilizing SER-active metal particles20-23 The sol-gels have been coated on the internal walls of standard glass vials to allow rapid SER analysis of small sample quantities without preparation We have measured over 100 chemicals with enhancements of 104 to 106 demonstrated reversible measurements in a flowing system reproducible measurements from vial-to-vial and measurements in multiple solvents including water20-23 Previously we used these vials to perform preliminary measurements of cyanide (CN) methylphosphonic acid (MPA) and dipicolinic acid (DPA) MPA is a hydrolysis product of the nerve agents (eg sarin Reaction 1) and may be a valuable indicator of nerve agent usage particularly since the alkyl methylphosphonic acids are relatively more stable than their corresponding parent complexes24 DPA is
SPIE -2003-5269 118
a chemical signature of spore forming bacteria such as Bacillus anthracis And in light of the inability to rapidly detect the anthrax spores distributed through the US mail in October 2001 a number of methods are being developed to extract and analyze this signature Reaction 1 Stepwise hydrolysis of Sarin to form hydrofluoric acid (HF) isopropyl methylphosphonic acid (IMPA) then methyl phosphonic acid (MPA) and 2-propanol In our previous SERS investigations MPA and DPA were measured at 50 and 100 mgL respectively In both cases limits of detection (LOD) were estimated at 100 microgL providing encouragement in that SERS may satisfy the needs of the JSAWM Since it has been shown that pH can substantially influence the intensity of SER bands25 which would clearly influence quantitative analysis we undertook the present study to determine the severity of these effects for cyanide methyl phosphonic acid and dipicolinic acid Furthermore we previously observed a band at 1050 cm-1 for MPA6 possibly due to an anion formed at basic pH Here we investigate the source of this spectral anomaly
2 EXPERIMENTAL All chemicals including potassium cyanide methyl phosphonic acid dipicolinic acid and those used to prepare the silver-doped and gold-doped sol-gel coated vials were obtained and used as received from Sigma-Aldrich All samples were prepared in HPLC grade water (Fischer Scientific Fair Lawn NJ) for SERS measurements The pH of these samples was adjusted using dilute nitric acid or potassium hydroxide and verified using a pH electrode (Corning Inc Corning NY) that had been calibrated with pH 400 700 and 1000 buffered standards from Fischer Scientific Once prepared the samples were transferred into the silver-doped sol-gel vials (Real-Time Analyzers Inc East Hartford CT) The vials were coated in a manner similar to that previously reported by adding ammonium hydroxide to a solution of silver nitrate tetramethyl orthosilicate (TMOS) and methanol20 Gold-vials were coated by adding nitric acid to a solution of gold tetrachloride TMOS and methanol The two precursor solutions were prepared mixed and transferred to 2-mL glass vials dried and heated After sol-gel formation the incorporated metal ions were reduced with dilute sodium borohydride (1mgmL) which was followed by a water wash to remove residual reducing agent After the resultant analyte solution was introduced the SER-active vial was immediately fixed horizontally to an XY positioning stage (Conix Research Springfield OR) just inside the focal point of an f07 aspheric lens The lens focused the beam into the sample and collected the scattered radiation back along the same axis A dichroic filter (Omega Optical Brattleborough VT) was used to reflect the excitation laser to the lens and pass the Raman scattered radiation collected by the lens An f2 achromat was used to collimate the laser beam exiting a 200 microm core diameter source fiber optic while a second f2 achromat was used to focus the scattered radiation into a 365 microm fiber optic (Spectran Avon CT) A short pass filter was placed in the excitation beam path to block the silicon Raman scattering generated in the source fiber from reflecting off sampling optics and reaching the detector A long pass filter was placed in the collection beam path to block the sample Rayleigh scattering from reaching the detector A 785 nm diode laser (Process Instruments Inc model 785-600 Salt Lake City UT) was used to deliver 100 to 150 mW of power to the sample A Fourier transform Raman spectrometer (Real-Time Analyzers model IRA-785 East Hartford CT) and a silicon photo-avalanche detector (Perkin Elmer model C30902S Stamford CT) were used to acquire the SER spectra
3 RESULTS AND DISCUSSION In a previous study of MPA6 aimed at developing a concentration calibration curve and determining limits of detection (LOD) we observed an anomaly at 1050 cm-1 Since it was found that the band intensity changed as a function of concentration the band must be associated with a sample parameter Two possible parameters photon flux and pH are examined here The first parameter was investigated by irradiating a 1mgmL MPA sample in a SER-active vial with laser powers of 200 mW and above and monitoring spectral changes It was immediately found that the 1050 cm-1 band
2O+ H HF + +OH
OH3H C
OP
OF
CH
CH3
3
3H C
OCP
OOH
CH
CH3
3
3H C
O
CPHO
CH
CH
3
3
C
Sarin IMPA MPA 2-propanol
SPIE -2003-5269 119
appeared and grew as a function of time and that the higher the laser power the faster the growth Figure 1 shows the growth of the 1050 cm-1 band over the course of 30 minutes when using 150 mW of 785 nm excitation while Figure 2A shows that the growth can be fit with a first order exponential equation namely I1050 = 03+05e-013t Furthermore the 760 cm-1 band could be fit with a first order decay equation with an identical exponential rate constant ie I760 = 08-
08e-013t The rates represent classical first order kinetics and their correspondence allows one to conclude that MPA is being transformed one-for-one into a photo-generated product At this time the photoproduct has not been positively identified but phosphonic (phosphorous) acid and phosphonate are likely candidates since the symmetric P(OH)3 stretch occurs at ~1050cm-1 Our studies show that a reduction in laser power to 100 mW at the sample for MPA essentially eliminates this degradation process This laser power or lower was used for further measurements
Methyl phosphonic acid is a diprotic acid that stepwise dissociates into two anions MPA- and MPA= according to the following reactions26
MPA MPA- + H+ pKa1 = 212 Reaction 2
MPA- MPA= + H+ pKa2 = 729 Reaction 3 The relative concentrations of MPA MPA- and MPA= can be determined at any pH by expressing [MPA] and [MPA=] in terms of [MPA-] using Reactions 2 and 3 and summing all three to equal the total starting concentration here 2 mgmL (0021M MW = 9602) viz
[MPA] + [MPA-] + [MPA=] = 0021M Equation 1 substituting from Reactions 1 and 2
([H+][MPA-])K1a + [MPA-] + (K2a[MPA-])[H+] = 0021M Equation 2 rearranging [MPA-] = 0021M(1+[H+]K1a + K2a[H+]) Equation 3 The relative concentrations of MPA MPA- and MPA= as a function of pH are shown in Figure 3 It is worth noting that near neutral pH both MPA- and MPA= will be present To confirm that the SER signal followed this pH dependence a starting solution consisting of 20 mg of MPA in 10 mL HPLC grade water was prepared and brought to pH of 20 using dilute nitric acid From this solution 2 mL were added to a SER-active vial and the SER spectrum recorded At this pH a peak at 760 cm-1 was barely discernable The 2 mL solution was returned to the starting solution and the pH was re-measured to correct for any changes that the silver-doped sol-gel vials might cause In most cases the change was less than 02 pH units and the pH is reported as the before and after average Next the pH of the
Figure 2 A) Exponential growth of 1050 cm-1 band and B) exponential decay of 760 cm-1 band for spectral series in Figure 1
Figure 1 Growth of 1050 cm-1 band as a function of time due to exposure to 150 mW of 785 nm Spectra are 5 sec each collected every 100-sec from 0 to 30-min
B
570 770 970 1170 1370Raman Shift (cm-1)
Arbit
rary
Unit
s
MPA Photodegradation
600 800 1000 1200 1400 Wavenumber (cm-1)
A
0 10 20 30 0 10 20 30 time (min) time (min)
Ram
an In
tens
ity (r
elat
ive)
30
min
0
SPIE -2003-5269 120
starting solution was adjusted to 325 using dilute KOH Again 2 mL were added to a vial and the SER spectrum recorded At this pH a reasonably strong 760 cm-1 band was observed This process was repeated as spectra were recorded at pHs of 70 74 75 79 85 and 100 A total of 1 mL of KOH was added diluting the total concentration by 10 Next the pH of the starting solution was made acidic by adding dilute nitric acid dropwise This time spectra were recorded at pHs of 72 69 64 and 37 Figure 4 shows the SER spectra for representative pHs (spectra were left out to simplify the figure) while Figure 3 shows the 760 cm-1 peak intensities as a function of pH (The band intensities were adjusted to compensate for dilution effects caused by the addition of HNO3 and KOH then normalized to 0021 M for the most intense band observed at pH 37) It is clear from Figure 3 that the 760 cm-1 band follows the MPA- concentration as a function of pH and must be assigned to this anion No bands were observed that corresponded to MPA or MPA= The lack of an MPA SER spectrum may be due to the absence of an attraction between the neutral analyte and the electropositive silver surface The same reasoning suggests that a strong SER spectrum should be observed for MPA= but it is not and a satisfactory explanation has not been found
0000
0005
0010
0015
0020
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14pH
Con
cent
ratio
n [M
]
pK1 = 212 pK2 = 729
MPA- MPA=
MPA
Figure 4 SER spectra of 002M MPA as a function of pH Conditions 100 mW of 785 nm 36 scans (1 min) 8 cm-1 recorded 2 min after sample introduction pH 19 and 20 not apparent on this scale pH 69 and 74 near identical to 70 and 75 and not shown for clarity
00
02
04
06
08
10
12
14
16
18
0 5 10 15 20 25 30Measurement Number
Ram
an In
tens
ity (7
60 c
m-1
)
Figure 3 Concentration dependence of MPA MPA- and MPA= as a function of pH for a 002M sample Intensity of 760 cm-1
band from Figure 6 as a function of pH ( for increasing basic adjustment for increasing acidic adjustment error was measured at ~10 for pH 64)
Figure 5 SER spectra of 002M MPA at pH 64 measured around a vial at three heights (9 points per height) Conditions as in Figure 4 but 10-sec scans
Since these measurements involved the removal and replacement of the SER-active vial in the sample holder to remove and add sample variation in the intensity as a function of vial position was minimized by illuminating the exact same height along the vial wall But this does not account for variability of the SERS response of the sol-gel coating around the vial To analyze this effect a vial containing MPA at pH 64 was rotated at ~ 40o
intervals at the original height and 18rdquo above and below this value Figure 5 shows the intensity of the 760 cm-1
band for the 27 positions It was found that the average value was 137plusmn014 an RSD of 10 overall and 5 for each height An error bar is included in Figure 4 for the pH 64 measurement
SPIE -2003-5269 121
With the above analyses in mind a preliminary investigation of the SER spectral response for MPA (as MPA-) as a function of concentration was performed A single vial was used for these measurements beginning with 1 mgL followed by measurements of 10 100 and 1000 mgL In all cases the pH was ~7 and 3 positions around the vial were measured per concentration Since the 760 cm-1 band was not observed for concentrations of 1 or 10 mgL using 100 mW of 785 nm the laser power at the sample was raised to 200 mW beginning with the 10 mgL concentration Photo-degradation was largely avoided (and not observed) by exposing the sample for only 33 seconds per spectral acquisition Representative spectra for 10 100 and 1000 mgL are shown in Figure 6 while a plot of the 760 cm-1 band intensity as a function of concentration is shown in Figure 7 These values were also used to estimate limits of detection based on the signal-to-noise ratio (SN) of the 760 cm-1 band using 100 mW of 785 nm laser excitation and a 10-min acquisition time As summarized in the Figure 7 inset the lower the measured concentration the lower the predicted LOD Even if as estimated 210 microgL could be measured using the silver-doped sol-gel vials an improvement of a factor of 70 is still required to meet JSAWM goals of measuring 3 microgL in 10-minutes Similar to MPA DPA is a diprotic acid (pKa1 = 216 and pKa2 = 692) and variations in pH will effect the relative concentrations of DPA DPA- and DPA= and possibly the SER spectra and band intensities This could prove significant if an acid or base is used to denature anthrax spores with the goal of extracting and analyzing DPA The MPA pH study described above was mimicked for DPA except that the starting solution consisted of 20 mg of DPA in 20 mL HPLC grade water (60x10-3M MW = 1671) The initial solution had a pH of 245 which was made basic by dilute KOH to pHs of 355 433 487 559 1069 and 1166 SER spectra were recorded at each pH using 100 mW of 785 nm and a 44-sec acquisition time Next one drop of concentrated nitric acid was used to remake the solution acidic at a pH of 200 Again sequential pH measurements were performed at 383 510 735 and 822 The solution pH was made acidic a third time but to pH 219 171 then 135 Throughout this process no more than 20 drops of acid or base were added and therefore the concentration was diluted by no more than 10 Most of the spectral bands showed a minor decrease in intensity as a function of increasing pH values However the bands at 525 and 795 cm-1 showed the most dramatic changes which occurred at acid pH Figure 8 shows the SER spectra of DPA for the spectral region and pH range of interest The identity of the DPA species was determined by plotting the normalized peak intensities with the lowest value set to 0 and the highest to 0006 M as a function of pH and overlaying these values on a plot of the relative concentrations for DPA DPA- and DPA= as previously done for MPA (Figure 9) As can be seen the 525 cm-1 band clearly corresponds to DPA The correspondence of the 795 cm-1 band to this species is less clear as the band retains intensity until pH of 55 This can be attributed to contribution to the overlapping band at 810 cm-1 which does not change as a function of pH The fact that most bands are observed at all pHs suggest that the primary interaction with silver is through the ring nitrogen This is supported by the fact that the most intense band occurs at 1008 cm-1 attributed to a symmetric ring breathing mode and that this interaction has been characterized for pyridine in numerous papers27
0
1
2
3
4
5
6
7
0 200 400 600 800 1000 1200
MPA Concentration (mgL)
760
Ban
d In
tens
ity
Figure 6 SER spectra of MPA in water at A) 1000 B) 100 and C) 10 mgL Conditions pH of 7 silver-doped sol-gel coated vial 200 mW of 785 nm 33-sec 8 cm-1 resolution
Figure 7 Plot of SER intensity of 760 cm-1 band of MPA as a function of concentration using 200 mW of 785 nm Inset table includes average intensity LOD standard deviation and percent deviation for each concentration but for 100 mW and 10 min
conc (mgL) LOD ave stddev dev10 021 002 1021100 072 010 14301000 312 040 1280
A
B C
SPIE -2003-5269 122
The ability to detect dipicolinic acid (DPA) by SERS is demonstrated in Figure 10 Even at 1 mgL the primary bands are visible The intensity of the dominant ring breathing mode observed at 1008 cm-1 for DPA in water (pH = 25 to 55) is plotted as a function of concentration in Figure 11 These values were also used to estimate limits of detection based on the SN of the 1008 cm-1 band using 100 mW of 785 nm laser excitation and a 10-min acquisition time Again the lower the measured concentration the lower the predicted LOD (see Figure 11 inset) and detection of 160 microgL is possible Gastrointestinal anthrax requires significant more spores than inhalation anthrax28 and a limit of detection might be placed at 1 million spores in 1 liter of water or 10 microgL Since each spore contains ~10 CaDPA by weight29 a goal for DPA might be 1 microgL indicating that the present measurements must be improved by nearly two orders of magnitude Due to the increased hazards of handling HCN gas KCN salt was used for these experiments Nevertheless all sample preparations were performed in a chemical hood KCN completely dissolves in water but its conjugate acid HCN is formed and has a Ka of 615x10-1030 viz
HCN CN- + H+ pKa = 921 Reaction 4
Figure 8 SER spectra of 1 mgmL DPA as a function of pH Conditions 100 mW 785 nm 100 scans (44-sec)
Figure 9 Concentration dependence of DPA DPA- and DPA= as a function of pH for a 0006M sample Intensity of 525 () and 795 (diams) cm-1 bands from Fig 8 as a function of pH
pH
135 171 219 383
0
0001
0002
0003
0004
0005
0006
0007
0 2 4 6 8 10 12 14pH
Con
cent
ratio
n [M
]
DPADPA-DPA=795525
DPA DPA=DPA-
pK2 = 692pK1 = 216
795 525
0
05
1
15
2
25
0 200 400 600 800 1000 1200
DPA Concentration (mgL)
1008
Ban
d In
tens
ity
conc (mgL) lod-10min-100mw1 017
10 016100 103
1000 355
Figure 10 SER spectra of DPA in water at A) 1000 B) 100 C) 10 and D) 1 mgL Conditions pH of 25-55 silver-doped sol-gel coated vial 175 mW of 785 nm 1-min 8 cm-1 D) has been multiplied by x10 to make bands visible
A
B C
D
Figure 11 Plot of SER intensity of 1008 cm-1 band of DPA as a function of concentration using 175 mW of 785 nm Inset table includes LOD in mgL for each concentration but for 100 mW and 10 min
SPIE -2003-5269 123
Consequently the cyanide concentration must be determined for each initial KCN concentration Specifically the samples prepared with concentrations of 01 1 10 100 and 1000 mgL of KCN produced CN- concentrations of 63x10-3 033 69 89 and 964 mgL at pHs of 816 90 967 102 and 107 respectively The pH dependence for the HCN and CN- concentrations are shown in Figure 12 Thus as the amount of KCN added to the solution decreases so does the pH of the solution (becomes less basic) and according to Reaction 4 the relative amount of CN- to HCN also decreases For example in the preparation of a 01 mgL solution of KCN the pH is shifted from 7 for pure water to only 816 and only 63 of the starting material becomes CN- or 63x10-3 mgL In comparison for a solution of 1000 mgL the pH is shifted from 7 to 107 and 96 of the starting material becomes CN- This is significant since the cyanide ion is better able to adsorb onto the silver particles and become SERS active SER spectra of 10 100 and 1000 mgL of KCN samples are shown in Figure 13 and a corresponding plot of the 2100 cm-1 band intensity as a function of concentration is shown in Figure 14
000102030405060708091011
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14pH
Con
cent
ratio
n [m
gm
L]
pKa = 921
CN -HCN
The SER spectra of cyanide are dominated by the single CN stretch at 2100 cm-1 which occurs in normal Raman spectra of solutions at 2080 cm-1 However a low frequency mode occurs at 135 cm-1 due to a Ag-CN stretch (not shown) indicative of a strong surface interaction It is also observed that as the concentration decreases the CN stretch shifts to 2140 cm-1 This shift has been attribute to the formation of a tetrahedral Ag(CN)3
2- surface structure31 as well
Figure 14 Concentration dependence of KCN SERS measured under conditions in Fig 11 Concentrations are 1 01 and 001 mgml Intensities are measured for the CN stretch at 2100 cm-1 Inset table includes LOD in mgL for each concentration in Figs 13 and 15 but for 100 mW and 10 min
Figure 13 SER spectra of KCN in water at A) 1000 B) 100 and C) 10 mgL Conditions pHs of 107 102 and 97 silver-doped sol-gel coated vial 100 mW of 785 nm 1-min 8 cm-1 C) has been multiplied by x10 to make band visible
Figure 12 Concentration dependence of HCN and CN- as a function of pH for a 1 mgmL sample Calculated intensity of 2100 () cm-1 band for a 1 mgml sample at pHs of 816 90 967 102 and 107
A
B
C
0
20
40
60
80
100
120
140
0 200 400 600 800 1000 1200CN Concentration (mgL)
2100
Ban
d In
tens
ity
Figure 15 SER spectra of KCN in water at A) 10 B) 1 and C) 01 mgL Conditions pHs of 97 12 and 12 gold-doped sol-gel coated vial 100 mW of 785 nm 1-min 8 cm-1 C) has been multiplied by x10 to make band visible
A
B
C
conc (mgL) lod-10min-100mw Condition01 001 Au-pH 121 007 Au-pH 12
69 003 Au-pH 9769 016 Ag-pH 9789 022 Ag-pH 102946 113 Ag-pH 107
SPIE -2003-5269 124
as to CN adsorbed to two different surface sites32 Alternatively the 2140 cm-1 band could be attributed to HCN since this species dominates at lower concentrations However it is unlikely that this species would be attracted to the electropositive silver surface Further both peaks should be present at pHs between 85 and 105 but this is not observed It has also been suggested that at concentrations near and above monolayer coverage the CN- species is forced to adsorb end-on due to crowding and at lower concentrations the molecule can reorient to lie flat33 This suggests that the 2100 and 2140 cm-1 bands correspond to the end-on and flat orientations respectively As Figures 13 and 14 show the intensity of the CN stretch for the 89 mgL sample is nearly as intense as the 964 mgmL band This suggests that the Raman signal for the flat orientation is more enhanced However more extensive measurements are required to verify this point Since resent research has suggested that cyanide may be more effectively detected on gold measurements of KCN solutions were also performed using gold-doped sol-gel vials Preliminary measurements are shown in Figure 15 for samples prepared from 01 1 and 10 mgL KCN Since the pHs are 816 90 967 the resultant CN- concentrations are 63x10-3 033 and 69 mgL Initially only the highest concentration was observed and the signal intensity was significantly better than the equivalent concentration measured using silver In an effort to shift Reaction 4 to the left transforming HCN to CN- (Le Chatelierrsquos principle) KOH was added to the lower concentration samples producing solutions with pH 12 This effectively forces all of the cyanide in solution to be CN- or 01 and 10 mgL respectively More importantly the CN stretch is now observed in the SER spectra The band appears at 2125 cm-1 as has been previously reported for gold12 As calculated for MPA and DPA LODs can be estimated from this data For the three concentrations of cyanide on silver the LODs are 016 to 11 mgL for 100 mW of 785 nm laser excitation and a 10-min acquisition time For gold pH adjusted the LODs are10 to 70 microgL an improvement of more than 10 times silver Nevertheless either substrate is sufficient to meet the JSAWM goals of measuring 3 mgL in 10-minutes as the requirements form cyanide are much less stringent than the nerve agents
4 CONCLUSIONS Here we examined the surface-enhanced Raman spectral response for methyl phosphonic acid dipicolinic acid and cyanide as a function of pH It was determined that the most prominent peak at 760 cm-1 reaches a maximum intensity between pH 3 and 7 and corresponds to the MPA- species Neither the MPA nor MPA= species appear to generate a SER spectrum and consequently no spectra were observed below pH 2 or above pH 8 In this study we also found that higher laser powers could cause photodegradation of MPA signified by the exponential growth of a band at 1050 cm-1 which is tentatively assigned to phosphorous acid Unlike MPA DPA was observed at all pHs This is attributed to the dominant interaction of the pyridine functional group with silver Minor spectral changes were observed at acid pHs and were assigned to neutral DPA Like MPA SER spectra of cyanide were pH dependent No spectra were observed for the HCN species while CN- was best observed at pHs more basic than 8 Preliminary concentration studies for the three analytes allowed estimating limits of detection for MPA DPA and CN using 100 mW of 785 nm and a 10-min acquisition time of 210 165 and 70 microgL respectively Although the latter value suggests that cyanide can be measured at sufficiently low concentrations to meet JSAWM goal improvements by 100 to 200 times are required for MPA and DPA It is clear from this study that pH of the sample is important and must be taken into account when developing concentration calibration curves as well as fieldable analyzers Future work will include tailoring the sol-gel with specific functional groups to dictate sample pH and thereby optimize sensitivity
ACKNOWLEDGMENTS The authors are grateful for the support of the US Army (DAAD13-02-C-0015 Joint Service Agent Water Monitor program) The authors would also like to thank Dr Steve Christensen of the US Army SBCCOM for helpful discussions and Mr Chetan Shende of Real-Time Analyzers for assistance in development of the gold-doped sol-gels
REFERENCES 1 Erickson B Analytical Chemistry News amp Features June 1 397A (1998)
SPIE -2003-5269 125
2 Johnston RL Hoefler CM Fargo JC and Moberley B AT-ONSITE 5-8 (1994) 3 Black RM RJ Clarke RW Read and MT Reid J Chromatography 662 301-321 (1994) 4 USA Army Center for Health Promotion and Preventive Medicine (USACHPPM) Technical Guideline 230A 5 Jensen JL NC Wong W R Loerop SPIE 4775-03 (2001) 6 Hoenig SL Handbook of Chemical Warfare and Terrorism Greenwood Press Wesport CT (2002) 7 Munro NB SS Talmage GD Griffin LC Waters AP Watson JF King and V Hauschild Env Health
Persp 107 933-974 (1999) 8 Holstege CP Kirk M Sidell FR Crit Care Clin13 923-42 (1997) 9 Farquharson S P Maksymiuk K Ong and S Christesen SPIE 4577 166-173 (2001) 10 Lee Y and S Farquharson SPIE 4378 21-26 (2001) 11 Spencer KM J Sylvia S Clauson and J Janni SPIE 4577 158-165 (2001) 12 Tessier P S Christesen K Ong E Clemente A Lenhoff E Kaler and O Velev Applied Spectroscopy 56
1524-1530 (2002) 13 Farquharson S WW Smith S Elliott and JF Sperry SPIE 3533 207-214 (1998) 14 Farquharson S WW Smith S Elliott and JF Sperry SPIE 3855110-116 (1999) 15 Farquharson S WW Smith YH Lee S Elliott and JF Sperry SPIE 4575 62-72 (2002) 16 Guzelian AA J Sylvia J Janni S Clauson and KM Spencer SPIE 4577 182-192 (2001) 17 Shende C F Inscore A Gift P Maksymiuk and S Farquharson in press 18 Weaver MJ S Farquharson and MA Tadayyoni J Chem Phys 82 4867-4874 (1985) 19 Alak AM and T Vo-Dinh Analytical Chemistry 59 2149-2153 (1987) 20 Lee Y and S Farquharson SPIE 4206 140-146 (2000) 21 Farquharson S and Y Lee SPIE 4200-16 (2000) 22 Lee Y S Farquharson and P M Rainey SPIE 3857 76-84 (1999) 23 Lee Y S Farquharson H Kwong and M Shahriari SPIE 3537 252-260 (1998) 24 Wang J M Pumera G Collins and A Mulchandani Analytical Chemistry 74 6121-6125 (2002) 25 Dou X YM Jung Z-Q Cao and Y Ozaki Applied Spectroscopy 53 1440-1447 (1999) 26 Data supplied by S Christesen and K Ewing 27 Kerker M and B Thompson Eds SPIE MS 10 (1990) 28 Inglesby TV DA Henderson JG Bartlett JAMA 287 2236 (2002) 29 FW Janssen AJ Lund and LE Anderson Science 127 26 (1958) 30 Lide DR Ed Handbook of Chemistry and Physics CRC Press 77th Ed (1996-7) 31 Billmann J G Kovacs and A Otto Surf Sci 92 153 (1980) 32 Murray CA and S Bodoff Phys Rev B 32 671 (1985) 33 Kellogg D and J Pemberton J Phys Chem 91 1120 (1987)
SPIE-2003-5269 16
Chemical agent detection by surface-enhanced Raman spectroscopy
Stuart Farquharson Alan Gift Paul Maksymiuk Frank Inscore and Wayne Smith
Real-Time Analyzers 87 Church Street East Hartford CT 06108
Kevin Morrisey and Steven D Christesen US Army SBCCOM Aberdeen Proving Ground MD 21010
ABSTRACT
In the past decade the Unites States and its allies have been challenged by a different kind of warfare exemplified by the terrorist attacks of September 11 2001 Although suicide bombings are the most often used form of terror military personnel must consider a wide range of attack scenarios Among these is the intentional poisoning of water supplies to obstruct military operations in Afghanistan and Iraq To counter such attacks the military is developing portable analyzers that can identify and quantify potential chemical agents in water supplies at microgram per liter concentrations within 10 minutes To aid this effort we have been investigating the value of a surface-enhanced Raman spectroscopy based portable analyzer In particular we have been developing silver-doped sol-gels to generate SER spectra of chemical agents and their hydrolysis products Here we present SER spectra of several chemical agents measured in a generic tap water Repeat measurements were performed to establish statistical error associated with SERS obtained using the sol-gel coated vials Keywords Chemical agents chemical agent detection SERS Raman spectroscopy
1 INTRODUCTION
In the autumn of 2001 terrorism within US borders became a sobering reality While extensive efforts are being implemented to secure the homeland US and Coalition forces in Afghanistan and Iraq are constantly faced with terrorist attacks In addition to car-bombings the military has identified several non-traditional attack scenarios including poisoning of water supplies by chemical warfare agents (CWAs) To counter this threat the Department of Defense has been investigating numerous approaches to detect such attacks under the auspices of the Joint Service Agent Water Monitor (JSAWM) program1 The JSAWM must be capable of identifying the chemical agents with no false-positives and quantifying the agents at microgram per liter concentrations within 10 minutes2 This includes the analysis of drinking water supplies distribution and storage systems Currently colorimetric paper is used to detect agents on-site while gas chromatography combined with mass spectrometry (GCMS) is used in mobile support laboratories However both methods have severe drawbacks The paper changes color in response to contact with many chemicals besides CWAs causing a high incidence of false positives3 while GCMS although very chemically specific requires up to an hour to perform and regular re-calibration456 Vibrational spectroscopy has also been investigated as a potential method of rapidly detecting CWAs7-
11 as early as 1985 in the case of infrared spectroscopy8 and 1988 in the case of Raman spectroscopy11 Again however these techniques also have limitations when it comes to measuring trace poisons in water Infrared spectra would be dominated by the very strong absorption of water which would obscure absorptions by most other chemicals present Whereas Raman spectroscopy is simply not a very sensitive technique and detection limits are typically grams per liter Surface-enhanced Raman spectroscopy (SERS) offers several advantages over conventional vibrational methods and that may provide the necessary sensitivity required for detecting trace quantities of chemical agents in water SERS employs the interaction of surface plasmon modes of metal particles with target analytes to increase scattering efficiency by as much as 1 million times12 In 1987 the potential of this approach to measure CWAs was demonstrated by measuring a series of pesticides13 Several of these organophosphonates have chemical structures similar to CWAs in particular P=O functional groups In the past few years we and others have further explored the ability of SERS to detect CWAs14-17 and even bioagents 18-21 We have been employing silver-doped sol-gels to promote the SER effect
SPIE-2003-5269 17
in these studies The porous silica network of the sol-gel matrix offers a unique environment for immobilizing and stabilizing SER-active metal particles22-25 The silver-doped sol-gels have been coated on the internal walls of standard glass vials to allow rapid SER analysis of small sample quantities without preparation We have measured over 100 chemicals with enhancements of 104 to 106 demonstrated reversible measurements in a flowing system reproducible measurements from vial-to-vial and measurements in multiple solvents including water21-25 Previously we used these vials to perform preliminary measurements of cyanide (CN) and methylphosphonic acid (MPA) Most of the nerve agents form MPA during hydrolysis while Tabun forms CN a chemical agent in its own right In another paper including in these proceedings we examined the limits of detection (LOD) for MPA by measuring a series of concentrations down to 50 mgL and estimated a limit of detection of 100 microgL26 These measurements provide encouragement in that SERS may satisfy the needs of the JSAWM To further establish the viability of SERS in particular silver-doped sol-gels here we present analysis of cyanide mustard and VX in tap water The measurements performed at the US Armyrsquos Edgewood Chemical Biological Center Aberdeen MD also included numerous repeat measurements to establish reproducibility
2 EXPERIMENTAL 2a General All chemicals including potassium cyanide 2-chloroethylethyl sulfide and those used to prepare the silver-doped and gold-doped sol-gel coated vials were obtained and used as received from Sigma-Aldrich All samples were prepared in a chemical hood using HPLC grade water unless otherwise noted (Fischer Scientific Fair Lawn NJ) for SERS measurements Once prepared the samples were transferred into the silver-doped sol-gel vials (Real-Time Analyzers Inc East Hartford CT) The vials were coated in a manner similar to that previously reported by adding ammonium hydroxide to a solution of silver nitrate tetramethyl orthosilicate (TMOS) and methanol22 The two precursor solutions were prepared mixed and transferred to 2-mL glass vials dried and heated After sol-gel formation the incorporated metal ions were reduced with dilute sodium borohydride (1mgmL) which is followed by a water wash to remove residual reducing agent After the resultant analyte solution was introduced the SER-active vial was fixed horizontally to an XY positioning stage (Conix Research Springfield OR) just inside the focal point of an f07 aspheric lens The lens focused the beam into the sample and collected the scattered radiation back along the same axis A dichroic filter (Omega Optical Brattleborough VT) was used to reflect the excitation laser to the lens and pass the Raman scattered radiation collected by the lens An f2 achromat was used to collimate the laser beam exiting a 200 microm core diameter source fiber optic while a second f2 achromat was used to focus the scattered radiation into a 365 microm fiber optic (Spectran Avon CT) A short pass filter was placed in the excitation beam path to block the silicon Raman scattering generated in the source fiber from reflecting off sampling optics and reaching the detector A long pass filter was placed in the collection beam path to block the sample Rayleigh scattering from reaching the detector A 785 nm diode laser (Process Instruments Inc model 785-600 Salt Lake City UT) was used to deliver 100 to 150 mW of power to the sample A Fourier transform Raman spectrometer (Real-Time Analyzers model IRA-785 East Hartford CT) and a silicon photo-avalanche detector (Perkin Elmer model C30902S Stamford CT) were used to acquire the SER spectra
2b Edgewood Chemical Biological Center The surface-enhanced Raman spectral measurements at the US Armyrsquos Chemical Biological center presented here were all performed on September 12 2003 To expedite measurements a plate was machined to hold up to 12 SER-active sample vials (Figure 1) The plate fit a standard XY plate reader that could be programmed Pure KCN bis-(2-chloroethyl)sulfide (distilled mustard gas HD) and ethyl S-2-diisopropylamino ethyl methylphosphonothioate (VX) were obtained on-site and used to prepare 1 gL tap water solutions in a chemical hood with appropriate safety equipment Simulated tap water was prepared by adding 10 chemicals most often found in tap water at appropriate concentrations to distilled water (Table 1) SER measurements were also performed in a chemical hood For added safety the FT-Raman instrument was placed outside the laboratory and 30 foot fiber optic and electrical cables were used to allow remote SERS measurements and plate manipulation For each experiment 1gL samples were prepared and added to 9 individual vials which were then loaded on the plate In some cases a tenth vial was included as a blank
Table 1 Generic Tap Water Composition
Compound mgL NaHCO3 100 CaSO4 27 MgSO4bull7H2O 67 NaNO3 1 Fulvic Acid 1 K2HPO4 07 KH2PO4 03 (NH4)2HSO4 001 NaCl 001 FeSO4 0001 pH 76-78
SPIE-2003-5269 18
A software program was written that allowed selecting the sequence that the vials were measured the number of positions along the length of the vials to measure (1 to 5) and the number of scans to co-add During sample analysis the program displayed the vial being analyzed the point being analyzed and the spectrum as it was being acquired Once all the data was collected a second software program was written to rapidly analyze the data The spectra collected for all the vials on a plate could be loaded at one time and then the spectra for each point could be displayed simultaneously or separately The user could then select the Raman peak to analyze in terms of peak height or area This was accomplished by selecting points on either side of the peak to define a baseline of zero The peak height or area could then be computed for all of the spectra loaded and then exported to a spreadsheet for statistical analysis
Figure 1 A) Vial Holder 6 slots to hold 2 vials each end-to-end B) Measurement Configuration Program user selects vials to measure sequence number of points per vial (1 to 5) and number of scans per point C) Spectral Acquisition Program shows spectrum being collected which vial and position D) Spectra Analysis Program user selects spectra to analyze by plate vial and point (s) as well as two wavenumbers defining the peak and the baseline to subtract The image is of 5 repeat measurements of 10 mgL KCN in generic tap water 16 sec each 100 mW of 785 nm
3 RESULTS AND DISCUSSION Raman and surface-enhanced Raman spectra were obtained for potassium cyanide bis-(2-chloroethyl)sulfide and ethyl S-2-diisopropylamino ethyl methylphosphonothioate representing three classes of chemical agents cyanides mustards and nerve agents respectively Spectra were also obtained for 2-chloroethyl ethyl sulfide (CEES) a structural analogue to HD which was included in the study to aid in assigning spectral bands KCN salt was used for cyanide experiments to avoid the increased hazards of handling HCN gas KCN completely dissolves in water forming its conjugate acid HCN according to its Ka of 615x10-1027 and at a concentration of 1 mgmL results in a pH 107 solution This is significant in that only CN- appears to interact sufficiently with silver to produce a SER spectrum and no spectral signal is observed below pH 726 Figure 2 shows the SER and normal Raman spectra for KCN The SERS spectrum of 1mgml KCN in water shows a single intense somewhat broad feature at
A B
C D
SPIE-2003-5269 19
2100 cm-1 assigned to the single CequivN stretch The band is much sharper in the normal Raman spectra of the solid KCN salt at 2074 cm-1 This band does broaden and shift to 2080 cm-1 in solution (not shown) However the observed SERS frequency is attributed to interaction with silver and low frequency mode at 135 cm-1 attributed to a Ag-CN stretch (not shown) supports this conclusion
cm-1 band to a C-S stretch but the authors concede that it is in fact more likely a C-Cl stretch28 It appears that the most intense Raman bands at 648 692 and 747 cm-1 shift to 620 660 and 720 cm-1 in the SER spectra and are tentatively assigned as above The width of these bands suggests that they overlap underlying spectral features Additional bands in the Raman spectra occur at 972 1034 1049 1263 1286 1423 1442 2865 2935 and 2960 cm-1 Corresponding bands occur in the SER spectra at 964 1015 1054 1286 1410 1447 2865 and 2935 cm-1 Most of these bands are associated with alkane modes specifically the bands at approximately 1040 cm-1 to a C-C stretch 1290 cm-1 to a CH2 in-phase twist 1440 cm-1 to a CH2 wag 2865 cm-1 to a symmetric CH2 stretch and 2965 cm-1 to an asymmetric CH2 stretch The Raman and SER spectra of sulfur mustard were measured at the Edgewood center (Figure 4) Both spectra are largely similar to CEES The C-Cl and C-S bands in the Raman spectrum of HD now occur at 640 655 700 739 and 760 cm-1 and are more resolved possibly due to the increased molecular symmetry Theoretical calculations indicate that the first three bands are due to C-Cl stretching modes and the latter two to C-S stretching modes28 Only the C-Cl bands maintain significant intensity in the SER spectra occurring at 624 and 643 cm-1 which is attributed to the
Prior to measurements of HD CEES was examined by Raman and SER spectroscopy (Figure 3) CEES also known as half-mustard is essentially identical to HD except one of the chlorine end atoms is replaced by a hydrogen atom Again although not as toxic as HD CEES is a blister agent and dilute aqueous samples were prepared using appropriate safety equipment Both the Raman and SER spectra of CEES are similar and dominated by bands between 600 and 800 cm-1 These are associated with C-Cl and C-S stretching modes which are tentatively assigned to 648 and 747 cm-1 in the Raman spectra respectively The shoulder at 630 cm-1 the overlapped band at 660 cm-1 and the strong band at 692 cm-1 could also be due to these modes or their asymmetric counterparts It is worth noting that theoretical calculations assign the 692
Figure 2 A) SER and B) NR spectra of KCN Conditions A) 1 mgml in tap water 100 mW of 785 nm at sample 1-min acquisition time B) solid 300 mW of 785 nm 5-min All spectra are 8 cm-1 resolution
A
B
Figure 3 A) SER and B) NR spectra of CEES Conditions A) 1 vv (10 mgml) in MeOH 100 mW of 785 nm 1-min acquisition time B) neat 300 mW of 785 nm 5-min
A
B
Cl-CH2-CH2-S-CH2-CH3
A
B
Figure 4 A) SER and B) NR of HD Conditions A) 1mgml in tap water B) pure both 100 mW of 785 nm 1-min
Cl-CH2-CH2-S-CH2-CH2-Cl
SPIE-2003-5269 20
expected strong interaction between chlorine and silver and adds support to the assignment of this band to a C-Cl stretch Weaker overlapping bands occur at 670 692 and 724 cm-1 the latter possibly due to C-S stretching modes Again the alkane modes are apparent in the normal Raman spectra of HD but only a broad feature at 1300 to 1450 cm-1 suggests CH2 contributions in the SER spectrum Although the observed bands in the VX spectrum have not been assigned (Figure 4) a computer generated Raman spectrum29 predicts many of the same features with surprising accuracy and are used here Two intense bands at 460 and 530 cm-1 closely match predicted bands at 463 and 546 cm-1 assigned to a CH3-P=O bend and a PO2CS wag Three highly overlapped bands occur at 694 745 and 771 cm-1 matching predicted bands at 713 730 and 760 cm-1 The first
Table 2 Measured SER peak heights for the CN stretch at 2100 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
Figure 5 A) SERS and B) NR spectra of VX Conditions A) 1 vv (10 mgml) in MeOH B) pure sample both 100 mW of 785 nm 1-min acquisition time
A
B
two have been assigned to a C-S stretch and CH2bend respectively while the latter has been attributed to either a P-C stretch or an O-C-C stretch Although the 745 cm-1 band may alternatively be assigned to a C-S stretch based on the previous measurements of CEES and HD The relatively intense bands at 890 1106 1218 1445 and 1465 cm-1 also match predicted bands at 880 1108 1216 1440 and 1464 cm-1 that are assigned to a C-C stretch CH3 rock N-C3 stretch various C-H3 bends and C-H bends respectively Both the computer generated and the measured spectra contain numerous other less intense bands One is worth mentioning A unique band appears at 370 cm-1 that is predicted at 368 cm-1 and corresponds to an O-P=O bend The surface-enhanced Raman spectrum of VX is also rich with spectral features It has the unique low frequency band at 370 cm-1 as well as a second band at 380 cm-1 that is assigned tothe S-P-O bend predicted in the normal Raman spectrum at 388 cm-1 Based on the measured and predicted normal Raman spectra the following SERS assignments are given 460 cm-1 to the CH3-P=O bend 544 cm-1 to the PO2CS wag 738 cm-1 to a C-S stretch (based on arguments above) 890 cm-1
to a C-C stretch 1101 cm-1 to a CH3 rock and 1456
cm-1 to a C-H bend The ability of SERS to measure chemical agents in water containing real-world chemical interferents was tested by using the generic tap water described in Table 1 The ability to reproduce measurements was accomplished by preparing three separate water stock solutions which were used to prepare three sample solutions each of 1 mgmL KCN in the generic tap water The samples defined as vials 1-9 were then measured at 4 points per vial in a semi-random fashion such that errors associated with stock solution preparation and errors associated with instrument drift could be identified No trends were apparent that signified such systematic errors Each spectrum collected consisted of 20 averaged scans taking 16 seconds at 8 cm-1 resolution The laser power at the sample was measured periodically during the day and it ranged from 102 to 105 mW spacccc
spot Vial 1 Vial 2 Vial 41 13754 13004 128192 13519 12692 12909
1 3 13541 12721 126394 13462 12648 12651
ave 13569 12766 12755Vial 6 Vial 3 Vial 5
1 11586 14068 10757 2 11236 14402 115122 3 11376 14546 11559
4 10894 11763 11214ave 11273 13695 11261
Vial 7 Vial8 Vial 91 11198 15804 11023
2 11243 15716 88753 3 11562 15216 11218
4 11655 15084 11077ave 11415 15455 10548
CN AVG STDEV ERR1 mgmL pk ht 12526 1560 1245
stock solution
SPIE-2003-5269 21
The CN measurements consisting of 36 data points produced an average height of 125 for the 2100 cm-1 peak with a standard deviation of 156 or 125 (Table 2) The HD and VX measurements were performed precisely the same way (Tables 3 and 4) For HD the 624 cm-1 peak was used for analysis and it had an average height of 53 with a standard deviation of 068 or 129 while for VX the 544 cm-1 peak was used for analysis and it had an average height of 1051 with a standard deviation of 308 or 293 The greater error in the VX measurements can be somewhat attributed to Vial 4 which produced lower SER signal intensities But removing this vial from the data set changes the standard deviation to 233 only a modest improvement
spot Vial 1 Vial 2 Vial 41 525 609 4682 45 675 484
1 3 527 644 5094 559 807 669
ave 51525 68375 5325Vial 6 Vial 3 Vial 5
1 521 574 575 2 536 53 4492 3 509 508 451
4 565 594 379ave 53275 5515 4635
Vial 7 Vial 8 Vial 91 457 589 413
2 583 497 4273 3 544 505 4
4 528 5 509ave 528 52275 43725
HD AVG STDEV ERR1 mgmL pk ht 530 068 1291
stock solution
spot Vial 1 Vial 2 Vial 41 1464 1034 4672 1485 99 61
1 3 1491 989 5684 1041 777 553
ave 137025 9475 5495Vial 6 Vial 3 Vial 5
1 1058 942 1293 2 697 121 965
2 3 727 1206 11124 689 1254 746
ave 79275 1153 1029Vial 7 Vial 8 Vial 9
1 1127 783 163 2 1358 812 1614
3 3 1371 908 15124 1097 875 1397
ave 123825 8445 153825
VX AVG STDEV ERR1 mgmL pk ht 1051 308 2925
stock solution
4 CONCLUSIONS In this paper we examined the ability of surface-enhanced Raman spectroscopy to reproducibly measure CN HD and VX in tap water without chemical interference Both normal and surface-enhanced Raman spectra were examined to select unique bands suitable to identify and quantify these chemical agents For SER measurements the 2100 cm-1 C-N stretch was used for CN the 624 cm-1 C-Cl stretch was used for HD and the 544 cm-1 PO2CS wag was used for VX It was determined that 1 mgmL samples of each of these chemicals measured 36 times in glass vials coated with a silver-doped sol-gel reproduced measurements with standard deviations of 125 129 and 293 It was further found that the 10 chemicals added to simulate generic tap water did not interfere with or alter the SER spectra It should be noted that the concentrations used in this study were considerably greater than those required by the JSAWM program Current work involves improving SER sensitivity and designing sampling systems with better reproducibility This includes the development of fractal silver and gold structures within the sol-gel matrix and the development of chemically selective sol-gels
ACKNOWLEDGMENTS The authors are grateful for the support of the US Army (DAAD13-02-C-0015 Joint Service Agent Water Monitor program) The authors would also like to thank Janet Jensen Ronald Crosier and Kristina Gonser for helpful discussions
Table 4 Measured SER peak heights for the VX band at 544 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
Table 3 Measured SER peak heights for the HD band at 624 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
SPIE-2003-5269 22
REFERENCES 1 Jensen JL NC Wong W R Loerop SPIE 4775-03 (2001) 2 JSAWM Requirements at wwwsbccomapgeaarmymilRDAecbcrtPRODSERJSAWMjsawmhtml 3 Erickson B ldquoThe Chemical Weapons Convention Redefines Analytical Challengerdquo Anal Chem 70 397A (1998) 4 Johnston RL Hoefler CM Fargo JC and Moberley B ldquoThe Defense Nuclear Agencyrsquos
ChemicalBiochemical Weapons Agreements Verification Technology Research Development Test and Evaluation Program and its Requirements for On-Site Analysisrdquo AT-ONSITE 5-8 (1994)
5 Black RM Clarke RJ Read RW and Reid MT ldquoApplication of GC-MS and GC-tandem MS to the analysis of chemical warfare samples found to contain residues of the nerve agent sarin sulphur mustard and their degradation productsrdquo J Chrom 662 301-321 (1994)
6 USA Army Center for Health Promotion and Preventive Medicine (USACHPPM) Technical Guideline 230A 7 Hoffland LD Piffath RJ Bouck JB rdquoSpectral signatures of chemical agents and simulantsrdquo Optical
Engineering 24 982-984 (1985) 8 Braue EHJ Pannella MGrdquoCIRCLE CELL FT-IR Analysis of Chemical Warfare Agents in Aqueous Solutionsrdquo
App Spectrosc 44 1513-1520 (1990) 9 Tseng C-H Mann CK and Vickers TJ ldquoDetermination of Organics on Metal Surfaces by Raman
Spectroscopyrdquo App Spectrosc 47 1767-1771 (1993) 10 Staff Reporter ldquoUS Army Tests Lidar to Detect Biological Toxinsrdquo Photonic Spectra pg 50 December 1998 11 Christesen SD Raman cross sections of chemical agents and simulants App Spectrosc 42 318-321 (1988) 12 Weaver MJ Farquharson S Tadayyoni MA ldquoSurface-enhancement factors for Raman scattering at silver electrodesrdquo J Chem Phys 82 4867-4874 (1985) 13 Alak AM and Vo-Dinh T ldquoSERS of Organophosphorous Chemical Agentsrdquo Anal Chem 59 2149-2153 (1987) 14 Lee Y Farquharson S ldquoRapid chemical agent identification by SERSrdquo SPIE 4378 21-26 (2001) 15 Farquharson S Maksymiuk P Ong K Christesen S ldquoChemical agent identification by surface-enhanced Raman spectroscopyrdquo SPIE 4577 166-173 (2001) 16 Spencer KM Sylvia J Clauson S and Janni J ldquoSurface Enhanced Raman as a Water Monitor for Warfare
Agents in Waterrdquo SPIE 4577 158-165 (2001) 17 Tessier P Christesen S Ong K Clemente E Lenhoff A Kaler E Velev O ldquoOn-line spectroscopic characterization of sodium cyanide with nanostructured gold surface-enhanced Raman spectroscopy substratesrdquo
App Spectrosc 56 1524-1530 (2002) 18 Farquharson S W W Smith S Elliott and J F Sperry Rapid biological agent identification by surface- enhanced Raman spectroscopy SPIE 3855110-116 (1999) 19 Farquharson S WW Smith YH Lee S Elliott and J F Sperry Detection of biological signatures A
comparison of electrolytic and metal-doped sol-gel surface-enhanced Raman media SPIE 4575 62-72 (2002) 20 Guzelian AA Sylvia J Janni J Clauson S and Spencer KM ldquoSERS of whole cell bacteria and trace levels of biological moleculesrdquo SPIE 4577 182-192 (2001) 21 Shende C Inscore F Gift A Maksymiuk P Farquharson S ldquoRapid dipicolinic acid extraction from Bacillus spores detected by surface-enhanced Raman spectroscopyrdquo App Spectrosc 58 accepted 22 Lee Y and Farquharson S ldquoSERS Sample Vials Based on Sol-Gel Process for Trace Pesticide Analysisrdquo SPIE
4206 140-146 (2000) 23 Farquharson S and Lee Y ldquoTrace Drug Analysis by SERSrdquo SPIE 4200-16 (2000) 24 Lee Y Farquharson S and Rainey P M Surface-Enhanced Raman Sensor for Trace Chemical Detection in
Water SPIE 3857 76-84 (1999) 25 Lee Y Farquharson S Kwong H and Shahriari M ldquoSol-Gel Chemical Sensor for Surface-Enhanced Raman
Spectroscopyrdquo SPIE 3537 252-260 (1998) 26 Farquharson S A Gift P Maksymiuk F Inscore and W Smith ldquopH dependence of methyl phosphonic acid
dipicolinic acid and cyanide by surface-enhanced Raman spectroscopyrdquo SPIE 5269-19 (2003) 27 Lide DR Ed Handbook of Chemistry and Physics CRC Press 77th Ed (1996-7) 28 Sosa C RJ Bartlett K KuBulat and WB Person ldquoA theoretical study of harmonic vibrational frequencies and
infrared intensities of XCH2CH2SCH2CH2X and XCH2CH2SH (= H Cl)rdquo J Phys Chem 93 577-588 (1993) 29 Hameka HF and JO Jenson ldquoComputer-generated predictions of the structure and the IR and Raman spectra of
VXrdquo ERDEC-TR-065 May 1993
- Chemagents Appendicespdf
-
- SPIE2001-4575-Bioagent-SERSapdf
-
- Detection of bioagent signatures A comparison of electrolytic and metal-doped sol-gel surface-enhanced Raman media
- Stuart Farquharson Wayne Smith and Yuan Lee
-
- Real-Time Analyzers 87 Church Street East Hartford CT 06108
-
- ABSTRACT
- 4 CONCLUSIONS
- 5 ACKNOWLEDGEMENTS
- 6 REFERENCES
-
352 Volume 58 Number 3 2004
speci c nucleic acid sequences (eg capsular protein en-coding gene for Bacillus anthracis)4 and polymerases toamplify the segment until it is detectable Recently am-pli cation times have been substantially reduced andcomplete analysis can now be performed in an hour orless Immunoassay methods are also being developed thatuse competitive binding of the bioagent (as an antigen)and its labeled conjugate for a limited number of anti-bodies Although analyses can be performed in under 30minutes a well-de ned anthrax antigen has not yet beenidenti ed5ndash7 and consequently the false-positive rate isunacceptably high8
A number of other methods are being developed witha focus on the detection of calcium dipicolinate (CaDPA)and its derivatives as a B anthracis signature This is sobecause only spore-forming bacteria contain CaDPA andthe most common potentially interfering spores such aspollen and mold spores do not Relatively fast methodshave been developed to chemically extract CaDPA andthen detect it directly by uorescence9 or indirectly byluminescence1011 In the latter case hot dodecylamine(DDA) has been used to extract dipicolinic acid (DPA)and terbium has been utilized to form a highly lumines-cent DPA complex11 Although measurements have beenperformed in as little as ve minutes it was found thatas many as three concentration-dependent complexes canform each with different lifetimes This coupled withthe fact that the Tb31 cation produces the same lumines-cence spectrum makes determinations of low spore con-centrations problematic
It has been long known that Raman spectra of Bacillispores are dominated by bands associated with CaDPA12
and that these spectra may provide a suitable anthrax sig-nature at the genus level13 Since that time considerableimprovements in Raman instrumentation have led to lab-oratory measurements of single Bacilli spores14 and to eld measurements of spores captured from a mail-sort-ing system15 However the single spore measurementsrequired complex instrumentation that is not ruggedwhile the eld measurements required milligrams of sam-ple Furthermore the Raman spectra of both measure-ments contained uorescence contributions that would in-crease uncertainty in quanti cation
In related research we demonstrated that nanogramquantities of DPA could be detected by uorescence-freesurface-enhanced Raman spectroscopy (SERS)16 We alsodemonstrated that microliter volumes of chemicals canbe detected by SERS using metal-doped sol-gel-packedglass capillaries17 Towards the goal of developing a rap-id eld SERS-based anthrax spore detector we havecombined our previous research and we now report thatDPA can be extracted from a 10 mg B cereus spore sam-ple using DDA in 1 minute and can be detected by SERSin an additional 1 minute
EXPERIMENTAL
Dipicolinic acid (26-pyridinedicarboxylic acid DPA)and dodecylamine (DDA) were used as received fromSigma-Aldrich (Milwaukee WI) Lyophilized B cereusspores prepared according to the literature13 were sup-plied by the University of Rhode Island and used as re-ceived Multiple particles approximately 01 mm3 each
were separated and weighed at 5 to 15 mg representing05 to 15 million spores The sample masses were con-sistent with a previous determination of spore density at0081 gmL that indicated a high degree of entrained air
All chemicals used to prepare the silver-doped sol-gelcoated capillaries were also obtained and used as receivedfrom Sigma-Aldrich According to previously publishedprocedures17 two precursor solutions were preparedmixed and then drawn into 1-mm-diameter glass capil-laries The silver amine precursor consisted of a 51 vvratio of 1 N AgNO 3 to 28 NH3OH while the alkoxideprecursor consisted of a 21 vv ratio of methanol to te-tramethyl orthosilicate The alkoxide precursors weremixed with silver amine precursor in an 81 vv ratioApproximately 015 mL was drawn into the capillarycoating a 15-mm length After sol-gel formation the in-corporated silver ions were reduced with dilute sodiumborohydride which was followed by a water wash to re-move residual reducing agent
A 100 mL drop of a 50 mM DDA solution in ethanolpre-heated to 78 8C was added to each of the B cereusparticles to digest the spore coat After 1 minute the re-sultant solution was drawn into a SER-active capillarythat was immediately xed horizontally to an XY posi-tioning stage (Conix Research Spring eld OR) just in-side the focal point of an f 07 aspheric lens The lensfocused the beam into the sample and collected the scat-tered radiation back along the same axis A dichroic lter(Omega Optical Brattleborough VT) was used to re ectthe excitation laser to the lens and pass the Raman scat-tered radiation collected by the lens An f 2 achromatwas used to collimate the laser beam exiting a 200-mm-core-diameter source ber optic while a second f 2 ach-romat was used to focus the scattered radiation into a 365mm ber optic (Spectran Avon CT) A short-pass lterwas placed in the excitation beam path to block the sil-icon Raman scattering generated in the source ber fromre ecting off sampling optics and reaching the detectorA long-pass lter was placed in the collection beam pathto block the sample Rayleigh scattering from reachingthe detector A 785 nm diode laser (Process InstrumentsInc model 785-600 Salt Lake City UT) was used todeliver 100 to 150 mW of power to the sample A Fouriertransform Raman spectrometer (Real-Time Analyzersmodel IRA-785 East Hartford CT) and a silicon photo-avalanche detector (Perkin Elmer model C30902S Stam-ford CT) were used to acquire the SER spectra
RESULTS AND DISCUSSION
As an initial experiment the SER spectrum of 1 gLof DPA in water was measured using the newly devel-oped silver-doped sol-gel-coated capillaries (Fig 1A) Atthis concentration a high signal-to-noise ratio (SN) isobtained in 1 min In fact a reasonable spectrum is ob-tained in the same time frame for 1 mgL (Fig 1B) TheSER spectra are reasonably similar to the normal Raman(NR) spectrum obtained for a saturated solution of DPAin 1 N KOH (Fig 1C) and the following band shifts areobserved (NR to SER) 647 to 657 cm21 817 to 815cm21 998 to 1008 cm21 1384 to 1382 cm21 1434 to1428 cm21 and 1569 to 1567 cm21 Many of these bandshave been previously assigned1213 such as 998 cm21 to
APPLIED SPECTROSCOPY 353
FIG 1 SERS of DPA in water using silver-doped sol-gel-coated glasscapillary for (A) 1 gL and (B) 1 mgL (C ) NR of saturated DPA in 1N KOH in a glass capillary Spectral conditions (A) and (B) 150 mWof 785 nm 1-min acquisition time (C) 450 mW of 785 nm 5-minacquisition time both 8 cm 21 resolution
FIG 2 SERS of DPA extracted from 10 mg B cereus particle using100 mL of 50 mM hot DDA acquired in (A) 1 minute and (B) 2 seconds(C ) Attempted SERS of 50 mM hot DDA in ethanol using silver-dopedsol-gel-coated glass capillary acquired in 1 min Spectral conditions150 mW of 785 nm 8 cm21 resolution
the symmetric ring stretch 1384 cm21 to the OndashCndashOsymmetric stretch 1428 cm21 to the symmetric ring CndashH bend and 1569 cm21 to the asymmetric OndashCndashOstretch
The rst B cereus samples consisted of 2 mg of sporesin 2 mL of 5 mM hot DDA The samples were main-tained at 78 8C for 40 min and while hot approximately10 mL was drawn into a SER-active capillary Since spec-tra of DPA were obtained for these initial samples small-er spore masses higher DDA concentrations and shorterheating periods were examined In due course it wasfound that 10 mg of spores could be digested by 100 mLof 50 mM hot DDA in one minute and detected (Fig2A) In fact the signal was suf ciently intense that it canbe observed in as little as two seconds (Fig 2B) Theamount of DPA that was extracted was estimated to bebetween 5 and 10 mgL by comparing the signal intensityof the 1008 cm21 band to that measured for DPA in waterThis is consistent with previous research that found thatthe majority of the DPA is extracted from spores usingDDA11 and that B cereus spores contain approximately10 DPA by weight18 The SN of 127 for the 1008 cm21
band in the 1-minute SER spectrum suggests a limit ofdetection of approximately 250 ng of B cereus sporesbased on a SN of 3 Finally it should be noted that DDAdid not produce a detectable SER spectrum as shown inFig 2C
CONCLUSION
We have demonstrated that by combining rapid extrac-tion of dipicolinic acid from Bacillus cereus spores withchemical identi cation by surface-enhanced Raman spec-troscopy as little as 10 mg of spores can be detected Infact the entire measurement from the time of adding hotdodecylamine to the spores to the time when the dipi-colinic acid SER spectrum is acquired and analyzedcould be performed in less than two minutes The abilityof this method to distinguish between spore-forming bac-teria such as Bacillus anthracis and non-DPA containingpowders could help prevent costly shutdowns associated
with the appearance of suspicious material or intentionalmailing of common substances as an anthrax hoax Thismethod could also prove useful in detecting the locationof anthrax endospores in mail distribution facilities if an-other veri ed attack should occur
Research continues to fully characterize the surface-enhanced Raman spectroscopy signal intensities as afunction of sample concentration and to explore otherextractants that do not require the use of elevated tem-perature
ACKNOWLEDGMENTS
The authors are grateful for the support of the National Science Foun-dation (DMI-0296116 and DMI-0215819) and the US Army(DAAD13-02-C-0015 Joint Service Agent Water Monitor program)The authors are indebted to Chetan Shende for preparing the sol-gelcapillaries The authors also thank James Gillespie Nicholas Fell andAugustus Fountain for providing important background informationMark Farquharson for laboratory support and Professor Jay Sperry ofthe University of Rhode Island for supplying B cereus spores
1 V A Pasechnik C C Shone and P Hambleton Bioseparations 3267 (1993)
2 P J Jackson M E Hugh-Jones D M Adair G Green K K HillC R Kuske L M Grinberg F A Abramova and P Keim ProcNatl Acad Sci USA 95 1224 (1998)
3 B R Glick and J J Pasternak Molecular Biology Principles andApplications of Recombinant DNA (ASM Press Washington DC1994)
4 C A Bell J R Uhl T L Had eld J C David R F Meyer TF Smith and F R Cockerill III J Clin Microbiol 40 2897(2002)
5 D L Gatto-Menking H Yu J G Bruno M T Goode M Millerand A W Zulich Biosens Bioelectron 10 501 (1995)
6 J J Quinlan and P M Foegeding J Rapid Methods AutomationMicrobiol 6 1 (1998)
7 A A Hindle and E A H Hall Analyst (Cambridge UK) 1241599 (1999)
8 M S Ascher US Department of Health amp Human Services(httpwwwhhsgovophppresentationsAscherdoc)
9 R Nudelman B V Bronk and S Efrima Appl Spectrosc 54 445(2000)
10 D L Rosen C Sharpless and L B McBrown Anal Chem 691082 (1997)
354 Volume 58 Number 3 2004
11 P M Pellegrino N F Fell Jr and J B Gillespie Anal ChimActa 455 167 (2002)
12 W H Woodruff T G Spiro and C Gilvarg Biochem BiophysRes Commun 58 197 (1974)
13 E Ghiamati R S Manoharan W H Nelson and J F SperryAppl Spectrosc 46 357 (1992)
14 A P Esposito C E Talley T Huser C W Hollars C M Schal-dach and S M Lane Appl Spectrosc 57 868 (2003)
15 S Farquharson L Grigely V Khitrov W W Smith J F Sperryand G Fenerty J Raman Spectrosc paper accep ted (2003)
16 S Farquharson W W Smith S Elliott and J F Sperry SPIE-IntSoc Opt Eng 3855 110 (1999)
17 S Farquharson and P Maksymiuk Appl Spectrosc 57 479(2003)
18 F W Janssen A J Lund and L E Anderson Science (Washing-ton DC) 127 26 (1958)
SPIE -2003-5269 117
pH dependence of methyl phosphonic acid dipicolinic acid and cyanide by surface-enhanced Raman spectroscopy
Stuart Farquharson Alan Gift Paul Maksymiuk Frank Inscore and Wayne Smith
Real-Time Analyzers 87 Church Street East Hartford CT 06108
ABSTRACT US and Coalition forces fighting terrorism in Afghanistan and Iraq must consider a wide range of attack scenarios in addition to car bombings Among these is the intentional poisoning of water supplies to obstruct military operations To counter such attacks the military is developing portable analyzers that can identify and quantify potential chemical agents in water supplies at microgram per liter concentrations within 10 minutes To aid this effort we have been investigating the value of a surface-enhanced Raman spectroscopy based portable analyzer In particular we have been developing silver-doped sol-gels to generate SER spectra of chemical agents and their hydrolysis products Here we present SER spectra of methyl phosphonic acid and cyanide as a function of pH an important factor affecting quantitation measurements which to our knowledge has not been examined In addition dipicolinic acid a chemical signature associated with anthrax-causing spores is also presented Keywords Chemical warfare agents agent detection agent hydrolysis SERS Raman spectroscopy homeland security
1 INTRODUCTION In the past decade the Unites States and its allies have been challenged by a different kind of warfare exemplified by the terrorist attacks of September 11 2001 Suicide bombings and the use of chemical agents are the norm and military personnel must consider a wide range of attack scenarios Among these is the intentional poisoning of water supplies to obstruct military operations in Afghanistan and Iraq Currently colorimetric paper is used to detect agents on-site while gas chromatography combined with mass spectrometry (GCMS) is used in mobile support laboratories However both methods have severe drawbacks The paper changes color in response to contact with many chemicals besides CWAs causing a high incidence of false positives1 while GCMS although very chemically specific requires hours to perform and constant re-calibration234 Military operations would be greatly aided by a portable analyzer that can identify and quantify potential chemical agents at concentrations that impact safety This includes the analysis of drinking water supplies distribution and storage systems To meet this goal the Department of Defense has been investigating numerous approaches under the auspices of the Joint Service Agent Water Monitor (JSAWM) program5 The JSAWM must be capable of identifying the chemical agents with no false-positives and quantifying the agents at microgram per liter concentrations within 10 minutes678 Recently we and others have been investigating the ability of surface-enhanced Raman spectroscopy (SERS) to measure chemical agents9-12 bioagents13-17 and their hydrolysis products in water SERS employs the interaction of surface plasmon modes of metal particles with target analytes to increase scattering efficiency by as much as 1 million times18 In 1987 the potential of this approach to measure CWAs was demonstrated by measuring a series of pesticides19 Several of these organophosphonates have chemical structures similar to CWAs in particular P=O functional groups In our studies we have been employing silver-doped sol-gels to promote the SER effect The porous silica network of the sol-gel matrix offers a unique environment for immobilizing and stabilizing SER-active metal particles20-23 The sol-gels have been coated on the internal walls of standard glass vials to allow rapid SER analysis of small sample quantities without preparation We have measured over 100 chemicals with enhancements of 104 to 106 demonstrated reversible measurements in a flowing system reproducible measurements from vial-to-vial and measurements in multiple solvents including water20-23 Previously we used these vials to perform preliminary measurements of cyanide (CN) methylphosphonic acid (MPA) and dipicolinic acid (DPA) MPA is a hydrolysis product of the nerve agents (eg sarin Reaction 1) and may be a valuable indicator of nerve agent usage particularly since the alkyl methylphosphonic acids are relatively more stable than their corresponding parent complexes24 DPA is
SPIE -2003-5269 118
a chemical signature of spore forming bacteria such as Bacillus anthracis And in light of the inability to rapidly detect the anthrax spores distributed through the US mail in October 2001 a number of methods are being developed to extract and analyze this signature Reaction 1 Stepwise hydrolysis of Sarin to form hydrofluoric acid (HF) isopropyl methylphosphonic acid (IMPA) then methyl phosphonic acid (MPA) and 2-propanol In our previous SERS investigations MPA and DPA were measured at 50 and 100 mgL respectively In both cases limits of detection (LOD) were estimated at 100 microgL providing encouragement in that SERS may satisfy the needs of the JSAWM Since it has been shown that pH can substantially influence the intensity of SER bands25 which would clearly influence quantitative analysis we undertook the present study to determine the severity of these effects for cyanide methyl phosphonic acid and dipicolinic acid Furthermore we previously observed a band at 1050 cm-1 for MPA6 possibly due to an anion formed at basic pH Here we investigate the source of this spectral anomaly
2 EXPERIMENTAL All chemicals including potassium cyanide methyl phosphonic acid dipicolinic acid and those used to prepare the silver-doped and gold-doped sol-gel coated vials were obtained and used as received from Sigma-Aldrich All samples were prepared in HPLC grade water (Fischer Scientific Fair Lawn NJ) for SERS measurements The pH of these samples was adjusted using dilute nitric acid or potassium hydroxide and verified using a pH electrode (Corning Inc Corning NY) that had been calibrated with pH 400 700 and 1000 buffered standards from Fischer Scientific Once prepared the samples were transferred into the silver-doped sol-gel vials (Real-Time Analyzers Inc East Hartford CT) The vials were coated in a manner similar to that previously reported by adding ammonium hydroxide to a solution of silver nitrate tetramethyl orthosilicate (TMOS) and methanol20 Gold-vials were coated by adding nitric acid to a solution of gold tetrachloride TMOS and methanol The two precursor solutions were prepared mixed and transferred to 2-mL glass vials dried and heated After sol-gel formation the incorporated metal ions were reduced with dilute sodium borohydride (1mgmL) which was followed by a water wash to remove residual reducing agent After the resultant analyte solution was introduced the SER-active vial was immediately fixed horizontally to an XY positioning stage (Conix Research Springfield OR) just inside the focal point of an f07 aspheric lens The lens focused the beam into the sample and collected the scattered radiation back along the same axis A dichroic filter (Omega Optical Brattleborough VT) was used to reflect the excitation laser to the lens and pass the Raman scattered radiation collected by the lens An f2 achromat was used to collimate the laser beam exiting a 200 microm core diameter source fiber optic while a second f2 achromat was used to focus the scattered radiation into a 365 microm fiber optic (Spectran Avon CT) A short pass filter was placed in the excitation beam path to block the silicon Raman scattering generated in the source fiber from reflecting off sampling optics and reaching the detector A long pass filter was placed in the collection beam path to block the sample Rayleigh scattering from reaching the detector A 785 nm diode laser (Process Instruments Inc model 785-600 Salt Lake City UT) was used to deliver 100 to 150 mW of power to the sample A Fourier transform Raman spectrometer (Real-Time Analyzers model IRA-785 East Hartford CT) and a silicon photo-avalanche detector (Perkin Elmer model C30902S Stamford CT) were used to acquire the SER spectra
3 RESULTS AND DISCUSSION In a previous study of MPA6 aimed at developing a concentration calibration curve and determining limits of detection (LOD) we observed an anomaly at 1050 cm-1 Since it was found that the band intensity changed as a function of concentration the band must be associated with a sample parameter Two possible parameters photon flux and pH are examined here The first parameter was investigated by irradiating a 1mgmL MPA sample in a SER-active vial with laser powers of 200 mW and above and monitoring spectral changes It was immediately found that the 1050 cm-1 band
2O+ H HF + +OH
OH3H C
OP
OF
CH
CH3
3
3H C
OCP
OOH
CH
CH3
3
3H C
O
CPHO
CH
CH
3
3
C
Sarin IMPA MPA 2-propanol
SPIE -2003-5269 119
appeared and grew as a function of time and that the higher the laser power the faster the growth Figure 1 shows the growth of the 1050 cm-1 band over the course of 30 minutes when using 150 mW of 785 nm excitation while Figure 2A shows that the growth can be fit with a first order exponential equation namely I1050 = 03+05e-013t Furthermore the 760 cm-1 band could be fit with a first order decay equation with an identical exponential rate constant ie I760 = 08-
08e-013t The rates represent classical first order kinetics and their correspondence allows one to conclude that MPA is being transformed one-for-one into a photo-generated product At this time the photoproduct has not been positively identified but phosphonic (phosphorous) acid and phosphonate are likely candidates since the symmetric P(OH)3 stretch occurs at ~1050cm-1 Our studies show that a reduction in laser power to 100 mW at the sample for MPA essentially eliminates this degradation process This laser power or lower was used for further measurements
Methyl phosphonic acid is a diprotic acid that stepwise dissociates into two anions MPA- and MPA= according to the following reactions26
MPA MPA- + H+ pKa1 = 212 Reaction 2
MPA- MPA= + H+ pKa2 = 729 Reaction 3 The relative concentrations of MPA MPA- and MPA= can be determined at any pH by expressing [MPA] and [MPA=] in terms of [MPA-] using Reactions 2 and 3 and summing all three to equal the total starting concentration here 2 mgmL (0021M MW = 9602) viz
[MPA] + [MPA-] + [MPA=] = 0021M Equation 1 substituting from Reactions 1 and 2
([H+][MPA-])K1a + [MPA-] + (K2a[MPA-])[H+] = 0021M Equation 2 rearranging [MPA-] = 0021M(1+[H+]K1a + K2a[H+]) Equation 3 The relative concentrations of MPA MPA- and MPA= as a function of pH are shown in Figure 3 It is worth noting that near neutral pH both MPA- and MPA= will be present To confirm that the SER signal followed this pH dependence a starting solution consisting of 20 mg of MPA in 10 mL HPLC grade water was prepared and brought to pH of 20 using dilute nitric acid From this solution 2 mL were added to a SER-active vial and the SER spectrum recorded At this pH a peak at 760 cm-1 was barely discernable The 2 mL solution was returned to the starting solution and the pH was re-measured to correct for any changes that the silver-doped sol-gel vials might cause In most cases the change was less than 02 pH units and the pH is reported as the before and after average Next the pH of the
Figure 2 A) Exponential growth of 1050 cm-1 band and B) exponential decay of 760 cm-1 band for spectral series in Figure 1
Figure 1 Growth of 1050 cm-1 band as a function of time due to exposure to 150 mW of 785 nm Spectra are 5 sec each collected every 100-sec from 0 to 30-min
B
570 770 970 1170 1370Raman Shift (cm-1)
Arbit
rary
Unit
s
MPA Photodegradation
600 800 1000 1200 1400 Wavenumber (cm-1)
A
0 10 20 30 0 10 20 30 time (min) time (min)
Ram
an In
tens
ity (r
elat
ive)
30
min
0
SPIE -2003-5269 120
starting solution was adjusted to 325 using dilute KOH Again 2 mL were added to a vial and the SER spectrum recorded At this pH a reasonably strong 760 cm-1 band was observed This process was repeated as spectra were recorded at pHs of 70 74 75 79 85 and 100 A total of 1 mL of KOH was added diluting the total concentration by 10 Next the pH of the starting solution was made acidic by adding dilute nitric acid dropwise This time spectra were recorded at pHs of 72 69 64 and 37 Figure 4 shows the SER spectra for representative pHs (spectra were left out to simplify the figure) while Figure 3 shows the 760 cm-1 peak intensities as a function of pH (The band intensities were adjusted to compensate for dilution effects caused by the addition of HNO3 and KOH then normalized to 0021 M for the most intense band observed at pH 37) It is clear from Figure 3 that the 760 cm-1 band follows the MPA- concentration as a function of pH and must be assigned to this anion No bands were observed that corresponded to MPA or MPA= The lack of an MPA SER spectrum may be due to the absence of an attraction between the neutral analyte and the electropositive silver surface The same reasoning suggests that a strong SER spectrum should be observed for MPA= but it is not and a satisfactory explanation has not been found
0000
0005
0010
0015
0020
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14pH
Con
cent
ratio
n [M
]
pK1 = 212 pK2 = 729
MPA- MPA=
MPA
Figure 4 SER spectra of 002M MPA as a function of pH Conditions 100 mW of 785 nm 36 scans (1 min) 8 cm-1 recorded 2 min after sample introduction pH 19 and 20 not apparent on this scale pH 69 and 74 near identical to 70 and 75 and not shown for clarity
00
02
04
06
08
10
12
14
16
18
0 5 10 15 20 25 30Measurement Number
Ram
an In
tens
ity (7
60 c
m-1
)
Figure 3 Concentration dependence of MPA MPA- and MPA= as a function of pH for a 002M sample Intensity of 760 cm-1
band from Figure 6 as a function of pH ( for increasing basic adjustment for increasing acidic adjustment error was measured at ~10 for pH 64)
Figure 5 SER spectra of 002M MPA at pH 64 measured around a vial at three heights (9 points per height) Conditions as in Figure 4 but 10-sec scans
Since these measurements involved the removal and replacement of the SER-active vial in the sample holder to remove and add sample variation in the intensity as a function of vial position was minimized by illuminating the exact same height along the vial wall But this does not account for variability of the SERS response of the sol-gel coating around the vial To analyze this effect a vial containing MPA at pH 64 was rotated at ~ 40o
intervals at the original height and 18rdquo above and below this value Figure 5 shows the intensity of the 760 cm-1
band for the 27 positions It was found that the average value was 137plusmn014 an RSD of 10 overall and 5 for each height An error bar is included in Figure 4 for the pH 64 measurement
SPIE -2003-5269 121
With the above analyses in mind a preliminary investigation of the SER spectral response for MPA (as MPA-) as a function of concentration was performed A single vial was used for these measurements beginning with 1 mgL followed by measurements of 10 100 and 1000 mgL In all cases the pH was ~7 and 3 positions around the vial were measured per concentration Since the 760 cm-1 band was not observed for concentrations of 1 or 10 mgL using 100 mW of 785 nm the laser power at the sample was raised to 200 mW beginning with the 10 mgL concentration Photo-degradation was largely avoided (and not observed) by exposing the sample for only 33 seconds per spectral acquisition Representative spectra for 10 100 and 1000 mgL are shown in Figure 6 while a plot of the 760 cm-1 band intensity as a function of concentration is shown in Figure 7 These values were also used to estimate limits of detection based on the signal-to-noise ratio (SN) of the 760 cm-1 band using 100 mW of 785 nm laser excitation and a 10-min acquisition time As summarized in the Figure 7 inset the lower the measured concentration the lower the predicted LOD Even if as estimated 210 microgL could be measured using the silver-doped sol-gel vials an improvement of a factor of 70 is still required to meet JSAWM goals of measuring 3 microgL in 10-minutes Similar to MPA DPA is a diprotic acid (pKa1 = 216 and pKa2 = 692) and variations in pH will effect the relative concentrations of DPA DPA- and DPA= and possibly the SER spectra and band intensities This could prove significant if an acid or base is used to denature anthrax spores with the goal of extracting and analyzing DPA The MPA pH study described above was mimicked for DPA except that the starting solution consisted of 20 mg of DPA in 20 mL HPLC grade water (60x10-3M MW = 1671) The initial solution had a pH of 245 which was made basic by dilute KOH to pHs of 355 433 487 559 1069 and 1166 SER spectra were recorded at each pH using 100 mW of 785 nm and a 44-sec acquisition time Next one drop of concentrated nitric acid was used to remake the solution acidic at a pH of 200 Again sequential pH measurements were performed at 383 510 735 and 822 The solution pH was made acidic a third time but to pH 219 171 then 135 Throughout this process no more than 20 drops of acid or base were added and therefore the concentration was diluted by no more than 10 Most of the spectral bands showed a minor decrease in intensity as a function of increasing pH values However the bands at 525 and 795 cm-1 showed the most dramatic changes which occurred at acid pH Figure 8 shows the SER spectra of DPA for the spectral region and pH range of interest The identity of the DPA species was determined by plotting the normalized peak intensities with the lowest value set to 0 and the highest to 0006 M as a function of pH and overlaying these values on a plot of the relative concentrations for DPA DPA- and DPA= as previously done for MPA (Figure 9) As can be seen the 525 cm-1 band clearly corresponds to DPA The correspondence of the 795 cm-1 band to this species is less clear as the band retains intensity until pH of 55 This can be attributed to contribution to the overlapping band at 810 cm-1 which does not change as a function of pH The fact that most bands are observed at all pHs suggest that the primary interaction with silver is through the ring nitrogen This is supported by the fact that the most intense band occurs at 1008 cm-1 attributed to a symmetric ring breathing mode and that this interaction has been characterized for pyridine in numerous papers27
0
1
2
3
4
5
6
7
0 200 400 600 800 1000 1200
MPA Concentration (mgL)
760
Ban
d In
tens
ity
Figure 6 SER spectra of MPA in water at A) 1000 B) 100 and C) 10 mgL Conditions pH of 7 silver-doped sol-gel coated vial 200 mW of 785 nm 33-sec 8 cm-1 resolution
Figure 7 Plot of SER intensity of 760 cm-1 band of MPA as a function of concentration using 200 mW of 785 nm Inset table includes average intensity LOD standard deviation and percent deviation for each concentration but for 100 mW and 10 min
conc (mgL) LOD ave stddev dev10 021 002 1021100 072 010 14301000 312 040 1280
A
B C
SPIE -2003-5269 122
The ability to detect dipicolinic acid (DPA) by SERS is demonstrated in Figure 10 Even at 1 mgL the primary bands are visible The intensity of the dominant ring breathing mode observed at 1008 cm-1 for DPA in water (pH = 25 to 55) is plotted as a function of concentration in Figure 11 These values were also used to estimate limits of detection based on the SN of the 1008 cm-1 band using 100 mW of 785 nm laser excitation and a 10-min acquisition time Again the lower the measured concentration the lower the predicted LOD (see Figure 11 inset) and detection of 160 microgL is possible Gastrointestinal anthrax requires significant more spores than inhalation anthrax28 and a limit of detection might be placed at 1 million spores in 1 liter of water or 10 microgL Since each spore contains ~10 CaDPA by weight29 a goal for DPA might be 1 microgL indicating that the present measurements must be improved by nearly two orders of magnitude Due to the increased hazards of handling HCN gas KCN salt was used for these experiments Nevertheless all sample preparations were performed in a chemical hood KCN completely dissolves in water but its conjugate acid HCN is formed and has a Ka of 615x10-1030 viz
HCN CN- + H+ pKa = 921 Reaction 4
Figure 8 SER spectra of 1 mgmL DPA as a function of pH Conditions 100 mW 785 nm 100 scans (44-sec)
Figure 9 Concentration dependence of DPA DPA- and DPA= as a function of pH for a 0006M sample Intensity of 525 () and 795 (diams) cm-1 bands from Fig 8 as a function of pH
pH
135 171 219 383
0
0001
0002
0003
0004
0005
0006
0007
0 2 4 6 8 10 12 14pH
Con
cent
ratio
n [M
]
DPADPA-DPA=795525
DPA DPA=DPA-
pK2 = 692pK1 = 216
795 525
0
05
1
15
2
25
0 200 400 600 800 1000 1200
DPA Concentration (mgL)
1008
Ban
d In
tens
ity
conc (mgL) lod-10min-100mw1 017
10 016100 103
1000 355
Figure 10 SER spectra of DPA in water at A) 1000 B) 100 C) 10 and D) 1 mgL Conditions pH of 25-55 silver-doped sol-gel coated vial 175 mW of 785 nm 1-min 8 cm-1 D) has been multiplied by x10 to make bands visible
A
B C
D
Figure 11 Plot of SER intensity of 1008 cm-1 band of DPA as a function of concentration using 175 mW of 785 nm Inset table includes LOD in mgL for each concentration but for 100 mW and 10 min
SPIE -2003-5269 123
Consequently the cyanide concentration must be determined for each initial KCN concentration Specifically the samples prepared with concentrations of 01 1 10 100 and 1000 mgL of KCN produced CN- concentrations of 63x10-3 033 69 89 and 964 mgL at pHs of 816 90 967 102 and 107 respectively The pH dependence for the HCN and CN- concentrations are shown in Figure 12 Thus as the amount of KCN added to the solution decreases so does the pH of the solution (becomes less basic) and according to Reaction 4 the relative amount of CN- to HCN also decreases For example in the preparation of a 01 mgL solution of KCN the pH is shifted from 7 for pure water to only 816 and only 63 of the starting material becomes CN- or 63x10-3 mgL In comparison for a solution of 1000 mgL the pH is shifted from 7 to 107 and 96 of the starting material becomes CN- This is significant since the cyanide ion is better able to adsorb onto the silver particles and become SERS active SER spectra of 10 100 and 1000 mgL of KCN samples are shown in Figure 13 and a corresponding plot of the 2100 cm-1 band intensity as a function of concentration is shown in Figure 14
000102030405060708091011
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14pH
Con
cent
ratio
n [m
gm
L]
pKa = 921
CN -HCN
The SER spectra of cyanide are dominated by the single CN stretch at 2100 cm-1 which occurs in normal Raman spectra of solutions at 2080 cm-1 However a low frequency mode occurs at 135 cm-1 due to a Ag-CN stretch (not shown) indicative of a strong surface interaction It is also observed that as the concentration decreases the CN stretch shifts to 2140 cm-1 This shift has been attribute to the formation of a tetrahedral Ag(CN)3
2- surface structure31 as well
Figure 14 Concentration dependence of KCN SERS measured under conditions in Fig 11 Concentrations are 1 01 and 001 mgml Intensities are measured for the CN stretch at 2100 cm-1 Inset table includes LOD in mgL for each concentration in Figs 13 and 15 but for 100 mW and 10 min
Figure 13 SER spectra of KCN in water at A) 1000 B) 100 and C) 10 mgL Conditions pHs of 107 102 and 97 silver-doped sol-gel coated vial 100 mW of 785 nm 1-min 8 cm-1 C) has been multiplied by x10 to make band visible
Figure 12 Concentration dependence of HCN and CN- as a function of pH for a 1 mgmL sample Calculated intensity of 2100 () cm-1 band for a 1 mgml sample at pHs of 816 90 967 102 and 107
A
B
C
0
20
40
60
80
100
120
140
0 200 400 600 800 1000 1200CN Concentration (mgL)
2100
Ban
d In
tens
ity
Figure 15 SER spectra of KCN in water at A) 10 B) 1 and C) 01 mgL Conditions pHs of 97 12 and 12 gold-doped sol-gel coated vial 100 mW of 785 nm 1-min 8 cm-1 C) has been multiplied by x10 to make band visible
A
B
C
conc (mgL) lod-10min-100mw Condition01 001 Au-pH 121 007 Au-pH 12
69 003 Au-pH 9769 016 Ag-pH 9789 022 Ag-pH 102946 113 Ag-pH 107
SPIE -2003-5269 124
as to CN adsorbed to two different surface sites32 Alternatively the 2140 cm-1 band could be attributed to HCN since this species dominates at lower concentrations However it is unlikely that this species would be attracted to the electropositive silver surface Further both peaks should be present at pHs between 85 and 105 but this is not observed It has also been suggested that at concentrations near and above monolayer coverage the CN- species is forced to adsorb end-on due to crowding and at lower concentrations the molecule can reorient to lie flat33 This suggests that the 2100 and 2140 cm-1 bands correspond to the end-on and flat orientations respectively As Figures 13 and 14 show the intensity of the CN stretch for the 89 mgL sample is nearly as intense as the 964 mgmL band This suggests that the Raman signal for the flat orientation is more enhanced However more extensive measurements are required to verify this point Since resent research has suggested that cyanide may be more effectively detected on gold measurements of KCN solutions were also performed using gold-doped sol-gel vials Preliminary measurements are shown in Figure 15 for samples prepared from 01 1 and 10 mgL KCN Since the pHs are 816 90 967 the resultant CN- concentrations are 63x10-3 033 and 69 mgL Initially only the highest concentration was observed and the signal intensity was significantly better than the equivalent concentration measured using silver In an effort to shift Reaction 4 to the left transforming HCN to CN- (Le Chatelierrsquos principle) KOH was added to the lower concentration samples producing solutions with pH 12 This effectively forces all of the cyanide in solution to be CN- or 01 and 10 mgL respectively More importantly the CN stretch is now observed in the SER spectra The band appears at 2125 cm-1 as has been previously reported for gold12 As calculated for MPA and DPA LODs can be estimated from this data For the three concentrations of cyanide on silver the LODs are 016 to 11 mgL for 100 mW of 785 nm laser excitation and a 10-min acquisition time For gold pH adjusted the LODs are10 to 70 microgL an improvement of more than 10 times silver Nevertheless either substrate is sufficient to meet the JSAWM goals of measuring 3 mgL in 10-minutes as the requirements form cyanide are much less stringent than the nerve agents
4 CONCLUSIONS Here we examined the surface-enhanced Raman spectral response for methyl phosphonic acid dipicolinic acid and cyanide as a function of pH It was determined that the most prominent peak at 760 cm-1 reaches a maximum intensity between pH 3 and 7 and corresponds to the MPA- species Neither the MPA nor MPA= species appear to generate a SER spectrum and consequently no spectra were observed below pH 2 or above pH 8 In this study we also found that higher laser powers could cause photodegradation of MPA signified by the exponential growth of a band at 1050 cm-1 which is tentatively assigned to phosphorous acid Unlike MPA DPA was observed at all pHs This is attributed to the dominant interaction of the pyridine functional group with silver Minor spectral changes were observed at acid pHs and were assigned to neutral DPA Like MPA SER spectra of cyanide were pH dependent No spectra were observed for the HCN species while CN- was best observed at pHs more basic than 8 Preliminary concentration studies for the three analytes allowed estimating limits of detection for MPA DPA and CN using 100 mW of 785 nm and a 10-min acquisition time of 210 165 and 70 microgL respectively Although the latter value suggests that cyanide can be measured at sufficiently low concentrations to meet JSAWM goal improvements by 100 to 200 times are required for MPA and DPA It is clear from this study that pH of the sample is important and must be taken into account when developing concentration calibration curves as well as fieldable analyzers Future work will include tailoring the sol-gel with specific functional groups to dictate sample pH and thereby optimize sensitivity
ACKNOWLEDGMENTS The authors are grateful for the support of the US Army (DAAD13-02-C-0015 Joint Service Agent Water Monitor program) The authors would also like to thank Dr Steve Christensen of the US Army SBCCOM for helpful discussions and Mr Chetan Shende of Real-Time Analyzers for assistance in development of the gold-doped sol-gels
REFERENCES 1 Erickson B Analytical Chemistry News amp Features June 1 397A (1998)
SPIE -2003-5269 125
2 Johnston RL Hoefler CM Fargo JC and Moberley B AT-ONSITE 5-8 (1994) 3 Black RM RJ Clarke RW Read and MT Reid J Chromatography 662 301-321 (1994) 4 USA Army Center for Health Promotion and Preventive Medicine (USACHPPM) Technical Guideline 230A 5 Jensen JL NC Wong W R Loerop SPIE 4775-03 (2001) 6 Hoenig SL Handbook of Chemical Warfare and Terrorism Greenwood Press Wesport CT (2002) 7 Munro NB SS Talmage GD Griffin LC Waters AP Watson JF King and V Hauschild Env Health
Persp 107 933-974 (1999) 8 Holstege CP Kirk M Sidell FR Crit Care Clin13 923-42 (1997) 9 Farquharson S P Maksymiuk K Ong and S Christesen SPIE 4577 166-173 (2001) 10 Lee Y and S Farquharson SPIE 4378 21-26 (2001) 11 Spencer KM J Sylvia S Clauson and J Janni SPIE 4577 158-165 (2001) 12 Tessier P S Christesen K Ong E Clemente A Lenhoff E Kaler and O Velev Applied Spectroscopy 56
1524-1530 (2002) 13 Farquharson S WW Smith S Elliott and JF Sperry SPIE 3533 207-214 (1998) 14 Farquharson S WW Smith S Elliott and JF Sperry SPIE 3855110-116 (1999) 15 Farquharson S WW Smith YH Lee S Elliott and JF Sperry SPIE 4575 62-72 (2002) 16 Guzelian AA J Sylvia J Janni S Clauson and KM Spencer SPIE 4577 182-192 (2001) 17 Shende C F Inscore A Gift P Maksymiuk and S Farquharson in press 18 Weaver MJ S Farquharson and MA Tadayyoni J Chem Phys 82 4867-4874 (1985) 19 Alak AM and T Vo-Dinh Analytical Chemistry 59 2149-2153 (1987) 20 Lee Y and S Farquharson SPIE 4206 140-146 (2000) 21 Farquharson S and Y Lee SPIE 4200-16 (2000) 22 Lee Y S Farquharson and P M Rainey SPIE 3857 76-84 (1999) 23 Lee Y S Farquharson H Kwong and M Shahriari SPIE 3537 252-260 (1998) 24 Wang J M Pumera G Collins and A Mulchandani Analytical Chemistry 74 6121-6125 (2002) 25 Dou X YM Jung Z-Q Cao and Y Ozaki Applied Spectroscopy 53 1440-1447 (1999) 26 Data supplied by S Christesen and K Ewing 27 Kerker M and B Thompson Eds SPIE MS 10 (1990) 28 Inglesby TV DA Henderson JG Bartlett JAMA 287 2236 (2002) 29 FW Janssen AJ Lund and LE Anderson Science 127 26 (1958) 30 Lide DR Ed Handbook of Chemistry and Physics CRC Press 77th Ed (1996-7) 31 Billmann J G Kovacs and A Otto Surf Sci 92 153 (1980) 32 Murray CA and S Bodoff Phys Rev B 32 671 (1985) 33 Kellogg D and J Pemberton J Phys Chem 91 1120 (1987)
SPIE-2003-5269 16
Chemical agent detection by surface-enhanced Raman spectroscopy
Stuart Farquharson Alan Gift Paul Maksymiuk Frank Inscore and Wayne Smith
Real-Time Analyzers 87 Church Street East Hartford CT 06108
Kevin Morrisey and Steven D Christesen US Army SBCCOM Aberdeen Proving Ground MD 21010
ABSTRACT
In the past decade the Unites States and its allies have been challenged by a different kind of warfare exemplified by the terrorist attacks of September 11 2001 Although suicide bombings are the most often used form of terror military personnel must consider a wide range of attack scenarios Among these is the intentional poisoning of water supplies to obstruct military operations in Afghanistan and Iraq To counter such attacks the military is developing portable analyzers that can identify and quantify potential chemical agents in water supplies at microgram per liter concentrations within 10 minutes To aid this effort we have been investigating the value of a surface-enhanced Raman spectroscopy based portable analyzer In particular we have been developing silver-doped sol-gels to generate SER spectra of chemical agents and their hydrolysis products Here we present SER spectra of several chemical agents measured in a generic tap water Repeat measurements were performed to establish statistical error associated with SERS obtained using the sol-gel coated vials Keywords Chemical agents chemical agent detection SERS Raman spectroscopy
1 INTRODUCTION
In the autumn of 2001 terrorism within US borders became a sobering reality While extensive efforts are being implemented to secure the homeland US and Coalition forces in Afghanistan and Iraq are constantly faced with terrorist attacks In addition to car-bombings the military has identified several non-traditional attack scenarios including poisoning of water supplies by chemical warfare agents (CWAs) To counter this threat the Department of Defense has been investigating numerous approaches to detect such attacks under the auspices of the Joint Service Agent Water Monitor (JSAWM) program1 The JSAWM must be capable of identifying the chemical agents with no false-positives and quantifying the agents at microgram per liter concentrations within 10 minutes2 This includes the analysis of drinking water supplies distribution and storage systems Currently colorimetric paper is used to detect agents on-site while gas chromatography combined with mass spectrometry (GCMS) is used in mobile support laboratories However both methods have severe drawbacks The paper changes color in response to contact with many chemicals besides CWAs causing a high incidence of false positives3 while GCMS although very chemically specific requires up to an hour to perform and regular re-calibration456 Vibrational spectroscopy has also been investigated as a potential method of rapidly detecting CWAs7-
11 as early as 1985 in the case of infrared spectroscopy8 and 1988 in the case of Raman spectroscopy11 Again however these techniques also have limitations when it comes to measuring trace poisons in water Infrared spectra would be dominated by the very strong absorption of water which would obscure absorptions by most other chemicals present Whereas Raman spectroscopy is simply not a very sensitive technique and detection limits are typically grams per liter Surface-enhanced Raman spectroscopy (SERS) offers several advantages over conventional vibrational methods and that may provide the necessary sensitivity required for detecting trace quantities of chemical agents in water SERS employs the interaction of surface plasmon modes of metal particles with target analytes to increase scattering efficiency by as much as 1 million times12 In 1987 the potential of this approach to measure CWAs was demonstrated by measuring a series of pesticides13 Several of these organophosphonates have chemical structures similar to CWAs in particular P=O functional groups In the past few years we and others have further explored the ability of SERS to detect CWAs14-17 and even bioagents 18-21 We have been employing silver-doped sol-gels to promote the SER effect
SPIE-2003-5269 17
in these studies The porous silica network of the sol-gel matrix offers a unique environment for immobilizing and stabilizing SER-active metal particles22-25 The silver-doped sol-gels have been coated on the internal walls of standard glass vials to allow rapid SER analysis of small sample quantities without preparation We have measured over 100 chemicals with enhancements of 104 to 106 demonstrated reversible measurements in a flowing system reproducible measurements from vial-to-vial and measurements in multiple solvents including water21-25 Previously we used these vials to perform preliminary measurements of cyanide (CN) and methylphosphonic acid (MPA) Most of the nerve agents form MPA during hydrolysis while Tabun forms CN a chemical agent in its own right In another paper including in these proceedings we examined the limits of detection (LOD) for MPA by measuring a series of concentrations down to 50 mgL and estimated a limit of detection of 100 microgL26 These measurements provide encouragement in that SERS may satisfy the needs of the JSAWM To further establish the viability of SERS in particular silver-doped sol-gels here we present analysis of cyanide mustard and VX in tap water The measurements performed at the US Armyrsquos Edgewood Chemical Biological Center Aberdeen MD also included numerous repeat measurements to establish reproducibility
2 EXPERIMENTAL 2a General All chemicals including potassium cyanide 2-chloroethylethyl sulfide and those used to prepare the silver-doped and gold-doped sol-gel coated vials were obtained and used as received from Sigma-Aldrich All samples were prepared in a chemical hood using HPLC grade water unless otherwise noted (Fischer Scientific Fair Lawn NJ) for SERS measurements Once prepared the samples were transferred into the silver-doped sol-gel vials (Real-Time Analyzers Inc East Hartford CT) The vials were coated in a manner similar to that previously reported by adding ammonium hydroxide to a solution of silver nitrate tetramethyl orthosilicate (TMOS) and methanol22 The two precursor solutions were prepared mixed and transferred to 2-mL glass vials dried and heated After sol-gel formation the incorporated metal ions were reduced with dilute sodium borohydride (1mgmL) which is followed by a water wash to remove residual reducing agent After the resultant analyte solution was introduced the SER-active vial was fixed horizontally to an XY positioning stage (Conix Research Springfield OR) just inside the focal point of an f07 aspheric lens The lens focused the beam into the sample and collected the scattered radiation back along the same axis A dichroic filter (Omega Optical Brattleborough VT) was used to reflect the excitation laser to the lens and pass the Raman scattered radiation collected by the lens An f2 achromat was used to collimate the laser beam exiting a 200 microm core diameter source fiber optic while a second f2 achromat was used to focus the scattered radiation into a 365 microm fiber optic (Spectran Avon CT) A short pass filter was placed in the excitation beam path to block the silicon Raman scattering generated in the source fiber from reflecting off sampling optics and reaching the detector A long pass filter was placed in the collection beam path to block the sample Rayleigh scattering from reaching the detector A 785 nm diode laser (Process Instruments Inc model 785-600 Salt Lake City UT) was used to deliver 100 to 150 mW of power to the sample A Fourier transform Raman spectrometer (Real-Time Analyzers model IRA-785 East Hartford CT) and a silicon photo-avalanche detector (Perkin Elmer model C30902S Stamford CT) were used to acquire the SER spectra
2b Edgewood Chemical Biological Center The surface-enhanced Raman spectral measurements at the US Armyrsquos Chemical Biological center presented here were all performed on September 12 2003 To expedite measurements a plate was machined to hold up to 12 SER-active sample vials (Figure 1) The plate fit a standard XY plate reader that could be programmed Pure KCN bis-(2-chloroethyl)sulfide (distilled mustard gas HD) and ethyl S-2-diisopropylamino ethyl methylphosphonothioate (VX) were obtained on-site and used to prepare 1 gL tap water solutions in a chemical hood with appropriate safety equipment Simulated tap water was prepared by adding 10 chemicals most often found in tap water at appropriate concentrations to distilled water (Table 1) SER measurements were also performed in a chemical hood For added safety the FT-Raman instrument was placed outside the laboratory and 30 foot fiber optic and electrical cables were used to allow remote SERS measurements and plate manipulation For each experiment 1gL samples were prepared and added to 9 individual vials which were then loaded on the plate In some cases a tenth vial was included as a blank
Table 1 Generic Tap Water Composition
Compound mgL NaHCO3 100 CaSO4 27 MgSO4bull7H2O 67 NaNO3 1 Fulvic Acid 1 K2HPO4 07 KH2PO4 03 (NH4)2HSO4 001 NaCl 001 FeSO4 0001 pH 76-78
SPIE-2003-5269 18
A software program was written that allowed selecting the sequence that the vials were measured the number of positions along the length of the vials to measure (1 to 5) and the number of scans to co-add During sample analysis the program displayed the vial being analyzed the point being analyzed and the spectrum as it was being acquired Once all the data was collected a second software program was written to rapidly analyze the data The spectra collected for all the vials on a plate could be loaded at one time and then the spectra for each point could be displayed simultaneously or separately The user could then select the Raman peak to analyze in terms of peak height or area This was accomplished by selecting points on either side of the peak to define a baseline of zero The peak height or area could then be computed for all of the spectra loaded and then exported to a spreadsheet for statistical analysis
Figure 1 A) Vial Holder 6 slots to hold 2 vials each end-to-end B) Measurement Configuration Program user selects vials to measure sequence number of points per vial (1 to 5) and number of scans per point C) Spectral Acquisition Program shows spectrum being collected which vial and position D) Spectra Analysis Program user selects spectra to analyze by plate vial and point (s) as well as two wavenumbers defining the peak and the baseline to subtract The image is of 5 repeat measurements of 10 mgL KCN in generic tap water 16 sec each 100 mW of 785 nm
3 RESULTS AND DISCUSSION Raman and surface-enhanced Raman spectra were obtained for potassium cyanide bis-(2-chloroethyl)sulfide and ethyl S-2-diisopropylamino ethyl methylphosphonothioate representing three classes of chemical agents cyanides mustards and nerve agents respectively Spectra were also obtained for 2-chloroethyl ethyl sulfide (CEES) a structural analogue to HD which was included in the study to aid in assigning spectral bands KCN salt was used for cyanide experiments to avoid the increased hazards of handling HCN gas KCN completely dissolves in water forming its conjugate acid HCN according to its Ka of 615x10-1027 and at a concentration of 1 mgmL results in a pH 107 solution This is significant in that only CN- appears to interact sufficiently with silver to produce a SER spectrum and no spectral signal is observed below pH 726 Figure 2 shows the SER and normal Raman spectra for KCN The SERS spectrum of 1mgml KCN in water shows a single intense somewhat broad feature at
A B
C D
SPIE-2003-5269 19
2100 cm-1 assigned to the single CequivN stretch The band is much sharper in the normal Raman spectra of the solid KCN salt at 2074 cm-1 This band does broaden and shift to 2080 cm-1 in solution (not shown) However the observed SERS frequency is attributed to interaction with silver and low frequency mode at 135 cm-1 attributed to a Ag-CN stretch (not shown) supports this conclusion
cm-1 band to a C-S stretch but the authors concede that it is in fact more likely a C-Cl stretch28 It appears that the most intense Raman bands at 648 692 and 747 cm-1 shift to 620 660 and 720 cm-1 in the SER spectra and are tentatively assigned as above The width of these bands suggests that they overlap underlying spectral features Additional bands in the Raman spectra occur at 972 1034 1049 1263 1286 1423 1442 2865 2935 and 2960 cm-1 Corresponding bands occur in the SER spectra at 964 1015 1054 1286 1410 1447 2865 and 2935 cm-1 Most of these bands are associated with alkane modes specifically the bands at approximately 1040 cm-1 to a C-C stretch 1290 cm-1 to a CH2 in-phase twist 1440 cm-1 to a CH2 wag 2865 cm-1 to a symmetric CH2 stretch and 2965 cm-1 to an asymmetric CH2 stretch The Raman and SER spectra of sulfur mustard were measured at the Edgewood center (Figure 4) Both spectra are largely similar to CEES The C-Cl and C-S bands in the Raman spectrum of HD now occur at 640 655 700 739 and 760 cm-1 and are more resolved possibly due to the increased molecular symmetry Theoretical calculations indicate that the first three bands are due to C-Cl stretching modes and the latter two to C-S stretching modes28 Only the C-Cl bands maintain significant intensity in the SER spectra occurring at 624 and 643 cm-1 which is attributed to the
Prior to measurements of HD CEES was examined by Raman and SER spectroscopy (Figure 3) CEES also known as half-mustard is essentially identical to HD except one of the chlorine end atoms is replaced by a hydrogen atom Again although not as toxic as HD CEES is a blister agent and dilute aqueous samples were prepared using appropriate safety equipment Both the Raman and SER spectra of CEES are similar and dominated by bands between 600 and 800 cm-1 These are associated with C-Cl and C-S stretching modes which are tentatively assigned to 648 and 747 cm-1 in the Raman spectra respectively The shoulder at 630 cm-1 the overlapped band at 660 cm-1 and the strong band at 692 cm-1 could also be due to these modes or their asymmetric counterparts It is worth noting that theoretical calculations assign the 692
Figure 2 A) SER and B) NR spectra of KCN Conditions A) 1 mgml in tap water 100 mW of 785 nm at sample 1-min acquisition time B) solid 300 mW of 785 nm 5-min All spectra are 8 cm-1 resolution
A
B
Figure 3 A) SER and B) NR spectra of CEES Conditions A) 1 vv (10 mgml) in MeOH 100 mW of 785 nm 1-min acquisition time B) neat 300 mW of 785 nm 5-min
A
B
Cl-CH2-CH2-S-CH2-CH3
A
B
Figure 4 A) SER and B) NR of HD Conditions A) 1mgml in tap water B) pure both 100 mW of 785 nm 1-min
Cl-CH2-CH2-S-CH2-CH2-Cl
SPIE-2003-5269 20
expected strong interaction between chlorine and silver and adds support to the assignment of this band to a C-Cl stretch Weaker overlapping bands occur at 670 692 and 724 cm-1 the latter possibly due to C-S stretching modes Again the alkane modes are apparent in the normal Raman spectra of HD but only a broad feature at 1300 to 1450 cm-1 suggests CH2 contributions in the SER spectrum Although the observed bands in the VX spectrum have not been assigned (Figure 4) a computer generated Raman spectrum29 predicts many of the same features with surprising accuracy and are used here Two intense bands at 460 and 530 cm-1 closely match predicted bands at 463 and 546 cm-1 assigned to a CH3-P=O bend and a PO2CS wag Three highly overlapped bands occur at 694 745 and 771 cm-1 matching predicted bands at 713 730 and 760 cm-1 The first
Table 2 Measured SER peak heights for the CN stretch at 2100 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
Figure 5 A) SERS and B) NR spectra of VX Conditions A) 1 vv (10 mgml) in MeOH B) pure sample both 100 mW of 785 nm 1-min acquisition time
A
B
two have been assigned to a C-S stretch and CH2bend respectively while the latter has been attributed to either a P-C stretch or an O-C-C stretch Although the 745 cm-1 band may alternatively be assigned to a C-S stretch based on the previous measurements of CEES and HD The relatively intense bands at 890 1106 1218 1445 and 1465 cm-1 also match predicted bands at 880 1108 1216 1440 and 1464 cm-1 that are assigned to a C-C stretch CH3 rock N-C3 stretch various C-H3 bends and C-H bends respectively Both the computer generated and the measured spectra contain numerous other less intense bands One is worth mentioning A unique band appears at 370 cm-1 that is predicted at 368 cm-1 and corresponds to an O-P=O bend The surface-enhanced Raman spectrum of VX is also rich with spectral features It has the unique low frequency band at 370 cm-1 as well as a second band at 380 cm-1 that is assigned tothe S-P-O bend predicted in the normal Raman spectrum at 388 cm-1 Based on the measured and predicted normal Raman spectra the following SERS assignments are given 460 cm-1 to the CH3-P=O bend 544 cm-1 to the PO2CS wag 738 cm-1 to a C-S stretch (based on arguments above) 890 cm-1
to a C-C stretch 1101 cm-1 to a CH3 rock and 1456
cm-1 to a C-H bend The ability of SERS to measure chemical agents in water containing real-world chemical interferents was tested by using the generic tap water described in Table 1 The ability to reproduce measurements was accomplished by preparing three separate water stock solutions which were used to prepare three sample solutions each of 1 mgmL KCN in the generic tap water The samples defined as vials 1-9 were then measured at 4 points per vial in a semi-random fashion such that errors associated with stock solution preparation and errors associated with instrument drift could be identified No trends were apparent that signified such systematic errors Each spectrum collected consisted of 20 averaged scans taking 16 seconds at 8 cm-1 resolution The laser power at the sample was measured periodically during the day and it ranged from 102 to 105 mW spacccc
spot Vial 1 Vial 2 Vial 41 13754 13004 128192 13519 12692 12909
1 3 13541 12721 126394 13462 12648 12651
ave 13569 12766 12755Vial 6 Vial 3 Vial 5
1 11586 14068 10757 2 11236 14402 115122 3 11376 14546 11559
4 10894 11763 11214ave 11273 13695 11261
Vial 7 Vial8 Vial 91 11198 15804 11023
2 11243 15716 88753 3 11562 15216 11218
4 11655 15084 11077ave 11415 15455 10548
CN AVG STDEV ERR1 mgmL pk ht 12526 1560 1245
stock solution
SPIE-2003-5269 21
The CN measurements consisting of 36 data points produced an average height of 125 for the 2100 cm-1 peak with a standard deviation of 156 or 125 (Table 2) The HD and VX measurements were performed precisely the same way (Tables 3 and 4) For HD the 624 cm-1 peak was used for analysis and it had an average height of 53 with a standard deviation of 068 or 129 while for VX the 544 cm-1 peak was used for analysis and it had an average height of 1051 with a standard deviation of 308 or 293 The greater error in the VX measurements can be somewhat attributed to Vial 4 which produced lower SER signal intensities But removing this vial from the data set changes the standard deviation to 233 only a modest improvement
spot Vial 1 Vial 2 Vial 41 525 609 4682 45 675 484
1 3 527 644 5094 559 807 669
ave 51525 68375 5325Vial 6 Vial 3 Vial 5
1 521 574 575 2 536 53 4492 3 509 508 451
4 565 594 379ave 53275 5515 4635
Vial 7 Vial 8 Vial 91 457 589 413
2 583 497 4273 3 544 505 4
4 528 5 509ave 528 52275 43725
HD AVG STDEV ERR1 mgmL pk ht 530 068 1291
stock solution
spot Vial 1 Vial 2 Vial 41 1464 1034 4672 1485 99 61
1 3 1491 989 5684 1041 777 553
ave 137025 9475 5495Vial 6 Vial 3 Vial 5
1 1058 942 1293 2 697 121 965
2 3 727 1206 11124 689 1254 746
ave 79275 1153 1029Vial 7 Vial 8 Vial 9
1 1127 783 163 2 1358 812 1614
3 3 1371 908 15124 1097 875 1397
ave 123825 8445 153825
VX AVG STDEV ERR1 mgmL pk ht 1051 308 2925
stock solution
4 CONCLUSIONS In this paper we examined the ability of surface-enhanced Raman spectroscopy to reproducibly measure CN HD and VX in tap water without chemical interference Both normal and surface-enhanced Raman spectra were examined to select unique bands suitable to identify and quantify these chemical agents For SER measurements the 2100 cm-1 C-N stretch was used for CN the 624 cm-1 C-Cl stretch was used for HD and the 544 cm-1 PO2CS wag was used for VX It was determined that 1 mgmL samples of each of these chemicals measured 36 times in glass vials coated with a silver-doped sol-gel reproduced measurements with standard deviations of 125 129 and 293 It was further found that the 10 chemicals added to simulate generic tap water did not interfere with or alter the SER spectra It should be noted that the concentrations used in this study were considerably greater than those required by the JSAWM program Current work involves improving SER sensitivity and designing sampling systems with better reproducibility This includes the development of fractal silver and gold structures within the sol-gel matrix and the development of chemically selective sol-gels
ACKNOWLEDGMENTS The authors are grateful for the support of the US Army (DAAD13-02-C-0015 Joint Service Agent Water Monitor program) The authors would also like to thank Janet Jensen Ronald Crosier and Kristina Gonser for helpful discussions
Table 4 Measured SER peak heights for the VX band at 544 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
Table 3 Measured SER peak heights for the HD band at 624 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
SPIE-2003-5269 22
REFERENCES 1 Jensen JL NC Wong W R Loerop SPIE 4775-03 (2001) 2 JSAWM Requirements at wwwsbccomapgeaarmymilRDAecbcrtPRODSERJSAWMjsawmhtml 3 Erickson B ldquoThe Chemical Weapons Convention Redefines Analytical Challengerdquo Anal Chem 70 397A (1998) 4 Johnston RL Hoefler CM Fargo JC and Moberley B ldquoThe Defense Nuclear Agencyrsquos
ChemicalBiochemical Weapons Agreements Verification Technology Research Development Test and Evaluation Program and its Requirements for On-Site Analysisrdquo AT-ONSITE 5-8 (1994)
5 Black RM Clarke RJ Read RW and Reid MT ldquoApplication of GC-MS and GC-tandem MS to the analysis of chemical warfare samples found to contain residues of the nerve agent sarin sulphur mustard and their degradation productsrdquo J Chrom 662 301-321 (1994)
6 USA Army Center for Health Promotion and Preventive Medicine (USACHPPM) Technical Guideline 230A 7 Hoffland LD Piffath RJ Bouck JB rdquoSpectral signatures of chemical agents and simulantsrdquo Optical
Engineering 24 982-984 (1985) 8 Braue EHJ Pannella MGrdquoCIRCLE CELL FT-IR Analysis of Chemical Warfare Agents in Aqueous Solutionsrdquo
App Spectrosc 44 1513-1520 (1990) 9 Tseng C-H Mann CK and Vickers TJ ldquoDetermination of Organics on Metal Surfaces by Raman
Spectroscopyrdquo App Spectrosc 47 1767-1771 (1993) 10 Staff Reporter ldquoUS Army Tests Lidar to Detect Biological Toxinsrdquo Photonic Spectra pg 50 December 1998 11 Christesen SD Raman cross sections of chemical agents and simulants App Spectrosc 42 318-321 (1988) 12 Weaver MJ Farquharson S Tadayyoni MA ldquoSurface-enhancement factors for Raman scattering at silver electrodesrdquo J Chem Phys 82 4867-4874 (1985) 13 Alak AM and Vo-Dinh T ldquoSERS of Organophosphorous Chemical Agentsrdquo Anal Chem 59 2149-2153 (1987) 14 Lee Y Farquharson S ldquoRapid chemical agent identification by SERSrdquo SPIE 4378 21-26 (2001) 15 Farquharson S Maksymiuk P Ong K Christesen S ldquoChemical agent identification by surface-enhanced Raman spectroscopyrdquo SPIE 4577 166-173 (2001) 16 Spencer KM Sylvia J Clauson S and Janni J ldquoSurface Enhanced Raman as a Water Monitor for Warfare
Agents in Waterrdquo SPIE 4577 158-165 (2001) 17 Tessier P Christesen S Ong K Clemente E Lenhoff A Kaler E Velev O ldquoOn-line spectroscopic characterization of sodium cyanide with nanostructured gold surface-enhanced Raman spectroscopy substratesrdquo
App Spectrosc 56 1524-1530 (2002) 18 Farquharson S W W Smith S Elliott and J F Sperry Rapid biological agent identification by surface- enhanced Raman spectroscopy SPIE 3855110-116 (1999) 19 Farquharson S WW Smith YH Lee S Elliott and J F Sperry Detection of biological signatures A
comparison of electrolytic and metal-doped sol-gel surface-enhanced Raman media SPIE 4575 62-72 (2002) 20 Guzelian AA Sylvia J Janni J Clauson S and Spencer KM ldquoSERS of whole cell bacteria and trace levels of biological moleculesrdquo SPIE 4577 182-192 (2001) 21 Shende C Inscore F Gift A Maksymiuk P Farquharson S ldquoRapid dipicolinic acid extraction from Bacillus spores detected by surface-enhanced Raman spectroscopyrdquo App Spectrosc 58 accepted 22 Lee Y and Farquharson S ldquoSERS Sample Vials Based on Sol-Gel Process for Trace Pesticide Analysisrdquo SPIE
4206 140-146 (2000) 23 Farquharson S and Lee Y ldquoTrace Drug Analysis by SERSrdquo SPIE 4200-16 (2000) 24 Lee Y Farquharson S and Rainey P M Surface-Enhanced Raman Sensor for Trace Chemical Detection in
Water SPIE 3857 76-84 (1999) 25 Lee Y Farquharson S Kwong H and Shahriari M ldquoSol-Gel Chemical Sensor for Surface-Enhanced Raman
Spectroscopyrdquo SPIE 3537 252-260 (1998) 26 Farquharson S A Gift P Maksymiuk F Inscore and W Smith ldquopH dependence of methyl phosphonic acid
dipicolinic acid and cyanide by surface-enhanced Raman spectroscopyrdquo SPIE 5269-19 (2003) 27 Lide DR Ed Handbook of Chemistry and Physics CRC Press 77th Ed (1996-7) 28 Sosa C RJ Bartlett K KuBulat and WB Person ldquoA theoretical study of harmonic vibrational frequencies and
infrared intensities of XCH2CH2SCH2CH2X and XCH2CH2SH (= H Cl)rdquo J Phys Chem 93 577-588 (1993) 29 Hameka HF and JO Jenson ldquoComputer-generated predictions of the structure and the IR and Raman spectra of
VXrdquo ERDEC-TR-065 May 1993
- Chemagents Appendicespdf
-
- SPIE2001-4575-Bioagent-SERSapdf
-
- Detection of bioagent signatures A comparison of electrolytic and metal-doped sol-gel surface-enhanced Raman media
- Stuart Farquharson Wayne Smith and Yuan Lee
-
- Real-Time Analyzers 87 Church Street East Hartford CT 06108
-
- ABSTRACT
- 4 CONCLUSIONS
- 5 ACKNOWLEDGEMENTS
- 6 REFERENCES
-
APPLIED SPECTROSCOPY 353
FIG 1 SERS of DPA in water using silver-doped sol-gel-coated glasscapillary for (A) 1 gL and (B) 1 mgL (C ) NR of saturated DPA in 1N KOH in a glass capillary Spectral conditions (A) and (B) 150 mWof 785 nm 1-min acquisition time (C) 450 mW of 785 nm 5-minacquisition time both 8 cm 21 resolution
FIG 2 SERS of DPA extracted from 10 mg B cereus particle using100 mL of 50 mM hot DDA acquired in (A) 1 minute and (B) 2 seconds(C ) Attempted SERS of 50 mM hot DDA in ethanol using silver-dopedsol-gel-coated glass capillary acquired in 1 min Spectral conditions150 mW of 785 nm 8 cm21 resolution
the symmetric ring stretch 1384 cm21 to the OndashCndashOsymmetric stretch 1428 cm21 to the symmetric ring CndashH bend and 1569 cm21 to the asymmetric OndashCndashOstretch
The rst B cereus samples consisted of 2 mg of sporesin 2 mL of 5 mM hot DDA The samples were main-tained at 78 8C for 40 min and while hot approximately10 mL was drawn into a SER-active capillary Since spec-tra of DPA were obtained for these initial samples small-er spore masses higher DDA concentrations and shorterheating periods were examined In due course it wasfound that 10 mg of spores could be digested by 100 mLof 50 mM hot DDA in one minute and detected (Fig2A) In fact the signal was suf ciently intense that it canbe observed in as little as two seconds (Fig 2B) Theamount of DPA that was extracted was estimated to bebetween 5 and 10 mgL by comparing the signal intensityof the 1008 cm21 band to that measured for DPA in waterThis is consistent with previous research that found thatthe majority of the DPA is extracted from spores usingDDA11 and that B cereus spores contain approximately10 DPA by weight18 The SN of 127 for the 1008 cm21
band in the 1-minute SER spectrum suggests a limit ofdetection of approximately 250 ng of B cereus sporesbased on a SN of 3 Finally it should be noted that DDAdid not produce a detectable SER spectrum as shown inFig 2C
CONCLUSION
We have demonstrated that by combining rapid extrac-tion of dipicolinic acid from Bacillus cereus spores withchemical identi cation by surface-enhanced Raman spec-troscopy as little as 10 mg of spores can be detected Infact the entire measurement from the time of adding hotdodecylamine to the spores to the time when the dipi-colinic acid SER spectrum is acquired and analyzedcould be performed in less than two minutes The abilityof this method to distinguish between spore-forming bac-teria such as Bacillus anthracis and non-DPA containingpowders could help prevent costly shutdowns associated
with the appearance of suspicious material or intentionalmailing of common substances as an anthrax hoax Thismethod could also prove useful in detecting the locationof anthrax endospores in mail distribution facilities if an-other veri ed attack should occur
Research continues to fully characterize the surface-enhanced Raman spectroscopy signal intensities as afunction of sample concentration and to explore otherextractants that do not require the use of elevated tem-perature
ACKNOWLEDGMENTS
The authors are grateful for the support of the National Science Foun-dation (DMI-0296116 and DMI-0215819) and the US Army(DAAD13-02-C-0015 Joint Service Agent Water Monitor program)The authors are indebted to Chetan Shende for preparing the sol-gelcapillaries The authors also thank James Gillespie Nicholas Fell andAugustus Fountain for providing important background informationMark Farquharson for laboratory support and Professor Jay Sperry ofthe University of Rhode Island for supplying B cereus spores
1 V A Pasechnik C C Shone and P Hambleton Bioseparations 3267 (1993)
2 P J Jackson M E Hugh-Jones D M Adair G Green K K HillC R Kuske L M Grinberg F A Abramova and P Keim ProcNatl Acad Sci USA 95 1224 (1998)
3 B R Glick and J J Pasternak Molecular Biology Principles andApplications of Recombinant DNA (ASM Press Washington DC1994)
4 C A Bell J R Uhl T L Had eld J C David R F Meyer TF Smith and F R Cockerill III J Clin Microbiol 40 2897(2002)
5 D L Gatto-Menking H Yu J G Bruno M T Goode M Millerand A W Zulich Biosens Bioelectron 10 501 (1995)
6 J J Quinlan and P M Foegeding J Rapid Methods AutomationMicrobiol 6 1 (1998)
7 A A Hindle and E A H Hall Analyst (Cambridge UK) 1241599 (1999)
8 M S Ascher US Department of Health amp Human Services(httpwwwhhsgovophppresentationsAscherdoc)
9 R Nudelman B V Bronk and S Efrima Appl Spectrosc 54 445(2000)
10 D L Rosen C Sharpless and L B McBrown Anal Chem 691082 (1997)
354 Volume 58 Number 3 2004
11 P M Pellegrino N F Fell Jr and J B Gillespie Anal ChimActa 455 167 (2002)
12 W H Woodruff T G Spiro and C Gilvarg Biochem BiophysRes Commun 58 197 (1974)
13 E Ghiamati R S Manoharan W H Nelson and J F SperryAppl Spectrosc 46 357 (1992)
14 A P Esposito C E Talley T Huser C W Hollars C M Schal-dach and S M Lane Appl Spectrosc 57 868 (2003)
15 S Farquharson L Grigely V Khitrov W W Smith J F Sperryand G Fenerty J Raman Spectrosc paper accep ted (2003)
16 S Farquharson W W Smith S Elliott and J F Sperry SPIE-IntSoc Opt Eng 3855 110 (1999)
17 S Farquharson and P Maksymiuk Appl Spectrosc 57 479(2003)
18 F W Janssen A J Lund and L E Anderson Science (Washing-ton DC) 127 26 (1958)
SPIE -2003-5269 117
pH dependence of methyl phosphonic acid dipicolinic acid and cyanide by surface-enhanced Raman spectroscopy
Stuart Farquharson Alan Gift Paul Maksymiuk Frank Inscore and Wayne Smith
Real-Time Analyzers 87 Church Street East Hartford CT 06108
ABSTRACT US and Coalition forces fighting terrorism in Afghanistan and Iraq must consider a wide range of attack scenarios in addition to car bombings Among these is the intentional poisoning of water supplies to obstruct military operations To counter such attacks the military is developing portable analyzers that can identify and quantify potential chemical agents in water supplies at microgram per liter concentrations within 10 minutes To aid this effort we have been investigating the value of a surface-enhanced Raman spectroscopy based portable analyzer In particular we have been developing silver-doped sol-gels to generate SER spectra of chemical agents and their hydrolysis products Here we present SER spectra of methyl phosphonic acid and cyanide as a function of pH an important factor affecting quantitation measurements which to our knowledge has not been examined In addition dipicolinic acid a chemical signature associated with anthrax-causing spores is also presented Keywords Chemical warfare agents agent detection agent hydrolysis SERS Raman spectroscopy homeland security
1 INTRODUCTION In the past decade the Unites States and its allies have been challenged by a different kind of warfare exemplified by the terrorist attacks of September 11 2001 Suicide bombings and the use of chemical agents are the norm and military personnel must consider a wide range of attack scenarios Among these is the intentional poisoning of water supplies to obstruct military operations in Afghanistan and Iraq Currently colorimetric paper is used to detect agents on-site while gas chromatography combined with mass spectrometry (GCMS) is used in mobile support laboratories However both methods have severe drawbacks The paper changes color in response to contact with many chemicals besides CWAs causing a high incidence of false positives1 while GCMS although very chemically specific requires hours to perform and constant re-calibration234 Military operations would be greatly aided by a portable analyzer that can identify and quantify potential chemical agents at concentrations that impact safety This includes the analysis of drinking water supplies distribution and storage systems To meet this goal the Department of Defense has been investigating numerous approaches under the auspices of the Joint Service Agent Water Monitor (JSAWM) program5 The JSAWM must be capable of identifying the chemical agents with no false-positives and quantifying the agents at microgram per liter concentrations within 10 minutes678 Recently we and others have been investigating the ability of surface-enhanced Raman spectroscopy (SERS) to measure chemical agents9-12 bioagents13-17 and their hydrolysis products in water SERS employs the interaction of surface plasmon modes of metal particles with target analytes to increase scattering efficiency by as much as 1 million times18 In 1987 the potential of this approach to measure CWAs was demonstrated by measuring a series of pesticides19 Several of these organophosphonates have chemical structures similar to CWAs in particular P=O functional groups In our studies we have been employing silver-doped sol-gels to promote the SER effect The porous silica network of the sol-gel matrix offers a unique environment for immobilizing and stabilizing SER-active metal particles20-23 The sol-gels have been coated on the internal walls of standard glass vials to allow rapid SER analysis of small sample quantities without preparation We have measured over 100 chemicals with enhancements of 104 to 106 demonstrated reversible measurements in a flowing system reproducible measurements from vial-to-vial and measurements in multiple solvents including water20-23 Previously we used these vials to perform preliminary measurements of cyanide (CN) methylphosphonic acid (MPA) and dipicolinic acid (DPA) MPA is a hydrolysis product of the nerve agents (eg sarin Reaction 1) and may be a valuable indicator of nerve agent usage particularly since the alkyl methylphosphonic acids are relatively more stable than their corresponding parent complexes24 DPA is
SPIE -2003-5269 118
a chemical signature of spore forming bacteria such as Bacillus anthracis And in light of the inability to rapidly detect the anthrax spores distributed through the US mail in October 2001 a number of methods are being developed to extract and analyze this signature Reaction 1 Stepwise hydrolysis of Sarin to form hydrofluoric acid (HF) isopropyl methylphosphonic acid (IMPA) then methyl phosphonic acid (MPA) and 2-propanol In our previous SERS investigations MPA and DPA were measured at 50 and 100 mgL respectively In both cases limits of detection (LOD) were estimated at 100 microgL providing encouragement in that SERS may satisfy the needs of the JSAWM Since it has been shown that pH can substantially influence the intensity of SER bands25 which would clearly influence quantitative analysis we undertook the present study to determine the severity of these effects for cyanide methyl phosphonic acid and dipicolinic acid Furthermore we previously observed a band at 1050 cm-1 for MPA6 possibly due to an anion formed at basic pH Here we investigate the source of this spectral anomaly
2 EXPERIMENTAL All chemicals including potassium cyanide methyl phosphonic acid dipicolinic acid and those used to prepare the silver-doped and gold-doped sol-gel coated vials were obtained and used as received from Sigma-Aldrich All samples were prepared in HPLC grade water (Fischer Scientific Fair Lawn NJ) for SERS measurements The pH of these samples was adjusted using dilute nitric acid or potassium hydroxide and verified using a pH electrode (Corning Inc Corning NY) that had been calibrated with pH 400 700 and 1000 buffered standards from Fischer Scientific Once prepared the samples were transferred into the silver-doped sol-gel vials (Real-Time Analyzers Inc East Hartford CT) The vials were coated in a manner similar to that previously reported by adding ammonium hydroxide to a solution of silver nitrate tetramethyl orthosilicate (TMOS) and methanol20 Gold-vials were coated by adding nitric acid to a solution of gold tetrachloride TMOS and methanol The two precursor solutions were prepared mixed and transferred to 2-mL glass vials dried and heated After sol-gel formation the incorporated metal ions were reduced with dilute sodium borohydride (1mgmL) which was followed by a water wash to remove residual reducing agent After the resultant analyte solution was introduced the SER-active vial was immediately fixed horizontally to an XY positioning stage (Conix Research Springfield OR) just inside the focal point of an f07 aspheric lens The lens focused the beam into the sample and collected the scattered radiation back along the same axis A dichroic filter (Omega Optical Brattleborough VT) was used to reflect the excitation laser to the lens and pass the Raman scattered radiation collected by the lens An f2 achromat was used to collimate the laser beam exiting a 200 microm core diameter source fiber optic while a second f2 achromat was used to focus the scattered radiation into a 365 microm fiber optic (Spectran Avon CT) A short pass filter was placed in the excitation beam path to block the silicon Raman scattering generated in the source fiber from reflecting off sampling optics and reaching the detector A long pass filter was placed in the collection beam path to block the sample Rayleigh scattering from reaching the detector A 785 nm diode laser (Process Instruments Inc model 785-600 Salt Lake City UT) was used to deliver 100 to 150 mW of power to the sample A Fourier transform Raman spectrometer (Real-Time Analyzers model IRA-785 East Hartford CT) and a silicon photo-avalanche detector (Perkin Elmer model C30902S Stamford CT) were used to acquire the SER spectra
3 RESULTS AND DISCUSSION In a previous study of MPA6 aimed at developing a concentration calibration curve and determining limits of detection (LOD) we observed an anomaly at 1050 cm-1 Since it was found that the band intensity changed as a function of concentration the band must be associated with a sample parameter Two possible parameters photon flux and pH are examined here The first parameter was investigated by irradiating a 1mgmL MPA sample in a SER-active vial with laser powers of 200 mW and above and monitoring spectral changes It was immediately found that the 1050 cm-1 band
2O+ H HF + +OH
OH3H C
OP
OF
CH
CH3
3
3H C
OCP
OOH
CH
CH3
3
3H C
O
CPHO
CH
CH
3
3
C
Sarin IMPA MPA 2-propanol
SPIE -2003-5269 119
appeared and grew as a function of time and that the higher the laser power the faster the growth Figure 1 shows the growth of the 1050 cm-1 band over the course of 30 minutes when using 150 mW of 785 nm excitation while Figure 2A shows that the growth can be fit with a first order exponential equation namely I1050 = 03+05e-013t Furthermore the 760 cm-1 band could be fit with a first order decay equation with an identical exponential rate constant ie I760 = 08-
08e-013t The rates represent classical first order kinetics and their correspondence allows one to conclude that MPA is being transformed one-for-one into a photo-generated product At this time the photoproduct has not been positively identified but phosphonic (phosphorous) acid and phosphonate are likely candidates since the symmetric P(OH)3 stretch occurs at ~1050cm-1 Our studies show that a reduction in laser power to 100 mW at the sample for MPA essentially eliminates this degradation process This laser power or lower was used for further measurements
Methyl phosphonic acid is a diprotic acid that stepwise dissociates into two anions MPA- and MPA= according to the following reactions26
MPA MPA- + H+ pKa1 = 212 Reaction 2
MPA- MPA= + H+ pKa2 = 729 Reaction 3 The relative concentrations of MPA MPA- and MPA= can be determined at any pH by expressing [MPA] and [MPA=] in terms of [MPA-] using Reactions 2 and 3 and summing all three to equal the total starting concentration here 2 mgmL (0021M MW = 9602) viz
[MPA] + [MPA-] + [MPA=] = 0021M Equation 1 substituting from Reactions 1 and 2
([H+][MPA-])K1a + [MPA-] + (K2a[MPA-])[H+] = 0021M Equation 2 rearranging [MPA-] = 0021M(1+[H+]K1a + K2a[H+]) Equation 3 The relative concentrations of MPA MPA- and MPA= as a function of pH are shown in Figure 3 It is worth noting that near neutral pH both MPA- and MPA= will be present To confirm that the SER signal followed this pH dependence a starting solution consisting of 20 mg of MPA in 10 mL HPLC grade water was prepared and brought to pH of 20 using dilute nitric acid From this solution 2 mL were added to a SER-active vial and the SER spectrum recorded At this pH a peak at 760 cm-1 was barely discernable The 2 mL solution was returned to the starting solution and the pH was re-measured to correct for any changes that the silver-doped sol-gel vials might cause In most cases the change was less than 02 pH units and the pH is reported as the before and after average Next the pH of the
Figure 2 A) Exponential growth of 1050 cm-1 band and B) exponential decay of 760 cm-1 band for spectral series in Figure 1
Figure 1 Growth of 1050 cm-1 band as a function of time due to exposure to 150 mW of 785 nm Spectra are 5 sec each collected every 100-sec from 0 to 30-min
B
570 770 970 1170 1370Raman Shift (cm-1)
Arbit
rary
Unit
s
MPA Photodegradation
600 800 1000 1200 1400 Wavenumber (cm-1)
A
0 10 20 30 0 10 20 30 time (min) time (min)
Ram
an In
tens
ity (r
elat
ive)
30
min
0
SPIE -2003-5269 120
starting solution was adjusted to 325 using dilute KOH Again 2 mL were added to a vial and the SER spectrum recorded At this pH a reasonably strong 760 cm-1 band was observed This process was repeated as spectra were recorded at pHs of 70 74 75 79 85 and 100 A total of 1 mL of KOH was added diluting the total concentration by 10 Next the pH of the starting solution was made acidic by adding dilute nitric acid dropwise This time spectra were recorded at pHs of 72 69 64 and 37 Figure 4 shows the SER spectra for representative pHs (spectra were left out to simplify the figure) while Figure 3 shows the 760 cm-1 peak intensities as a function of pH (The band intensities were adjusted to compensate for dilution effects caused by the addition of HNO3 and KOH then normalized to 0021 M for the most intense band observed at pH 37) It is clear from Figure 3 that the 760 cm-1 band follows the MPA- concentration as a function of pH and must be assigned to this anion No bands were observed that corresponded to MPA or MPA= The lack of an MPA SER spectrum may be due to the absence of an attraction between the neutral analyte and the electropositive silver surface The same reasoning suggests that a strong SER spectrum should be observed for MPA= but it is not and a satisfactory explanation has not been found
0000
0005
0010
0015
0020
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14pH
Con
cent
ratio
n [M
]
pK1 = 212 pK2 = 729
MPA- MPA=
MPA
Figure 4 SER spectra of 002M MPA as a function of pH Conditions 100 mW of 785 nm 36 scans (1 min) 8 cm-1 recorded 2 min after sample introduction pH 19 and 20 not apparent on this scale pH 69 and 74 near identical to 70 and 75 and not shown for clarity
00
02
04
06
08
10
12
14
16
18
0 5 10 15 20 25 30Measurement Number
Ram
an In
tens
ity (7
60 c
m-1
)
Figure 3 Concentration dependence of MPA MPA- and MPA= as a function of pH for a 002M sample Intensity of 760 cm-1
band from Figure 6 as a function of pH ( for increasing basic adjustment for increasing acidic adjustment error was measured at ~10 for pH 64)
Figure 5 SER spectra of 002M MPA at pH 64 measured around a vial at three heights (9 points per height) Conditions as in Figure 4 but 10-sec scans
Since these measurements involved the removal and replacement of the SER-active vial in the sample holder to remove and add sample variation in the intensity as a function of vial position was minimized by illuminating the exact same height along the vial wall But this does not account for variability of the SERS response of the sol-gel coating around the vial To analyze this effect a vial containing MPA at pH 64 was rotated at ~ 40o
intervals at the original height and 18rdquo above and below this value Figure 5 shows the intensity of the 760 cm-1
band for the 27 positions It was found that the average value was 137plusmn014 an RSD of 10 overall and 5 for each height An error bar is included in Figure 4 for the pH 64 measurement
SPIE -2003-5269 121
With the above analyses in mind a preliminary investigation of the SER spectral response for MPA (as MPA-) as a function of concentration was performed A single vial was used for these measurements beginning with 1 mgL followed by measurements of 10 100 and 1000 mgL In all cases the pH was ~7 and 3 positions around the vial were measured per concentration Since the 760 cm-1 band was not observed for concentrations of 1 or 10 mgL using 100 mW of 785 nm the laser power at the sample was raised to 200 mW beginning with the 10 mgL concentration Photo-degradation was largely avoided (and not observed) by exposing the sample for only 33 seconds per spectral acquisition Representative spectra for 10 100 and 1000 mgL are shown in Figure 6 while a plot of the 760 cm-1 band intensity as a function of concentration is shown in Figure 7 These values were also used to estimate limits of detection based on the signal-to-noise ratio (SN) of the 760 cm-1 band using 100 mW of 785 nm laser excitation and a 10-min acquisition time As summarized in the Figure 7 inset the lower the measured concentration the lower the predicted LOD Even if as estimated 210 microgL could be measured using the silver-doped sol-gel vials an improvement of a factor of 70 is still required to meet JSAWM goals of measuring 3 microgL in 10-minutes Similar to MPA DPA is a diprotic acid (pKa1 = 216 and pKa2 = 692) and variations in pH will effect the relative concentrations of DPA DPA- and DPA= and possibly the SER spectra and band intensities This could prove significant if an acid or base is used to denature anthrax spores with the goal of extracting and analyzing DPA The MPA pH study described above was mimicked for DPA except that the starting solution consisted of 20 mg of DPA in 20 mL HPLC grade water (60x10-3M MW = 1671) The initial solution had a pH of 245 which was made basic by dilute KOH to pHs of 355 433 487 559 1069 and 1166 SER spectra were recorded at each pH using 100 mW of 785 nm and a 44-sec acquisition time Next one drop of concentrated nitric acid was used to remake the solution acidic at a pH of 200 Again sequential pH measurements were performed at 383 510 735 and 822 The solution pH was made acidic a third time but to pH 219 171 then 135 Throughout this process no more than 20 drops of acid or base were added and therefore the concentration was diluted by no more than 10 Most of the spectral bands showed a minor decrease in intensity as a function of increasing pH values However the bands at 525 and 795 cm-1 showed the most dramatic changes which occurred at acid pH Figure 8 shows the SER spectra of DPA for the spectral region and pH range of interest The identity of the DPA species was determined by plotting the normalized peak intensities with the lowest value set to 0 and the highest to 0006 M as a function of pH and overlaying these values on a plot of the relative concentrations for DPA DPA- and DPA= as previously done for MPA (Figure 9) As can be seen the 525 cm-1 band clearly corresponds to DPA The correspondence of the 795 cm-1 band to this species is less clear as the band retains intensity until pH of 55 This can be attributed to contribution to the overlapping band at 810 cm-1 which does not change as a function of pH The fact that most bands are observed at all pHs suggest that the primary interaction with silver is through the ring nitrogen This is supported by the fact that the most intense band occurs at 1008 cm-1 attributed to a symmetric ring breathing mode and that this interaction has been characterized for pyridine in numerous papers27
0
1
2
3
4
5
6
7
0 200 400 600 800 1000 1200
MPA Concentration (mgL)
760
Ban
d In
tens
ity
Figure 6 SER spectra of MPA in water at A) 1000 B) 100 and C) 10 mgL Conditions pH of 7 silver-doped sol-gel coated vial 200 mW of 785 nm 33-sec 8 cm-1 resolution
Figure 7 Plot of SER intensity of 760 cm-1 band of MPA as a function of concentration using 200 mW of 785 nm Inset table includes average intensity LOD standard deviation and percent deviation for each concentration but for 100 mW and 10 min
conc (mgL) LOD ave stddev dev10 021 002 1021100 072 010 14301000 312 040 1280
A
B C
SPIE -2003-5269 122
The ability to detect dipicolinic acid (DPA) by SERS is demonstrated in Figure 10 Even at 1 mgL the primary bands are visible The intensity of the dominant ring breathing mode observed at 1008 cm-1 for DPA in water (pH = 25 to 55) is plotted as a function of concentration in Figure 11 These values were also used to estimate limits of detection based on the SN of the 1008 cm-1 band using 100 mW of 785 nm laser excitation and a 10-min acquisition time Again the lower the measured concentration the lower the predicted LOD (see Figure 11 inset) and detection of 160 microgL is possible Gastrointestinal anthrax requires significant more spores than inhalation anthrax28 and a limit of detection might be placed at 1 million spores in 1 liter of water or 10 microgL Since each spore contains ~10 CaDPA by weight29 a goal for DPA might be 1 microgL indicating that the present measurements must be improved by nearly two orders of magnitude Due to the increased hazards of handling HCN gas KCN salt was used for these experiments Nevertheless all sample preparations were performed in a chemical hood KCN completely dissolves in water but its conjugate acid HCN is formed and has a Ka of 615x10-1030 viz
HCN CN- + H+ pKa = 921 Reaction 4
Figure 8 SER spectra of 1 mgmL DPA as a function of pH Conditions 100 mW 785 nm 100 scans (44-sec)
Figure 9 Concentration dependence of DPA DPA- and DPA= as a function of pH for a 0006M sample Intensity of 525 () and 795 (diams) cm-1 bands from Fig 8 as a function of pH
pH
135 171 219 383
0
0001
0002
0003
0004
0005
0006
0007
0 2 4 6 8 10 12 14pH
Con
cent
ratio
n [M
]
DPADPA-DPA=795525
DPA DPA=DPA-
pK2 = 692pK1 = 216
795 525
0
05
1
15
2
25
0 200 400 600 800 1000 1200
DPA Concentration (mgL)
1008
Ban
d In
tens
ity
conc (mgL) lod-10min-100mw1 017
10 016100 103
1000 355
Figure 10 SER spectra of DPA in water at A) 1000 B) 100 C) 10 and D) 1 mgL Conditions pH of 25-55 silver-doped sol-gel coated vial 175 mW of 785 nm 1-min 8 cm-1 D) has been multiplied by x10 to make bands visible
A
B C
D
Figure 11 Plot of SER intensity of 1008 cm-1 band of DPA as a function of concentration using 175 mW of 785 nm Inset table includes LOD in mgL for each concentration but for 100 mW and 10 min
SPIE -2003-5269 123
Consequently the cyanide concentration must be determined for each initial KCN concentration Specifically the samples prepared with concentrations of 01 1 10 100 and 1000 mgL of KCN produced CN- concentrations of 63x10-3 033 69 89 and 964 mgL at pHs of 816 90 967 102 and 107 respectively The pH dependence for the HCN and CN- concentrations are shown in Figure 12 Thus as the amount of KCN added to the solution decreases so does the pH of the solution (becomes less basic) and according to Reaction 4 the relative amount of CN- to HCN also decreases For example in the preparation of a 01 mgL solution of KCN the pH is shifted from 7 for pure water to only 816 and only 63 of the starting material becomes CN- or 63x10-3 mgL In comparison for a solution of 1000 mgL the pH is shifted from 7 to 107 and 96 of the starting material becomes CN- This is significant since the cyanide ion is better able to adsorb onto the silver particles and become SERS active SER spectra of 10 100 and 1000 mgL of KCN samples are shown in Figure 13 and a corresponding plot of the 2100 cm-1 band intensity as a function of concentration is shown in Figure 14
000102030405060708091011
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14pH
Con
cent
ratio
n [m
gm
L]
pKa = 921
CN -HCN
The SER spectra of cyanide are dominated by the single CN stretch at 2100 cm-1 which occurs in normal Raman spectra of solutions at 2080 cm-1 However a low frequency mode occurs at 135 cm-1 due to a Ag-CN stretch (not shown) indicative of a strong surface interaction It is also observed that as the concentration decreases the CN stretch shifts to 2140 cm-1 This shift has been attribute to the formation of a tetrahedral Ag(CN)3
2- surface structure31 as well
Figure 14 Concentration dependence of KCN SERS measured under conditions in Fig 11 Concentrations are 1 01 and 001 mgml Intensities are measured for the CN stretch at 2100 cm-1 Inset table includes LOD in mgL for each concentration in Figs 13 and 15 but for 100 mW and 10 min
Figure 13 SER spectra of KCN in water at A) 1000 B) 100 and C) 10 mgL Conditions pHs of 107 102 and 97 silver-doped sol-gel coated vial 100 mW of 785 nm 1-min 8 cm-1 C) has been multiplied by x10 to make band visible
Figure 12 Concentration dependence of HCN and CN- as a function of pH for a 1 mgmL sample Calculated intensity of 2100 () cm-1 band for a 1 mgml sample at pHs of 816 90 967 102 and 107
A
B
C
0
20
40
60
80
100
120
140
0 200 400 600 800 1000 1200CN Concentration (mgL)
2100
Ban
d In
tens
ity
Figure 15 SER spectra of KCN in water at A) 10 B) 1 and C) 01 mgL Conditions pHs of 97 12 and 12 gold-doped sol-gel coated vial 100 mW of 785 nm 1-min 8 cm-1 C) has been multiplied by x10 to make band visible
A
B
C
conc (mgL) lod-10min-100mw Condition01 001 Au-pH 121 007 Au-pH 12
69 003 Au-pH 9769 016 Ag-pH 9789 022 Ag-pH 102946 113 Ag-pH 107
SPIE -2003-5269 124
as to CN adsorbed to two different surface sites32 Alternatively the 2140 cm-1 band could be attributed to HCN since this species dominates at lower concentrations However it is unlikely that this species would be attracted to the electropositive silver surface Further both peaks should be present at pHs between 85 and 105 but this is not observed It has also been suggested that at concentrations near and above monolayer coverage the CN- species is forced to adsorb end-on due to crowding and at lower concentrations the molecule can reorient to lie flat33 This suggests that the 2100 and 2140 cm-1 bands correspond to the end-on and flat orientations respectively As Figures 13 and 14 show the intensity of the CN stretch for the 89 mgL sample is nearly as intense as the 964 mgmL band This suggests that the Raman signal for the flat orientation is more enhanced However more extensive measurements are required to verify this point Since resent research has suggested that cyanide may be more effectively detected on gold measurements of KCN solutions were also performed using gold-doped sol-gel vials Preliminary measurements are shown in Figure 15 for samples prepared from 01 1 and 10 mgL KCN Since the pHs are 816 90 967 the resultant CN- concentrations are 63x10-3 033 and 69 mgL Initially only the highest concentration was observed and the signal intensity was significantly better than the equivalent concentration measured using silver In an effort to shift Reaction 4 to the left transforming HCN to CN- (Le Chatelierrsquos principle) KOH was added to the lower concentration samples producing solutions with pH 12 This effectively forces all of the cyanide in solution to be CN- or 01 and 10 mgL respectively More importantly the CN stretch is now observed in the SER spectra The band appears at 2125 cm-1 as has been previously reported for gold12 As calculated for MPA and DPA LODs can be estimated from this data For the three concentrations of cyanide on silver the LODs are 016 to 11 mgL for 100 mW of 785 nm laser excitation and a 10-min acquisition time For gold pH adjusted the LODs are10 to 70 microgL an improvement of more than 10 times silver Nevertheless either substrate is sufficient to meet the JSAWM goals of measuring 3 mgL in 10-minutes as the requirements form cyanide are much less stringent than the nerve agents
4 CONCLUSIONS Here we examined the surface-enhanced Raman spectral response for methyl phosphonic acid dipicolinic acid and cyanide as a function of pH It was determined that the most prominent peak at 760 cm-1 reaches a maximum intensity between pH 3 and 7 and corresponds to the MPA- species Neither the MPA nor MPA= species appear to generate a SER spectrum and consequently no spectra were observed below pH 2 or above pH 8 In this study we also found that higher laser powers could cause photodegradation of MPA signified by the exponential growth of a band at 1050 cm-1 which is tentatively assigned to phosphorous acid Unlike MPA DPA was observed at all pHs This is attributed to the dominant interaction of the pyridine functional group with silver Minor spectral changes were observed at acid pHs and were assigned to neutral DPA Like MPA SER spectra of cyanide were pH dependent No spectra were observed for the HCN species while CN- was best observed at pHs more basic than 8 Preliminary concentration studies for the three analytes allowed estimating limits of detection for MPA DPA and CN using 100 mW of 785 nm and a 10-min acquisition time of 210 165 and 70 microgL respectively Although the latter value suggests that cyanide can be measured at sufficiently low concentrations to meet JSAWM goal improvements by 100 to 200 times are required for MPA and DPA It is clear from this study that pH of the sample is important and must be taken into account when developing concentration calibration curves as well as fieldable analyzers Future work will include tailoring the sol-gel with specific functional groups to dictate sample pH and thereby optimize sensitivity
ACKNOWLEDGMENTS The authors are grateful for the support of the US Army (DAAD13-02-C-0015 Joint Service Agent Water Monitor program) The authors would also like to thank Dr Steve Christensen of the US Army SBCCOM for helpful discussions and Mr Chetan Shende of Real-Time Analyzers for assistance in development of the gold-doped sol-gels
REFERENCES 1 Erickson B Analytical Chemistry News amp Features June 1 397A (1998)
SPIE -2003-5269 125
2 Johnston RL Hoefler CM Fargo JC and Moberley B AT-ONSITE 5-8 (1994) 3 Black RM RJ Clarke RW Read and MT Reid J Chromatography 662 301-321 (1994) 4 USA Army Center for Health Promotion and Preventive Medicine (USACHPPM) Technical Guideline 230A 5 Jensen JL NC Wong W R Loerop SPIE 4775-03 (2001) 6 Hoenig SL Handbook of Chemical Warfare and Terrorism Greenwood Press Wesport CT (2002) 7 Munro NB SS Talmage GD Griffin LC Waters AP Watson JF King and V Hauschild Env Health
Persp 107 933-974 (1999) 8 Holstege CP Kirk M Sidell FR Crit Care Clin13 923-42 (1997) 9 Farquharson S P Maksymiuk K Ong and S Christesen SPIE 4577 166-173 (2001) 10 Lee Y and S Farquharson SPIE 4378 21-26 (2001) 11 Spencer KM J Sylvia S Clauson and J Janni SPIE 4577 158-165 (2001) 12 Tessier P S Christesen K Ong E Clemente A Lenhoff E Kaler and O Velev Applied Spectroscopy 56
1524-1530 (2002) 13 Farquharson S WW Smith S Elliott and JF Sperry SPIE 3533 207-214 (1998) 14 Farquharson S WW Smith S Elliott and JF Sperry SPIE 3855110-116 (1999) 15 Farquharson S WW Smith YH Lee S Elliott and JF Sperry SPIE 4575 62-72 (2002) 16 Guzelian AA J Sylvia J Janni S Clauson and KM Spencer SPIE 4577 182-192 (2001) 17 Shende C F Inscore A Gift P Maksymiuk and S Farquharson in press 18 Weaver MJ S Farquharson and MA Tadayyoni J Chem Phys 82 4867-4874 (1985) 19 Alak AM and T Vo-Dinh Analytical Chemistry 59 2149-2153 (1987) 20 Lee Y and S Farquharson SPIE 4206 140-146 (2000) 21 Farquharson S and Y Lee SPIE 4200-16 (2000) 22 Lee Y S Farquharson and P M Rainey SPIE 3857 76-84 (1999) 23 Lee Y S Farquharson H Kwong and M Shahriari SPIE 3537 252-260 (1998) 24 Wang J M Pumera G Collins and A Mulchandani Analytical Chemistry 74 6121-6125 (2002) 25 Dou X YM Jung Z-Q Cao and Y Ozaki Applied Spectroscopy 53 1440-1447 (1999) 26 Data supplied by S Christesen and K Ewing 27 Kerker M and B Thompson Eds SPIE MS 10 (1990) 28 Inglesby TV DA Henderson JG Bartlett JAMA 287 2236 (2002) 29 FW Janssen AJ Lund and LE Anderson Science 127 26 (1958) 30 Lide DR Ed Handbook of Chemistry and Physics CRC Press 77th Ed (1996-7) 31 Billmann J G Kovacs and A Otto Surf Sci 92 153 (1980) 32 Murray CA and S Bodoff Phys Rev B 32 671 (1985) 33 Kellogg D and J Pemberton J Phys Chem 91 1120 (1987)
SPIE-2003-5269 16
Chemical agent detection by surface-enhanced Raman spectroscopy
Stuart Farquharson Alan Gift Paul Maksymiuk Frank Inscore and Wayne Smith
Real-Time Analyzers 87 Church Street East Hartford CT 06108
Kevin Morrisey and Steven D Christesen US Army SBCCOM Aberdeen Proving Ground MD 21010
ABSTRACT
In the past decade the Unites States and its allies have been challenged by a different kind of warfare exemplified by the terrorist attacks of September 11 2001 Although suicide bombings are the most often used form of terror military personnel must consider a wide range of attack scenarios Among these is the intentional poisoning of water supplies to obstruct military operations in Afghanistan and Iraq To counter such attacks the military is developing portable analyzers that can identify and quantify potential chemical agents in water supplies at microgram per liter concentrations within 10 minutes To aid this effort we have been investigating the value of a surface-enhanced Raman spectroscopy based portable analyzer In particular we have been developing silver-doped sol-gels to generate SER spectra of chemical agents and their hydrolysis products Here we present SER spectra of several chemical agents measured in a generic tap water Repeat measurements were performed to establish statistical error associated with SERS obtained using the sol-gel coated vials Keywords Chemical agents chemical agent detection SERS Raman spectroscopy
1 INTRODUCTION
In the autumn of 2001 terrorism within US borders became a sobering reality While extensive efforts are being implemented to secure the homeland US and Coalition forces in Afghanistan and Iraq are constantly faced with terrorist attacks In addition to car-bombings the military has identified several non-traditional attack scenarios including poisoning of water supplies by chemical warfare agents (CWAs) To counter this threat the Department of Defense has been investigating numerous approaches to detect such attacks under the auspices of the Joint Service Agent Water Monitor (JSAWM) program1 The JSAWM must be capable of identifying the chemical agents with no false-positives and quantifying the agents at microgram per liter concentrations within 10 minutes2 This includes the analysis of drinking water supplies distribution and storage systems Currently colorimetric paper is used to detect agents on-site while gas chromatography combined with mass spectrometry (GCMS) is used in mobile support laboratories However both methods have severe drawbacks The paper changes color in response to contact with many chemicals besides CWAs causing a high incidence of false positives3 while GCMS although very chemically specific requires up to an hour to perform and regular re-calibration456 Vibrational spectroscopy has also been investigated as a potential method of rapidly detecting CWAs7-
11 as early as 1985 in the case of infrared spectroscopy8 and 1988 in the case of Raman spectroscopy11 Again however these techniques also have limitations when it comes to measuring trace poisons in water Infrared spectra would be dominated by the very strong absorption of water which would obscure absorptions by most other chemicals present Whereas Raman spectroscopy is simply not a very sensitive technique and detection limits are typically grams per liter Surface-enhanced Raman spectroscopy (SERS) offers several advantages over conventional vibrational methods and that may provide the necessary sensitivity required for detecting trace quantities of chemical agents in water SERS employs the interaction of surface plasmon modes of metal particles with target analytes to increase scattering efficiency by as much as 1 million times12 In 1987 the potential of this approach to measure CWAs was demonstrated by measuring a series of pesticides13 Several of these organophosphonates have chemical structures similar to CWAs in particular P=O functional groups In the past few years we and others have further explored the ability of SERS to detect CWAs14-17 and even bioagents 18-21 We have been employing silver-doped sol-gels to promote the SER effect
SPIE-2003-5269 17
in these studies The porous silica network of the sol-gel matrix offers a unique environment for immobilizing and stabilizing SER-active metal particles22-25 The silver-doped sol-gels have been coated on the internal walls of standard glass vials to allow rapid SER analysis of small sample quantities without preparation We have measured over 100 chemicals with enhancements of 104 to 106 demonstrated reversible measurements in a flowing system reproducible measurements from vial-to-vial and measurements in multiple solvents including water21-25 Previously we used these vials to perform preliminary measurements of cyanide (CN) and methylphosphonic acid (MPA) Most of the nerve agents form MPA during hydrolysis while Tabun forms CN a chemical agent in its own right In another paper including in these proceedings we examined the limits of detection (LOD) for MPA by measuring a series of concentrations down to 50 mgL and estimated a limit of detection of 100 microgL26 These measurements provide encouragement in that SERS may satisfy the needs of the JSAWM To further establish the viability of SERS in particular silver-doped sol-gels here we present analysis of cyanide mustard and VX in tap water The measurements performed at the US Armyrsquos Edgewood Chemical Biological Center Aberdeen MD also included numerous repeat measurements to establish reproducibility
2 EXPERIMENTAL 2a General All chemicals including potassium cyanide 2-chloroethylethyl sulfide and those used to prepare the silver-doped and gold-doped sol-gel coated vials were obtained and used as received from Sigma-Aldrich All samples were prepared in a chemical hood using HPLC grade water unless otherwise noted (Fischer Scientific Fair Lawn NJ) for SERS measurements Once prepared the samples were transferred into the silver-doped sol-gel vials (Real-Time Analyzers Inc East Hartford CT) The vials were coated in a manner similar to that previously reported by adding ammonium hydroxide to a solution of silver nitrate tetramethyl orthosilicate (TMOS) and methanol22 The two precursor solutions were prepared mixed and transferred to 2-mL glass vials dried and heated After sol-gel formation the incorporated metal ions were reduced with dilute sodium borohydride (1mgmL) which is followed by a water wash to remove residual reducing agent After the resultant analyte solution was introduced the SER-active vial was fixed horizontally to an XY positioning stage (Conix Research Springfield OR) just inside the focal point of an f07 aspheric lens The lens focused the beam into the sample and collected the scattered radiation back along the same axis A dichroic filter (Omega Optical Brattleborough VT) was used to reflect the excitation laser to the lens and pass the Raman scattered radiation collected by the lens An f2 achromat was used to collimate the laser beam exiting a 200 microm core diameter source fiber optic while a second f2 achromat was used to focus the scattered radiation into a 365 microm fiber optic (Spectran Avon CT) A short pass filter was placed in the excitation beam path to block the silicon Raman scattering generated in the source fiber from reflecting off sampling optics and reaching the detector A long pass filter was placed in the collection beam path to block the sample Rayleigh scattering from reaching the detector A 785 nm diode laser (Process Instruments Inc model 785-600 Salt Lake City UT) was used to deliver 100 to 150 mW of power to the sample A Fourier transform Raman spectrometer (Real-Time Analyzers model IRA-785 East Hartford CT) and a silicon photo-avalanche detector (Perkin Elmer model C30902S Stamford CT) were used to acquire the SER spectra
2b Edgewood Chemical Biological Center The surface-enhanced Raman spectral measurements at the US Armyrsquos Chemical Biological center presented here were all performed on September 12 2003 To expedite measurements a plate was machined to hold up to 12 SER-active sample vials (Figure 1) The plate fit a standard XY plate reader that could be programmed Pure KCN bis-(2-chloroethyl)sulfide (distilled mustard gas HD) and ethyl S-2-diisopropylamino ethyl methylphosphonothioate (VX) were obtained on-site and used to prepare 1 gL tap water solutions in a chemical hood with appropriate safety equipment Simulated tap water was prepared by adding 10 chemicals most often found in tap water at appropriate concentrations to distilled water (Table 1) SER measurements were also performed in a chemical hood For added safety the FT-Raman instrument was placed outside the laboratory and 30 foot fiber optic and electrical cables were used to allow remote SERS measurements and plate manipulation For each experiment 1gL samples were prepared and added to 9 individual vials which were then loaded on the plate In some cases a tenth vial was included as a blank
Table 1 Generic Tap Water Composition
Compound mgL NaHCO3 100 CaSO4 27 MgSO4bull7H2O 67 NaNO3 1 Fulvic Acid 1 K2HPO4 07 KH2PO4 03 (NH4)2HSO4 001 NaCl 001 FeSO4 0001 pH 76-78
SPIE-2003-5269 18
A software program was written that allowed selecting the sequence that the vials were measured the number of positions along the length of the vials to measure (1 to 5) and the number of scans to co-add During sample analysis the program displayed the vial being analyzed the point being analyzed and the spectrum as it was being acquired Once all the data was collected a second software program was written to rapidly analyze the data The spectra collected for all the vials on a plate could be loaded at one time and then the spectra for each point could be displayed simultaneously or separately The user could then select the Raman peak to analyze in terms of peak height or area This was accomplished by selecting points on either side of the peak to define a baseline of zero The peak height or area could then be computed for all of the spectra loaded and then exported to a spreadsheet for statistical analysis
Figure 1 A) Vial Holder 6 slots to hold 2 vials each end-to-end B) Measurement Configuration Program user selects vials to measure sequence number of points per vial (1 to 5) and number of scans per point C) Spectral Acquisition Program shows spectrum being collected which vial and position D) Spectra Analysis Program user selects spectra to analyze by plate vial and point (s) as well as two wavenumbers defining the peak and the baseline to subtract The image is of 5 repeat measurements of 10 mgL KCN in generic tap water 16 sec each 100 mW of 785 nm
3 RESULTS AND DISCUSSION Raman and surface-enhanced Raman spectra were obtained for potassium cyanide bis-(2-chloroethyl)sulfide and ethyl S-2-diisopropylamino ethyl methylphosphonothioate representing three classes of chemical agents cyanides mustards and nerve agents respectively Spectra were also obtained for 2-chloroethyl ethyl sulfide (CEES) a structural analogue to HD which was included in the study to aid in assigning spectral bands KCN salt was used for cyanide experiments to avoid the increased hazards of handling HCN gas KCN completely dissolves in water forming its conjugate acid HCN according to its Ka of 615x10-1027 and at a concentration of 1 mgmL results in a pH 107 solution This is significant in that only CN- appears to interact sufficiently with silver to produce a SER spectrum and no spectral signal is observed below pH 726 Figure 2 shows the SER and normal Raman spectra for KCN The SERS spectrum of 1mgml KCN in water shows a single intense somewhat broad feature at
A B
C D
SPIE-2003-5269 19
2100 cm-1 assigned to the single CequivN stretch The band is much sharper in the normal Raman spectra of the solid KCN salt at 2074 cm-1 This band does broaden and shift to 2080 cm-1 in solution (not shown) However the observed SERS frequency is attributed to interaction with silver and low frequency mode at 135 cm-1 attributed to a Ag-CN stretch (not shown) supports this conclusion
cm-1 band to a C-S stretch but the authors concede that it is in fact more likely a C-Cl stretch28 It appears that the most intense Raman bands at 648 692 and 747 cm-1 shift to 620 660 and 720 cm-1 in the SER spectra and are tentatively assigned as above The width of these bands suggests that they overlap underlying spectral features Additional bands in the Raman spectra occur at 972 1034 1049 1263 1286 1423 1442 2865 2935 and 2960 cm-1 Corresponding bands occur in the SER spectra at 964 1015 1054 1286 1410 1447 2865 and 2935 cm-1 Most of these bands are associated with alkane modes specifically the bands at approximately 1040 cm-1 to a C-C stretch 1290 cm-1 to a CH2 in-phase twist 1440 cm-1 to a CH2 wag 2865 cm-1 to a symmetric CH2 stretch and 2965 cm-1 to an asymmetric CH2 stretch The Raman and SER spectra of sulfur mustard were measured at the Edgewood center (Figure 4) Both spectra are largely similar to CEES The C-Cl and C-S bands in the Raman spectrum of HD now occur at 640 655 700 739 and 760 cm-1 and are more resolved possibly due to the increased molecular symmetry Theoretical calculations indicate that the first three bands are due to C-Cl stretching modes and the latter two to C-S stretching modes28 Only the C-Cl bands maintain significant intensity in the SER spectra occurring at 624 and 643 cm-1 which is attributed to the
Prior to measurements of HD CEES was examined by Raman and SER spectroscopy (Figure 3) CEES also known as half-mustard is essentially identical to HD except one of the chlorine end atoms is replaced by a hydrogen atom Again although not as toxic as HD CEES is a blister agent and dilute aqueous samples were prepared using appropriate safety equipment Both the Raman and SER spectra of CEES are similar and dominated by bands between 600 and 800 cm-1 These are associated with C-Cl and C-S stretching modes which are tentatively assigned to 648 and 747 cm-1 in the Raman spectra respectively The shoulder at 630 cm-1 the overlapped band at 660 cm-1 and the strong band at 692 cm-1 could also be due to these modes or their asymmetric counterparts It is worth noting that theoretical calculations assign the 692
Figure 2 A) SER and B) NR spectra of KCN Conditions A) 1 mgml in tap water 100 mW of 785 nm at sample 1-min acquisition time B) solid 300 mW of 785 nm 5-min All spectra are 8 cm-1 resolution
A
B
Figure 3 A) SER and B) NR spectra of CEES Conditions A) 1 vv (10 mgml) in MeOH 100 mW of 785 nm 1-min acquisition time B) neat 300 mW of 785 nm 5-min
A
B
Cl-CH2-CH2-S-CH2-CH3
A
B
Figure 4 A) SER and B) NR of HD Conditions A) 1mgml in tap water B) pure both 100 mW of 785 nm 1-min
Cl-CH2-CH2-S-CH2-CH2-Cl
SPIE-2003-5269 20
expected strong interaction between chlorine and silver and adds support to the assignment of this band to a C-Cl stretch Weaker overlapping bands occur at 670 692 and 724 cm-1 the latter possibly due to C-S stretching modes Again the alkane modes are apparent in the normal Raman spectra of HD but only a broad feature at 1300 to 1450 cm-1 suggests CH2 contributions in the SER spectrum Although the observed bands in the VX spectrum have not been assigned (Figure 4) a computer generated Raman spectrum29 predicts many of the same features with surprising accuracy and are used here Two intense bands at 460 and 530 cm-1 closely match predicted bands at 463 and 546 cm-1 assigned to a CH3-P=O bend and a PO2CS wag Three highly overlapped bands occur at 694 745 and 771 cm-1 matching predicted bands at 713 730 and 760 cm-1 The first
Table 2 Measured SER peak heights for the CN stretch at 2100 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
Figure 5 A) SERS and B) NR spectra of VX Conditions A) 1 vv (10 mgml) in MeOH B) pure sample both 100 mW of 785 nm 1-min acquisition time
A
B
two have been assigned to a C-S stretch and CH2bend respectively while the latter has been attributed to either a P-C stretch or an O-C-C stretch Although the 745 cm-1 band may alternatively be assigned to a C-S stretch based on the previous measurements of CEES and HD The relatively intense bands at 890 1106 1218 1445 and 1465 cm-1 also match predicted bands at 880 1108 1216 1440 and 1464 cm-1 that are assigned to a C-C stretch CH3 rock N-C3 stretch various C-H3 bends and C-H bends respectively Both the computer generated and the measured spectra contain numerous other less intense bands One is worth mentioning A unique band appears at 370 cm-1 that is predicted at 368 cm-1 and corresponds to an O-P=O bend The surface-enhanced Raman spectrum of VX is also rich with spectral features It has the unique low frequency band at 370 cm-1 as well as a second band at 380 cm-1 that is assigned tothe S-P-O bend predicted in the normal Raman spectrum at 388 cm-1 Based on the measured and predicted normal Raman spectra the following SERS assignments are given 460 cm-1 to the CH3-P=O bend 544 cm-1 to the PO2CS wag 738 cm-1 to a C-S stretch (based on arguments above) 890 cm-1
to a C-C stretch 1101 cm-1 to a CH3 rock and 1456
cm-1 to a C-H bend The ability of SERS to measure chemical agents in water containing real-world chemical interferents was tested by using the generic tap water described in Table 1 The ability to reproduce measurements was accomplished by preparing three separate water stock solutions which were used to prepare three sample solutions each of 1 mgmL KCN in the generic tap water The samples defined as vials 1-9 were then measured at 4 points per vial in a semi-random fashion such that errors associated with stock solution preparation and errors associated with instrument drift could be identified No trends were apparent that signified such systematic errors Each spectrum collected consisted of 20 averaged scans taking 16 seconds at 8 cm-1 resolution The laser power at the sample was measured periodically during the day and it ranged from 102 to 105 mW spacccc
spot Vial 1 Vial 2 Vial 41 13754 13004 128192 13519 12692 12909
1 3 13541 12721 126394 13462 12648 12651
ave 13569 12766 12755Vial 6 Vial 3 Vial 5
1 11586 14068 10757 2 11236 14402 115122 3 11376 14546 11559
4 10894 11763 11214ave 11273 13695 11261
Vial 7 Vial8 Vial 91 11198 15804 11023
2 11243 15716 88753 3 11562 15216 11218
4 11655 15084 11077ave 11415 15455 10548
CN AVG STDEV ERR1 mgmL pk ht 12526 1560 1245
stock solution
SPIE-2003-5269 21
The CN measurements consisting of 36 data points produced an average height of 125 for the 2100 cm-1 peak with a standard deviation of 156 or 125 (Table 2) The HD and VX measurements were performed precisely the same way (Tables 3 and 4) For HD the 624 cm-1 peak was used for analysis and it had an average height of 53 with a standard deviation of 068 or 129 while for VX the 544 cm-1 peak was used for analysis and it had an average height of 1051 with a standard deviation of 308 or 293 The greater error in the VX measurements can be somewhat attributed to Vial 4 which produced lower SER signal intensities But removing this vial from the data set changes the standard deviation to 233 only a modest improvement
spot Vial 1 Vial 2 Vial 41 525 609 4682 45 675 484
1 3 527 644 5094 559 807 669
ave 51525 68375 5325Vial 6 Vial 3 Vial 5
1 521 574 575 2 536 53 4492 3 509 508 451
4 565 594 379ave 53275 5515 4635
Vial 7 Vial 8 Vial 91 457 589 413
2 583 497 4273 3 544 505 4
4 528 5 509ave 528 52275 43725
HD AVG STDEV ERR1 mgmL pk ht 530 068 1291
stock solution
spot Vial 1 Vial 2 Vial 41 1464 1034 4672 1485 99 61
1 3 1491 989 5684 1041 777 553
ave 137025 9475 5495Vial 6 Vial 3 Vial 5
1 1058 942 1293 2 697 121 965
2 3 727 1206 11124 689 1254 746
ave 79275 1153 1029Vial 7 Vial 8 Vial 9
1 1127 783 163 2 1358 812 1614
3 3 1371 908 15124 1097 875 1397
ave 123825 8445 153825
VX AVG STDEV ERR1 mgmL pk ht 1051 308 2925
stock solution
4 CONCLUSIONS In this paper we examined the ability of surface-enhanced Raman spectroscopy to reproducibly measure CN HD and VX in tap water without chemical interference Both normal and surface-enhanced Raman spectra were examined to select unique bands suitable to identify and quantify these chemical agents For SER measurements the 2100 cm-1 C-N stretch was used for CN the 624 cm-1 C-Cl stretch was used for HD and the 544 cm-1 PO2CS wag was used for VX It was determined that 1 mgmL samples of each of these chemicals measured 36 times in glass vials coated with a silver-doped sol-gel reproduced measurements with standard deviations of 125 129 and 293 It was further found that the 10 chemicals added to simulate generic tap water did not interfere with or alter the SER spectra It should be noted that the concentrations used in this study were considerably greater than those required by the JSAWM program Current work involves improving SER sensitivity and designing sampling systems with better reproducibility This includes the development of fractal silver and gold structures within the sol-gel matrix and the development of chemically selective sol-gels
ACKNOWLEDGMENTS The authors are grateful for the support of the US Army (DAAD13-02-C-0015 Joint Service Agent Water Monitor program) The authors would also like to thank Janet Jensen Ronald Crosier and Kristina Gonser for helpful discussions
Table 4 Measured SER peak heights for the VX band at 544 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
Table 3 Measured SER peak heights for the HD band at 624 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
SPIE-2003-5269 22
REFERENCES 1 Jensen JL NC Wong W R Loerop SPIE 4775-03 (2001) 2 JSAWM Requirements at wwwsbccomapgeaarmymilRDAecbcrtPRODSERJSAWMjsawmhtml 3 Erickson B ldquoThe Chemical Weapons Convention Redefines Analytical Challengerdquo Anal Chem 70 397A (1998) 4 Johnston RL Hoefler CM Fargo JC and Moberley B ldquoThe Defense Nuclear Agencyrsquos
ChemicalBiochemical Weapons Agreements Verification Technology Research Development Test and Evaluation Program and its Requirements for On-Site Analysisrdquo AT-ONSITE 5-8 (1994)
5 Black RM Clarke RJ Read RW and Reid MT ldquoApplication of GC-MS and GC-tandem MS to the analysis of chemical warfare samples found to contain residues of the nerve agent sarin sulphur mustard and their degradation productsrdquo J Chrom 662 301-321 (1994)
6 USA Army Center for Health Promotion and Preventive Medicine (USACHPPM) Technical Guideline 230A 7 Hoffland LD Piffath RJ Bouck JB rdquoSpectral signatures of chemical agents and simulantsrdquo Optical
Engineering 24 982-984 (1985) 8 Braue EHJ Pannella MGrdquoCIRCLE CELL FT-IR Analysis of Chemical Warfare Agents in Aqueous Solutionsrdquo
App Spectrosc 44 1513-1520 (1990) 9 Tseng C-H Mann CK and Vickers TJ ldquoDetermination of Organics on Metal Surfaces by Raman
Spectroscopyrdquo App Spectrosc 47 1767-1771 (1993) 10 Staff Reporter ldquoUS Army Tests Lidar to Detect Biological Toxinsrdquo Photonic Spectra pg 50 December 1998 11 Christesen SD Raman cross sections of chemical agents and simulants App Spectrosc 42 318-321 (1988) 12 Weaver MJ Farquharson S Tadayyoni MA ldquoSurface-enhancement factors for Raman scattering at silver electrodesrdquo J Chem Phys 82 4867-4874 (1985) 13 Alak AM and Vo-Dinh T ldquoSERS of Organophosphorous Chemical Agentsrdquo Anal Chem 59 2149-2153 (1987) 14 Lee Y Farquharson S ldquoRapid chemical agent identification by SERSrdquo SPIE 4378 21-26 (2001) 15 Farquharson S Maksymiuk P Ong K Christesen S ldquoChemical agent identification by surface-enhanced Raman spectroscopyrdquo SPIE 4577 166-173 (2001) 16 Spencer KM Sylvia J Clauson S and Janni J ldquoSurface Enhanced Raman as a Water Monitor for Warfare
Agents in Waterrdquo SPIE 4577 158-165 (2001) 17 Tessier P Christesen S Ong K Clemente E Lenhoff A Kaler E Velev O ldquoOn-line spectroscopic characterization of sodium cyanide with nanostructured gold surface-enhanced Raman spectroscopy substratesrdquo
App Spectrosc 56 1524-1530 (2002) 18 Farquharson S W W Smith S Elliott and J F Sperry Rapid biological agent identification by surface- enhanced Raman spectroscopy SPIE 3855110-116 (1999) 19 Farquharson S WW Smith YH Lee S Elliott and J F Sperry Detection of biological signatures A
comparison of electrolytic and metal-doped sol-gel surface-enhanced Raman media SPIE 4575 62-72 (2002) 20 Guzelian AA Sylvia J Janni J Clauson S and Spencer KM ldquoSERS of whole cell bacteria and trace levels of biological moleculesrdquo SPIE 4577 182-192 (2001) 21 Shende C Inscore F Gift A Maksymiuk P Farquharson S ldquoRapid dipicolinic acid extraction from Bacillus spores detected by surface-enhanced Raman spectroscopyrdquo App Spectrosc 58 accepted 22 Lee Y and Farquharson S ldquoSERS Sample Vials Based on Sol-Gel Process for Trace Pesticide Analysisrdquo SPIE
4206 140-146 (2000) 23 Farquharson S and Lee Y ldquoTrace Drug Analysis by SERSrdquo SPIE 4200-16 (2000) 24 Lee Y Farquharson S and Rainey P M Surface-Enhanced Raman Sensor for Trace Chemical Detection in
Water SPIE 3857 76-84 (1999) 25 Lee Y Farquharson S Kwong H and Shahriari M ldquoSol-Gel Chemical Sensor for Surface-Enhanced Raman
Spectroscopyrdquo SPIE 3537 252-260 (1998) 26 Farquharson S A Gift P Maksymiuk F Inscore and W Smith ldquopH dependence of methyl phosphonic acid
dipicolinic acid and cyanide by surface-enhanced Raman spectroscopyrdquo SPIE 5269-19 (2003) 27 Lide DR Ed Handbook of Chemistry and Physics CRC Press 77th Ed (1996-7) 28 Sosa C RJ Bartlett K KuBulat and WB Person ldquoA theoretical study of harmonic vibrational frequencies and
infrared intensities of XCH2CH2SCH2CH2X and XCH2CH2SH (= H Cl)rdquo J Phys Chem 93 577-588 (1993) 29 Hameka HF and JO Jenson ldquoComputer-generated predictions of the structure and the IR and Raman spectra of
VXrdquo ERDEC-TR-065 May 1993
- Chemagents Appendicespdf
-
- SPIE2001-4575-Bioagent-SERSapdf
-
- Detection of bioagent signatures A comparison of electrolytic and metal-doped sol-gel surface-enhanced Raman media
- Stuart Farquharson Wayne Smith and Yuan Lee
-
- Real-Time Analyzers 87 Church Street East Hartford CT 06108
-
- ABSTRACT
- 4 CONCLUSIONS
- 5 ACKNOWLEDGEMENTS
- 6 REFERENCES
-
354 Volume 58 Number 3 2004
11 P M Pellegrino N F Fell Jr and J B Gillespie Anal ChimActa 455 167 (2002)
12 W H Woodruff T G Spiro and C Gilvarg Biochem BiophysRes Commun 58 197 (1974)
13 E Ghiamati R S Manoharan W H Nelson and J F SperryAppl Spectrosc 46 357 (1992)
14 A P Esposito C E Talley T Huser C W Hollars C M Schal-dach and S M Lane Appl Spectrosc 57 868 (2003)
15 S Farquharson L Grigely V Khitrov W W Smith J F Sperryand G Fenerty J Raman Spectrosc paper accep ted (2003)
16 S Farquharson W W Smith S Elliott and J F Sperry SPIE-IntSoc Opt Eng 3855 110 (1999)
17 S Farquharson and P Maksymiuk Appl Spectrosc 57 479(2003)
18 F W Janssen A J Lund and L E Anderson Science (Washing-ton DC) 127 26 (1958)
SPIE -2003-5269 117
pH dependence of methyl phosphonic acid dipicolinic acid and cyanide by surface-enhanced Raman spectroscopy
Stuart Farquharson Alan Gift Paul Maksymiuk Frank Inscore and Wayne Smith
Real-Time Analyzers 87 Church Street East Hartford CT 06108
ABSTRACT US and Coalition forces fighting terrorism in Afghanistan and Iraq must consider a wide range of attack scenarios in addition to car bombings Among these is the intentional poisoning of water supplies to obstruct military operations To counter such attacks the military is developing portable analyzers that can identify and quantify potential chemical agents in water supplies at microgram per liter concentrations within 10 minutes To aid this effort we have been investigating the value of a surface-enhanced Raman spectroscopy based portable analyzer In particular we have been developing silver-doped sol-gels to generate SER spectra of chemical agents and their hydrolysis products Here we present SER spectra of methyl phosphonic acid and cyanide as a function of pH an important factor affecting quantitation measurements which to our knowledge has not been examined In addition dipicolinic acid a chemical signature associated with anthrax-causing spores is also presented Keywords Chemical warfare agents agent detection agent hydrolysis SERS Raman spectroscopy homeland security
1 INTRODUCTION In the past decade the Unites States and its allies have been challenged by a different kind of warfare exemplified by the terrorist attacks of September 11 2001 Suicide bombings and the use of chemical agents are the norm and military personnel must consider a wide range of attack scenarios Among these is the intentional poisoning of water supplies to obstruct military operations in Afghanistan and Iraq Currently colorimetric paper is used to detect agents on-site while gas chromatography combined with mass spectrometry (GCMS) is used in mobile support laboratories However both methods have severe drawbacks The paper changes color in response to contact with many chemicals besides CWAs causing a high incidence of false positives1 while GCMS although very chemically specific requires hours to perform and constant re-calibration234 Military operations would be greatly aided by a portable analyzer that can identify and quantify potential chemical agents at concentrations that impact safety This includes the analysis of drinking water supplies distribution and storage systems To meet this goal the Department of Defense has been investigating numerous approaches under the auspices of the Joint Service Agent Water Monitor (JSAWM) program5 The JSAWM must be capable of identifying the chemical agents with no false-positives and quantifying the agents at microgram per liter concentrations within 10 minutes678 Recently we and others have been investigating the ability of surface-enhanced Raman spectroscopy (SERS) to measure chemical agents9-12 bioagents13-17 and their hydrolysis products in water SERS employs the interaction of surface plasmon modes of metal particles with target analytes to increase scattering efficiency by as much as 1 million times18 In 1987 the potential of this approach to measure CWAs was demonstrated by measuring a series of pesticides19 Several of these organophosphonates have chemical structures similar to CWAs in particular P=O functional groups In our studies we have been employing silver-doped sol-gels to promote the SER effect The porous silica network of the sol-gel matrix offers a unique environment for immobilizing and stabilizing SER-active metal particles20-23 The sol-gels have been coated on the internal walls of standard glass vials to allow rapid SER analysis of small sample quantities without preparation We have measured over 100 chemicals with enhancements of 104 to 106 demonstrated reversible measurements in a flowing system reproducible measurements from vial-to-vial and measurements in multiple solvents including water20-23 Previously we used these vials to perform preliminary measurements of cyanide (CN) methylphosphonic acid (MPA) and dipicolinic acid (DPA) MPA is a hydrolysis product of the nerve agents (eg sarin Reaction 1) and may be a valuable indicator of nerve agent usage particularly since the alkyl methylphosphonic acids are relatively more stable than their corresponding parent complexes24 DPA is
SPIE -2003-5269 118
a chemical signature of spore forming bacteria such as Bacillus anthracis And in light of the inability to rapidly detect the anthrax spores distributed through the US mail in October 2001 a number of methods are being developed to extract and analyze this signature Reaction 1 Stepwise hydrolysis of Sarin to form hydrofluoric acid (HF) isopropyl methylphosphonic acid (IMPA) then methyl phosphonic acid (MPA) and 2-propanol In our previous SERS investigations MPA and DPA were measured at 50 and 100 mgL respectively In both cases limits of detection (LOD) were estimated at 100 microgL providing encouragement in that SERS may satisfy the needs of the JSAWM Since it has been shown that pH can substantially influence the intensity of SER bands25 which would clearly influence quantitative analysis we undertook the present study to determine the severity of these effects for cyanide methyl phosphonic acid and dipicolinic acid Furthermore we previously observed a band at 1050 cm-1 for MPA6 possibly due to an anion formed at basic pH Here we investigate the source of this spectral anomaly
2 EXPERIMENTAL All chemicals including potassium cyanide methyl phosphonic acid dipicolinic acid and those used to prepare the silver-doped and gold-doped sol-gel coated vials were obtained and used as received from Sigma-Aldrich All samples were prepared in HPLC grade water (Fischer Scientific Fair Lawn NJ) for SERS measurements The pH of these samples was adjusted using dilute nitric acid or potassium hydroxide and verified using a pH electrode (Corning Inc Corning NY) that had been calibrated with pH 400 700 and 1000 buffered standards from Fischer Scientific Once prepared the samples were transferred into the silver-doped sol-gel vials (Real-Time Analyzers Inc East Hartford CT) The vials were coated in a manner similar to that previously reported by adding ammonium hydroxide to a solution of silver nitrate tetramethyl orthosilicate (TMOS) and methanol20 Gold-vials were coated by adding nitric acid to a solution of gold tetrachloride TMOS and methanol The two precursor solutions were prepared mixed and transferred to 2-mL glass vials dried and heated After sol-gel formation the incorporated metal ions were reduced with dilute sodium borohydride (1mgmL) which was followed by a water wash to remove residual reducing agent After the resultant analyte solution was introduced the SER-active vial was immediately fixed horizontally to an XY positioning stage (Conix Research Springfield OR) just inside the focal point of an f07 aspheric lens The lens focused the beam into the sample and collected the scattered radiation back along the same axis A dichroic filter (Omega Optical Brattleborough VT) was used to reflect the excitation laser to the lens and pass the Raman scattered radiation collected by the lens An f2 achromat was used to collimate the laser beam exiting a 200 microm core diameter source fiber optic while a second f2 achromat was used to focus the scattered radiation into a 365 microm fiber optic (Spectran Avon CT) A short pass filter was placed in the excitation beam path to block the silicon Raman scattering generated in the source fiber from reflecting off sampling optics and reaching the detector A long pass filter was placed in the collection beam path to block the sample Rayleigh scattering from reaching the detector A 785 nm diode laser (Process Instruments Inc model 785-600 Salt Lake City UT) was used to deliver 100 to 150 mW of power to the sample A Fourier transform Raman spectrometer (Real-Time Analyzers model IRA-785 East Hartford CT) and a silicon photo-avalanche detector (Perkin Elmer model C30902S Stamford CT) were used to acquire the SER spectra
3 RESULTS AND DISCUSSION In a previous study of MPA6 aimed at developing a concentration calibration curve and determining limits of detection (LOD) we observed an anomaly at 1050 cm-1 Since it was found that the band intensity changed as a function of concentration the band must be associated with a sample parameter Two possible parameters photon flux and pH are examined here The first parameter was investigated by irradiating a 1mgmL MPA sample in a SER-active vial with laser powers of 200 mW and above and monitoring spectral changes It was immediately found that the 1050 cm-1 band
2O+ H HF + +OH
OH3H C
OP
OF
CH
CH3
3
3H C
OCP
OOH
CH
CH3
3
3H C
O
CPHO
CH
CH
3
3
C
Sarin IMPA MPA 2-propanol
SPIE -2003-5269 119
appeared and grew as a function of time and that the higher the laser power the faster the growth Figure 1 shows the growth of the 1050 cm-1 band over the course of 30 minutes when using 150 mW of 785 nm excitation while Figure 2A shows that the growth can be fit with a first order exponential equation namely I1050 = 03+05e-013t Furthermore the 760 cm-1 band could be fit with a first order decay equation with an identical exponential rate constant ie I760 = 08-
08e-013t The rates represent classical first order kinetics and their correspondence allows one to conclude that MPA is being transformed one-for-one into a photo-generated product At this time the photoproduct has not been positively identified but phosphonic (phosphorous) acid and phosphonate are likely candidates since the symmetric P(OH)3 stretch occurs at ~1050cm-1 Our studies show that a reduction in laser power to 100 mW at the sample for MPA essentially eliminates this degradation process This laser power or lower was used for further measurements
Methyl phosphonic acid is a diprotic acid that stepwise dissociates into two anions MPA- and MPA= according to the following reactions26
MPA MPA- + H+ pKa1 = 212 Reaction 2
MPA- MPA= + H+ pKa2 = 729 Reaction 3 The relative concentrations of MPA MPA- and MPA= can be determined at any pH by expressing [MPA] and [MPA=] in terms of [MPA-] using Reactions 2 and 3 and summing all three to equal the total starting concentration here 2 mgmL (0021M MW = 9602) viz
[MPA] + [MPA-] + [MPA=] = 0021M Equation 1 substituting from Reactions 1 and 2
([H+][MPA-])K1a + [MPA-] + (K2a[MPA-])[H+] = 0021M Equation 2 rearranging [MPA-] = 0021M(1+[H+]K1a + K2a[H+]) Equation 3 The relative concentrations of MPA MPA- and MPA= as a function of pH are shown in Figure 3 It is worth noting that near neutral pH both MPA- and MPA= will be present To confirm that the SER signal followed this pH dependence a starting solution consisting of 20 mg of MPA in 10 mL HPLC grade water was prepared and brought to pH of 20 using dilute nitric acid From this solution 2 mL were added to a SER-active vial and the SER spectrum recorded At this pH a peak at 760 cm-1 was barely discernable The 2 mL solution was returned to the starting solution and the pH was re-measured to correct for any changes that the silver-doped sol-gel vials might cause In most cases the change was less than 02 pH units and the pH is reported as the before and after average Next the pH of the
Figure 2 A) Exponential growth of 1050 cm-1 band and B) exponential decay of 760 cm-1 band for spectral series in Figure 1
Figure 1 Growth of 1050 cm-1 band as a function of time due to exposure to 150 mW of 785 nm Spectra are 5 sec each collected every 100-sec from 0 to 30-min
B
570 770 970 1170 1370Raman Shift (cm-1)
Arbit
rary
Unit
s
MPA Photodegradation
600 800 1000 1200 1400 Wavenumber (cm-1)
A
0 10 20 30 0 10 20 30 time (min) time (min)
Ram
an In
tens
ity (r
elat
ive)
30
min
0
SPIE -2003-5269 120
starting solution was adjusted to 325 using dilute KOH Again 2 mL were added to a vial and the SER spectrum recorded At this pH a reasonably strong 760 cm-1 band was observed This process was repeated as spectra were recorded at pHs of 70 74 75 79 85 and 100 A total of 1 mL of KOH was added diluting the total concentration by 10 Next the pH of the starting solution was made acidic by adding dilute nitric acid dropwise This time spectra were recorded at pHs of 72 69 64 and 37 Figure 4 shows the SER spectra for representative pHs (spectra were left out to simplify the figure) while Figure 3 shows the 760 cm-1 peak intensities as a function of pH (The band intensities were adjusted to compensate for dilution effects caused by the addition of HNO3 and KOH then normalized to 0021 M for the most intense band observed at pH 37) It is clear from Figure 3 that the 760 cm-1 band follows the MPA- concentration as a function of pH and must be assigned to this anion No bands were observed that corresponded to MPA or MPA= The lack of an MPA SER spectrum may be due to the absence of an attraction between the neutral analyte and the electropositive silver surface The same reasoning suggests that a strong SER spectrum should be observed for MPA= but it is not and a satisfactory explanation has not been found
0000
0005
0010
0015
0020
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14pH
Con
cent
ratio
n [M
]
pK1 = 212 pK2 = 729
MPA- MPA=
MPA
Figure 4 SER spectra of 002M MPA as a function of pH Conditions 100 mW of 785 nm 36 scans (1 min) 8 cm-1 recorded 2 min after sample introduction pH 19 and 20 not apparent on this scale pH 69 and 74 near identical to 70 and 75 and not shown for clarity
00
02
04
06
08
10
12
14
16
18
0 5 10 15 20 25 30Measurement Number
Ram
an In
tens
ity (7
60 c
m-1
)
Figure 3 Concentration dependence of MPA MPA- and MPA= as a function of pH for a 002M sample Intensity of 760 cm-1
band from Figure 6 as a function of pH ( for increasing basic adjustment for increasing acidic adjustment error was measured at ~10 for pH 64)
Figure 5 SER spectra of 002M MPA at pH 64 measured around a vial at three heights (9 points per height) Conditions as in Figure 4 but 10-sec scans
Since these measurements involved the removal and replacement of the SER-active vial in the sample holder to remove and add sample variation in the intensity as a function of vial position was minimized by illuminating the exact same height along the vial wall But this does not account for variability of the SERS response of the sol-gel coating around the vial To analyze this effect a vial containing MPA at pH 64 was rotated at ~ 40o
intervals at the original height and 18rdquo above and below this value Figure 5 shows the intensity of the 760 cm-1
band for the 27 positions It was found that the average value was 137plusmn014 an RSD of 10 overall and 5 for each height An error bar is included in Figure 4 for the pH 64 measurement
SPIE -2003-5269 121
With the above analyses in mind a preliminary investigation of the SER spectral response for MPA (as MPA-) as a function of concentration was performed A single vial was used for these measurements beginning with 1 mgL followed by measurements of 10 100 and 1000 mgL In all cases the pH was ~7 and 3 positions around the vial were measured per concentration Since the 760 cm-1 band was not observed for concentrations of 1 or 10 mgL using 100 mW of 785 nm the laser power at the sample was raised to 200 mW beginning with the 10 mgL concentration Photo-degradation was largely avoided (and not observed) by exposing the sample for only 33 seconds per spectral acquisition Representative spectra for 10 100 and 1000 mgL are shown in Figure 6 while a plot of the 760 cm-1 band intensity as a function of concentration is shown in Figure 7 These values were also used to estimate limits of detection based on the signal-to-noise ratio (SN) of the 760 cm-1 band using 100 mW of 785 nm laser excitation and a 10-min acquisition time As summarized in the Figure 7 inset the lower the measured concentration the lower the predicted LOD Even if as estimated 210 microgL could be measured using the silver-doped sol-gel vials an improvement of a factor of 70 is still required to meet JSAWM goals of measuring 3 microgL in 10-minutes Similar to MPA DPA is a diprotic acid (pKa1 = 216 and pKa2 = 692) and variations in pH will effect the relative concentrations of DPA DPA- and DPA= and possibly the SER spectra and band intensities This could prove significant if an acid or base is used to denature anthrax spores with the goal of extracting and analyzing DPA The MPA pH study described above was mimicked for DPA except that the starting solution consisted of 20 mg of DPA in 20 mL HPLC grade water (60x10-3M MW = 1671) The initial solution had a pH of 245 which was made basic by dilute KOH to pHs of 355 433 487 559 1069 and 1166 SER spectra were recorded at each pH using 100 mW of 785 nm and a 44-sec acquisition time Next one drop of concentrated nitric acid was used to remake the solution acidic at a pH of 200 Again sequential pH measurements were performed at 383 510 735 and 822 The solution pH was made acidic a third time but to pH 219 171 then 135 Throughout this process no more than 20 drops of acid or base were added and therefore the concentration was diluted by no more than 10 Most of the spectral bands showed a minor decrease in intensity as a function of increasing pH values However the bands at 525 and 795 cm-1 showed the most dramatic changes which occurred at acid pH Figure 8 shows the SER spectra of DPA for the spectral region and pH range of interest The identity of the DPA species was determined by plotting the normalized peak intensities with the lowest value set to 0 and the highest to 0006 M as a function of pH and overlaying these values on a plot of the relative concentrations for DPA DPA- and DPA= as previously done for MPA (Figure 9) As can be seen the 525 cm-1 band clearly corresponds to DPA The correspondence of the 795 cm-1 band to this species is less clear as the band retains intensity until pH of 55 This can be attributed to contribution to the overlapping band at 810 cm-1 which does not change as a function of pH The fact that most bands are observed at all pHs suggest that the primary interaction with silver is through the ring nitrogen This is supported by the fact that the most intense band occurs at 1008 cm-1 attributed to a symmetric ring breathing mode and that this interaction has been characterized for pyridine in numerous papers27
0
1
2
3
4
5
6
7
0 200 400 600 800 1000 1200
MPA Concentration (mgL)
760
Ban
d In
tens
ity
Figure 6 SER spectra of MPA in water at A) 1000 B) 100 and C) 10 mgL Conditions pH of 7 silver-doped sol-gel coated vial 200 mW of 785 nm 33-sec 8 cm-1 resolution
Figure 7 Plot of SER intensity of 760 cm-1 band of MPA as a function of concentration using 200 mW of 785 nm Inset table includes average intensity LOD standard deviation and percent deviation for each concentration but for 100 mW and 10 min
conc (mgL) LOD ave stddev dev10 021 002 1021100 072 010 14301000 312 040 1280
A
B C
SPIE -2003-5269 122
The ability to detect dipicolinic acid (DPA) by SERS is demonstrated in Figure 10 Even at 1 mgL the primary bands are visible The intensity of the dominant ring breathing mode observed at 1008 cm-1 for DPA in water (pH = 25 to 55) is plotted as a function of concentration in Figure 11 These values were also used to estimate limits of detection based on the SN of the 1008 cm-1 band using 100 mW of 785 nm laser excitation and a 10-min acquisition time Again the lower the measured concentration the lower the predicted LOD (see Figure 11 inset) and detection of 160 microgL is possible Gastrointestinal anthrax requires significant more spores than inhalation anthrax28 and a limit of detection might be placed at 1 million spores in 1 liter of water or 10 microgL Since each spore contains ~10 CaDPA by weight29 a goal for DPA might be 1 microgL indicating that the present measurements must be improved by nearly two orders of magnitude Due to the increased hazards of handling HCN gas KCN salt was used for these experiments Nevertheless all sample preparations were performed in a chemical hood KCN completely dissolves in water but its conjugate acid HCN is formed and has a Ka of 615x10-1030 viz
HCN CN- + H+ pKa = 921 Reaction 4
Figure 8 SER spectra of 1 mgmL DPA as a function of pH Conditions 100 mW 785 nm 100 scans (44-sec)
Figure 9 Concentration dependence of DPA DPA- and DPA= as a function of pH for a 0006M sample Intensity of 525 () and 795 (diams) cm-1 bands from Fig 8 as a function of pH
pH
135 171 219 383
0
0001
0002
0003
0004
0005
0006
0007
0 2 4 6 8 10 12 14pH
Con
cent
ratio
n [M
]
DPADPA-DPA=795525
DPA DPA=DPA-
pK2 = 692pK1 = 216
795 525
0
05
1
15
2
25
0 200 400 600 800 1000 1200
DPA Concentration (mgL)
1008
Ban
d In
tens
ity
conc (mgL) lod-10min-100mw1 017
10 016100 103
1000 355
Figure 10 SER spectra of DPA in water at A) 1000 B) 100 C) 10 and D) 1 mgL Conditions pH of 25-55 silver-doped sol-gel coated vial 175 mW of 785 nm 1-min 8 cm-1 D) has been multiplied by x10 to make bands visible
A
B C
D
Figure 11 Plot of SER intensity of 1008 cm-1 band of DPA as a function of concentration using 175 mW of 785 nm Inset table includes LOD in mgL for each concentration but for 100 mW and 10 min
SPIE -2003-5269 123
Consequently the cyanide concentration must be determined for each initial KCN concentration Specifically the samples prepared with concentrations of 01 1 10 100 and 1000 mgL of KCN produced CN- concentrations of 63x10-3 033 69 89 and 964 mgL at pHs of 816 90 967 102 and 107 respectively The pH dependence for the HCN and CN- concentrations are shown in Figure 12 Thus as the amount of KCN added to the solution decreases so does the pH of the solution (becomes less basic) and according to Reaction 4 the relative amount of CN- to HCN also decreases For example in the preparation of a 01 mgL solution of KCN the pH is shifted from 7 for pure water to only 816 and only 63 of the starting material becomes CN- or 63x10-3 mgL In comparison for a solution of 1000 mgL the pH is shifted from 7 to 107 and 96 of the starting material becomes CN- This is significant since the cyanide ion is better able to adsorb onto the silver particles and become SERS active SER spectra of 10 100 and 1000 mgL of KCN samples are shown in Figure 13 and a corresponding plot of the 2100 cm-1 band intensity as a function of concentration is shown in Figure 14
000102030405060708091011
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14pH
Con
cent
ratio
n [m
gm
L]
pKa = 921
CN -HCN
The SER spectra of cyanide are dominated by the single CN stretch at 2100 cm-1 which occurs in normal Raman spectra of solutions at 2080 cm-1 However a low frequency mode occurs at 135 cm-1 due to a Ag-CN stretch (not shown) indicative of a strong surface interaction It is also observed that as the concentration decreases the CN stretch shifts to 2140 cm-1 This shift has been attribute to the formation of a tetrahedral Ag(CN)3
2- surface structure31 as well
Figure 14 Concentration dependence of KCN SERS measured under conditions in Fig 11 Concentrations are 1 01 and 001 mgml Intensities are measured for the CN stretch at 2100 cm-1 Inset table includes LOD in mgL for each concentration in Figs 13 and 15 but for 100 mW and 10 min
Figure 13 SER spectra of KCN in water at A) 1000 B) 100 and C) 10 mgL Conditions pHs of 107 102 and 97 silver-doped sol-gel coated vial 100 mW of 785 nm 1-min 8 cm-1 C) has been multiplied by x10 to make band visible
Figure 12 Concentration dependence of HCN and CN- as a function of pH for a 1 mgmL sample Calculated intensity of 2100 () cm-1 band for a 1 mgml sample at pHs of 816 90 967 102 and 107
A
B
C
0
20
40
60
80
100
120
140
0 200 400 600 800 1000 1200CN Concentration (mgL)
2100
Ban
d In
tens
ity
Figure 15 SER spectra of KCN in water at A) 10 B) 1 and C) 01 mgL Conditions pHs of 97 12 and 12 gold-doped sol-gel coated vial 100 mW of 785 nm 1-min 8 cm-1 C) has been multiplied by x10 to make band visible
A
B
C
conc (mgL) lod-10min-100mw Condition01 001 Au-pH 121 007 Au-pH 12
69 003 Au-pH 9769 016 Ag-pH 9789 022 Ag-pH 102946 113 Ag-pH 107
SPIE -2003-5269 124
as to CN adsorbed to two different surface sites32 Alternatively the 2140 cm-1 band could be attributed to HCN since this species dominates at lower concentrations However it is unlikely that this species would be attracted to the electropositive silver surface Further both peaks should be present at pHs between 85 and 105 but this is not observed It has also been suggested that at concentrations near and above monolayer coverage the CN- species is forced to adsorb end-on due to crowding and at lower concentrations the molecule can reorient to lie flat33 This suggests that the 2100 and 2140 cm-1 bands correspond to the end-on and flat orientations respectively As Figures 13 and 14 show the intensity of the CN stretch for the 89 mgL sample is nearly as intense as the 964 mgmL band This suggests that the Raman signal for the flat orientation is more enhanced However more extensive measurements are required to verify this point Since resent research has suggested that cyanide may be more effectively detected on gold measurements of KCN solutions were also performed using gold-doped sol-gel vials Preliminary measurements are shown in Figure 15 for samples prepared from 01 1 and 10 mgL KCN Since the pHs are 816 90 967 the resultant CN- concentrations are 63x10-3 033 and 69 mgL Initially only the highest concentration was observed and the signal intensity was significantly better than the equivalent concentration measured using silver In an effort to shift Reaction 4 to the left transforming HCN to CN- (Le Chatelierrsquos principle) KOH was added to the lower concentration samples producing solutions with pH 12 This effectively forces all of the cyanide in solution to be CN- or 01 and 10 mgL respectively More importantly the CN stretch is now observed in the SER spectra The band appears at 2125 cm-1 as has been previously reported for gold12 As calculated for MPA and DPA LODs can be estimated from this data For the three concentrations of cyanide on silver the LODs are 016 to 11 mgL for 100 mW of 785 nm laser excitation and a 10-min acquisition time For gold pH adjusted the LODs are10 to 70 microgL an improvement of more than 10 times silver Nevertheless either substrate is sufficient to meet the JSAWM goals of measuring 3 mgL in 10-minutes as the requirements form cyanide are much less stringent than the nerve agents
4 CONCLUSIONS Here we examined the surface-enhanced Raman spectral response for methyl phosphonic acid dipicolinic acid and cyanide as a function of pH It was determined that the most prominent peak at 760 cm-1 reaches a maximum intensity between pH 3 and 7 and corresponds to the MPA- species Neither the MPA nor MPA= species appear to generate a SER spectrum and consequently no spectra were observed below pH 2 or above pH 8 In this study we also found that higher laser powers could cause photodegradation of MPA signified by the exponential growth of a band at 1050 cm-1 which is tentatively assigned to phosphorous acid Unlike MPA DPA was observed at all pHs This is attributed to the dominant interaction of the pyridine functional group with silver Minor spectral changes were observed at acid pHs and were assigned to neutral DPA Like MPA SER spectra of cyanide were pH dependent No spectra were observed for the HCN species while CN- was best observed at pHs more basic than 8 Preliminary concentration studies for the three analytes allowed estimating limits of detection for MPA DPA and CN using 100 mW of 785 nm and a 10-min acquisition time of 210 165 and 70 microgL respectively Although the latter value suggests that cyanide can be measured at sufficiently low concentrations to meet JSAWM goal improvements by 100 to 200 times are required for MPA and DPA It is clear from this study that pH of the sample is important and must be taken into account when developing concentration calibration curves as well as fieldable analyzers Future work will include tailoring the sol-gel with specific functional groups to dictate sample pH and thereby optimize sensitivity
ACKNOWLEDGMENTS The authors are grateful for the support of the US Army (DAAD13-02-C-0015 Joint Service Agent Water Monitor program) The authors would also like to thank Dr Steve Christensen of the US Army SBCCOM for helpful discussions and Mr Chetan Shende of Real-Time Analyzers for assistance in development of the gold-doped sol-gels
REFERENCES 1 Erickson B Analytical Chemistry News amp Features June 1 397A (1998)
SPIE -2003-5269 125
2 Johnston RL Hoefler CM Fargo JC and Moberley B AT-ONSITE 5-8 (1994) 3 Black RM RJ Clarke RW Read and MT Reid J Chromatography 662 301-321 (1994) 4 USA Army Center for Health Promotion and Preventive Medicine (USACHPPM) Technical Guideline 230A 5 Jensen JL NC Wong W R Loerop SPIE 4775-03 (2001) 6 Hoenig SL Handbook of Chemical Warfare and Terrorism Greenwood Press Wesport CT (2002) 7 Munro NB SS Talmage GD Griffin LC Waters AP Watson JF King and V Hauschild Env Health
Persp 107 933-974 (1999) 8 Holstege CP Kirk M Sidell FR Crit Care Clin13 923-42 (1997) 9 Farquharson S P Maksymiuk K Ong and S Christesen SPIE 4577 166-173 (2001) 10 Lee Y and S Farquharson SPIE 4378 21-26 (2001) 11 Spencer KM J Sylvia S Clauson and J Janni SPIE 4577 158-165 (2001) 12 Tessier P S Christesen K Ong E Clemente A Lenhoff E Kaler and O Velev Applied Spectroscopy 56
1524-1530 (2002) 13 Farquharson S WW Smith S Elliott and JF Sperry SPIE 3533 207-214 (1998) 14 Farquharson S WW Smith S Elliott and JF Sperry SPIE 3855110-116 (1999) 15 Farquharson S WW Smith YH Lee S Elliott and JF Sperry SPIE 4575 62-72 (2002) 16 Guzelian AA J Sylvia J Janni S Clauson and KM Spencer SPIE 4577 182-192 (2001) 17 Shende C F Inscore A Gift P Maksymiuk and S Farquharson in press 18 Weaver MJ S Farquharson and MA Tadayyoni J Chem Phys 82 4867-4874 (1985) 19 Alak AM and T Vo-Dinh Analytical Chemistry 59 2149-2153 (1987) 20 Lee Y and S Farquharson SPIE 4206 140-146 (2000) 21 Farquharson S and Y Lee SPIE 4200-16 (2000) 22 Lee Y S Farquharson and P M Rainey SPIE 3857 76-84 (1999) 23 Lee Y S Farquharson H Kwong and M Shahriari SPIE 3537 252-260 (1998) 24 Wang J M Pumera G Collins and A Mulchandani Analytical Chemistry 74 6121-6125 (2002) 25 Dou X YM Jung Z-Q Cao and Y Ozaki Applied Spectroscopy 53 1440-1447 (1999) 26 Data supplied by S Christesen and K Ewing 27 Kerker M and B Thompson Eds SPIE MS 10 (1990) 28 Inglesby TV DA Henderson JG Bartlett JAMA 287 2236 (2002) 29 FW Janssen AJ Lund and LE Anderson Science 127 26 (1958) 30 Lide DR Ed Handbook of Chemistry and Physics CRC Press 77th Ed (1996-7) 31 Billmann J G Kovacs and A Otto Surf Sci 92 153 (1980) 32 Murray CA and S Bodoff Phys Rev B 32 671 (1985) 33 Kellogg D and J Pemberton J Phys Chem 91 1120 (1987)
SPIE-2003-5269 16
Chemical agent detection by surface-enhanced Raman spectroscopy
Stuart Farquharson Alan Gift Paul Maksymiuk Frank Inscore and Wayne Smith
Real-Time Analyzers 87 Church Street East Hartford CT 06108
Kevin Morrisey and Steven D Christesen US Army SBCCOM Aberdeen Proving Ground MD 21010
ABSTRACT
In the past decade the Unites States and its allies have been challenged by a different kind of warfare exemplified by the terrorist attacks of September 11 2001 Although suicide bombings are the most often used form of terror military personnel must consider a wide range of attack scenarios Among these is the intentional poisoning of water supplies to obstruct military operations in Afghanistan and Iraq To counter such attacks the military is developing portable analyzers that can identify and quantify potential chemical agents in water supplies at microgram per liter concentrations within 10 minutes To aid this effort we have been investigating the value of a surface-enhanced Raman spectroscopy based portable analyzer In particular we have been developing silver-doped sol-gels to generate SER spectra of chemical agents and their hydrolysis products Here we present SER spectra of several chemical agents measured in a generic tap water Repeat measurements were performed to establish statistical error associated with SERS obtained using the sol-gel coated vials Keywords Chemical agents chemical agent detection SERS Raman spectroscopy
1 INTRODUCTION
In the autumn of 2001 terrorism within US borders became a sobering reality While extensive efforts are being implemented to secure the homeland US and Coalition forces in Afghanistan and Iraq are constantly faced with terrorist attacks In addition to car-bombings the military has identified several non-traditional attack scenarios including poisoning of water supplies by chemical warfare agents (CWAs) To counter this threat the Department of Defense has been investigating numerous approaches to detect such attacks under the auspices of the Joint Service Agent Water Monitor (JSAWM) program1 The JSAWM must be capable of identifying the chemical agents with no false-positives and quantifying the agents at microgram per liter concentrations within 10 minutes2 This includes the analysis of drinking water supplies distribution and storage systems Currently colorimetric paper is used to detect agents on-site while gas chromatography combined with mass spectrometry (GCMS) is used in mobile support laboratories However both methods have severe drawbacks The paper changes color in response to contact with many chemicals besides CWAs causing a high incidence of false positives3 while GCMS although very chemically specific requires up to an hour to perform and regular re-calibration456 Vibrational spectroscopy has also been investigated as a potential method of rapidly detecting CWAs7-
11 as early as 1985 in the case of infrared spectroscopy8 and 1988 in the case of Raman spectroscopy11 Again however these techniques also have limitations when it comes to measuring trace poisons in water Infrared spectra would be dominated by the very strong absorption of water which would obscure absorptions by most other chemicals present Whereas Raman spectroscopy is simply not a very sensitive technique and detection limits are typically grams per liter Surface-enhanced Raman spectroscopy (SERS) offers several advantages over conventional vibrational methods and that may provide the necessary sensitivity required for detecting trace quantities of chemical agents in water SERS employs the interaction of surface plasmon modes of metal particles with target analytes to increase scattering efficiency by as much as 1 million times12 In 1987 the potential of this approach to measure CWAs was demonstrated by measuring a series of pesticides13 Several of these organophosphonates have chemical structures similar to CWAs in particular P=O functional groups In the past few years we and others have further explored the ability of SERS to detect CWAs14-17 and even bioagents 18-21 We have been employing silver-doped sol-gels to promote the SER effect
SPIE-2003-5269 17
in these studies The porous silica network of the sol-gel matrix offers a unique environment for immobilizing and stabilizing SER-active metal particles22-25 The silver-doped sol-gels have been coated on the internal walls of standard glass vials to allow rapid SER analysis of small sample quantities without preparation We have measured over 100 chemicals with enhancements of 104 to 106 demonstrated reversible measurements in a flowing system reproducible measurements from vial-to-vial and measurements in multiple solvents including water21-25 Previously we used these vials to perform preliminary measurements of cyanide (CN) and methylphosphonic acid (MPA) Most of the nerve agents form MPA during hydrolysis while Tabun forms CN a chemical agent in its own right In another paper including in these proceedings we examined the limits of detection (LOD) for MPA by measuring a series of concentrations down to 50 mgL and estimated a limit of detection of 100 microgL26 These measurements provide encouragement in that SERS may satisfy the needs of the JSAWM To further establish the viability of SERS in particular silver-doped sol-gels here we present analysis of cyanide mustard and VX in tap water The measurements performed at the US Armyrsquos Edgewood Chemical Biological Center Aberdeen MD also included numerous repeat measurements to establish reproducibility
2 EXPERIMENTAL 2a General All chemicals including potassium cyanide 2-chloroethylethyl sulfide and those used to prepare the silver-doped and gold-doped sol-gel coated vials were obtained and used as received from Sigma-Aldrich All samples were prepared in a chemical hood using HPLC grade water unless otherwise noted (Fischer Scientific Fair Lawn NJ) for SERS measurements Once prepared the samples were transferred into the silver-doped sol-gel vials (Real-Time Analyzers Inc East Hartford CT) The vials were coated in a manner similar to that previously reported by adding ammonium hydroxide to a solution of silver nitrate tetramethyl orthosilicate (TMOS) and methanol22 The two precursor solutions were prepared mixed and transferred to 2-mL glass vials dried and heated After sol-gel formation the incorporated metal ions were reduced with dilute sodium borohydride (1mgmL) which is followed by a water wash to remove residual reducing agent After the resultant analyte solution was introduced the SER-active vial was fixed horizontally to an XY positioning stage (Conix Research Springfield OR) just inside the focal point of an f07 aspheric lens The lens focused the beam into the sample and collected the scattered radiation back along the same axis A dichroic filter (Omega Optical Brattleborough VT) was used to reflect the excitation laser to the lens and pass the Raman scattered radiation collected by the lens An f2 achromat was used to collimate the laser beam exiting a 200 microm core diameter source fiber optic while a second f2 achromat was used to focus the scattered radiation into a 365 microm fiber optic (Spectran Avon CT) A short pass filter was placed in the excitation beam path to block the silicon Raman scattering generated in the source fiber from reflecting off sampling optics and reaching the detector A long pass filter was placed in the collection beam path to block the sample Rayleigh scattering from reaching the detector A 785 nm diode laser (Process Instruments Inc model 785-600 Salt Lake City UT) was used to deliver 100 to 150 mW of power to the sample A Fourier transform Raman spectrometer (Real-Time Analyzers model IRA-785 East Hartford CT) and a silicon photo-avalanche detector (Perkin Elmer model C30902S Stamford CT) were used to acquire the SER spectra
2b Edgewood Chemical Biological Center The surface-enhanced Raman spectral measurements at the US Armyrsquos Chemical Biological center presented here were all performed on September 12 2003 To expedite measurements a plate was machined to hold up to 12 SER-active sample vials (Figure 1) The plate fit a standard XY plate reader that could be programmed Pure KCN bis-(2-chloroethyl)sulfide (distilled mustard gas HD) and ethyl S-2-diisopropylamino ethyl methylphosphonothioate (VX) were obtained on-site and used to prepare 1 gL tap water solutions in a chemical hood with appropriate safety equipment Simulated tap water was prepared by adding 10 chemicals most often found in tap water at appropriate concentrations to distilled water (Table 1) SER measurements were also performed in a chemical hood For added safety the FT-Raman instrument was placed outside the laboratory and 30 foot fiber optic and electrical cables were used to allow remote SERS measurements and plate manipulation For each experiment 1gL samples were prepared and added to 9 individual vials which were then loaded on the plate In some cases a tenth vial was included as a blank
Table 1 Generic Tap Water Composition
Compound mgL NaHCO3 100 CaSO4 27 MgSO4bull7H2O 67 NaNO3 1 Fulvic Acid 1 K2HPO4 07 KH2PO4 03 (NH4)2HSO4 001 NaCl 001 FeSO4 0001 pH 76-78
SPIE-2003-5269 18
A software program was written that allowed selecting the sequence that the vials were measured the number of positions along the length of the vials to measure (1 to 5) and the number of scans to co-add During sample analysis the program displayed the vial being analyzed the point being analyzed and the spectrum as it was being acquired Once all the data was collected a second software program was written to rapidly analyze the data The spectra collected for all the vials on a plate could be loaded at one time and then the spectra for each point could be displayed simultaneously or separately The user could then select the Raman peak to analyze in terms of peak height or area This was accomplished by selecting points on either side of the peak to define a baseline of zero The peak height or area could then be computed for all of the spectra loaded and then exported to a spreadsheet for statistical analysis
Figure 1 A) Vial Holder 6 slots to hold 2 vials each end-to-end B) Measurement Configuration Program user selects vials to measure sequence number of points per vial (1 to 5) and number of scans per point C) Spectral Acquisition Program shows spectrum being collected which vial and position D) Spectra Analysis Program user selects spectra to analyze by plate vial and point (s) as well as two wavenumbers defining the peak and the baseline to subtract The image is of 5 repeat measurements of 10 mgL KCN in generic tap water 16 sec each 100 mW of 785 nm
3 RESULTS AND DISCUSSION Raman and surface-enhanced Raman spectra were obtained for potassium cyanide bis-(2-chloroethyl)sulfide and ethyl S-2-diisopropylamino ethyl methylphosphonothioate representing three classes of chemical agents cyanides mustards and nerve agents respectively Spectra were also obtained for 2-chloroethyl ethyl sulfide (CEES) a structural analogue to HD which was included in the study to aid in assigning spectral bands KCN salt was used for cyanide experiments to avoid the increased hazards of handling HCN gas KCN completely dissolves in water forming its conjugate acid HCN according to its Ka of 615x10-1027 and at a concentration of 1 mgmL results in a pH 107 solution This is significant in that only CN- appears to interact sufficiently with silver to produce a SER spectrum and no spectral signal is observed below pH 726 Figure 2 shows the SER and normal Raman spectra for KCN The SERS spectrum of 1mgml KCN in water shows a single intense somewhat broad feature at
A B
C D
SPIE-2003-5269 19
2100 cm-1 assigned to the single CequivN stretch The band is much sharper in the normal Raman spectra of the solid KCN salt at 2074 cm-1 This band does broaden and shift to 2080 cm-1 in solution (not shown) However the observed SERS frequency is attributed to interaction with silver and low frequency mode at 135 cm-1 attributed to a Ag-CN stretch (not shown) supports this conclusion
cm-1 band to a C-S stretch but the authors concede that it is in fact more likely a C-Cl stretch28 It appears that the most intense Raman bands at 648 692 and 747 cm-1 shift to 620 660 and 720 cm-1 in the SER spectra and are tentatively assigned as above The width of these bands suggests that they overlap underlying spectral features Additional bands in the Raman spectra occur at 972 1034 1049 1263 1286 1423 1442 2865 2935 and 2960 cm-1 Corresponding bands occur in the SER spectra at 964 1015 1054 1286 1410 1447 2865 and 2935 cm-1 Most of these bands are associated with alkane modes specifically the bands at approximately 1040 cm-1 to a C-C stretch 1290 cm-1 to a CH2 in-phase twist 1440 cm-1 to a CH2 wag 2865 cm-1 to a symmetric CH2 stretch and 2965 cm-1 to an asymmetric CH2 stretch The Raman and SER spectra of sulfur mustard were measured at the Edgewood center (Figure 4) Both spectra are largely similar to CEES The C-Cl and C-S bands in the Raman spectrum of HD now occur at 640 655 700 739 and 760 cm-1 and are more resolved possibly due to the increased molecular symmetry Theoretical calculations indicate that the first three bands are due to C-Cl stretching modes and the latter two to C-S stretching modes28 Only the C-Cl bands maintain significant intensity in the SER spectra occurring at 624 and 643 cm-1 which is attributed to the
Prior to measurements of HD CEES was examined by Raman and SER spectroscopy (Figure 3) CEES also known as half-mustard is essentially identical to HD except one of the chlorine end atoms is replaced by a hydrogen atom Again although not as toxic as HD CEES is a blister agent and dilute aqueous samples were prepared using appropriate safety equipment Both the Raman and SER spectra of CEES are similar and dominated by bands between 600 and 800 cm-1 These are associated with C-Cl and C-S stretching modes which are tentatively assigned to 648 and 747 cm-1 in the Raman spectra respectively The shoulder at 630 cm-1 the overlapped band at 660 cm-1 and the strong band at 692 cm-1 could also be due to these modes or their asymmetric counterparts It is worth noting that theoretical calculations assign the 692
Figure 2 A) SER and B) NR spectra of KCN Conditions A) 1 mgml in tap water 100 mW of 785 nm at sample 1-min acquisition time B) solid 300 mW of 785 nm 5-min All spectra are 8 cm-1 resolution
A
B
Figure 3 A) SER and B) NR spectra of CEES Conditions A) 1 vv (10 mgml) in MeOH 100 mW of 785 nm 1-min acquisition time B) neat 300 mW of 785 nm 5-min
A
B
Cl-CH2-CH2-S-CH2-CH3
A
B
Figure 4 A) SER and B) NR of HD Conditions A) 1mgml in tap water B) pure both 100 mW of 785 nm 1-min
Cl-CH2-CH2-S-CH2-CH2-Cl
SPIE-2003-5269 20
expected strong interaction between chlorine and silver and adds support to the assignment of this band to a C-Cl stretch Weaker overlapping bands occur at 670 692 and 724 cm-1 the latter possibly due to C-S stretching modes Again the alkane modes are apparent in the normal Raman spectra of HD but only a broad feature at 1300 to 1450 cm-1 suggests CH2 contributions in the SER spectrum Although the observed bands in the VX spectrum have not been assigned (Figure 4) a computer generated Raman spectrum29 predicts many of the same features with surprising accuracy and are used here Two intense bands at 460 and 530 cm-1 closely match predicted bands at 463 and 546 cm-1 assigned to a CH3-P=O bend and a PO2CS wag Three highly overlapped bands occur at 694 745 and 771 cm-1 matching predicted bands at 713 730 and 760 cm-1 The first
Table 2 Measured SER peak heights for the CN stretch at 2100 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
Figure 5 A) SERS and B) NR spectra of VX Conditions A) 1 vv (10 mgml) in MeOH B) pure sample both 100 mW of 785 nm 1-min acquisition time
A
B
two have been assigned to a C-S stretch and CH2bend respectively while the latter has been attributed to either a P-C stretch or an O-C-C stretch Although the 745 cm-1 band may alternatively be assigned to a C-S stretch based on the previous measurements of CEES and HD The relatively intense bands at 890 1106 1218 1445 and 1465 cm-1 also match predicted bands at 880 1108 1216 1440 and 1464 cm-1 that are assigned to a C-C stretch CH3 rock N-C3 stretch various C-H3 bends and C-H bends respectively Both the computer generated and the measured spectra contain numerous other less intense bands One is worth mentioning A unique band appears at 370 cm-1 that is predicted at 368 cm-1 and corresponds to an O-P=O bend The surface-enhanced Raman spectrum of VX is also rich with spectral features It has the unique low frequency band at 370 cm-1 as well as a second band at 380 cm-1 that is assigned tothe S-P-O bend predicted in the normal Raman spectrum at 388 cm-1 Based on the measured and predicted normal Raman spectra the following SERS assignments are given 460 cm-1 to the CH3-P=O bend 544 cm-1 to the PO2CS wag 738 cm-1 to a C-S stretch (based on arguments above) 890 cm-1
to a C-C stretch 1101 cm-1 to a CH3 rock and 1456
cm-1 to a C-H bend The ability of SERS to measure chemical agents in water containing real-world chemical interferents was tested by using the generic tap water described in Table 1 The ability to reproduce measurements was accomplished by preparing three separate water stock solutions which were used to prepare three sample solutions each of 1 mgmL KCN in the generic tap water The samples defined as vials 1-9 were then measured at 4 points per vial in a semi-random fashion such that errors associated with stock solution preparation and errors associated with instrument drift could be identified No trends were apparent that signified such systematic errors Each spectrum collected consisted of 20 averaged scans taking 16 seconds at 8 cm-1 resolution The laser power at the sample was measured periodically during the day and it ranged from 102 to 105 mW spacccc
spot Vial 1 Vial 2 Vial 41 13754 13004 128192 13519 12692 12909
1 3 13541 12721 126394 13462 12648 12651
ave 13569 12766 12755Vial 6 Vial 3 Vial 5
1 11586 14068 10757 2 11236 14402 115122 3 11376 14546 11559
4 10894 11763 11214ave 11273 13695 11261
Vial 7 Vial8 Vial 91 11198 15804 11023
2 11243 15716 88753 3 11562 15216 11218
4 11655 15084 11077ave 11415 15455 10548
CN AVG STDEV ERR1 mgmL pk ht 12526 1560 1245
stock solution
SPIE-2003-5269 21
The CN measurements consisting of 36 data points produced an average height of 125 for the 2100 cm-1 peak with a standard deviation of 156 or 125 (Table 2) The HD and VX measurements were performed precisely the same way (Tables 3 and 4) For HD the 624 cm-1 peak was used for analysis and it had an average height of 53 with a standard deviation of 068 or 129 while for VX the 544 cm-1 peak was used for analysis and it had an average height of 1051 with a standard deviation of 308 or 293 The greater error in the VX measurements can be somewhat attributed to Vial 4 which produced lower SER signal intensities But removing this vial from the data set changes the standard deviation to 233 only a modest improvement
spot Vial 1 Vial 2 Vial 41 525 609 4682 45 675 484
1 3 527 644 5094 559 807 669
ave 51525 68375 5325Vial 6 Vial 3 Vial 5
1 521 574 575 2 536 53 4492 3 509 508 451
4 565 594 379ave 53275 5515 4635
Vial 7 Vial 8 Vial 91 457 589 413
2 583 497 4273 3 544 505 4
4 528 5 509ave 528 52275 43725
HD AVG STDEV ERR1 mgmL pk ht 530 068 1291
stock solution
spot Vial 1 Vial 2 Vial 41 1464 1034 4672 1485 99 61
1 3 1491 989 5684 1041 777 553
ave 137025 9475 5495Vial 6 Vial 3 Vial 5
1 1058 942 1293 2 697 121 965
2 3 727 1206 11124 689 1254 746
ave 79275 1153 1029Vial 7 Vial 8 Vial 9
1 1127 783 163 2 1358 812 1614
3 3 1371 908 15124 1097 875 1397
ave 123825 8445 153825
VX AVG STDEV ERR1 mgmL pk ht 1051 308 2925
stock solution
4 CONCLUSIONS In this paper we examined the ability of surface-enhanced Raman spectroscopy to reproducibly measure CN HD and VX in tap water without chemical interference Both normal and surface-enhanced Raman spectra were examined to select unique bands suitable to identify and quantify these chemical agents For SER measurements the 2100 cm-1 C-N stretch was used for CN the 624 cm-1 C-Cl stretch was used for HD and the 544 cm-1 PO2CS wag was used for VX It was determined that 1 mgmL samples of each of these chemicals measured 36 times in glass vials coated with a silver-doped sol-gel reproduced measurements with standard deviations of 125 129 and 293 It was further found that the 10 chemicals added to simulate generic tap water did not interfere with or alter the SER spectra It should be noted that the concentrations used in this study were considerably greater than those required by the JSAWM program Current work involves improving SER sensitivity and designing sampling systems with better reproducibility This includes the development of fractal silver and gold structures within the sol-gel matrix and the development of chemically selective sol-gels
ACKNOWLEDGMENTS The authors are grateful for the support of the US Army (DAAD13-02-C-0015 Joint Service Agent Water Monitor program) The authors would also like to thank Janet Jensen Ronald Crosier and Kristina Gonser for helpful discussions
Table 4 Measured SER peak heights for the VX band at 544 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
Table 3 Measured SER peak heights for the HD band at 624 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
SPIE-2003-5269 22
REFERENCES 1 Jensen JL NC Wong W R Loerop SPIE 4775-03 (2001) 2 JSAWM Requirements at wwwsbccomapgeaarmymilRDAecbcrtPRODSERJSAWMjsawmhtml 3 Erickson B ldquoThe Chemical Weapons Convention Redefines Analytical Challengerdquo Anal Chem 70 397A (1998) 4 Johnston RL Hoefler CM Fargo JC and Moberley B ldquoThe Defense Nuclear Agencyrsquos
ChemicalBiochemical Weapons Agreements Verification Technology Research Development Test and Evaluation Program and its Requirements for On-Site Analysisrdquo AT-ONSITE 5-8 (1994)
5 Black RM Clarke RJ Read RW and Reid MT ldquoApplication of GC-MS and GC-tandem MS to the analysis of chemical warfare samples found to contain residues of the nerve agent sarin sulphur mustard and their degradation productsrdquo J Chrom 662 301-321 (1994)
6 USA Army Center for Health Promotion and Preventive Medicine (USACHPPM) Technical Guideline 230A 7 Hoffland LD Piffath RJ Bouck JB rdquoSpectral signatures of chemical agents and simulantsrdquo Optical
Engineering 24 982-984 (1985) 8 Braue EHJ Pannella MGrdquoCIRCLE CELL FT-IR Analysis of Chemical Warfare Agents in Aqueous Solutionsrdquo
App Spectrosc 44 1513-1520 (1990) 9 Tseng C-H Mann CK and Vickers TJ ldquoDetermination of Organics on Metal Surfaces by Raman
Spectroscopyrdquo App Spectrosc 47 1767-1771 (1993) 10 Staff Reporter ldquoUS Army Tests Lidar to Detect Biological Toxinsrdquo Photonic Spectra pg 50 December 1998 11 Christesen SD Raman cross sections of chemical agents and simulants App Spectrosc 42 318-321 (1988) 12 Weaver MJ Farquharson S Tadayyoni MA ldquoSurface-enhancement factors for Raman scattering at silver electrodesrdquo J Chem Phys 82 4867-4874 (1985) 13 Alak AM and Vo-Dinh T ldquoSERS of Organophosphorous Chemical Agentsrdquo Anal Chem 59 2149-2153 (1987) 14 Lee Y Farquharson S ldquoRapid chemical agent identification by SERSrdquo SPIE 4378 21-26 (2001) 15 Farquharson S Maksymiuk P Ong K Christesen S ldquoChemical agent identification by surface-enhanced Raman spectroscopyrdquo SPIE 4577 166-173 (2001) 16 Spencer KM Sylvia J Clauson S and Janni J ldquoSurface Enhanced Raman as a Water Monitor for Warfare
Agents in Waterrdquo SPIE 4577 158-165 (2001) 17 Tessier P Christesen S Ong K Clemente E Lenhoff A Kaler E Velev O ldquoOn-line spectroscopic characterization of sodium cyanide with nanostructured gold surface-enhanced Raman spectroscopy substratesrdquo
App Spectrosc 56 1524-1530 (2002) 18 Farquharson S W W Smith S Elliott and J F Sperry Rapid biological agent identification by surface- enhanced Raman spectroscopy SPIE 3855110-116 (1999) 19 Farquharson S WW Smith YH Lee S Elliott and J F Sperry Detection of biological signatures A
comparison of electrolytic and metal-doped sol-gel surface-enhanced Raman media SPIE 4575 62-72 (2002) 20 Guzelian AA Sylvia J Janni J Clauson S and Spencer KM ldquoSERS of whole cell bacteria and trace levels of biological moleculesrdquo SPIE 4577 182-192 (2001) 21 Shende C Inscore F Gift A Maksymiuk P Farquharson S ldquoRapid dipicolinic acid extraction from Bacillus spores detected by surface-enhanced Raman spectroscopyrdquo App Spectrosc 58 accepted 22 Lee Y and Farquharson S ldquoSERS Sample Vials Based on Sol-Gel Process for Trace Pesticide Analysisrdquo SPIE
4206 140-146 (2000) 23 Farquharson S and Lee Y ldquoTrace Drug Analysis by SERSrdquo SPIE 4200-16 (2000) 24 Lee Y Farquharson S and Rainey P M Surface-Enhanced Raman Sensor for Trace Chemical Detection in
Water SPIE 3857 76-84 (1999) 25 Lee Y Farquharson S Kwong H and Shahriari M ldquoSol-Gel Chemical Sensor for Surface-Enhanced Raman
Spectroscopyrdquo SPIE 3537 252-260 (1998) 26 Farquharson S A Gift P Maksymiuk F Inscore and W Smith ldquopH dependence of methyl phosphonic acid
dipicolinic acid and cyanide by surface-enhanced Raman spectroscopyrdquo SPIE 5269-19 (2003) 27 Lide DR Ed Handbook of Chemistry and Physics CRC Press 77th Ed (1996-7) 28 Sosa C RJ Bartlett K KuBulat and WB Person ldquoA theoretical study of harmonic vibrational frequencies and
infrared intensities of XCH2CH2SCH2CH2X and XCH2CH2SH (= H Cl)rdquo J Phys Chem 93 577-588 (1993) 29 Hameka HF and JO Jenson ldquoComputer-generated predictions of the structure and the IR and Raman spectra of
VXrdquo ERDEC-TR-065 May 1993
- Chemagents Appendicespdf
-
- SPIE2001-4575-Bioagent-SERSapdf
-
- Detection of bioagent signatures A comparison of electrolytic and metal-doped sol-gel surface-enhanced Raman media
- Stuart Farquharson Wayne Smith and Yuan Lee
-
- Real-Time Analyzers 87 Church Street East Hartford CT 06108
-
- ABSTRACT
- 4 CONCLUSIONS
- 5 ACKNOWLEDGEMENTS
- 6 REFERENCES
-
SPIE -2003-5269 117
pH dependence of methyl phosphonic acid dipicolinic acid and cyanide by surface-enhanced Raman spectroscopy
Stuart Farquharson Alan Gift Paul Maksymiuk Frank Inscore and Wayne Smith
Real-Time Analyzers 87 Church Street East Hartford CT 06108
ABSTRACT US and Coalition forces fighting terrorism in Afghanistan and Iraq must consider a wide range of attack scenarios in addition to car bombings Among these is the intentional poisoning of water supplies to obstruct military operations To counter such attacks the military is developing portable analyzers that can identify and quantify potential chemical agents in water supplies at microgram per liter concentrations within 10 minutes To aid this effort we have been investigating the value of a surface-enhanced Raman spectroscopy based portable analyzer In particular we have been developing silver-doped sol-gels to generate SER spectra of chemical agents and their hydrolysis products Here we present SER spectra of methyl phosphonic acid and cyanide as a function of pH an important factor affecting quantitation measurements which to our knowledge has not been examined In addition dipicolinic acid a chemical signature associated with anthrax-causing spores is also presented Keywords Chemical warfare agents agent detection agent hydrolysis SERS Raman spectroscopy homeland security
1 INTRODUCTION In the past decade the Unites States and its allies have been challenged by a different kind of warfare exemplified by the terrorist attacks of September 11 2001 Suicide bombings and the use of chemical agents are the norm and military personnel must consider a wide range of attack scenarios Among these is the intentional poisoning of water supplies to obstruct military operations in Afghanistan and Iraq Currently colorimetric paper is used to detect agents on-site while gas chromatography combined with mass spectrometry (GCMS) is used in mobile support laboratories However both methods have severe drawbacks The paper changes color in response to contact with many chemicals besides CWAs causing a high incidence of false positives1 while GCMS although very chemically specific requires hours to perform and constant re-calibration234 Military operations would be greatly aided by a portable analyzer that can identify and quantify potential chemical agents at concentrations that impact safety This includes the analysis of drinking water supplies distribution and storage systems To meet this goal the Department of Defense has been investigating numerous approaches under the auspices of the Joint Service Agent Water Monitor (JSAWM) program5 The JSAWM must be capable of identifying the chemical agents with no false-positives and quantifying the agents at microgram per liter concentrations within 10 minutes678 Recently we and others have been investigating the ability of surface-enhanced Raman spectroscopy (SERS) to measure chemical agents9-12 bioagents13-17 and their hydrolysis products in water SERS employs the interaction of surface plasmon modes of metal particles with target analytes to increase scattering efficiency by as much as 1 million times18 In 1987 the potential of this approach to measure CWAs was demonstrated by measuring a series of pesticides19 Several of these organophosphonates have chemical structures similar to CWAs in particular P=O functional groups In our studies we have been employing silver-doped sol-gels to promote the SER effect The porous silica network of the sol-gel matrix offers a unique environment for immobilizing and stabilizing SER-active metal particles20-23 The sol-gels have been coated on the internal walls of standard glass vials to allow rapid SER analysis of small sample quantities without preparation We have measured over 100 chemicals with enhancements of 104 to 106 demonstrated reversible measurements in a flowing system reproducible measurements from vial-to-vial and measurements in multiple solvents including water20-23 Previously we used these vials to perform preliminary measurements of cyanide (CN) methylphosphonic acid (MPA) and dipicolinic acid (DPA) MPA is a hydrolysis product of the nerve agents (eg sarin Reaction 1) and may be a valuable indicator of nerve agent usage particularly since the alkyl methylphosphonic acids are relatively more stable than their corresponding parent complexes24 DPA is
SPIE -2003-5269 118
a chemical signature of spore forming bacteria such as Bacillus anthracis And in light of the inability to rapidly detect the anthrax spores distributed through the US mail in October 2001 a number of methods are being developed to extract and analyze this signature Reaction 1 Stepwise hydrolysis of Sarin to form hydrofluoric acid (HF) isopropyl methylphosphonic acid (IMPA) then methyl phosphonic acid (MPA) and 2-propanol In our previous SERS investigations MPA and DPA were measured at 50 and 100 mgL respectively In both cases limits of detection (LOD) were estimated at 100 microgL providing encouragement in that SERS may satisfy the needs of the JSAWM Since it has been shown that pH can substantially influence the intensity of SER bands25 which would clearly influence quantitative analysis we undertook the present study to determine the severity of these effects for cyanide methyl phosphonic acid and dipicolinic acid Furthermore we previously observed a band at 1050 cm-1 for MPA6 possibly due to an anion formed at basic pH Here we investigate the source of this spectral anomaly
2 EXPERIMENTAL All chemicals including potassium cyanide methyl phosphonic acid dipicolinic acid and those used to prepare the silver-doped and gold-doped sol-gel coated vials were obtained and used as received from Sigma-Aldrich All samples were prepared in HPLC grade water (Fischer Scientific Fair Lawn NJ) for SERS measurements The pH of these samples was adjusted using dilute nitric acid or potassium hydroxide and verified using a pH electrode (Corning Inc Corning NY) that had been calibrated with pH 400 700 and 1000 buffered standards from Fischer Scientific Once prepared the samples were transferred into the silver-doped sol-gel vials (Real-Time Analyzers Inc East Hartford CT) The vials were coated in a manner similar to that previously reported by adding ammonium hydroxide to a solution of silver nitrate tetramethyl orthosilicate (TMOS) and methanol20 Gold-vials were coated by adding nitric acid to a solution of gold tetrachloride TMOS and methanol The two precursor solutions were prepared mixed and transferred to 2-mL glass vials dried and heated After sol-gel formation the incorporated metal ions were reduced with dilute sodium borohydride (1mgmL) which was followed by a water wash to remove residual reducing agent After the resultant analyte solution was introduced the SER-active vial was immediately fixed horizontally to an XY positioning stage (Conix Research Springfield OR) just inside the focal point of an f07 aspheric lens The lens focused the beam into the sample and collected the scattered radiation back along the same axis A dichroic filter (Omega Optical Brattleborough VT) was used to reflect the excitation laser to the lens and pass the Raman scattered radiation collected by the lens An f2 achromat was used to collimate the laser beam exiting a 200 microm core diameter source fiber optic while a second f2 achromat was used to focus the scattered radiation into a 365 microm fiber optic (Spectran Avon CT) A short pass filter was placed in the excitation beam path to block the silicon Raman scattering generated in the source fiber from reflecting off sampling optics and reaching the detector A long pass filter was placed in the collection beam path to block the sample Rayleigh scattering from reaching the detector A 785 nm diode laser (Process Instruments Inc model 785-600 Salt Lake City UT) was used to deliver 100 to 150 mW of power to the sample A Fourier transform Raman spectrometer (Real-Time Analyzers model IRA-785 East Hartford CT) and a silicon photo-avalanche detector (Perkin Elmer model C30902S Stamford CT) were used to acquire the SER spectra
3 RESULTS AND DISCUSSION In a previous study of MPA6 aimed at developing a concentration calibration curve and determining limits of detection (LOD) we observed an anomaly at 1050 cm-1 Since it was found that the band intensity changed as a function of concentration the band must be associated with a sample parameter Two possible parameters photon flux and pH are examined here The first parameter was investigated by irradiating a 1mgmL MPA sample in a SER-active vial with laser powers of 200 mW and above and monitoring spectral changes It was immediately found that the 1050 cm-1 band
2O+ H HF + +OH
OH3H C
OP
OF
CH
CH3
3
3H C
OCP
OOH
CH
CH3
3
3H C
O
CPHO
CH
CH
3
3
C
Sarin IMPA MPA 2-propanol
SPIE -2003-5269 119
appeared and grew as a function of time and that the higher the laser power the faster the growth Figure 1 shows the growth of the 1050 cm-1 band over the course of 30 minutes when using 150 mW of 785 nm excitation while Figure 2A shows that the growth can be fit with a first order exponential equation namely I1050 = 03+05e-013t Furthermore the 760 cm-1 band could be fit with a first order decay equation with an identical exponential rate constant ie I760 = 08-
08e-013t The rates represent classical first order kinetics and their correspondence allows one to conclude that MPA is being transformed one-for-one into a photo-generated product At this time the photoproduct has not been positively identified but phosphonic (phosphorous) acid and phosphonate are likely candidates since the symmetric P(OH)3 stretch occurs at ~1050cm-1 Our studies show that a reduction in laser power to 100 mW at the sample for MPA essentially eliminates this degradation process This laser power or lower was used for further measurements
Methyl phosphonic acid is a diprotic acid that stepwise dissociates into two anions MPA- and MPA= according to the following reactions26
MPA MPA- + H+ pKa1 = 212 Reaction 2
MPA- MPA= + H+ pKa2 = 729 Reaction 3 The relative concentrations of MPA MPA- and MPA= can be determined at any pH by expressing [MPA] and [MPA=] in terms of [MPA-] using Reactions 2 and 3 and summing all three to equal the total starting concentration here 2 mgmL (0021M MW = 9602) viz
[MPA] + [MPA-] + [MPA=] = 0021M Equation 1 substituting from Reactions 1 and 2
([H+][MPA-])K1a + [MPA-] + (K2a[MPA-])[H+] = 0021M Equation 2 rearranging [MPA-] = 0021M(1+[H+]K1a + K2a[H+]) Equation 3 The relative concentrations of MPA MPA- and MPA= as a function of pH are shown in Figure 3 It is worth noting that near neutral pH both MPA- and MPA= will be present To confirm that the SER signal followed this pH dependence a starting solution consisting of 20 mg of MPA in 10 mL HPLC grade water was prepared and brought to pH of 20 using dilute nitric acid From this solution 2 mL were added to a SER-active vial and the SER spectrum recorded At this pH a peak at 760 cm-1 was barely discernable The 2 mL solution was returned to the starting solution and the pH was re-measured to correct for any changes that the silver-doped sol-gel vials might cause In most cases the change was less than 02 pH units and the pH is reported as the before and after average Next the pH of the
Figure 2 A) Exponential growth of 1050 cm-1 band and B) exponential decay of 760 cm-1 band for spectral series in Figure 1
Figure 1 Growth of 1050 cm-1 band as a function of time due to exposure to 150 mW of 785 nm Spectra are 5 sec each collected every 100-sec from 0 to 30-min
B
570 770 970 1170 1370Raman Shift (cm-1)
Arbit
rary
Unit
s
MPA Photodegradation
600 800 1000 1200 1400 Wavenumber (cm-1)
A
0 10 20 30 0 10 20 30 time (min) time (min)
Ram
an In
tens
ity (r
elat
ive)
30
min
0
SPIE -2003-5269 120
starting solution was adjusted to 325 using dilute KOH Again 2 mL were added to a vial and the SER spectrum recorded At this pH a reasonably strong 760 cm-1 band was observed This process was repeated as spectra were recorded at pHs of 70 74 75 79 85 and 100 A total of 1 mL of KOH was added diluting the total concentration by 10 Next the pH of the starting solution was made acidic by adding dilute nitric acid dropwise This time spectra were recorded at pHs of 72 69 64 and 37 Figure 4 shows the SER spectra for representative pHs (spectra were left out to simplify the figure) while Figure 3 shows the 760 cm-1 peak intensities as a function of pH (The band intensities were adjusted to compensate for dilution effects caused by the addition of HNO3 and KOH then normalized to 0021 M for the most intense band observed at pH 37) It is clear from Figure 3 that the 760 cm-1 band follows the MPA- concentration as a function of pH and must be assigned to this anion No bands were observed that corresponded to MPA or MPA= The lack of an MPA SER spectrum may be due to the absence of an attraction between the neutral analyte and the electropositive silver surface The same reasoning suggests that a strong SER spectrum should be observed for MPA= but it is not and a satisfactory explanation has not been found
0000
0005
0010
0015
0020
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14pH
Con
cent
ratio
n [M
]
pK1 = 212 pK2 = 729
MPA- MPA=
MPA
Figure 4 SER spectra of 002M MPA as a function of pH Conditions 100 mW of 785 nm 36 scans (1 min) 8 cm-1 recorded 2 min after sample introduction pH 19 and 20 not apparent on this scale pH 69 and 74 near identical to 70 and 75 and not shown for clarity
00
02
04
06
08
10
12
14
16
18
0 5 10 15 20 25 30Measurement Number
Ram
an In
tens
ity (7
60 c
m-1
)
Figure 3 Concentration dependence of MPA MPA- and MPA= as a function of pH for a 002M sample Intensity of 760 cm-1
band from Figure 6 as a function of pH ( for increasing basic adjustment for increasing acidic adjustment error was measured at ~10 for pH 64)
Figure 5 SER spectra of 002M MPA at pH 64 measured around a vial at three heights (9 points per height) Conditions as in Figure 4 but 10-sec scans
Since these measurements involved the removal and replacement of the SER-active vial in the sample holder to remove and add sample variation in the intensity as a function of vial position was minimized by illuminating the exact same height along the vial wall But this does not account for variability of the SERS response of the sol-gel coating around the vial To analyze this effect a vial containing MPA at pH 64 was rotated at ~ 40o
intervals at the original height and 18rdquo above and below this value Figure 5 shows the intensity of the 760 cm-1
band for the 27 positions It was found that the average value was 137plusmn014 an RSD of 10 overall and 5 for each height An error bar is included in Figure 4 for the pH 64 measurement
SPIE -2003-5269 121
With the above analyses in mind a preliminary investigation of the SER spectral response for MPA (as MPA-) as a function of concentration was performed A single vial was used for these measurements beginning with 1 mgL followed by measurements of 10 100 and 1000 mgL In all cases the pH was ~7 and 3 positions around the vial were measured per concentration Since the 760 cm-1 band was not observed for concentrations of 1 or 10 mgL using 100 mW of 785 nm the laser power at the sample was raised to 200 mW beginning with the 10 mgL concentration Photo-degradation was largely avoided (and not observed) by exposing the sample for only 33 seconds per spectral acquisition Representative spectra for 10 100 and 1000 mgL are shown in Figure 6 while a plot of the 760 cm-1 band intensity as a function of concentration is shown in Figure 7 These values were also used to estimate limits of detection based on the signal-to-noise ratio (SN) of the 760 cm-1 band using 100 mW of 785 nm laser excitation and a 10-min acquisition time As summarized in the Figure 7 inset the lower the measured concentration the lower the predicted LOD Even if as estimated 210 microgL could be measured using the silver-doped sol-gel vials an improvement of a factor of 70 is still required to meet JSAWM goals of measuring 3 microgL in 10-minutes Similar to MPA DPA is a diprotic acid (pKa1 = 216 and pKa2 = 692) and variations in pH will effect the relative concentrations of DPA DPA- and DPA= and possibly the SER spectra and band intensities This could prove significant if an acid or base is used to denature anthrax spores with the goal of extracting and analyzing DPA The MPA pH study described above was mimicked for DPA except that the starting solution consisted of 20 mg of DPA in 20 mL HPLC grade water (60x10-3M MW = 1671) The initial solution had a pH of 245 which was made basic by dilute KOH to pHs of 355 433 487 559 1069 and 1166 SER spectra were recorded at each pH using 100 mW of 785 nm and a 44-sec acquisition time Next one drop of concentrated nitric acid was used to remake the solution acidic at a pH of 200 Again sequential pH measurements were performed at 383 510 735 and 822 The solution pH was made acidic a third time but to pH 219 171 then 135 Throughout this process no more than 20 drops of acid or base were added and therefore the concentration was diluted by no more than 10 Most of the spectral bands showed a minor decrease in intensity as a function of increasing pH values However the bands at 525 and 795 cm-1 showed the most dramatic changes which occurred at acid pH Figure 8 shows the SER spectra of DPA for the spectral region and pH range of interest The identity of the DPA species was determined by plotting the normalized peak intensities with the lowest value set to 0 and the highest to 0006 M as a function of pH and overlaying these values on a plot of the relative concentrations for DPA DPA- and DPA= as previously done for MPA (Figure 9) As can be seen the 525 cm-1 band clearly corresponds to DPA The correspondence of the 795 cm-1 band to this species is less clear as the band retains intensity until pH of 55 This can be attributed to contribution to the overlapping band at 810 cm-1 which does not change as a function of pH The fact that most bands are observed at all pHs suggest that the primary interaction with silver is through the ring nitrogen This is supported by the fact that the most intense band occurs at 1008 cm-1 attributed to a symmetric ring breathing mode and that this interaction has been characterized for pyridine in numerous papers27
0
1
2
3
4
5
6
7
0 200 400 600 800 1000 1200
MPA Concentration (mgL)
760
Ban
d In
tens
ity
Figure 6 SER spectra of MPA in water at A) 1000 B) 100 and C) 10 mgL Conditions pH of 7 silver-doped sol-gel coated vial 200 mW of 785 nm 33-sec 8 cm-1 resolution
Figure 7 Plot of SER intensity of 760 cm-1 band of MPA as a function of concentration using 200 mW of 785 nm Inset table includes average intensity LOD standard deviation and percent deviation for each concentration but for 100 mW and 10 min
conc (mgL) LOD ave stddev dev10 021 002 1021100 072 010 14301000 312 040 1280
A
B C
SPIE -2003-5269 122
The ability to detect dipicolinic acid (DPA) by SERS is demonstrated in Figure 10 Even at 1 mgL the primary bands are visible The intensity of the dominant ring breathing mode observed at 1008 cm-1 for DPA in water (pH = 25 to 55) is plotted as a function of concentration in Figure 11 These values were also used to estimate limits of detection based on the SN of the 1008 cm-1 band using 100 mW of 785 nm laser excitation and a 10-min acquisition time Again the lower the measured concentration the lower the predicted LOD (see Figure 11 inset) and detection of 160 microgL is possible Gastrointestinal anthrax requires significant more spores than inhalation anthrax28 and a limit of detection might be placed at 1 million spores in 1 liter of water or 10 microgL Since each spore contains ~10 CaDPA by weight29 a goal for DPA might be 1 microgL indicating that the present measurements must be improved by nearly two orders of magnitude Due to the increased hazards of handling HCN gas KCN salt was used for these experiments Nevertheless all sample preparations were performed in a chemical hood KCN completely dissolves in water but its conjugate acid HCN is formed and has a Ka of 615x10-1030 viz
HCN CN- + H+ pKa = 921 Reaction 4
Figure 8 SER spectra of 1 mgmL DPA as a function of pH Conditions 100 mW 785 nm 100 scans (44-sec)
Figure 9 Concentration dependence of DPA DPA- and DPA= as a function of pH for a 0006M sample Intensity of 525 () and 795 (diams) cm-1 bands from Fig 8 as a function of pH
pH
135 171 219 383
0
0001
0002
0003
0004
0005
0006
0007
0 2 4 6 8 10 12 14pH
Con
cent
ratio
n [M
]
DPADPA-DPA=795525
DPA DPA=DPA-
pK2 = 692pK1 = 216
795 525
0
05
1
15
2
25
0 200 400 600 800 1000 1200
DPA Concentration (mgL)
1008
Ban
d In
tens
ity
conc (mgL) lod-10min-100mw1 017
10 016100 103
1000 355
Figure 10 SER spectra of DPA in water at A) 1000 B) 100 C) 10 and D) 1 mgL Conditions pH of 25-55 silver-doped sol-gel coated vial 175 mW of 785 nm 1-min 8 cm-1 D) has been multiplied by x10 to make bands visible
A
B C
D
Figure 11 Plot of SER intensity of 1008 cm-1 band of DPA as a function of concentration using 175 mW of 785 nm Inset table includes LOD in mgL for each concentration but for 100 mW and 10 min
SPIE -2003-5269 123
Consequently the cyanide concentration must be determined for each initial KCN concentration Specifically the samples prepared with concentrations of 01 1 10 100 and 1000 mgL of KCN produced CN- concentrations of 63x10-3 033 69 89 and 964 mgL at pHs of 816 90 967 102 and 107 respectively The pH dependence for the HCN and CN- concentrations are shown in Figure 12 Thus as the amount of KCN added to the solution decreases so does the pH of the solution (becomes less basic) and according to Reaction 4 the relative amount of CN- to HCN also decreases For example in the preparation of a 01 mgL solution of KCN the pH is shifted from 7 for pure water to only 816 and only 63 of the starting material becomes CN- or 63x10-3 mgL In comparison for a solution of 1000 mgL the pH is shifted from 7 to 107 and 96 of the starting material becomes CN- This is significant since the cyanide ion is better able to adsorb onto the silver particles and become SERS active SER spectra of 10 100 and 1000 mgL of KCN samples are shown in Figure 13 and a corresponding plot of the 2100 cm-1 band intensity as a function of concentration is shown in Figure 14
000102030405060708091011
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14pH
Con
cent
ratio
n [m
gm
L]
pKa = 921
CN -HCN
The SER spectra of cyanide are dominated by the single CN stretch at 2100 cm-1 which occurs in normal Raman spectra of solutions at 2080 cm-1 However a low frequency mode occurs at 135 cm-1 due to a Ag-CN stretch (not shown) indicative of a strong surface interaction It is also observed that as the concentration decreases the CN stretch shifts to 2140 cm-1 This shift has been attribute to the formation of a tetrahedral Ag(CN)3
2- surface structure31 as well
Figure 14 Concentration dependence of KCN SERS measured under conditions in Fig 11 Concentrations are 1 01 and 001 mgml Intensities are measured for the CN stretch at 2100 cm-1 Inset table includes LOD in mgL for each concentration in Figs 13 and 15 but for 100 mW and 10 min
Figure 13 SER spectra of KCN in water at A) 1000 B) 100 and C) 10 mgL Conditions pHs of 107 102 and 97 silver-doped sol-gel coated vial 100 mW of 785 nm 1-min 8 cm-1 C) has been multiplied by x10 to make band visible
Figure 12 Concentration dependence of HCN and CN- as a function of pH for a 1 mgmL sample Calculated intensity of 2100 () cm-1 band for a 1 mgml sample at pHs of 816 90 967 102 and 107
A
B
C
0
20
40
60
80
100
120
140
0 200 400 600 800 1000 1200CN Concentration (mgL)
2100
Ban
d In
tens
ity
Figure 15 SER spectra of KCN in water at A) 10 B) 1 and C) 01 mgL Conditions pHs of 97 12 and 12 gold-doped sol-gel coated vial 100 mW of 785 nm 1-min 8 cm-1 C) has been multiplied by x10 to make band visible
A
B
C
conc (mgL) lod-10min-100mw Condition01 001 Au-pH 121 007 Au-pH 12
69 003 Au-pH 9769 016 Ag-pH 9789 022 Ag-pH 102946 113 Ag-pH 107
SPIE -2003-5269 124
as to CN adsorbed to two different surface sites32 Alternatively the 2140 cm-1 band could be attributed to HCN since this species dominates at lower concentrations However it is unlikely that this species would be attracted to the electropositive silver surface Further both peaks should be present at pHs between 85 and 105 but this is not observed It has also been suggested that at concentrations near and above monolayer coverage the CN- species is forced to adsorb end-on due to crowding and at lower concentrations the molecule can reorient to lie flat33 This suggests that the 2100 and 2140 cm-1 bands correspond to the end-on and flat orientations respectively As Figures 13 and 14 show the intensity of the CN stretch for the 89 mgL sample is nearly as intense as the 964 mgmL band This suggests that the Raman signal for the flat orientation is more enhanced However more extensive measurements are required to verify this point Since resent research has suggested that cyanide may be more effectively detected on gold measurements of KCN solutions were also performed using gold-doped sol-gel vials Preliminary measurements are shown in Figure 15 for samples prepared from 01 1 and 10 mgL KCN Since the pHs are 816 90 967 the resultant CN- concentrations are 63x10-3 033 and 69 mgL Initially only the highest concentration was observed and the signal intensity was significantly better than the equivalent concentration measured using silver In an effort to shift Reaction 4 to the left transforming HCN to CN- (Le Chatelierrsquos principle) KOH was added to the lower concentration samples producing solutions with pH 12 This effectively forces all of the cyanide in solution to be CN- or 01 and 10 mgL respectively More importantly the CN stretch is now observed in the SER spectra The band appears at 2125 cm-1 as has been previously reported for gold12 As calculated for MPA and DPA LODs can be estimated from this data For the three concentrations of cyanide on silver the LODs are 016 to 11 mgL for 100 mW of 785 nm laser excitation and a 10-min acquisition time For gold pH adjusted the LODs are10 to 70 microgL an improvement of more than 10 times silver Nevertheless either substrate is sufficient to meet the JSAWM goals of measuring 3 mgL in 10-minutes as the requirements form cyanide are much less stringent than the nerve agents
4 CONCLUSIONS Here we examined the surface-enhanced Raman spectral response for methyl phosphonic acid dipicolinic acid and cyanide as a function of pH It was determined that the most prominent peak at 760 cm-1 reaches a maximum intensity between pH 3 and 7 and corresponds to the MPA- species Neither the MPA nor MPA= species appear to generate a SER spectrum and consequently no spectra were observed below pH 2 or above pH 8 In this study we also found that higher laser powers could cause photodegradation of MPA signified by the exponential growth of a band at 1050 cm-1 which is tentatively assigned to phosphorous acid Unlike MPA DPA was observed at all pHs This is attributed to the dominant interaction of the pyridine functional group with silver Minor spectral changes were observed at acid pHs and were assigned to neutral DPA Like MPA SER spectra of cyanide were pH dependent No spectra were observed for the HCN species while CN- was best observed at pHs more basic than 8 Preliminary concentration studies for the three analytes allowed estimating limits of detection for MPA DPA and CN using 100 mW of 785 nm and a 10-min acquisition time of 210 165 and 70 microgL respectively Although the latter value suggests that cyanide can be measured at sufficiently low concentrations to meet JSAWM goal improvements by 100 to 200 times are required for MPA and DPA It is clear from this study that pH of the sample is important and must be taken into account when developing concentration calibration curves as well as fieldable analyzers Future work will include tailoring the sol-gel with specific functional groups to dictate sample pH and thereby optimize sensitivity
ACKNOWLEDGMENTS The authors are grateful for the support of the US Army (DAAD13-02-C-0015 Joint Service Agent Water Monitor program) The authors would also like to thank Dr Steve Christensen of the US Army SBCCOM for helpful discussions and Mr Chetan Shende of Real-Time Analyzers for assistance in development of the gold-doped sol-gels
REFERENCES 1 Erickson B Analytical Chemistry News amp Features June 1 397A (1998)
SPIE -2003-5269 125
2 Johnston RL Hoefler CM Fargo JC and Moberley B AT-ONSITE 5-8 (1994) 3 Black RM RJ Clarke RW Read and MT Reid J Chromatography 662 301-321 (1994) 4 USA Army Center for Health Promotion and Preventive Medicine (USACHPPM) Technical Guideline 230A 5 Jensen JL NC Wong W R Loerop SPIE 4775-03 (2001) 6 Hoenig SL Handbook of Chemical Warfare and Terrorism Greenwood Press Wesport CT (2002) 7 Munro NB SS Talmage GD Griffin LC Waters AP Watson JF King and V Hauschild Env Health
Persp 107 933-974 (1999) 8 Holstege CP Kirk M Sidell FR Crit Care Clin13 923-42 (1997) 9 Farquharson S P Maksymiuk K Ong and S Christesen SPIE 4577 166-173 (2001) 10 Lee Y and S Farquharson SPIE 4378 21-26 (2001) 11 Spencer KM J Sylvia S Clauson and J Janni SPIE 4577 158-165 (2001) 12 Tessier P S Christesen K Ong E Clemente A Lenhoff E Kaler and O Velev Applied Spectroscopy 56
1524-1530 (2002) 13 Farquharson S WW Smith S Elliott and JF Sperry SPIE 3533 207-214 (1998) 14 Farquharson S WW Smith S Elliott and JF Sperry SPIE 3855110-116 (1999) 15 Farquharson S WW Smith YH Lee S Elliott and JF Sperry SPIE 4575 62-72 (2002) 16 Guzelian AA J Sylvia J Janni S Clauson and KM Spencer SPIE 4577 182-192 (2001) 17 Shende C F Inscore A Gift P Maksymiuk and S Farquharson in press 18 Weaver MJ S Farquharson and MA Tadayyoni J Chem Phys 82 4867-4874 (1985) 19 Alak AM and T Vo-Dinh Analytical Chemistry 59 2149-2153 (1987) 20 Lee Y and S Farquharson SPIE 4206 140-146 (2000) 21 Farquharson S and Y Lee SPIE 4200-16 (2000) 22 Lee Y S Farquharson and P M Rainey SPIE 3857 76-84 (1999) 23 Lee Y S Farquharson H Kwong and M Shahriari SPIE 3537 252-260 (1998) 24 Wang J M Pumera G Collins and A Mulchandani Analytical Chemistry 74 6121-6125 (2002) 25 Dou X YM Jung Z-Q Cao and Y Ozaki Applied Spectroscopy 53 1440-1447 (1999) 26 Data supplied by S Christesen and K Ewing 27 Kerker M and B Thompson Eds SPIE MS 10 (1990) 28 Inglesby TV DA Henderson JG Bartlett JAMA 287 2236 (2002) 29 FW Janssen AJ Lund and LE Anderson Science 127 26 (1958) 30 Lide DR Ed Handbook of Chemistry and Physics CRC Press 77th Ed (1996-7) 31 Billmann J G Kovacs and A Otto Surf Sci 92 153 (1980) 32 Murray CA and S Bodoff Phys Rev B 32 671 (1985) 33 Kellogg D and J Pemberton J Phys Chem 91 1120 (1987)
SPIE-2003-5269 16
Chemical agent detection by surface-enhanced Raman spectroscopy
Stuart Farquharson Alan Gift Paul Maksymiuk Frank Inscore and Wayne Smith
Real-Time Analyzers 87 Church Street East Hartford CT 06108
Kevin Morrisey and Steven D Christesen US Army SBCCOM Aberdeen Proving Ground MD 21010
ABSTRACT
In the past decade the Unites States and its allies have been challenged by a different kind of warfare exemplified by the terrorist attacks of September 11 2001 Although suicide bombings are the most often used form of terror military personnel must consider a wide range of attack scenarios Among these is the intentional poisoning of water supplies to obstruct military operations in Afghanistan and Iraq To counter such attacks the military is developing portable analyzers that can identify and quantify potential chemical agents in water supplies at microgram per liter concentrations within 10 minutes To aid this effort we have been investigating the value of a surface-enhanced Raman spectroscopy based portable analyzer In particular we have been developing silver-doped sol-gels to generate SER spectra of chemical agents and their hydrolysis products Here we present SER spectra of several chemical agents measured in a generic tap water Repeat measurements were performed to establish statistical error associated with SERS obtained using the sol-gel coated vials Keywords Chemical agents chemical agent detection SERS Raman spectroscopy
1 INTRODUCTION
In the autumn of 2001 terrorism within US borders became a sobering reality While extensive efforts are being implemented to secure the homeland US and Coalition forces in Afghanistan and Iraq are constantly faced with terrorist attacks In addition to car-bombings the military has identified several non-traditional attack scenarios including poisoning of water supplies by chemical warfare agents (CWAs) To counter this threat the Department of Defense has been investigating numerous approaches to detect such attacks under the auspices of the Joint Service Agent Water Monitor (JSAWM) program1 The JSAWM must be capable of identifying the chemical agents with no false-positives and quantifying the agents at microgram per liter concentrations within 10 minutes2 This includes the analysis of drinking water supplies distribution and storage systems Currently colorimetric paper is used to detect agents on-site while gas chromatography combined with mass spectrometry (GCMS) is used in mobile support laboratories However both methods have severe drawbacks The paper changes color in response to contact with many chemicals besides CWAs causing a high incidence of false positives3 while GCMS although very chemically specific requires up to an hour to perform and regular re-calibration456 Vibrational spectroscopy has also been investigated as a potential method of rapidly detecting CWAs7-
11 as early as 1985 in the case of infrared spectroscopy8 and 1988 in the case of Raman spectroscopy11 Again however these techniques also have limitations when it comes to measuring trace poisons in water Infrared spectra would be dominated by the very strong absorption of water which would obscure absorptions by most other chemicals present Whereas Raman spectroscopy is simply not a very sensitive technique and detection limits are typically grams per liter Surface-enhanced Raman spectroscopy (SERS) offers several advantages over conventional vibrational methods and that may provide the necessary sensitivity required for detecting trace quantities of chemical agents in water SERS employs the interaction of surface plasmon modes of metal particles with target analytes to increase scattering efficiency by as much as 1 million times12 In 1987 the potential of this approach to measure CWAs was demonstrated by measuring a series of pesticides13 Several of these organophosphonates have chemical structures similar to CWAs in particular P=O functional groups In the past few years we and others have further explored the ability of SERS to detect CWAs14-17 and even bioagents 18-21 We have been employing silver-doped sol-gels to promote the SER effect
SPIE-2003-5269 17
in these studies The porous silica network of the sol-gel matrix offers a unique environment for immobilizing and stabilizing SER-active metal particles22-25 The silver-doped sol-gels have been coated on the internal walls of standard glass vials to allow rapid SER analysis of small sample quantities without preparation We have measured over 100 chemicals with enhancements of 104 to 106 demonstrated reversible measurements in a flowing system reproducible measurements from vial-to-vial and measurements in multiple solvents including water21-25 Previously we used these vials to perform preliminary measurements of cyanide (CN) and methylphosphonic acid (MPA) Most of the nerve agents form MPA during hydrolysis while Tabun forms CN a chemical agent in its own right In another paper including in these proceedings we examined the limits of detection (LOD) for MPA by measuring a series of concentrations down to 50 mgL and estimated a limit of detection of 100 microgL26 These measurements provide encouragement in that SERS may satisfy the needs of the JSAWM To further establish the viability of SERS in particular silver-doped sol-gels here we present analysis of cyanide mustard and VX in tap water The measurements performed at the US Armyrsquos Edgewood Chemical Biological Center Aberdeen MD also included numerous repeat measurements to establish reproducibility
2 EXPERIMENTAL 2a General All chemicals including potassium cyanide 2-chloroethylethyl sulfide and those used to prepare the silver-doped and gold-doped sol-gel coated vials were obtained and used as received from Sigma-Aldrich All samples were prepared in a chemical hood using HPLC grade water unless otherwise noted (Fischer Scientific Fair Lawn NJ) for SERS measurements Once prepared the samples were transferred into the silver-doped sol-gel vials (Real-Time Analyzers Inc East Hartford CT) The vials were coated in a manner similar to that previously reported by adding ammonium hydroxide to a solution of silver nitrate tetramethyl orthosilicate (TMOS) and methanol22 The two precursor solutions were prepared mixed and transferred to 2-mL glass vials dried and heated After sol-gel formation the incorporated metal ions were reduced with dilute sodium borohydride (1mgmL) which is followed by a water wash to remove residual reducing agent After the resultant analyte solution was introduced the SER-active vial was fixed horizontally to an XY positioning stage (Conix Research Springfield OR) just inside the focal point of an f07 aspheric lens The lens focused the beam into the sample and collected the scattered radiation back along the same axis A dichroic filter (Omega Optical Brattleborough VT) was used to reflect the excitation laser to the lens and pass the Raman scattered radiation collected by the lens An f2 achromat was used to collimate the laser beam exiting a 200 microm core diameter source fiber optic while a second f2 achromat was used to focus the scattered radiation into a 365 microm fiber optic (Spectran Avon CT) A short pass filter was placed in the excitation beam path to block the silicon Raman scattering generated in the source fiber from reflecting off sampling optics and reaching the detector A long pass filter was placed in the collection beam path to block the sample Rayleigh scattering from reaching the detector A 785 nm diode laser (Process Instruments Inc model 785-600 Salt Lake City UT) was used to deliver 100 to 150 mW of power to the sample A Fourier transform Raman spectrometer (Real-Time Analyzers model IRA-785 East Hartford CT) and a silicon photo-avalanche detector (Perkin Elmer model C30902S Stamford CT) were used to acquire the SER spectra
2b Edgewood Chemical Biological Center The surface-enhanced Raman spectral measurements at the US Armyrsquos Chemical Biological center presented here were all performed on September 12 2003 To expedite measurements a plate was machined to hold up to 12 SER-active sample vials (Figure 1) The plate fit a standard XY plate reader that could be programmed Pure KCN bis-(2-chloroethyl)sulfide (distilled mustard gas HD) and ethyl S-2-diisopropylamino ethyl methylphosphonothioate (VX) were obtained on-site and used to prepare 1 gL tap water solutions in a chemical hood with appropriate safety equipment Simulated tap water was prepared by adding 10 chemicals most often found in tap water at appropriate concentrations to distilled water (Table 1) SER measurements were also performed in a chemical hood For added safety the FT-Raman instrument was placed outside the laboratory and 30 foot fiber optic and electrical cables were used to allow remote SERS measurements and plate manipulation For each experiment 1gL samples were prepared and added to 9 individual vials which were then loaded on the plate In some cases a tenth vial was included as a blank
Table 1 Generic Tap Water Composition
Compound mgL NaHCO3 100 CaSO4 27 MgSO4bull7H2O 67 NaNO3 1 Fulvic Acid 1 K2HPO4 07 KH2PO4 03 (NH4)2HSO4 001 NaCl 001 FeSO4 0001 pH 76-78
SPIE-2003-5269 18
A software program was written that allowed selecting the sequence that the vials were measured the number of positions along the length of the vials to measure (1 to 5) and the number of scans to co-add During sample analysis the program displayed the vial being analyzed the point being analyzed and the spectrum as it was being acquired Once all the data was collected a second software program was written to rapidly analyze the data The spectra collected for all the vials on a plate could be loaded at one time and then the spectra for each point could be displayed simultaneously or separately The user could then select the Raman peak to analyze in terms of peak height or area This was accomplished by selecting points on either side of the peak to define a baseline of zero The peak height or area could then be computed for all of the spectra loaded and then exported to a spreadsheet for statistical analysis
Figure 1 A) Vial Holder 6 slots to hold 2 vials each end-to-end B) Measurement Configuration Program user selects vials to measure sequence number of points per vial (1 to 5) and number of scans per point C) Spectral Acquisition Program shows spectrum being collected which vial and position D) Spectra Analysis Program user selects spectra to analyze by plate vial and point (s) as well as two wavenumbers defining the peak and the baseline to subtract The image is of 5 repeat measurements of 10 mgL KCN in generic tap water 16 sec each 100 mW of 785 nm
3 RESULTS AND DISCUSSION Raman and surface-enhanced Raman spectra were obtained for potassium cyanide bis-(2-chloroethyl)sulfide and ethyl S-2-diisopropylamino ethyl methylphosphonothioate representing three classes of chemical agents cyanides mustards and nerve agents respectively Spectra were also obtained for 2-chloroethyl ethyl sulfide (CEES) a structural analogue to HD which was included in the study to aid in assigning spectral bands KCN salt was used for cyanide experiments to avoid the increased hazards of handling HCN gas KCN completely dissolves in water forming its conjugate acid HCN according to its Ka of 615x10-1027 and at a concentration of 1 mgmL results in a pH 107 solution This is significant in that only CN- appears to interact sufficiently with silver to produce a SER spectrum and no spectral signal is observed below pH 726 Figure 2 shows the SER and normal Raman spectra for KCN The SERS spectrum of 1mgml KCN in water shows a single intense somewhat broad feature at
A B
C D
SPIE-2003-5269 19
2100 cm-1 assigned to the single CequivN stretch The band is much sharper in the normal Raman spectra of the solid KCN salt at 2074 cm-1 This band does broaden and shift to 2080 cm-1 in solution (not shown) However the observed SERS frequency is attributed to interaction with silver and low frequency mode at 135 cm-1 attributed to a Ag-CN stretch (not shown) supports this conclusion
cm-1 band to a C-S stretch but the authors concede that it is in fact more likely a C-Cl stretch28 It appears that the most intense Raman bands at 648 692 and 747 cm-1 shift to 620 660 and 720 cm-1 in the SER spectra and are tentatively assigned as above The width of these bands suggests that they overlap underlying spectral features Additional bands in the Raman spectra occur at 972 1034 1049 1263 1286 1423 1442 2865 2935 and 2960 cm-1 Corresponding bands occur in the SER spectra at 964 1015 1054 1286 1410 1447 2865 and 2935 cm-1 Most of these bands are associated with alkane modes specifically the bands at approximately 1040 cm-1 to a C-C stretch 1290 cm-1 to a CH2 in-phase twist 1440 cm-1 to a CH2 wag 2865 cm-1 to a symmetric CH2 stretch and 2965 cm-1 to an asymmetric CH2 stretch The Raman and SER spectra of sulfur mustard were measured at the Edgewood center (Figure 4) Both spectra are largely similar to CEES The C-Cl and C-S bands in the Raman spectrum of HD now occur at 640 655 700 739 and 760 cm-1 and are more resolved possibly due to the increased molecular symmetry Theoretical calculations indicate that the first three bands are due to C-Cl stretching modes and the latter two to C-S stretching modes28 Only the C-Cl bands maintain significant intensity in the SER spectra occurring at 624 and 643 cm-1 which is attributed to the
Prior to measurements of HD CEES was examined by Raman and SER spectroscopy (Figure 3) CEES also known as half-mustard is essentially identical to HD except one of the chlorine end atoms is replaced by a hydrogen atom Again although not as toxic as HD CEES is a blister agent and dilute aqueous samples were prepared using appropriate safety equipment Both the Raman and SER spectra of CEES are similar and dominated by bands between 600 and 800 cm-1 These are associated with C-Cl and C-S stretching modes which are tentatively assigned to 648 and 747 cm-1 in the Raman spectra respectively The shoulder at 630 cm-1 the overlapped band at 660 cm-1 and the strong band at 692 cm-1 could also be due to these modes or their asymmetric counterparts It is worth noting that theoretical calculations assign the 692
Figure 2 A) SER and B) NR spectra of KCN Conditions A) 1 mgml in tap water 100 mW of 785 nm at sample 1-min acquisition time B) solid 300 mW of 785 nm 5-min All spectra are 8 cm-1 resolution
A
B
Figure 3 A) SER and B) NR spectra of CEES Conditions A) 1 vv (10 mgml) in MeOH 100 mW of 785 nm 1-min acquisition time B) neat 300 mW of 785 nm 5-min
A
B
Cl-CH2-CH2-S-CH2-CH3
A
B
Figure 4 A) SER and B) NR of HD Conditions A) 1mgml in tap water B) pure both 100 mW of 785 nm 1-min
Cl-CH2-CH2-S-CH2-CH2-Cl
SPIE-2003-5269 20
expected strong interaction between chlorine and silver and adds support to the assignment of this band to a C-Cl stretch Weaker overlapping bands occur at 670 692 and 724 cm-1 the latter possibly due to C-S stretching modes Again the alkane modes are apparent in the normal Raman spectra of HD but only a broad feature at 1300 to 1450 cm-1 suggests CH2 contributions in the SER spectrum Although the observed bands in the VX spectrum have not been assigned (Figure 4) a computer generated Raman spectrum29 predicts many of the same features with surprising accuracy and are used here Two intense bands at 460 and 530 cm-1 closely match predicted bands at 463 and 546 cm-1 assigned to a CH3-P=O bend and a PO2CS wag Three highly overlapped bands occur at 694 745 and 771 cm-1 matching predicted bands at 713 730 and 760 cm-1 The first
Table 2 Measured SER peak heights for the CN stretch at 2100 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
Figure 5 A) SERS and B) NR spectra of VX Conditions A) 1 vv (10 mgml) in MeOH B) pure sample both 100 mW of 785 nm 1-min acquisition time
A
B
two have been assigned to a C-S stretch and CH2bend respectively while the latter has been attributed to either a P-C stretch or an O-C-C stretch Although the 745 cm-1 band may alternatively be assigned to a C-S stretch based on the previous measurements of CEES and HD The relatively intense bands at 890 1106 1218 1445 and 1465 cm-1 also match predicted bands at 880 1108 1216 1440 and 1464 cm-1 that are assigned to a C-C stretch CH3 rock N-C3 stretch various C-H3 bends and C-H bends respectively Both the computer generated and the measured spectra contain numerous other less intense bands One is worth mentioning A unique band appears at 370 cm-1 that is predicted at 368 cm-1 and corresponds to an O-P=O bend The surface-enhanced Raman spectrum of VX is also rich with spectral features It has the unique low frequency band at 370 cm-1 as well as a second band at 380 cm-1 that is assigned tothe S-P-O bend predicted in the normal Raman spectrum at 388 cm-1 Based on the measured and predicted normal Raman spectra the following SERS assignments are given 460 cm-1 to the CH3-P=O bend 544 cm-1 to the PO2CS wag 738 cm-1 to a C-S stretch (based on arguments above) 890 cm-1
to a C-C stretch 1101 cm-1 to a CH3 rock and 1456
cm-1 to a C-H bend The ability of SERS to measure chemical agents in water containing real-world chemical interferents was tested by using the generic tap water described in Table 1 The ability to reproduce measurements was accomplished by preparing three separate water stock solutions which were used to prepare three sample solutions each of 1 mgmL KCN in the generic tap water The samples defined as vials 1-9 were then measured at 4 points per vial in a semi-random fashion such that errors associated with stock solution preparation and errors associated with instrument drift could be identified No trends were apparent that signified such systematic errors Each spectrum collected consisted of 20 averaged scans taking 16 seconds at 8 cm-1 resolution The laser power at the sample was measured periodically during the day and it ranged from 102 to 105 mW spacccc
spot Vial 1 Vial 2 Vial 41 13754 13004 128192 13519 12692 12909
1 3 13541 12721 126394 13462 12648 12651
ave 13569 12766 12755Vial 6 Vial 3 Vial 5
1 11586 14068 10757 2 11236 14402 115122 3 11376 14546 11559
4 10894 11763 11214ave 11273 13695 11261
Vial 7 Vial8 Vial 91 11198 15804 11023
2 11243 15716 88753 3 11562 15216 11218
4 11655 15084 11077ave 11415 15455 10548
CN AVG STDEV ERR1 mgmL pk ht 12526 1560 1245
stock solution
SPIE-2003-5269 21
The CN measurements consisting of 36 data points produced an average height of 125 for the 2100 cm-1 peak with a standard deviation of 156 or 125 (Table 2) The HD and VX measurements were performed precisely the same way (Tables 3 and 4) For HD the 624 cm-1 peak was used for analysis and it had an average height of 53 with a standard deviation of 068 or 129 while for VX the 544 cm-1 peak was used for analysis and it had an average height of 1051 with a standard deviation of 308 or 293 The greater error in the VX measurements can be somewhat attributed to Vial 4 which produced lower SER signal intensities But removing this vial from the data set changes the standard deviation to 233 only a modest improvement
spot Vial 1 Vial 2 Vial 41 525 609 4682 45 675 484
1 3 527 644 5094 559 807 669
ave 51525 68375 5325Vial 6 Vial 3 Vial 5
1 521 574 575 2 536 53 4492 3 509 508 451
4 565 594 379ave 53275 5515 4635
Vial 7 Vial 8 Vial 91 457 589 413
2 583 497 4273 3 544 505 4
4 528 5 509ave 528 52275 43725
HD AVG STDEV ERR1 mgmL pk ht 530 068 1291
stock solution
spot Vial 1 Vial 2 Vial 41 1464 1034 4672 1485 99 61
1 3 1491 989 5684 1041 777 553
ave 137025 9475 5495Vial 6 Vial 3 Vial 5
1 1058 942 1293 2 697 121 965
2 3 727 1206 11124 689 1254 746
ave 79275 1153 1029Vial 7 Vial 8 Vial 9
1 1127 783 163 2 1358 812 1614
3 3 1371 908 15124 1097 875 1397
ave 123825 8445 153825
VX AVG STDEV ERR1 mgmL pk ht 1051 308 2925
stock solution
4 CONCLUSIONS In this paper we examined the ability of surface-enhanced Raman spectroscopy to reproducibly measure CN HD and VX in tap water without chemical interference Both normal and surface-enhanced Raman spectra were examined to select unique bands suitable to identify and quantify these chemical agents For SER measurements the 2100 cm-1 C-N stretch was used for CN the 624 cm-1 C-Cl stretch was used for HD and the 544 cm-1 PO2CS wag was used for VX It was determined that 1 mgmL samples of each of these chemicals measured 36 times in glass vials coated with a silver-doped sol-gel reproduced measurements with standard deviations of 125 129 and 293 It was further found that the 10 chemicals added to simulate generic tap water did not interfere with or alter the SER spectra It should be noted that the concentrations used in this study were considerably greater than those required by the JSAWM program Current work involves improving SER sensitivity and designing sampling systems with better reproducibility This includes the development of fractal silver and gold structures within the sol-gel matrix and the development of chemically selective sol-gels
ACKNOWLEDGMENTS The authors are grateful for the support of the US Army (DAAD13-02-C-0015 Joint Service Agent Water Monitor program) The authors would also like to thank Janet Jensen Ronald Crosier and Kristina Gonser for helpful discussions
Table 4 Measured SER peak heights for the VX band at 544 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
Table 3 Measured SER peak heights for the HD band at 624 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
SPIE-2003-5269 22
REFERENCES 1 Jensen JL NC Wong W R Loerop SPIE 4775-03 (2001) 2 JSAWM Requirements at wwwsbccomapgeaarmymilRDAecbcrtPRODSERJSAWMjsawmhtml 3 Erickson B ldquoThe Chemical Weapons Convention Redefines Analytical Challengerdquo Anal Chem 70 397A (1998) 4 Johnston RL Hoefler CM Fargo JC and Moberley B ldquoThe Defense Nuclear Agencyrsquos
ChemicalBiochemical Weapons Agreements Verification Technology Research Development Test and Evaluation Program and its Requirements for On-Site Analysisrdquo AT-ONSITE 5-8 (1994)
5 Black RM Clarke RJ Read RW and Reid MT ldquoApplication of GC-MS and GC-tandem MS to the analysis of chemical warfare samples found to contain residues of the nerve agent sarin sulphur mustard and their degradation productsrdquo J Chrom 662 301-321 (1994)
6 USA Army Center for Health Promotion and Preventive Medicine (USACHPPM) Technical Guideline 230A 7 Hoffland LD Piffath RJ Bouck JB rdquoSpectral signatures of chemical agents and simulantsrdquo Optical
Engineering 24 982-984 (1985) 8 Braue EHJ Pannella MGrdquoCIRCLE CELL FT-IR Analysis of Chemical Warfare Agents in Aqueous Solutionsrdquo
App Spectrosc 44 1513-1520 (1990) 9 Tseng C-H Mann CK and Vickers TJ ldquoDetermination of Organics on Metal Surfaces by Raman
Spectroscopyrdquo App Spectrosc 47 1767-1771 (1993) 10 Staff Reporter ldquoUS Army Tests Lidar to Detect Biological Toxinsrdquo Photonic Spectra pg 50 December 1998 11 Christesen SD Raman cross sections of chemical agents and simulants App Spectrosc 42 318-321 (1988) 12 Weaver MJ Farquharson S Tadayyoni MA ldquoSurface-enhancement factors for Raman scattering at silver electrodesrdquo J Chem Phys 82 4867-4874 (1985) 13 Alak AM and Vo-Dinh T ldquoSERS of Organophosphorous Chemical Agentsrdquo Anal Chem 59 2149-2153 (1987) 14 Lee Y Farquharson S ldquoRapid chemical agent identification by SERSrdquo SPIE 4378 21-26 (2001) 15 Farquharson S Maksymiuk P Ong K Christesen S ldquoChemical agent identification by surface-enhanced Raman spectroscopyrdquo SPIE 4577 166-173 (2001) 16 Spencer KM Sylvia J Clauson S and Janni J ldquoSurface Enhanced Raman as a Water Monitor for Warfare
Agents in Waterrdquo SPIE 4577 158-165 (2001) 17 Tessier P Christesen S Ong K Clemente E Lenhoff A Kaler E Velev O ldquoOn-line spectroscopic characterization of sodium cyanide with nanostructured gold surface-enhanced Raman spectroscopy substratesrdquo
App Spectrosc 56 1524-1530 (2002) 18 Farquharson S W W Smith S Elliott and J F Sperry Rapid biological agent identification by surface- enhanced Raman spectroscopy SPIE 3855110-116 (1999) 19 Farquharson S WW Smith YH Lee S Elliott and J F Sperry Detection of biological signatures A
comparison of electrolytic and metal-doped sol-gel surface-enhanced Raman media SPIE 4575 62-72 (2002) 20 Guzelian AA Sylvia J Janni J Clauson S and Spencer KM ldquoSERS of whole cell bacteria and trace levels of biological moleculesrdquo SPIE 4577 182-192 (2001) 21 Shende C Inscore F Gift A Maksymiuk P Farquharson S ldquoRapid dipicolinic acid extraction from Bacillus spores detected by surface-enhanced Raman spectroscopyrdquo App Spectrosc 58 accepted 22 Lee Y and Farquharson S ldquoSERS Sample Vials Based on Sol-Gel Process for Trace Pesticide Analysisrdquo SPIE
4206 140-146 (2000) 23 Farquharson S and Lee Y ldquoTrace Drug Analysis by SERSrdquo SPIE 4200-16 (2000) 24 Lee Y Farquharson S and Rainey P M Surface-Enhanced Raman Sensor for Trace Chemical Detection in
Water SPIE 3857 76-84 (1999) 25 Lee Y Farquharson S Kwong H and Shahriari M ldquoSol-Gel Chemical Sensor for Surface-Enhanced Raman
Spectroscopyrdquo SPIE 3537 252-260 (1998) 26 Farquharson S A Gift P Maksymiuk F Inscore and W Smith ldquopH dependence of methyl phosphonic acid
dipicolinic acid and cyanide by surface-enhanced Raman spectroscopyrdquo SPIE 5269-19 (2003) 27 Lide DR Ed Handbook of Chemistry and Physics CRC Press 77th Ed (1996-7) 28 Sosa C RJ Bartlett K KuBulat and WB Person ldquoA theoretical study of harmonic vibrational frequencies and
infrared intensities of XCH2CH2SCH2CH2X and XCH2CH2SH (= H Cl)rdquo J Phys Chem 93 577-588 (1993) 29 Hameka HF and JO Jenson ldquoComputer-generated predictions of the structure and the IR and Raman spectra of
VXrdquo ERDEC-TR-065 May 1993
- Chemagents Appendicespdf
-
- SPIE2001-4575-Bioagent-SERSapdf
-
- Detection of bioagent signatures A comparison of electrolytic and metal-doped sol-gel surface-enhanced Raman media
- Stuart Farquharson Wayne Smith and Yuan Lee
-
- Real-Time Analyzers 87 Church Street East Hartford CT 06108
-
- ABSTRACT
- 4 CONCLUSIONS
- 5 ACKNOWLEDGEMENTS
- 6 REFERENCES
-
SPIE -2003-5269 118
a chemical signature of spore forming bacteria such as Bacillus anthracis And in light of the inability to rapidly detect the anthrax spores distributed through the US mail in October 2001 a number of methods are being developed to extract and analyze this signature Reaction 1 Stepwise hydrolysis of Sarin to form hydrofluoric acid (HF) isopropyl methylphosphonic acid (IMPA) then methyl phosphonic acid (MPA) and 2-propanol In our previous SERS investigations MPA and DPA were measured at 50 and 100 mgL respectively In both cases limits of detection (LOD) were estimated at 100 microgL providing encouragement in that SERS may satisfy the needs of the JSAWM Since it has been shown that pH can substantially influence the intensity of SER bands25 which would clearly influence quantitative analysis we undertook the present study to determine the severity of these effects for cyanide methyl phosphonic acid and dipicolinic acid Furthermore we previously observed a band at 1050 cm-1 for MPA6 possibly due to an anion formed at basic pH Here we investigate the source of this spectral anomaly
2 EXPERIMENTAL All chemicals including potassium cyanide methyl phosphonic acid dipicolinic acid and those used to prepare the silver-doped and gold-doped sol-gel coated vials were obtained and used as received from Sigma-Aldrich All samples were prepared in HPLC grade water (Fischer Scientific Fair Lawn NJ) for SERS measurements The pH of these samples was adjusted using dilute nitric acid or potassium hydroxide and verified using a pH electrode (Corning Inc Corning NY) that had been calibrated with pH 400 700 and 1000 buffered standards from Fischer Scientific Once prepared the samples were transferred into the silver-doped sol-gel vials (Real-Time Analyzers Inc East Hartford CT) The vials were coated in a manner similar to that previously reported by adding ammonium hydroxide to a solution of silver nitrate tetramethyl orthosilicate (TMOS) and methanol20 Gold-vials were coated by adding nitric acid to a solution of gold tetrachloride TMOS and methanol The two precursor solutions were prepared mixed and transferred to 2-mL glass vials dried and heated After sol-gel formation the incorporated metal ions were reduced with dilute sodium borohydride (1mgmL) which was followed by a water wash to remove residual reducing agent After the resultant analyte solution was introduced the SER-active vial was immediately fixed horizontally to an XY positioning stage (Conix Research Springfield OR) just inside the focal point of an f07 aspheric lens The lens focused the beam into the sample and collected the scattered radiation back along the same axis A dichroic filter (Omega Optical Brattleborough VT) was used to reflect the excitation laser to the lens and pass the Raman scattered radiation collected by the lens An f2 achromat was used to collimate the laser beam exiting a 200 microm core diameter source fiber optic while a second f2 achromat was used to focus the scattered radiation into a 365 microm fiber optic (Spectran Avon CT) A short pass filter was placed in the excitation beam path to block the silicon Raman scattering generated in the source fiber from reflecting off sampling optics and reaching the detector A long pass filter was placed in the collection beam path to block the sample Rayleigh scattering from reaching the detector A 785 nm diode laser (Process Instruments Inc model 785-600 Salt Lake City UT) was used to deliver 100 to 150 mW of power to the sample A Fourier transform Raman spectrometer (Real-Time Analyzers model IRA-785 East Hartford CT) and a silicon photo-avalanche detector (Perkin Elmer model C30902S Stamford CT) were used to acquire the SER spectra
3 RESULTS AND DISCUSSION In a previous study of MPA6 aimed at developing a concentration calibration curve and determining limits of detection (LOD) we observed an anomaly at 1050 cm-1 Since it was found that the band intensity changed as a function of concentration the band must be associated with a sample parameter Two possible parameters photon flux and pH are examined here The first parameter was investigated by irradiating a 1mgmL MPA sample in a SER-active vial with laser powers of 200 mW and above and monitoring spectral changes It was immediately found that the 1050 cm-1 band
2O+ H HF + +OH
OH3H C
OP
OF
CH
CH3
3
3H C
OCP
OOH
CH
CH3
3
3H C
O
CPHO
CH
CH
3
3
C
Sarin IMPA MPA 2-propanol
SPIE -2003-5269 119
appeared and grew as a function of time and that the higher the laser power the faster the growth Figure 1 shows the growth of the 1050 cm-1 band over the course of 30 minutes when using 150 mW of 785 nm excitation while Figure 2A shows that the growth can be fit with a first order exponential equation namely I1050 = 03+05e-013t Furthermore the 760 cm-1 band could be fit with a first order decay equation with an identical exponential rate constant ie I760 = 08-
08e-013t The rates represent classical first order kinetics and their correspondence allows one to conclude that MPA is being transformed one-for-one into a photo-generated product At this time the photoproduct has not been positively identified but phosphonic (phosphorous) acid and phosphonate are likely candidates since the symmetric P(OH)3 stretch occurs at ~1050cm-1 Our studies show that a reduction in laser power to 100 mW at the sample for MPA essentially eliminates this degradation process This laser power or lower was used for further measurements
Methyl phosphonic acid is a diprotic acid that stepwise dissociates into two anions MPA- and MPA= according to the following reactions26
MPA MPA- + H+ pKa1 = 212 Reaction 2
MPA- MPA= + H+ pKa2 = 729 Reaction 3 The relative concentrations of MPA MPA- and MPA= can be determined at any pH by expressing [MPA] and [MPA=] in terms of [MPA-] using Reactions 2 and 3 and summing all three to equal the total starting concentration here 2 mgmL (0021M MW = 9602) viz
[MPA] + [MPA-] + [MPA=] = 0021M Equation 1 substituting from Reactions 1 and 2
([H+][MPA-])K1a + [MPA-] + (K2a[MPA-])[H+] = 0021M Equation 2 rearranging [MPA-] = 0021M(1+[H+]K1a + K2a[H+]) Equation 3 The relative concentrations of MPA MPA- and MPA= as a function of pH are shown in Figure 3 It is worth noting that near neutral pH both MPA- and MPA= will be present To confirm that the SER signal followed this pH dependence a starting solution consisting of 20 mg of MPA in 10 mL HPLC grade water was prepared and brought to pH of 20 using dilute nitric acid From this solution 2 mL were added to a SER-active vial and the SER spectrum recorded At this pH a peak at 760 cm-1 was barely discernable The 2 mL solution was returned to the starting solution and the pH was re-measured to correct for any changes that the silver-doped sol-gel vials might cause In most cases the change was less than 02 pH units and the pH is reported as the before and after average Next the pH of the
Figure 2 A) Exponential growth of 1050 cm-1 band and B) exponential decay of 760 cm-1 band for spectral series in Figure 1
Figure 1 Growth of 1050 cm-1 band as a function of time due to exposure to 150 mW of 785 nm Spectra are 5 sec each collected every 100-sec from 0 to 30-min
B
570 770 970 1170 1370Raman Shift (cm-1)
Arbit
rary
Unit
s
MPA Photodegradation
600 800 1000 1200 1400 Wavenumber (cm-1)
A
0 10 20 30 0 10 20 30 time (min) time (min)
Ram
an In
tens
ity (r
elat
ive)
30
min
0
SPIE -2003-5269 120
starting solution was adjusted to 325 using dilute KOH Again 2 mL were added to a vial and the SER spectrum recorded At this pH a reasonably strong 760 cm-1 band was observed This process was repeated as spectra were recorded at pHs of 70 74 75 79 85 and 100 A total of 1 mL of KOH was added diluting the total concentration by 10 Next the pH of the starting solution was made acidic by adding dilute nitric acid dropwise This time spectra were recorded at pHs of 72 69 64 and 37 Figure 4 shows the SER spectra for representative pHs (spectra were left out to simplify the figure) while Figure 3 shows the 760 cm-1 peak intensities as a function of pH (The band intensities were adjusted to compensate for dilution effects caused by the addition of HNO3 and KOH then normalized to 0021 M for the most intense band observed at pH 37) It is clear from Figure 3 that the 760 cm-1 band follows the MPA- concentration as a function of pH and must be assigned to this anion No bands were observed that corresponded to MPA or MPA= The lack of an MPA SER spectrum may be due to the absence of an attraction between the neutral analyte and the electropositive silver surface The same reasoning suggests that a strong SER spectrum should be observed for MPA= but it is not and a satisfactory explanation has not been found
0000
0005
0010
0015
0020
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14pH
Con
cent
ratio
n [M
]
pK1 = 212 pK2 = 729
MPA- MPA=
MPA
Figure 4 SER spectra of 002M MPA as a function of pH Conditions 100 mW of 785 nm 36 scans (1 min) 8 cm-1 recorded 2 min after sample introduction pH 19 and 20 not apparent on this scale pH 69 and 74 near identical to 70 and 75 and not shown for clarity
00
02
04
06
08
10
12
14
16
18
0 5 10 15 20 25 30Measurement Number
Ram
an In
tens
ity (7
60 c
m-1
)
Figure 3 Concentration dependence of MPA MPA- and MPA= as a function of pH for a 002M sample Intensity of 760 cm-1
band from Figure 6 as a function of pH ( for increasing basic adjustment for increasing acidic adjustment error was measured at ~10 for pH 64)
Figure 5 SER spectra of 002M MPA at pH 64 measured around a vial at three heights (9 points per height) Conditions as in Figure 4 but 10-sec scans
Since these measurements involved the removal and replacement of the SER-active vial in the sample holder to remove and add sample variation in the intensity as a function of vial position was minimized by illuminating the exact same height along the vial wall But this does not account for variability of the SERS response of the sol-gel coating around the vial To analyze this effect a vial containing MPA at pH 64 was rotated at ~ 40o
intervals at the original height and 18rdquo above and below this value Figure 5 shows the intensity of the 760 cm-1
band for the 27 positions It was found that the average value was 137plusmn014 an RSD of 10 overall and 5 for each height An error bar is included in Figure 4 for the pH 64 measurement
SPIE -2003-5269 121
With the above analyses in mind a preliminary investigation of the SER spectral response for MPA (as MPA-) as a function of concentration was performed A single vial was used for these measurements beginning with 1 mgL followed by measurements of 10 100 and 1000 mgL In all cases the pH was ~7 and 3 positions around the vial were measured per concentration Since the 760 cm-1 band was not observed for concentrations of 1 or 10 mgL using 100 mW of 785 nm the laser power at the sample was raised to 200 mW beginning with the 10 mgL concentration Photo-degradation was largely avoided (and not observed) by exposing the sample for only 33 seconds per spectral acquisition Representative spectra for 10 100 and 1000 mgL are shown in Figure 6 while a plot of the 760 cm-1 band intensity as a function of concentration is shown in Figure 7 These values were also used to estimate limits of detection based on the signal-to-noise ratio (SN) of the 760 cm-1 band using 100 mW of 785 nm laser excitation and a 10-min acquisition time As summarized in the Figure 7 inset the lower the measured concentration the lower the predicted LOD Even if as estimated 210 microgL could be measured using the silver-doped sol-gel vials an improvement of a factor of 70 is still required to meet JSAWM goals of measuring 3 microgL in 10-minutes Similar to MPA DPA is a diprotic acid (pKa1 = 216 and pKa2 = 692) and variations in pH will effect the relative concentrations of DPA DPA- and DPA= and possibly the SER spectra and band intensities This could prove significant if an acid or base is used to denature anthrax spores with the goal of extracting and analyzing DPA The MPA pH study described above was mimicked for DPA except that the starting solution consisted of 20 mg of DPA in 20 mL HPLC grade water (60x10-3M MW = 1671) The initial solution had a pH of 245 which was made basic by dilute KOH to pHs of 355 433 487 559 1069 and 1166 SER spectra were recorded at each pH using 100 mW of 785 nm and a 44-sec acquisition time Next one drop of concentrated nitric acid was used to remake the solution acidic at a pH of 200 Again sequential pH measurements were performed at 383 510 735 and 822 The solution pH was made acidic a third time but to pH 219 171 then 135 Throughout this process no more than 20 drops of acid or base were added and therefore the concentration was diluted by no more than 10 Most of the spectral bands showed a minor decrease in intensity as a function of increasing pH values However the bands at 525 and 795 cm-1 showed the most dramatic changes which occurred at acid pH Figure 8 shows the SER spectra of DPA for the spectral region and pH range of interest The identity of the DPA species was determined by plotting the normalized peak intensities with the lowest value set to 0 and the highest to 0006 M as a function of pH and overlaying these values on a plot of the relative concentrations for DPA DPA- and DPA= as previously done for MPA (Figure 9) As can be seen the 525 cm-1 band clearly corresponds to DPA The correspondence of the 795 cm-1 band to this species is less clear as the band retains intensity until pH of 55 This can be attributed to contribution to the overlapping band at 810 cm-1 which does not change as a function of pH The fact that most bands are observed at all pHs suggest that the primary interaction with silver is through the ring nitrogen This is supported by the fact that the most intense band occurs at 1008 cm-1 attributed to a symmetric ring breathing mode and that this interaction has been characterized for pyridine in numerous papers27
0
1
2
3
4
5
6
7
0 200 400 600 800 1000 1200
MPA Concentration (mgL)
760
Ban
d In
tens
ity
Figure 6 SER spectra of MPA in water at A) 1000 B) 100 and C) 10 mgL Conditions pH of 7 silver-doped sol-gel coated vial 200 mW of 785 nm 33-sec 8 cm-1 resolution
Figure 7 Plot of SER intensity of 760 cm-1 band of MPA as a function of concentration using 200 mW of 785 nm Inset table includes average intensity LOD standard deviation and percent deviation for each concentration but for 100 mW and 10 min
conc (mgL) LOD ave stddev dev10 021 002 1021100 072 010 14301000 312 040 1280
A
B C
SPIE -2003-5269 122
The ability to detect dipicolinic acid (DPA) by SERS is demonstrated in Figure 10 Even at 1 mgL the primary bands are visible The intensity of the dominant ring breathing mode observed at 1008 cm-1 for DPA in water (pH = 25 to 55) is plotted as a function of concentration in Figure 11 These values were also used to estimate limits of detection based on the SN of the 1008 cm-1 band using 100 mW of 785 nm laser excitation and a 10-min acquisition time Again the lower the measured concentration the lower the predicted LOD (see Figure 11 inset) and detection of 160 microgL is possible Gastrointestinal anthrax requires significant more spores than inhalation anthrax28 and a limit of detection might be placed at 1 million spores in 1 liter of water or 10 microgL Since each spore contains ~10 CaDPA by weight29 a goal for DPA might be 1 microgL indicating that the present measurements must be improved by nearly two orders of magnitude Due to the increased hazards of handling HCN gas KCN salt was used for these experiments Nevertheless all sample preparations were performed in a chemical hood KCN completely dissolves in water but its conjugate acid HCN is formed and has a Ka of 615x10-1030 viz
HCN CN- + H+ pKa = 921 Reaction 4
Figure 8 SER spectra of 1 mgmL DPA as a function of pH Conditions 100 mW 785 nm 100 scans (44-sec)
Figure 9 Concentration dependence of DPA DPA- and DPA= as a function of pH for a 0006M sample Intensity of 525 () and 795 (diams) cm-1 bands from Fig 8 as a function of pH
pH
135 171 219 383
0
0001
0002
0003
0004
0005
0006
0007
0 2 4 6 8 10 12 14pH
Con
cent
ratio
n [M
]
DPADPA-DPA=795525
DPA DPA=DPA-
pK2 = 692pK1 = 216
795 525
0
05
1
15
2
25
0 200 400 600 800 1000 1200
DPA Concentration (mgL)
1008
Ban
d In
tens
ity
conc (mgL) lod-10min-100mw1 017
10 016100 103
1000 355
Figure 10 SER spectra of DPA in water at A) 1000 B) 100 C) 10 and D) 1 mgL Conditions pH of 25-55 silver-doped sol-gel coated vial 175 mW of 785 nm 1-min 8 cm-1 D) has been multiplied by x10 to make bands visible
A
B C
D
Figure 11 Plot of SER intensity of 1008 cm-1 band of DPA as a function of concentration using 175 mW of 785 nm Inset table includes LOD in mgL for each concentration but for 100 mW and 10 min
SPIE -2003-5269 123
Consequently the cyanide concentration must be determined for each initial KCN concentration Specifically the samples prepared with concentrations of 01 1 10 100 and 1000 mgL of KCN produced CN- concentrations of 63x10-3 033 69 89 and 964 mgL at pHs of 816 90 967 102 and 107 respectively The pH dependence for the HCN and CN- concentrations are shown in Figure 12 Thus as the amount of KCN added to the solution decreases so does the pH of the solution (becomes less basic) and according to Reaction 4 the relative amount of CN- to HCN also decreases For example in the preparation of a 01 mgL solution of KCN the pH is shifted from 7 for pure water to only 816 and only 63 of the starting material becomes CN- or 63x10-3 mgL In comparison for a solution of 1000 mgL the pH is shifted from 7 to 107 and 96 of the starting material becomes CN- This is significant since the cyanide ion is better able to adsorb onto the silver particles and become SERS active SER spectra of 10 100 and 1000 mgL of KCN samples are shown in Figure 13 and a corresponding plot of the 2100 cm-1 band intensity as a function of concentration is shown in Figure 14
000102030405060708091011
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14pH
Con
cent
ratio
n [m
gm
L]
pKa = 921
CN -HCN
The SER spectra of cyanide are dominated by the single CN stretch at 2100 cm-1 which occurs in normal Raman spectra of solutions at 2080 cm-1 However a low frequency mode occurs at 135 cm-1 due to a Ag-CN stretch (not shown) indicative of a strong surface interaction It is also observed that as the concentration decreases the CN stretch shifts to 2140 cm-1 This shift has been attribute to the formation of a tetrahedral Ag(CN)3
2- surface structure31 as well
Figure 14 Concentration dependence of KCN SERS measured under conditions in Fig 11 Concentrations are 1 01 and 001 mgml Intensities are measured for the CN stretch at 2100 cm-1 Inset table includes LOD in mgL for each concentration in Figs 13 and 15 but for 100 mW and 10 min
Figure 13 SER spectra of KCN in water at A) 1000 B) 100 and C) 10 mgL Conditions pHs of 107 102 and 97 silver-doped sol-gel coated vial 100 mW of 785 nm 1-min 8 cm-1 C) has been multiplied by x10 to make band visible
Figure 12 Concentration dependence of HCN and CN- as a function of pH for a 1 mgmL sample Calculated intensity of 2100 () cm-1 band for a 1 mgml sample at pHs of 816 90 967 102 and 107
A
B
C
0
20
40
60
80
100
120
140
0 200 400 600 800 1000 1200CN Concentration (mgL)
2100
Ban
d In
tens
ity
Figure 15 SER spectra of KCN in water at A) 10 B) 1 and C) 01 mgL Conditions pHs of 97 12 and 12 gold-doped sol-gel coated vial 100 mW of 785 nm 1-min 8 cm-1 C) has been multiplied by x10 to make band visible
A
B
C
conc (mgL) lod-10min-100mw Condition01 001 Au-pH 121 007 Au-pH 12
69 003 Au-pH 9769 016 Ag-pH 9789 022 Ag-pH 102946 113 Ag-pH 107
SPIE -2003-5269 124
as to CN adsorbed to two different surface sites32 Alternatively the 2140 cm-1 band could be attributed to HCN since this species dominates at lower concentrations However it is unlikely that this species would be attracted to the electropositive silver surface Further both peaks should be present at pHs between 85 and 105 but this is not observed It has also been suggested that at concentrations near and above monolayer coverage the CN- species is forced to adsorb end-on due to crowding and at lower concentrations the molecule can reorient to lie flat33 This suggests that the 2100 and 2140 cm-1 bands correspond to the end-on and flat orientations respectively As Figures 13 and 14 show the intensity of the CN stretch for the 89 mgL sample is nearly as intense as the 964 mgmL band This suggests that the Raman signal for the flat orientation is more enhanced However more extensive measurements are required to verify this point Since resent research has suggested that cyanide may be more effectively detected on gold measurements of KCN solutions were also performed using gold-doped sol-gel vials Preliminary measurements are shown in Figure 15 for samples prepared from 01 1 and 10 mgL KCN Since the pHs are 816 90 967 the resultant CN- concentrations are 63x10-3 033 and 69 mgL Initially only the highest concentration was observed and the signal intensity was significantly better than the equivalent concentration measured using silver In an effort to shift Reaction 4 to the left transforming HCN to CN- (Le Chatelierrsquos principle) KOH was added to the lower concentration samples producing solutions with pH 12 This effectively forces all of the cyanide in solution to be CN- or 01 and 10 mgL respectively More importantly the CN stretch is now observed in the SER spectra The band appears at 2125 cm-1 as has been previously reported for gold12 As calculated for MPA and DPA LODs can be estimated from this data For the three concentrations of cyanide on silver the LODs are 016 to 11 mgL for 100 mW of 785 nm laser excitation and a 10-min acquisition time For gold pH adjusted the LODs are10 to 70 microgL an improvement of more than 10 times silver Nevertheless either substrate is sufficient to meet the JSAWM goals of measuring 3 mgL in 10-minutes as the requirements form cyanide are much less stringent than the nerve agents
4 CONCLUSIONS Here we examined the surface-enhanced Raman spectral response for methyl phosphonic acid dipicolinic acid and cyanide as a function of pH It was determined that the most prominent peak at 760 cm-1 reaches a maximum intensity between pH 3 and 7 and corresponds to the MPA- species Neither the MPA nor MPA= species appear to generate a SER spectrum and consequently no spectra were observed below pH 2 or above pH 8 In this study we also found that higher laser powers could cause photodegradation of MPA signified by the exponential growth of a band at 1050 cm-1 which is tentatively assigned to phosphorous acid Unlike MPA DPA was observed at all pHs This is attributed to the dominant interaction of the pyridine functional group with silver Minor spectral changes were observed at acid pHs and were assigned to neutral DPA Like MPA SER spectra of cyanide were pH dependent No spectra were observed for the HCN species while CN- was best observed at pHs more basic than 8 Preliminary concentration studies for the three analytes allowed estimating limits of detection for MPA DPA and CN using 100 mW of 785 nm and a 10-min acquisition time of 210 165 and 70 microgL respectively Although the latter value suggests that cyanide can be measured at sufficiently low concentrations to meet JSAWM goal improvements by 100 to 200 times are required for MPA and DPA It is clear from this study that pH of the sample is important and must be taken into account when developing concentration calibration curves as well as fieldable analyzers Future work will include tailoring the sol-gel with specific functional groups to dictate sample pH and thereby optimize sensitivity
ACKNOWLEDGMENTS The authors are grateful for the support of the US Army (DAAD13-02-C-0015 Joint Service Agent Water Monitor program) The authors would also like to thank Dr Steve Christensen of the US Army SBCCOM for helpful discussions and Mr Chetan Shende of Real-Time Analyzers for assistance in development of the gold-doped sol-gels
REFERENCES 1 Erickson B Analytical Chemistry News amp Features June 1 397A (1998)
SPIE -2003-5269 125
2 Johnston RL Hoefler CM Fargo JC and Moberley B AT-ONSITE 5-8 (1994) 3 Black RM RJ Clarke RW Read and MT Reid J Chromatography 662 301-321 (1994) 4 USA Army Center for Health Promotion and Preventive Medicine (USACHPPM) Technical Guideline 230A 5 Jensen JL NC Wong W R Loerop SPIE 4775-03 (2001) 6 Hoenig SL Handbook of Chemical Warfare and Terrorism Greenwood Press Wesport CT (2002) 7 Munro NB SS Talmage GD Griffin LC Waters AP Watson JF King and V Hauschild Env Health
Persp 107 933-974 (1999) 8 Holstege CP Kirk M Sidell FR Crit Care Clin13 923-42 (1997) 9 Farquharson S P Maksymiuk K Ong and S Christesen SPIE 4577 166-173 (2001) 10 Lee Y and S Farquharson SPIE 4378 21-26 (2001) 11 Spencer KM J Sylvia S Clauson and J Janni SPIE 4577 158-165 (2001) 12 Tessier P S Christesen K Ong E Clemente A Lenhoff E Kaler and O Velev Applied Spectroscopy 56
1524-1530 (2002) 13 Farquharson S WW Smith S Elliott and JF Sperry SPIE 3533 207-214 (1998) 14 Farquharson S WW Smith S Elliott and JF Sperry SPIE 3855110-116 (1999) 15 Farquharson S WW Smith YH Lee S Elliott and JF Sperry SPIE 4575 62-72 (2002) 16 Guzelian AA J Sylvia J Janni S Clauson and KM Spencer SPIE 4577 182-192 (2001) 17 Shende C F Inscore A Gift P Maksymiuk and S Farquharson in press 18 Weaver MJ S Farquharson and MA Tadayyoni J Chem Phys 82 4867-4874 (1985) 19 Alak AM and T Vo-Dinh Analytical Chemistry 59 2149-2153 (1987) 20 Lee Y and S Farquharson SPIE 4206 140-146 (2000) 21 Farquharson S and Y Lee SPIE 4200-16 (2000) 22 Lee Y S Farquharson and P M Rainey SPIE 3857 76-84 (1999) 23 Lee Y S Farquharson H Kwong and M Shahriari SPIE 3537 252-260 (1998) 24 Wang J M Pumera G Collins and A Mulchandani Analytical Chemistry 74 6121-6125 (2002) 25 Dou X YM Jung Z-Q Cao and Y Ozaki Applied Spectroscopy 53 1440-1447 (1999) 26 Data supplied by S Christesen and K Ewing 27 Kerker M and B Thompson Eds SPIE MS 10 (1990) 28 Inglesby TV DA Henderson JG Bartlett JAMA 287 2236 (2002) 29 FW Janssen AJ Lund and LE Anderson Science 127 26 (1958) 30 Lide DR Ed Handbook of Chemistry and Physics CRC Press 77th Ed (1996-7) 31 Billmann J G Kovacs and A Otto Surf Sci 92 153 (1980) 32 Murray CA and S Bodoff Phys Rev B 32 671 (1985) 33 Kellogg D and J Pemberton J Phys Chem 91 1120 (1987)
SPIE-2003-5269 16
Chemical agent detection by surface-enhanced Raman spectroscopy
Stuart Farquharson Alan Gift Paul Maksymiuk Frank Inscore and Wayne Smith
Real-Time Analyzers 87 Church Street East Hartford CT 06108
Kevin Morrisey and Steven D Christesen US Army SBCCOM Aberdeen Proving Ground MD 21010
ABSTRACT
In the past decade the Unites States and its allies have been challenged by a different kind of warfare exemplified by the terrorist attacks of September 11 2001 Although suicide bombings are the most often used form of terror military personnel must consider a wide range of attack scenarios Among these is the intentional poisoning of water supplies to obstruct military operations in Afghanistan and Iraq To counter such attacks the military is developing portable analyzers that can identify and quantify potential chemical agents in water supplies at microgram per liter concentrations within 10 minutes To aid this effort we have been investigating the value of a surface-enhanced Raman spectroscopy based portable analyzer In particular we have been developing silver-doped sol-gels to generate SER spectra of chemical agents and their hydrolysis products Here we present SER spectra of several chemical agents measured in a generic tap water Repeat measurements were performed to establish statistical error associated with SERS obtained using the sol-gel coated vials Keywords Chemical agents chemical agent detection SERS Raman spectroscopy
1 INTRODUCTION
In the autumn of 2001 terrorism within US borders became a sobering reality While extensive efforts are being implemented to secure the homeland US and Coalition forces in Afghanistan and Iraq are constantly faced with terrorist attacks In addition to car-bombings the military has identified several non-traditional attack scenarios including poisoning of water supplies by chemical warfare agents (CWAs) To counter this threat the Department of Defense has been investigating numerous approaches to detect such attacks under the auspices of the Joint Service Agent Water Monitor (JSAWM) program1 The JSAWM must be capable of identifying the chemical agents with no false-positives and quantifying the agents at microgram per liter concentrations within 10 minutes2 This includes the analysis of drinking water supplies distribution and storage systems Currently colorimetric paper is used to detect agents on-site while gas chromatography combined with mass spectrometry (GCMS) is used in mobile support laboratories However both methods have severe drawbacks The paper changes color in response to contact with many chemicals besides CWAs causing a high incidence of false positives3 while GCMS although very chemically specific requires up to an hour to perform and regular re-calibration456 Vibrational spectroscopy has also been investigated as a potential method of rapidly detecting CWAs7-
11 as early as 1985 in the case of infrared spectroscopy8 and 1988 in the case of Raman spectroscopy11 Again however these techniques also have limitations when it comes to measuring trace poisons in water Infrared spectra would be dominated by the very strong absorption of water which would obscure absorptions by most other chemicals present Whereas Raman spectroscopy is simply not a very sensitive technique and detection limits are typically grams per liter Surface-enhanced Raman spectroscopy (SERS) offers several advantages over conventional vibrational methods and that may provide the necessary sensitivity required for detecting trace quantities of chemical agents in water SERS employs the interaction of surface plasmon modes of metal particles with target analytes to increase scattering efficiency by as much as 1 million times12 In 1987 the potential of this approach to measure CWAs was demonstrated by measuring a series of pesticides13 Several of these organophosphonates have chemical structures similar to CWAs in particular P=O functional groups In the past few years we and others have further explored the ability of SERS to detect CWAs14-17 and even bioagents 18-21 We have been employing silver-doped sol-gels to promote the SER effect
SPIE-2003-5269 17
in these studies The porous silica network of the sol-gel matrix offers a unique environment for immobilizing and stabilizing SER-active metal particles22-25 The silver-doped sol-gels have been coated on the internal walls of standard glass vials to allow rapid SER analysis of small sample quantities without preparation We have measured over 100 chemicals with enhancements of 104 to 106 demonstrated reversible measurements in a flowing system reproducible measurements from vial-to-vial and measurements in multiple solvents including water21-25 Previously we used these vials to perform preliminary measurements of cyanide (CN) and methylphosphonic acid (MPA) Most of the nerve agents form MPA during hydrolysis while Tabun forms CN a chemical agent in its own right In another paper including in these proceedings we examined the limits of detection (LOD) for MPA by measuring a series of concentrations down to 50 mgL and estimated a limit of detection of 100 microgL26 These measurements provide encouragement in that SERS may satisfy the needs of the JSAWM To further establish the viability of SERS in particular silver-doped sol-gels here we present analysis of cyanide mustard and VX in tap water The measurements performed at the US Armyrsquos Edgewood Chemical Biological Center Aberdeen MD also included numerous repeat measurements to establish reproducibility
2 EXPERIMENTAL 2a General All chemicals including potassium cyanide 2-chloroethylethyl sulfide and those used to prepare the silver-doped and gold-doped sol-gel coated vials were obtained and used as received from Sigma-Aldrich All samples were prepared in a chemical hood using HPLC grade water unless otherwise noted (Fischer Scientific Fair Lawn NJ) for SERS measurements Once prepared the samples were transferred into the silver-doped sol-gel vials (Real-Time Analyzers Inc East Hartford CT) The vials were coated in a manner similar to that previously reported by adding ammonium hydroxide to a solution of silver nitrate tetramethyl orthosilicate (TMOS) and methanol22 The two precursor solutions were prepared mixed and transferred to 2-mL glass vials dried and heated After sol-gel formation the incorporated metal ions were reduced with dilute sodium borohydride (1mgmL) which is followed by a water wash to remove residual reducing agent After the resultant analyte solution was introduced the SER-active vial was fixed horizontally to an XY positioning stage (Conix Research Springfield OR) just inside the focal point of an f07 aspheric lens The lens focused the beam into the sample and collected the scattered radiation back along the same axis A dichroic filter (Omega Optical Brattleborough VT) was used to reflect the excitation laser to the lens and pass the Raman scattered radiation collected by the lens An f2 achromat was used to collimate the laser beam exiting a 200 microm core diameter source fiber optic while a second f2 achromat was used to focus the scattered radiation into a 365 microm fiber optic (Spectran Avon CT) A short pass filter was placed in the excitation beam path to block the silicon Raman scattering generated in the source fiber from reflecting off sampling optics and reaching the detector A long pass filter was placed in the collection beam path to block the sample Rayleigh scattering from reaching the detector A 785 nm diode laser (Process Instruments Inc model 785-600 Salt Lake City UT) was used to deliver 100 to 150 mW of power to the sample A Fourier transform Raman spectrometer (Real-Time Analyzers model IRA-785 East Hartford CT) and a silicon photo-avalanche detector (Perkin Elmer model C30902S Stamford CT) were used to acquire the SER spectra
2b Edgewood Chemical Biological Center The surface-enhanced Raman spectral measurements at the US Armyrsquos Chemical Biological center presented here were all performed on September 12 2003 To expedite measurements a plate was machined to hold up to 12 SER-active sample vials (Figure 1) The plate fit a standard XY plate reader that could be programmed Pure KCN bis-(2-chloroethyl)sulfide (distilled mustard gas HD) and ethyl S-2-diisopropylamino ethyl methylphosphonothioate (VX) were obtained on-site and used to prepare 1 gL tap water solutions in a chemical hood with appropriate safety equipment Simulated tap water was prepared by adding 10 chemicals most often found in tap water at appropriate concentrations to distilled water (Table 1) SER measurements were also performed in a chemical hood For added safety the FT-Raman instrument was placed outside the laboratory and 30 foot fiber optic and electrical cables were used to allow remote SERS measurements and plate manipulation For each experiment 1gL samples were prepared and added to 9 individual vials which were then loaded on the plate In some cases a tenth vial was included as a blank
Table 1 Generic Tap Water Composition
Compound mgL NaHCO3 100 CaSO4 27 MgSO4bull7H2O 67 NaNO3 1 Fulvic Acid 1 K2HPO4 07 KH2PO4 03 (NH4)2HSO4 001 NaCl 001 FeSO4 0001 pH 76-78
SPIE-2003-5269 18
A software program was written that allowed selecting the sequence that the vials were measured the number of positions along the length of the vials to measure (1 to 5) and the number of scans to co-add During sample analysis the program displayed the vial being analyzed the point being analyzed and the spectrum as it was being acquired Once all the data was collected a second software program was written to rapidly analyze the data The spectra collected for all the vials on a plate could be loaded at one time and then the spectra for each point could be displayed simultaneously or separately The user could then select the Raman peak to analyze in terms of peak height or area This was accomplished by selecting points on either side of the peak to define a baseline of zero The peak height or area could then be computed for all of the spectra loaded and then exported to a spreadsheet for statistical analysis
Figure 1 A) Vial Holder 6 slots to hold 2 vials each end-to-end B) Measurement Configuration Program user selects vials to measure sequence number of points per vial (1 to 5) and number of scans per point C) Spectral Acquisition Program shows spectrum being collected which vial and position D) Spectra Analysis Program user selects spectra to analyze by plate vial and point (s) as well as two wavenumbers defining the peak and the baseline to subtract The image is of 5 repeat measurements of 10 mgL KCN in generic tap water 16 sec each 100 mW of 785 nm
3 RESULTS AND DISCUSSION Raman and surface-enhanced Raman spectra were obtained for potassium cyanide bis-(2-chloroethyl)sulfide and ethyl S-2-diisopropylamino ethyl methylphosphonothioate representing three classes of chemical agents cyanides mustards and nerve agents respectively Spectra were also obtained for 2-chloroethyl ethyl sulfide (CEES) a structural analogue to HD which was included in the study to aid in assigning spectral bands KCN salt was used for cyanide experiments to avoid the increased hazards of handling HCN gas KCN completely dissolves in water forming its conjugate acid HCN according to its Ka of 615x10-1027 and at a concentration of 1 mgmL results in a pH 107 solution This is significant in that only CN- appears to interact sufficiently with silver to produce a SER spectrum and no spectral signal is observed below pH 726 Figure 2 shows the SER and normal Raman spectra for KCN The SERS spectrum of 1mgml KCN in water shows a single intense somewhat broad feature at
A B
C D
SPIE-2003-5269 19
2100 cm-1 assigned to the single CequivN stretch The band is much sharper in the normal Raman spectra of the solid KCN salt at 2074 cm-1 This band does broaden and shift to 2080 cm-1 in solution (not shown) However the observed SERS frequency is attributed to interaction with silver and low frequency mode at 135 cm-1 attributed to a Ag-CN stretch (not shown) supports this conclusion
cm-1 band to a C-S stretch but the authors concede that it is in fact more likely a C-Cl stretch28 It appears that the most intense Raman bands at 648 692 and 747 cm-1 shift to 620 660 and 720 cm-1 in the SER spectra and are tentatively assigned as above The width of these bands suggests that they overlap underlying spectral features Additional bands in the Raman spectra occur at 972 1034 1049 1263 1286 1423 1442 2865 2935 and 2960 cm-1 Corresponding bands occur in the SER spectra at 964 1015 1054 1286 1410 1447 2865 and 2935 cm-1 Most of these bands are associated with alkane modes specifically the bands at approximately 1040 cm-1 to a C-C stretch 1290 cm-1 to a CH2 in-phase twist 1440 cm-1 to a CH2 wag 2865 cm-1 to a symmetric CH2 stretch and 2965 cm-1 to an asymmetric CH2 stretch The Raman and SER spectra of sulfur mustard were measured at the Edgewood center (Figure 4) Both spectra are largely similar to CEES The C-Cl and C-S bands in the Raman spectrum of HD now occur at 640 655 700 739 and 760 cm-1 and are more resolved possibly due to the increased molecular symmetry Theoretical calculations indicate that the first three bands are due to C-Cl stretching modes and the latter two to C-S stretching modes28 Only the C-Cl bands maintain significant intensity in the SER spectra occurring at 624 and 643 cm-1 which is attributed to the
Prior to measurements of HD CEES was examined by Raman and SER spectroscopy (Figure 3) CEES also known as half-mustard is essentially identical to HD except one of the chlorine end atoms is replaced by a hydrogen atom Again although not as toxic as HD CEES is a blister agent and dilute aqueous samples were prepared using appropriate safety equipment Both the Raman and SER spectra of CEES are similar and dominated by bands between 600 and 800 cm-1 These are associated with C-Cl and C-S stretching modes which are tentatively assigned to 648 and 747 cm-1 in the Raman spectra respectively The shoulder at 630 cm-1 the overlapped band at 660 cm-1 and the strong band at 692 cm-1 could also be due to these modes or their asymmetric counterparts It is worth noting that theoretical calculations assign the 692
Figure 2 A) SER and B) NR spectra of KCN Conditions A) 1 mgml in tap water 100 mW of 785 nm at sample 1-min acquisition time B) solid 300 mW of 785 nm 5-min All spectra are 8 cm-1 resolution
A
B
Figure 3 A) SER and B) NR spectra of CEES Conditions A) 1 vv (10 mgml) in MeOH 100 mW of 785 nm 1-min acquisition time B) neat 300 mW of 785 nm 5-min
A
B
Cl-CH2-CH2-S-CH2-CH3
A
B
Figure 4 A) SER and B) NR of HD Conditions A) 1mgml in tap water B) pure both 100 mW of 785 nm 1-min
Cl-CH2-CH2-S-CH2-CH2-Cl
SPIE-2003-5269 20
expected strong interaction between chlorine and silver and adds support to the assignment of this band to a C-Cl stretch Weaker overlapping bands occur at 670 692 and 724 cm-1 the latter possibly due to C-S stretching modes Again the alkane modes are apparent in the normal Raman spectra of HD but only a broad feature at 1300 to 1450 cm-1 suggests CH2 contributions in the SER spectrum Although the observed bands in the VX spectrum have not been assigned (Figure 4) a computer generated Raman spectrum29 predicts many of the same features with surprising accuracy and are used here Two intense bands at 460 and 530 cm-1 closely match predicted bands at 463 and 546 cm-1 assigned to a CH3-P=O bend and a PO2CS wag Three highly overlapped bands occur at 694 745 and 771 cm-1 matching predicted bands at 713 730 and 760 cm-1 The first
Table 2 Measured SER peak heights for the CN stretch at 2100 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
Figure 5 A) SERS and B) NR spectra of VX Conditions A) 1 vv (10 mgml) in MeOH B) pure sample both 100 mW of 785 nm 1-min acquisition time
A
B
two have been assigned to a C-S stretch and CH2bend respectively while the latter has been attributed to either a P-C stretch or an O-C-C stretch Although the 745 cm-1 band may alternatively be assigned to a C-S stretch based on the previous measurements of CEES and HD The relatively intense bands at 890 1106 1218 1445 and 1465 cm-1 also match predicted bands at 880 1108 1216 1440 and 1464 cm-1 that are assigned to a C-C stretch CH3 rock N-C3 stretch various C-H3 bends and C-H bends respectively Both the computer generated and the measured spectra contain numerous other less intense bands One is worth mentioning A unique band appears at 370 cm-1 that is predicted at 368 cm-1 and corresponds to an O-P=O bend The surface-enhanced Raman spectrum of VX is also rich with spectral features It has the unique low frequency band at 370 cm-1 as well as a second band at 380 cm-1 that is assigned tothe S-P-O bend predicted in the normal Raman spectrum at 388 cm-1 Based on the measured and predicted normal Raman spectra the following SERS assignments are given 460 cm-1 to the CH3-P=O bend 544 cm-1 to the PO2CS wag 738 cm-1 to a C-S stretch (based on arguments above) 890 cm-1
to a C-C stretch 1101 cm-1 to a CH3 rock and 1456
cm-1 to a C-H bend The ability of SERS to measure chemical agents in water containing real-world chemical interferents was tested by using the generic tap water described in Table 1 The ability to reproduce measurements was accomplished by preparing three separate water stock solutions which were used to prepare three sample solutions each of 1 mgmL KCN in the generic tap water The samples defined as vials 1-9 were then measured at 4 points per vial in a semi-random fashion such that errors associated with stock solution preparation and errors associated with instrument drift could be identified No trends were apparent that signified such systematic errors Each spectrum collected consisted of 20 averaged scans taking 16 seconds at 8 cm-1 resolution The laser power at the sample was measured periodically during the day and it ranged from 102 to 105 mW spacccc
spot Vial 1 Vial 2 Vial 41 13754 13004 128192 13519 12692 12909
1 3 13541 12721 126394 13462 12648 12651
ave 13569 12766 12755Vial 6 Vial 3 Vial 5
1 11586 14068 10757 2 11236 14402 115122 3 11376 14546 11559
4 10894 11763 11214ave 11273 13695 11261
Vial 7 Vial8 Vial 91 11198 15804 11023
2 11243 15716 88753 3 11562 15216 11218
4 11655 15084 11077ave 11415 15455 10548
CN AVG STDEV ERR1 mgmL pk ht 12526 1560 1245
stock solution
SPIE-2003-5269 21
The CN measurements consisting of 36 data points produced an average height of 125 for the 2100 cm-1 peak with a standard deviation of 156 or 125 (Table 2) The HD and VX measurements were performed precisely the same way (Tables 3 and 4) For HD the 624 cm-1 peak was used for analysis and it had an average height of 53 with a standard deviation of 068 or 129 while for VX the 544 cm-1 peak was used for analysis and it had an average height of 1051 with a standard deviation of 308 or 293 The greater error in the VX measurements can be somewhat attributed to Vial 4 which produced lower SER signal intensities But removing this vial from the data set changes the standard deviation to 233 only a modest improvement
spot Vial 1 Vial 2 Vial 41 525 609 4682 45 675 484
1 3 527 644 5094 559 807 669
ave 51525 68375 5325Vial 6 Vial 3 Vial 5
1 521 574 575 2 536 53 4492 3 509 508 451
4 565 594 379ave 53275 5515 4635
Vial 7 Vial 8 Vial 91 457 589 413
2 583 497 4273 3 544 505 4
4 528 5 509ave 528 52275 43725
HD AVG STDEV ERR1 mgmL pk ht 530 068 1291
stock solution
spot Vial 1 Vial 2 Vial 41 1464 1034 4672 1485 99 61
1 3 1491 989 5684 1041 777 553
ave 137025 9475 5495Vial 6 Vial 3 Vial 5
1 1058 942 1293 2 697 121 965
2 3 727 1206 11124 689 1254 746
ave 79275 1153 1029Vial 7 Vial 8 Vial 9
1 1127 783 163 2 1358 812 1614
3 3 1371 908 15124 1097 875 1397
ave 123825 8445 153825
VX AVG STDEV ERR1 mgmL pk ht 1051 308 2925
stock solution
4 CONCLUSIONS In this paper we examined the ability of surface-enhanced Raman spectroscopy to reproducibly measure CN HD and VX in tap water without chemical interference Both normal and surface-enhanced Raman spectra were examined to select unique bands suitable to identify and quantify these chemical agents For SER measurements the 2100 cm-1 C-N stretch was used for CN the 624 cm-1 C-Cl stretch was used for HD and the 544 cm-1 PO2CS wag was used for VX It was determined that 1 mgmL samples of each of these chemicals measured 36 times in glass vials coated with a silver-doped sol-gel reproduced measurements with standard deviations of 125 129 and 293 It was further found that the 10 chemicals added to simulate generic tap water did not interfere with or alter the SER spectra It should be noted that the concentrations used in this study were considerably greater than those required by the JSAWM program Current work involves improving SER sensitivity and designing sampling systems with better reproducibility This includes the development of fractal silver and gold structures within the sol-gel matrix and the development of chemically selective sol-gels
ACKNOWLEDGMENTS The authors are grateful for the support of the US Army (DAAD13-02-C-0015 Joint Service Agent Water Monitor program) The authors would also like to thank Janet Jensen Ronald Crosier and Kristina Gonser for helpful discussions
Table 4 Measured SER peak heights for the VX band at 544 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
Table 3 Measured SER peak heights for the HD band at 624 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
SPIE-2003-5269 22
REFERENCES 1 Jensen JL NC Wong W R Loerop SPIE 4775-03 (2001) 2 JSAWM Requirements at wwwsbccomapgeaarmymilRDAecbcrtPRODSERJSAWMjsawmhtml 3 Erickson B ldquoThe Chemical Weapons Convention Redefines Analytical Challengerdquo Anal Chem 70 397A (1998) 4 Johnston RL Hoefler CM Fargo JC and Moberley B ldquoThe Defense Nuclear Agencyrsquos
ChemicalBiochemical Weapons Agreements Verification Technology Research Development Test and Evaluation Program and its Requirements for On-Site Analysisrdquo AT-ONSITE 5-8 (1994)
5 Black RM Clarke RJ Read RW and Reid MT ldquoApplication of GC-MS and GC-tandem MS to the analysis of chemical warfare samples found to contain residues of the nerve agent sarin sulphur mustard and their degradation productsrdquo J Chrom 662 301-321 (1994)
6 USA Army Center for Health Promotion and Preventive Medicine (USACHPPM) Technical Guideline 230A 7 Hoffland LD Piffath RJ Bouck JB rdquoSpectral signatures of chemical agents and simulantsrdquo Optical
Engineering 24 982-984 (1985) 8 Braue EHJ Pannella MGrdquoCIRCLE CELL FT-IR Analysis of Chemical Warfare Agents in Aqueous Solutionsrdquo
App Spectrosc 44 1513-1520 (1990) 9 Tseng C-H Mann CK and Vickers TJ ldquoDetermination of Organics on Metal Surfaces by Raman
Spectroscopyrdquo App Spectrosc 47 1767-1771 (1993) 10 Staff Reporter ldquoUS Army Tests Lidar to Detect Biological Toxinsrdquo Photonic Spectra pg 50 December 1998 11 Christesen SD Raman cross sections of chemical agents and simulants App Spectrosc 42 318-321 (1988) 12 Weaver MJ Farquharson S Tadayyoni MA ldquoSurface-enhancement factors for Raman scattering at silver electrodesrdquo J Chem Phys 82 4867-4874 (1985) 13 Alak AM and Vo-Dinh T ldquoSERS of Organophosphorous Chemical Agentsrdquo Anal Chem 59 2149-2153 (1987) 14 Lee Y Farquharson S ldquoRapid chemical agent identification by SERSrdquo SPIE 4378 21-26 (2001) 15 Farquharson S Maksymiuk P Ong K Christesen S ldquoChemical agent identification by surface-enhanced Raman spectroscopyrdquo SPIE 4577 166-173 (2001) 16 Spencer KM Sylvia J Clauson S and Janni J ldquoSurface Enhanced Raman as a Water Monitor for Warfare
Agents in Waterrdquo SPIE 4577 158-165 (2001) 17 Tessier P Christesen S Ong K Clemente E Lenhoff A Kaler E Velev O ldquoOn-line spectroscopic characterization of sodium cyanide with nanostructured gold surface-enhanced Raman spectroscopy substratesrdquo
App Spectrosc 56 1524-1530 (2002) 18 Farquharson S W W Smith S Elliott and J F Sperry Rapid biological agent identification by surface- enhanced Raman spectroscopy SPIE 3855110-116 (1999) 19 Farquharson S WW Smith YH Lee S Elliott and J F Sperry Detection of biological signatures A
comparison of electrolytic and metal-doped sol-gel surface-enhanced Raman media SPIE 4575 62-72 (2002) 20 Guzelian AA Sylvia J Janni J Clauson S and Spencer KM ldquoSERS of whole cell bacteria and trace levels of biological moleculesrdquo SPIE 4577 182-192 (2001) 21 Shende C Inscore F Gift A Maksymiuk P Farquharson S ldquoRapid dipicolinic acid extraction from Bacillus spores detected by surface-enhanced Raman spectroscopyrdquo App Spectrosc 58 accepted 22 Lee Y and Farquharson S ldquoSERS Sample Vials Based on Sol-Gel Process for Trace Pesticide Analysisrdquo SPIE
4206 140-146 (2000) 23 Farquharson S and Lee Y ldquoTrace Drug Analysis by SERSrdquo SPIE 4200-16 (2000) 24 Lee Y Farquharson S and Rainey P M Surface-Enhanced Raman Sensor for Trace Chemical Detection in
Water SPIE 3857 76-84 (1999) 25 Lee Y Farquharson S Kwong H and Shahriari M ldquoSol-Gel Chemical Sensor for Surface-Enhanced Raman
Spectroscopyrdquo SPIE 3537 252-260 (1998) 26 Farquharson S A Gift P Maksymiuk F Inscore and W Smith ldquopH dependence of methyl phosphonic acid
dipicolinic acid and cyanide by surface-enhanced Raman spectroscopyrdquo SPIE 5269-19 (2003) 27 Lide DR Ed Handbook of Chemistry and Physics CRC Press 77th Ed (1996-7) 28 Sosa C RJ Bartlett K KuBulat and WB Person ldquoA theoretical study of harmonic vibrational frequencies and
infrared intensities of XCH2CH2SCH2CH2X and XCH2CH2SH (= H Cl)rdquo J Phys Chem 93 577-588 (1993) 29 Hameka HF and JO Jenson ldquoComputer-generated predictions of the structure and the IR and Raman spectra of
VXrdquo ERDEC-TR-065 May 1993
- Chemagents Appendicespdf
-
- SPIE2001-4575-Bioagent-SERSapdf
-
- Detection of bioagent signatures A comparison of electrolytic and metal-doped sol-gel surface-enhanced Raman media
- Stuart Farquharson Wayne Smith and Yuan Lee
-
- Real-Time Analyzers 87 Church Street East Hartford CT 06108
-
- ABSTRACT
- 4 CONCLUSIONS
- 5 ACKNOWLEDGEMENTS
- 6 REFERENCES
-
SPIE -2003-5269 119
appeared and grew as a function of time and that the higher the laser power the faster the growth Figure 1 shows the growth of the 1050 cm-1 band over the course of 30 minutes when using 150 mW of 785 nm excitation while Figure 2A shows that the growth can be fit with a first order exponential equation namely I1050 = 03+05e-013t Furthermore the 760 cm-1 band could be fit with a first order decay equation with an identical exponential rate constant ie I760 = 08-
08e-013t The rates represent classical first order kinetics and their correspondence allows one to conclude that MPA is being transformed one-for-one into a photo-generated product At this time the photoproduct has not been positively identified but phosphonic (phosphorous) acid and phosphonate are likely candidates since the symmetric P(OH)3 stretch occurs at ~1050cm-1 Our studies show that a reduction in laser power to 100 mW at the sample for MPA essentially eliminates this degradation process This laser power or lower was used for further measurements
Methyl phosphonic acid is a diprotic acid that stepwise dissociates into two anions MPA- and MPA= according to the following reactions26
MPA MPA- + H+ pKa1 = 212 Reaction 2
MPA- MPA= + H+ pKa2 = 729 Reaction 3 The relative concentrations of MPA MPA- and MPA= can be determined at any pH by expressing [MPA] and [MPA=] in terms of [MPA-] using Reactions 2 and 3 and summing all three to equal the total starting concentration here 2 mgmL (0021M MW = 9602) viz
[MPA] + [MPA-] + [MPA=] = 0021M Equation 1 substituting from Reactions 1 and 2
([H+][MPA-])K1a + [MPA-] + (K2a[MPA-])[H+] = 0021M Equation 2 rearranging [MPA-] = 0021M(1+[H+]K1a + K2a[H+]) Equation 3 The relative concentrations of MPA MPA- and MPA= as a function of pH are shown in Figure 3 It is worth noting that near neutral pH both MPA- and MPA= will be present To confirm that the SER signal followed this pH dependence a starting solution consisting of 20 mg of MPA in 10 mL HPLC grade water was prepared and brought to pH of 20 using dilute nitric acid From this solution 2 mL were added to a SER-active vial and the SER spectrum recorded At this pH a peak at 760 cm-1 was barely discernable The 2 mL solution was returned to the starting solution and the pH was re-measured to correct for any changes that the silver-doped sol-gel vials might cause In most cases the change was less than 02 pH units and the pH is reported as the before and after average Next the pH of the
Figure 2 A) Exponential growth of 1050 cm-1 band and B) exponential decay of 760 cm-1 band for spectral series in Figure 1
Figure 1 Growth of 1050 cm-1 band as a function of time due to exposure to 150 mW of 785 nm Spectra are 5 sec each collected every 100-sec from 0 to 30-min
B
570 770 970 1170 1370Raman Shift (cm-1)
Arbit
rary
Unit
s
MPA Photodegradation
600 800 1000 1200 1400 Wavenumber (cm-1)
A
0 10 20 30 0 10 20 30 time (min) time (min)
Ram
an In
tens
ity (r
elat
ive)
30
min
0
SPIE -2003-5269 120
starting solution was adjusted to 325 using dilute KOH Again 2 mL were added to a vial and the SER spectrum recorded At this pH a reasonably strong 760 cm-1 band was observed This process was repeated as spectra were recorded at pHs of 70 74 75 79 85 and 100 A total of 1 mL of KOH was added diluting the total concentration by 10 Next the pH of the starting solution was made acidic by adding dilute nitric acid dropwise This time spectra were recorded at pHs of 72 69 64 and 37 Figure 4 shows the SER spectra for representative pHs (spectra were left out to simplify the figure) while Figure 3 shows the 760 cm-1 peak intensities as a function of pH (The band intensities were adjusted to compensate for dilution effects caused by the addition of HNO3 and KOH then normalized to 0021 M for the most intense band observed at pH 37) It is clear from Figure 3 that the 760 cm-1 band follows the MPA- concentration as a function of pH and must be assigned to this anion No bands were observed that corresponded to MPA or MPA= The lack of an MPA SER spectrum may be due to the absence of an attraction between the neutral analyte and the electropositive silver surface The same reasoning suggests that a strong SER spectrum should be observed for MPA= but it is not and a satisfactory explanation has not been found
0000
0005
0010
0015
0020
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14pH
Con
cent
ratio
n [M
]
pK1 = 212 pK2 = 729
MPA- MPA=
MPA
Figure 4 SER spectra of 002M MPA as a function of pH Conditions 100 mW of 785 nm 36 scans (1 min) 8 cm-1 recorded 2 min after sample introduction pH 19 and 20 not apparent on this scale pH 69 and 74 near identical to 70 and 75 and not shown for clarity
00
02
04
06
08
10
12
14
16
18
0 5 10 15 20 25 30Measurement Number
Ram
an In
tens
ity (7
60 c
m-1
)
Figure 3 Concentration dependence of MPA MPA- and MPA= as a function of pH for a 002M sample Intensity of 760 cm-1
band from Figure 6 as a function of pH ( for increasing basic adjustment for increasing acidic adjustment error was measured at ~10 for pH 64)
Figure 5 SER spectra of 002M MPA at pH 64 measured around a vial at three heights (9 points per height) Conditions as in Figure 4 but 10-sec scans
Since these measurements involved the removal and replacement of the SER-active vial in the sample holder to remove and add sample variation in the intensity as a function of vial position was minimized by illuminating the exact same height along the vial wall But this does not account for variability of the SERS response of the sol-gel coating around the vial To analyze this effect a vial containing MPA at pH 64 was rotated at ~ 40o
intervals at the original height and 18rdquo above and below this value Figure 5 shows the intensity of the 760 cm-1
band for the 27 positions It was found that the average value was 137plusmn014 an RSD of 10 overall and 5 for each height An error bar is included in Figure 4 for the pH 64 measurement
SPIE -2003-5269 121
With the above analyses in mind a preliminary investigation of the SER spectral response for MPA (as MPA-) as a function of concentration was performed A single vial was used for these measurements beginning with 1 mgL followed by measurements of 10 100 and 1000 mgL In all cases the pH was ~7 and 3 positions around the vial were measured per concentration Since the 760 cm-1 band was not observed for concentrations of 1 or 10 mgL using 100 mW of 785 nm the laser power at the sample was raised to 200 mW beginning with the 10 mgL concentration Photo-degradation was largely avoided (and not observed) by exposing the sample for only 33 seconds per spectral acquisition Representative spectra for 10 100 and 1000 mgL are shown in Figure 6 while a plot of the 760 cm-1 band intensity as a function of concentration is shown in Figure 7 These values were also used to estimate limits of detection based on the signal-to-noise ratio (SN) of the 760 cm-1 band using 100 mW of 785 nm laser excitation and a 10-min acquisition time As summarized in the Figure 7 inset the lower the measured concentration the lower the predicted LOD Even if as estimated 210 microgL could be measured using the silver-doped sol-gel vials an improvement of a factor of 70 is still required to meet JSAWM goals of measuring 3 microgL in 10-minutes Similar to MPA DPA is a diprotic acid (pKa1 = 216 and pKa2 = 692) and variations in pH will effect the relative concentrations of DPA DPA- and DPA= and possibly the SER spectra and band intensities This could prove significant if an acid or base is used to denature anthrax spores with the goal of extracting and analyzing DPA The MPA pH study described above was mimicked for DPA except that the starting solution consisted of 20 mg of DPA in 20 mL HPLC grade water (60x10-3M MW = 1671) The initial solution had a pH of 245 which was made basic by dilute KOH to pHs of 355 433 487 559 1069 and 1166 SER spectra were recorded at each pH using 100 mW of 785 nm and a 44-sec acquisition time Next one drop of concentrated nitric acid was used to remake the solution acidic at a pH of 200 Again sequential pH measurements were performed at 383 510 735 and 822 The solution pH was made acidic a third time but to pH 219 171 then 135 Throughout this process no more than 20 drops of acid or base were added and therefore the concentration was diluted by no more than 10 Most of the spectral bands showed a minor decrease in intensity as a function of increasing pH values However the bands at 525 and 795 cm-1 showed the most dramatic changes which occurred at acid pH Figure 8 shows the SER spectra of DPA for the spectral region and pH range of interest The identity of the DPA species was determined by plotting the normalized peak intensities with the lowest value set to 0 and the highest to 0006 M as a function of pH and overlaying these values on a plot of the relative concentrations for DPA DPA- and DPA= as previously done for MPA (Figure 9) As can be seen the 525 cm-1 band clearly corresponds to DPA The correspondence of the 795 cm-1 band to this species is less clear as the band retains intensity until pH of 55 This can be attributed to contribution to the overlapping band at 810 cm-1 which does not change as a function of pH The fact that most bands are observed at all pHs suggest that the primary interaction with silver is through the ring nitrogen This is supported by the fact that the most intense band occurs at 1008 cm-1 attributed to a symmetric ring breathing mode and that this interaction has been characterized for pyridine in numerous papers27
0
1
2
3
4
5
6
7
0 200 400 600 800 1000 1200
MPA Concentration (mgL)
760
Ban
d In
tens
ity
Figure 6 SER spectra of MPA in water at A) 1000 B) 100 and C) 10 mgL Conditions pH of 7 silver-doped sol-gel coated vial 200 mW of 785 nm 33-sec 8 cm-1 resolution
Figure 7 Plot of SER intensity of 760 cm-1 band of MPA as a function of concentration using 200 mW of 785 nm Inset table includes average intensity LOD standard deviation and percent deviation for each concentration but for 100 mW and 10 min
conc (mgL) LOD ave stddev dev10 021 002 1021100 072 010 14301000 312 040 1280
A
B C
SPIE -2003-5269 122
The ability to detect dipicolinic acid (DPA) by SERS is demonstrated in Figure 10 Even at 1 mgL the primary bands are visible The intensity of the dominant ring breathing mode observed at 1008 cm-1 for DPA in water (pH = 25 to 55) is plotted as a function of concentration in Figure 11 These values were also used to estimate limits of detection based on the SN of the 1008 cm-1 band using 100 mW of 785 nm laser excitation and a 10-min acquisition time Again the lower the measured concentration the lower the predicted LOD (see Figure 11 inset) and detection of 160 microgL is possible Gastrointestinal anthrax requires significant more spores than inhalation anthrax28 and a limit of detection might be placed at 1 million spores in 1 liter of water or 10 microgL Since each spore contains ~10 CaDPA by weight29 a goal for DPA might be 1 microgL indicating that the present measurements must be improved by nearly two orders of magnitude Due to the increased hazards of handling HCN gas KCN salt was used for these experiments Nevertheless all sample preparations were performed in a chemical hood KCN completely dissolves in water but its conjugate acid HCN is formed and has a Ka of 615x10-1030 viz
HCN CN- + H+ pKa = 921 Reaction 4
Figure 8 SER spectra of 1 mgmL DPA as a function of pH Conditions 100 mW 785 nm 100 scans (44-sec)
Figure 9 Concentration dependence of DPA DPA- and DPA= as a function of pH for a 0006M sample Intensity of 525 () and 795 (diams) cm-1 bands from Fig 8 as a function of pH
pH
135 171 219 383
0
0001
0002
0003
0004
0005
0006
0007
0 2 4 6 8 10 12 14pH
Con
cent
ratio
n [M
]
DPADPA-DPA=795525
DPA DPA=DPA-
pK2 = 692pK1 = 216
795 525
0
05
1
15
2
25
0 200 400 600 800 1000 1200
DPA Concentration (mgL)
1008
Ban
d In
tens
ity
conc (mgL) lod-10min-100mw1 017
10 016100 103
1000 355
Figure 10 SER spectra of DPA in water at A) 1000 B) 100 C) 10 and D) 1 mgL Conditions pH of 25-55 silver-doped sol-gel coated vial 175 mW of 785 nm 1-min 8 cm-1 D) has been multiplied by x10 to make bands visible
A
B C
D
Figure 11 Plot of SER intensity of 1008 cm-1 band of DPA as a function of concentration using 175 mW of 785 nm Inset table includes LOD in mgL for each concentration but for 100 mW and 10 min
SPIE -2003-5269 123
Consequently the cyanide concentration must be determined for each initial KCN concentration Specifically the samples prepared with concentrations of 01 1 10 100 and 1000 mgL of KCN produced CN- concentrations of 63x10-3 033 69 89 and 964 mgL at pHs of 816 90 967 102 and 107 respectively The pH dependence for the HCN and CN- concentrations are shown in Figure 12 Thus as the amount of KCN added to the solution decreases so does the pH of the solution (becomes less basic) and according to Reaction 4 the relative amount of CN- to HCN also decreases For example in the preparation of a 01 mgL solution of KCN the pH is shifted from 7 for pure water to only 816 and only 63 of the starting material becomes CN- or 63x10-3 mgL In comparison for a solution of 1000 mgL the pH is shifted from 7 to 107 and 96 of the starting material becomes CN- This is significant since the cyanide ion is better able to adsorb onto the silver particles and become SERS active SER spectra of 10 100 and 1000 mgL of KCN samples are shown in Figure 13 and a corresponding plot of the 2100 cm-1 band intensity as a function of concentration is shown in Figure 14
000102030405060708091011
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14pH
Con
cent
ratio
n [m
gm
L]
pKa = 921
CN -HCN
The SER spectra of cyanide are dominated by the single CN stretch at 2100 cm-1 which occurs in normal Raman spectra of solutions at 2080 cm-1 However a low frequency mode occurs at 135 cm-1 due to a Ag-CN stretch (not shown) indicative of a strong surface interaction It is also observed that as the concentration decreases the CN stretch shifts to 2140 cm-1 This shift has been attribute to the formation of a tetrahedral Ag(CN)3
2- surface structure31 as well
Figure 14 Concentration dependence of KCN SERS measured under conditions in Fig 11 Concentrations are 1 01 and 001 mgml Intensities are measured for the CN stretch at 2100 cm-1 Inset table includes LOD in mgL for each concentration in Figs 13 and 15 but for 100 mW and 10 min
Figure 13 SER spectra of KCN in water at A) 1000 B) 100 and C) 10 mgL Conditions pHs of 107 102 and 97 silver-doped sol-gel coated vial 100 mW of 785 nm 1-min 8 cm-1 C) has been multiplied by x10 to make band visible
Figure 12 Concentration dependence of HCN and CN- as a function of pH for a 1 mgmL sample Calculated intensity of 2100 () cm-1 band for a 1 mgml sample at pHs of 816 90 967 102 and 107
A
B
C
0
20
40
60
80
100
120
140
0 200 400 600 800 1000 1200CN Concentration (mgL)
2100
Ban
d In
tens
ity
Figure 15 SER spectra of KCN in water at A) 10 B) 1 and C) 01 mgL Conditions pHs of 97 12 and 12 gold-doped sol-gel coated vial 100 mW of 785 nm 1-min 8 cm-1 C) has been multiplied by x10 to make band visible
A
B
C
conc (mgL) lod-10min-100mw Condition01 001 Au-pH 121 007 Au-pH 12
69 003 Au-pH 9769 016 Ag-pH 9789 022 Ag-pH 102946 113 Ag-pH 107
SPIE -2003-5269 124
as to CN adsorbed to two different surface sites32 Alternatively the 2140 cm-1 band could be attributed to HCN since this species dominates at lower concentrations However it is unlikely that this species would be attracted to the electropositive silver surface Further both peaks should be present at pHs between 85 and 105 but this is not observed It has also been suggested that at concentrations near and above monolayer coverage the CN- species is forced to adsorb end-on due to crowding and at lower concentrations the molecule can reorient to lie flat33 This suggests that the 2100 and 2140 cm-1 bands correspond to the end-on and flat orientations respectively As Figures 13 and 14 show the intensity of the CN stretch for the 89 mgL sample is nearly as intense as the 964 mgmL band This suggests that the Raman signal for the flat orientation is more enhanced However more extensive measurements are required to verify this point Since resent research has suggested that cyanide may be more effectively detected on gold measurements of KCN solutions were also performed using gold-doped sol-gel vials Preliminary measurements are shown in Figure 15 for samples prepared from 01 1 and 10 mgL KCN Since the pHs are 816 90 967 the resultant CN- concentrations are 63x10-3 033 and 69 mgL Initially only the highest concentration was observed and the signal intensity was significantly better than the equivalent concentration measured using silver In an effort to shift Reaction 4 to the left transforming HCN to CN- (Le Chatelierrsquos principle) KOH was added to the lower concentration samples producing solutions with pH 12 This effectively forces all of the cyanide in solution to be CN- or 01 and 10 mgL respectively More importantly the CN stretch is now observed in the SER spectra The band appears at 2125 cm-1 as has been previously reported for gold12 As calculated for MPA and DPA LODs can be estimated from this data For the three concentrations of cyanide on silver the LODs are 016 to 11 mgL for 100 mW of 785 nm laser excitation and a 10-min acquisition time For gold pH adjusted the LODs are10 to 70 microgL an improvement of more than 10 times silver Nevertheless either substrate is sufficient to meet the JSAWM goals of measuring 3 mgL in 10-minutes as the requirements form cyanide are much less stringent than the nerve agents
4 CONCLUSIONS Here we examined the surface-enhanced Raman spectral response for methyl phosphonic acid dipicolinic acid and cyanide as a function of pH It was determined that the most prominent peak at 760 cm-1 reaches a maximum intensity between pH 3 and 7 and corresponds to the MPA- species Neither the MPA nor MPA= species appear to generate a SER spectrum and consequently no spectra were observed below pH 2 or above pH 8 In this study we also found that higher laser powers could cause photodegradation of MPA signified by the exponential growth of a band at 1050 cm-1 which is tentatively assigned to phosphorous acid Unlike MPA DPA was observed at all pHs This is attributed to the dominant interaction of the pyridine functional group with silver Minor spectral changes were observed at acid pHs and were assigned to neutral DPA Like MPA SER spectra of cyanide were pH dependent No spectra were observed for the HCN species while CN- was best observed at pHs more basic than 8 Preliminary concentration studies for the three analytes allowed estimating limits of detection for MPA DPA and CN using 100 mW of 785 nm and a 10-min acquisition time of 210 165 and 70 microgL respectively Although the latter value suggests that cyanide can be measured at sufficiently low concentrations to meet JSAWM goal improvements by 100 to 200 times are required for MPA and DPA It is clear from this study that pH of the sample is important and must be taken into account when developing concentration calibration curves as well as fieldable analyzers Future work will include tailoring the sol-gel with specific functional groups to dictate sample pH and thereby optimize sensitivity
ACKNOWLEDGMENTS The authors are grateful for the support of the US Army (DAAD13-02-C-0015 Joint Service Agent Water Monitor program) The authors would also like to thank Dr Steve Christensen of the US Army SBCCOM for helpful discussions and Mr Chetan Shende of Real-Time Analyzers for assistance in development of the gold-doped sol-gels
REFERENCES 1 Erickson B Analytical Chemistry News amp Features June 1 397A (1998)
SPIE -2003-5269 125
2 Johnston RL Hoefler CM Fargo JC and Moberley B AT-ONSITE 5-8 (1994) 3 Black RM RJ Clarke RW Read and MT Reid J Chromatography 662 301-321 (1994) 4 USA Army Center for Health Promotion and Preventive Medicine (USACHPPM) Technical Guideline 230A 5 Jensen JL NC Wong W R Loerop SPIE 4775-03 (2001) 6 Hoenig SL Handbook of Chemical Warfare and Terrorism Greenwood Press Wesport CT (2002) 7 Munro NB SS Talmage GD Griffin LC Waters AP Watson JF King and V Hauschild Env Health
Persp 107 933-974 (1999) 8 Holstege CP Kirk M Sidell FR Crit Care Clin13 923-42 (1997) 9 Farquharson S P Maksymiuk K Ong and S Christesen SPIE 4577 166-173 (2001) 10 Lee Y and S Farquharson SPIE 4378 21-26 (2001) 11 Spencer KM J Sylvia S Clauson and J Janni SPIE 4577 158-165 (2001) 12 Tessier P S Christesen K Ong E Clemente A Lenhoff E Kaler and O Velev Applied Spectroscopy 56
1524-1530 (2002) 13 Farquharson S WW Smith S Elliott and JF Sperry SPIE 3533 207-214 (1998) 14 Farquharson S WW Smith S Elliott and JF Sperry SPIE 3855110-116 (1999) 15 Farquharson S WW Smith YH Lee S Elliott and JF Sperry SPIE 4575 62-72 (2002) 16 Guzelian AA J Sylvia J Janni S Clauson and KM Spencer SPIE 4577 182-192 (2001) 17 Shende C F Inscore A Gift P Maksymiuk and S Farquharson in press 18 Weaver MJ S Farquharson and MA Tadayyoni J Chem Phys 82 4867-4874 (1985) 19 Alak AM and T Vo-Dinh Analytical Chemistry 59 2149-2153 (1987) 20 Lee Y and S Farquharson SPIE 4206 140-146 (2000) 21 Farquharson S and Y Lee SPIE 4200-16 (2000) 22 Lee Y S Farquharson and P M Rainey SPIE 3857 76-84 (1999) 23 Lee Y S Farquharson H Kwong and M Shahriari SPIE 3537 252-260 (1998) 24 Wang J M Pumera G Collins and A Mulchandani Analytical Chemistry 74 6121-6125 (2002) 25 Dou X YM Jung Z-Q Cao and Y Ozaki Applied Spectroscopy 53 1440-1447 (1999) 26 Data supplied by S Christesen and K Ewing 27 Kerker M and B Thompson Eds SPIE MS 10 (1990) 28 Inglesby TV DA Henderson JG Bartlett JAMA 287 2236 (2002) 29 FW Janssen AJ Lund and LE Anderson Science 127 26 (1958) 30 Lide DR Ed Handbook of Chemistry and Physics CRC Press 77th Ed (1996-7) 31 Billmann J G Kovacs and A Otto Surf Sci 92 153 (1980) 32 Murray CA and S Bodoff Phys Rev B 32 671 (1985) 33 Kellogg D and J Pemberton J Phys Chem 91 1120 (1987)
SPIE-2003-5269 16
Chemical agent detection by surface-enhanced Raman spectroscopy
Stuart Farquharson Alan Gift Paul Maksymiuk Frank Inscore and Wayne Smith
Real-Time Analyzers 87 Church Street East Hartford CT 06108
Kevin Morrisey and Steven D Christesen US Army SBCCOM Aberdeen Proving Ground MD 21010
ABSTRACT
In the past decade the Unites States and its allies have been challenged by a different kind of warfare exemplified by the terrorist attacks of September 11 2001 Although suicide bombings are the most often used form of terror military personnel must consider a wide range of attack scenarios Among these is the intentional poisoning of water supplies to obstruct military operations in Afghanistan and Iraq To counter such attacks the military is developing portable analyzers that can identify and quantify potential chemical agents in water supplies at microgram per liter concentrations within 10 minutes To aid this effort we have been investigating the value of a surface-enhanced Raman spectroscopy based portable analyzer In particular we have been developing silver-doped sol-gels to generate SER spectra of chemical agents and their hydrolysis products Here we present SER spectra of several chemical agents measured in a generic tap water Repeat measurements were performed to establish statistical error associated with SERS obtained using the sol-gel coated vials Keywords Chemical agents chemical agent detection SERS Raman spectroscopy
1 INTRODUCTION
In the autumn of 2001 terrorism within US borders became a sobering reality While extensive efforts are being implemented to secure the homeland US and Coalition forces in Afghanistan and Iraq are constantly faced with terrorist attacks In addition to car-bombings the military has identified several non-traditional attack scenarios including poisoning of water supplies by chemical warfare agents (CWAs) To counter this threat the Department of Defense has been investigating numerous approaches to detect such attacks under the auspices of the Joint Service Agent Water Monitor (JSAWM) program1 The JSAWM must be capable of identifying the chemical agents with no false-positives and quantifying the agents at microgram per liter concentrations within 10 minutes2 This includes the analysis of drinking water supplies distribution and storage systems Currently colorimetric paper is used to detect agents on-site while gas chromatography combined with mass spectrometry (GCMS) is used in mobile support laboratories However both methods have severe drawbacks The paper changes color in response to contact with many chemicals besides CWAs causing a high incidence of false positives3 while GCMS although very chemically specific requires up to an hour to perform and regular re-calibration456 Vibrational spectroscopy has also been investigated as a potential method of rapidly detecting CWAs7-
11 as early as 1985 in the case of infrared spectroscopy8 and 1988 in the case of Raman spectroscopy11 Again however these techniques also have limitations when it comes to measuring trace poisons in water Infrared spectra would be dominated by the very strong absorption of water which would obscure absorptions by most other chemicals present Whereas Raman spectroscopy is simply not a very sensitive technique and detection limits are typically grams per liter Surface-enhanced Raman spectroscopy (SERS) offers several advantages over conventional vibrational methods and that may provide the necessary sensitivity required for detecting trace quantities of chemical agents in water SERS employs the interaction of surface plasmon modes of metal particles with target analytes to increase scattering efficiency by as much as 1 million times12 In 1987 the potential of this approach to measure CWAs was demonstrated by measuring a series of pesticides13 Several of these organophosphonates have chemical structures similar to CWAs in particular P=O functional groups In the past few years we and others have further explored the ability of SERS to detect CWAs14-17 and even bioagents 18-21 We have been employing silver-doped sol-gels to promote the SER effect
SPIE-2003-5269 17
in these studies The porous silica network of the sol-gel matrix offers a unique environment for immobilizing and stabilizing SER-active metal particles22-25 The silver-doped sol-gels have been coated on the internal walls of standard glass vials to allow rapid SER analysis of small sample quantities without preparation We have measured over 100 chemicals with enhancements of 104 to 106 demonstrated reversible measurements in a flowing system reproducible measurements from vial-to-vial and measurements in multiple solvents including water21-25 Previously we used these vials to perform preliminary measurements of cyanide (CN) and methylphosphonic acid (MPA) Most of the nerve agents form MPA during hydrolysis while Tabun forms CN a chemical agent in its own right In another paper including in these proceedings we examined the limits of detection (LOD) for MPA by measuring a series of concentrations down to 50 mgL and estimated a limit of detection of 100 microgL26 These measurements provide encouragement in that SERS may satisfy the needs of the JSAWM To further establish the viability of SERS in particular silver-doped sol-gels here we present analysis of cyanide mustard and VX in tap water The measurements performed at the US Armyrsquos Edgewood Chemical Biological Center Aberdeen MD also included numerous repeat measurements to establish reproducibility
2 EXPERIMENTAL 2a General All chemicals including potassium cyanide 2-chloroethylethyl sulfide and those used to prepare the silver-doped and gold-doped sol-gel coated vials were obtained and used as received from Sigma-Aldrich All samples were prepared in a chemical hood using HPLC grade water unless otherwise noted (Fischer Scientific Fair Lawn NJ) for SERS measurements Once prepared the samples were transferred into the silver-doped sol-gel vials (Real-Time Analyzers Inc East Hartford CT) The vials were coated in a manner similar to that previously reported by adding ammonium hydroxide to a solution of silver nitrate tetramethyl orthosilicate (TMOS) and methanol22 The two precursor solutions were prepared mixed and transferred to 2-mL glass vials dried and heated After sol-gel formation the incorporated metal ions were reduced with dilute sodium borohydride (1mgmL) which is followed by a water wash to remove residual reducing agent After the resultant analyte solution was introduced the SER-active vial was fixed horizontally to an XY positioning stage (Conix Research Springfield OR) just inside the focal point of an f07 aspheric lens The lens focused the beam into the sample and collected the scattered radiation back along the same axis A dichroic filter (Omega Optical Brattleborough VT) was used to reflect the excitation laser to the lens and pass the Raman scattered radiation collected by the lens An f2 achromat was used to collimate the laser beam exiting a 200 microm core diameter source fiber optic while a second f2 achromat was used to focus the scattered radiation into a 365 microm fiber optic (Spectran Avon CT) A short pass filter was placed in the excitation beam path to block the silicon Raman scattering generated in the source fiber from reflecting off sampling optics and reaching the detector A long pass filter was placed in the collection beam path to block the sample Rayleigh scattering from reaching the detector A 785 nm diode laser (Process Instruments Inc model 785-600 Salt Lake City UT) was used to deliver 100 to 150 mW of power to the sample A Fourier transform Raman spectrometer (Real-Time Analyzers model IRA-785 East Hartford CT) and a silicon photo-avalanche detector (Perkin Elmer model C30902S Stamford CT) were used to acquire the SER spectra
2b Edgewood Chemical Biological Center The surface-enhanced Raman spectral measurements at the US Armyrsquos Chemical Biological center presented here were all performed on September 12 2003 To expedite measurements a plate was machined to hold up to 12 SER-active sample vials (Figure 1) The plate fit a standard XY plate reader that could be programmed Pure KCN bis-(2-chloroethyl)sulfide (distilled mustard gas HD) and ethyl S-2-diisopropylamino ethyl methylphosphonothioate (VX) were obtained on-site and used to prepare 1 gL tap water solutions in a chemical hood with appropriate safety equipment Simulated tap water was prepared by adding 10 chemicals most often found in tap water at appropriate concentrations to distilled water (Table 1) SER measurements were also performed in a chemical hood For added safety the FT-Raman instrument was placed outside the laboratory and 30 foot fiber optic and electrical cables were used to allow remote SERS measurements and plate manipulation For each experiment 1gL samples were prepared and added to 9 individual vials which were then loaded on the plate In some cases a tenth vial was included as a blank
Table 1 Generic Tap Water Composition
Compound mgL NaHCO3 100 CaSO4 27 MgSO4bull7H2O 67 NaNO3 1 Fulvic Acid 1 K2HPO4 07 KH2PO4 03 (NH4)2HSO4 001 NaCl 001 FeSO4 0001 pH 76-78
SPIE-2003-5269 18
A software program was written that allowed selecting the sequence that the vials were measured the number of positions along the length of the vials to measure (1 to 5) and the number of scans to co-add During sample analysis the program displayed the vial being analyzed the point being analyzed and the spectrum as it was being acquired Once all the data was collected a second software program was written to rapidly analyze the data The spectra collected for all the vials on a plate could be loaded at one time and then the spectra for each point could be displayed simultaneously or separately The user could then select the Raman peak to analyze in terms of peak height or area This was accomplished by selecting points on either side of the peak to define a baseline of zero The peak height or area could then be computed for all of the spectra loaded and then exported to a spreadsheet for statistical analysis
Figure 1 A) Vial Holder 6 slots to hold 2 vials each end-to-end B) Measurement Configuration Program user selects vials to measure sequence number of points per vial (1 to 5) and number of scans per point C) Spectral Acquisition Program shows spectrum being collected which vial and position D) Spectra Analysis Program user selects spectra to analyze by plate vial and point (s) as well as two wavenumbers defining the peak and the baseline to subtract The image is of 5 repeat measurements of 10 mgL KCN in generic tap water 16 sec each 100 mW of 785 nm
3 RESULTS AND DISCUSSION Raman and surface-enhanced Raman spectra were obtained for potassium cyanide bis-(2-chloroethyl)sulfide and ethyl S-2-diisopropylamino ethyl methylphosphonothioate representing three classes of chemical agents cyanides mustards and nerve agents respectively Spectra were also obtained for 2-chloroethyl ethyl sulfide (CEES) a structural analogue to HD which was included in the study to aid in assigning spectral bands KCN salt was used for cyanide experiments to avoid the increased hazards of handling HCN gas KCN completely dissolves in water forming its conjugate acid HCN according to its Ka of 615x10-1027 and at a concentration of 1 mgmL results in a pH 107 solution This is significant in that only CN- appears to interact sufficiently with silver to produce a SER spectrum and no spectral signal is observed below pH 726 Figure 2 shows the SER and normal Raman spectra for KCN The SERS spectrum of 1mgml KCN in water shows a single intense somewhat broad feature at
A B
C D
SPIE-2003-5269 19
2100 cm-1 assigned to the single CequivN stretch The band is much sharper in the normal Raman spectra of the solid KCN salt at 2074 cm-1 This band does broaden and shift to 2080 cm-1 in solution (not shown) However the observed SERS frequency is attributed to interaction with silver and low frequency mode at 135 cm-1 attributed to a Ag-CN stretch (not shown) supports this conclusion
cm-1 band to a C-S stretch but the authors concede that it is in fact more likely a C-Cl stretch28 It appears that the most intense Raman bands at 648 692 and 747 cm-1 shift to 620 660 and 720 cm-1 in the SER spectra and are tentatively assigned as above The width of these bands suggests that they overlap underlying spectral features Additional bands in the Raman spectra occur at 972 1034 1049 1263 1286 1423 1442 2865 2935 and 2960 cm-1 Corresponding bands occur in the SER spectra at 964 1015 1054 1286 1410 1447 2865 and 2935 cm-1 Most of these bands are associated with alkane modes specifically the bands at approximately 1040 cm-1 to a C-C stretch 1290 cm-1 to a CH2 in-phase twist 1440 cm-1 to a CH2 wag 2865 cm-1 to a symmetric CH2 stretch and 2965 cm-1 to an asymmetric CH2 stretch The Raman and SER spectra of sulfur mustard were measured at the Edgewood center (Figure 4) Both spectra are largely similar to CEES The C-Cl and C-S bands in the Raman spectrum of HD now occur at 640 655 700 739 and 760 cm-1 and are more resolved possibly due to the increased molecular symmetry Theoretical calculations indicate that the first three bands are due to C-Cl stretching modes and the latter two to C-S stretching modes28 Only the C-Cl bands maintain significant intensity in the SER spectra occurring at 624 and 643 cm-1 which is attributed to the
Prior to measurements of HD CEES was examined by Raman and SER spectroscopy (Figure 3) CEES also known as half-mustard is essentially identical to HD except one of the chlorine end atoms is replaced by a hydrogen atom Again although not as toxic as HD CEES is a blister agent and dilute aqueous samples were prepared using appropriate safety equipment Both the Raman and SER spectra of CEES are similar and dominated by bands between 600 and 800 cm-1 These are associated with C-Cl and C-S stretching modes which are tentatively assigned to 648 and 747 cm-1 in the Raman spectra respectively The shoulder at 630 cm-1 the overlapped band at 660 cm-1 and the strong band at 692 cm-1 could also be due to these modes or their asymmetric counterparts It is worth noting that theoretical calculations assign the 692
Figure 2 A) SER and B) NR spectra of KCN Conditions A) 1 mgml in tap water 100 mW of 785 nm at sample 1-min acquisition time B) solid 300 mW of 785 nm 5-min All spectra are 8 cm-1 resolution
A
B
Figure 3 A) SER and B) NR spectra of CEES Conditions A) 1 vv (10 mgml) in MeOH 100 mW of 785 nm 1-min acquisition time B) neat 300 mW of 785 nm 5-min
A
B
Cl-CH2-CH2-S-CH2-CH3
A
B
Figure 4 A) SER and B) NR of HD Conditions A) 1mgml in tap water B) pure both 100 mW of 785 nm 1-min
Cl-CH2-CH2-S-CH2-CH2-Cl
SPIE-2003-5269 20
expected strong interaction between chlorine and silver and adds support to the assignment of this band to a C-Cl stretch Weaker overlapping bands occur at 670 692 and 724 cm-1 the latter possibly due to C-S stretching modes Again the alkane modes are apparent in the normal Raman spectra of HD but only a broad feature at 1300 to 1450 cm-1 suggests CH2 contributions in the SER spectrum Although the observed bands in the VX spectrum have not been assigned (Figure 4) a computer generated Raman spectrum29 predicts many of the same features with surprising accuracy and are used here Two intense bands at 460 and 530 cm-1 closely match predicted bands at 463 and 546 cm-1 assigned to a CH3-P=O bend and a PO2CS wag Three highly overlapped bands occur at 694 745 and 771 cm-1 matching predicted bands at 713 730 and 760 cm-1 The first
Table 2 Measured SER peak heights for the CN stretch at 2100 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
Figure 5 A) SERS and B) NR spectra of VX Conditions A) 1 vv (10 mgml) in MeOH B) pure sample both 100 mW of 785 nm 1-min acquisition time
A
B
two have been assigned to a C-S stretch and CH2bend respectively while the latter has been attributed to either a P-C stretch or an O-C-C stretch Although the 745 cm-1 band may alternatively be assigned to a C-S stretch based on the previous measurements of CEES and HD The relatively intense bands at 890 1106 1218 1445 and 1465 cm-1 also match predicted bands at 880 1108 1216 1440 and 1464 cm-1 that are assigned to a C-C stretch CH3 rock N-C3 stretch various C-H3 bends and C-H bends respectively Both the computer generated and the measured spectra contain numerous other less intense bands One is worth mentioning A unique band appears at 370 cm-1 that is predicted at 368 cm-1 and corresponds to an O-P=O bend The surface-enhanced Raman spectrum of VX is also rich with spectral features It has the unique low frequency band at 370 cm-1 as well as a second band at 380 cm-1 that is assigned tothe S-P-O bend predicted in the normal Raman spectrum at 388 cm-1 Based on the measured and predicted normal Raman spectra the following SERS assignments are given 460 cm-1 to the CH3-P=O bend 544 cm-1 to the PO2CS wag 738 cm-1 to a C-S stretch (based on arguments above) 890 cm-1
to a C-C stretch 1101 cm-1 to a CH3 rock and 1456
cm-1 to a C-H bend The ability of SERS to measure chemical agents in water containing real-world chemical interferents was tested by using the generic tap water described in Table 1 The ability to reproduce measurements was accomplished by preparing three separate water stock solutions which were used to prepare three sample solutions each of 1 mgmL KCN in the generic tap water The samples defined as vials 1-9 were then measured at 4 points per vial in a semi-random fashion such that errors associated with stock solution preparation and errors associated with instrument drift could be identified No trends were apparent that signified such systematic errors Each spectrum collected consisted of 20 averaged scans taking 16 seconds at 8 cm-1 resolution The laser power at the sample was measured periodically during the day and it ranged from 102 to 105 mW spacccc
spot Vial 1 Vial 2 Vial 41 13754 13004 128192 13519 12692 12909
1 3 13541 12721 126394 13462 12648 12651
ave 13569 12766 12755Vial 6 Vial 3 Vial 5
1 11586 14068 10757 2 11236 14402 115122 3 11376 14546 11559
4 10894 11763 11214ave 11273 13695 11261
Vial 7 Vial8 Vial 91 11198 15804 11023
2 11243 15716 88753 3 11562 15216 11218
4 11655 15084 11077ave 11415 15455 10548
CN AVG STDEV ERR1 mgmL pk ht 12526 1560 1245
stock solution
SPIE-2003-5269 21
The CN measurements consisting of 36 data points produced an average height of 125 for the 2100 cm-1 peak with a standard deviation of 156 or 125 (Table 2) The HD and VX measurements were performed precisely the same way (Tables 3 and 4) For HD the 624 cm-1 peak was used for analysis and it had an average height of 53 with a standard deviation of 068 or 129 while for VX the 544 cm-1 peak was used for analysis and it had an average height of 1051 with a standard deviation of 308 or 293 The greater error in the VX measurements can be somewhat attributed to Vial 4 which produced lower SER signal intensities But removing this vial from the data set changes the standard deviation to 233 only a modest improvement
spot Vial 1 Vial 2 Vial 41 525 609 4682 45 675 484
1 3 527 644 5094 559 807 669
ave 51525 68375 5325Vial 6 Vial 3 Vial 5
1 521 574 575 2 536 53 4492 3 509 508 451
4 565 594 379ave 53275 5515 4635
Vial 7 Vial 8 Vial 91 457 589 413
2 583 497 4273 3 544 505 4
4 528 5 509ave 528 52275 43725
HD AVG STDEV ERR1 mgmL pk ht 530 068 1291
stock solution
spot Vial 1 Vial 2 Vial 41 1464 1034 4672 1485 99 61
1 3 1491 989 5684 1041 777 553
ave 137025 9475 5495Vial 6 Vial 3 Vial 5
1 1058 942 1293 2 697 121 965
2 3 727 1206 11124 689 1254 746
ave 79275 1153 1029Vial 7 Vial 8 Vial 9
1 1127 783 163 2 1358 812 1614
3 3 1371 908 15124 1097 875 1397
ave 123825 8445 153825
VX AVG STDEV ERR1 mgmL pk ht 1051 308 2925
stock solution
4 CONCLUSIONS In this paper we examined the ability of surface-enhanced Raman spectroscopy to reproducibly measure CN HD and VX in tap water without chemical interference Both normal and surface-enhanced Raman spectra were examined to select unique bands suitable to identify and quantify these chemical agents For SER measurements the 2100 cm-1 C-N stretch was used for CN the 624 cm-1 C-Cl stretch was used for HD and the 544 cm-1 PO2CS wag was used for VX It was determined that 1 mgmL samples of each of these chemicals measured 36 times in glass vials coated with a silver-doped sol-gel reproduced measurements with standard deviations of 125 129 and 293 It was further found that the 10 chemicals added to simulate generic tap water did not interfere with or alter the SER spectra It should be noted that the concentrations used in this study were considerably greater than those required by the JSAWM program Current work involves improving SER sensitivity and designing sampling systems with better reproducibility This includes the development of fractal silver and gold structures within the sol-gel matrix and the development of chemically selective sol-gels
ACKNOWLEDGMENTS The authors are grateful for the support of the US Army (DAAD13-02-C-0015 Joint Service Agent Water Monitor program) The authors would also like to thank Janet Jensen Ronald Crosier and Kristina Gonser for helpful discussions
Table 4 Measured SER peak heights for the VX band at 544 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
Table 3 Measured SER peak heights for the HD band at 624 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
SPIE-2003-5269 22
REFERENCES 1 Jensen JL NC Wong W R Loerop SPIE 4775-03 (2001) 2 JSAWM Requirements at wwwsbccomapgeaarmymilRDAecbcrtPRODSERJSAWMjsawmhtml 3 Erickson B ldquoThe Chemical Weapons Convention Redefines Analytical Challengerdquo Anal Chem 70 397A (1998) 4 Johnston RL Hoefler CM Fargo JC and Moberley B ldquoThe Defense Nuclear Agencyrsquos
ChemicalBiochemical Weapons Agreements Verification Technology Research Development Test and Evaluation Program and its Requirements for On-Site Analysisrdquo AT-ONSITE 5-8 (1994)
5 Black RM Clarke RJ Read RW and Reid MT ldquoApplication of GC-MS and GC-tandem MS to the analysis of chemical warfare samples found to contain residues of the nerve agent sarin sulphur mustard and their degradation productsrdquo J Chrom 662 301-321 (1994)
6 USA Army Center for Health Promotion and Preventive Medicine (USACHPPM) Technical Guideline 230A 7 Hoffland LD Piffath RJ Bouck JB rdquoSpectral signatures of chemical agents and simulantsrdquo Optical
Engineering 24 982-984 (1985) 8 Braue EHJ Pannella MGrdquoCIRCLE CELL FT-IR Analysis of Chemical Warfare Agents in Aqueous Solutionsrdquo
App Spectrosc 44 1513-1520 (1990) 9 Tseng C-H Mann CK and Vickers TJ ldquoDetermination of Organics on Metal Surfaces by Raman
Spectroscopyrdquo App Spectrosc 47 1767-1771 (1993) 10 Staff Reporter ldquoUS Army Tests Lidar to Detect Biological Toxinsrdquo Photonic Spectra pg 50 December 1998 11 Christesen SD Raman cross sections of chemical agents and simulants App Spectrosc 42 318-321 (1988) 12 Weaver MJ Farquharson S Tadayyoni MA ldquoSurface-enhancement factors for Raman scattering at silver electrodesrdquo J Chem Phys 82 4867-4874 (1985) 13 Alak AM and Vo-Dinh T ldquoSERS of Organophosphorous Chemical Agentsrdquo Anal Chem 59 2149-2153 (1987) 14 Lee Y Farquharson S ldquoRapid chemical agent identification by SERSrdquo SPIE 4378 21-26 (2001) 15 Farquharson S Maksymiuk P Ong K Christesen S ldquoChemical agent identification by surface-enhanced Raman spectroscopyrdquo SPIE 4577 166-173 (2001) 16 Spencer KM Sylvia J Clauson S and Janni J ldquoSurface Enhanced Raman as a Water Monitor for Warfare
Agents in Waterrdquo SPIE 4577 158-165 (2001) 17 Tessier P Christesen S Ong K Clemente E Lenhoff A Kaler E Velev O ldquoOn-line spectroscopic characterization of sodium cyanide with nanostructured gold surface-enhanced Raman spectroscopy substratesrdquo
App Spectrosc 56 1524-1530 (2002) 18 Farquharson S W W Smith S Elliott and J F Sperry Rapid biological agent identification by surface- enhanced Raman spectroscopy SPIE 3855110-116 (1999) 19 Farquharson S WW Smith YH Lee S Elliott and J F Sperry Detection of biological signatures A
comparison of electrolytic and metal-doped sol-gel surface-enhanced Raman media SPIE 4575 62-72 (2002) 20 Guzelian AA Sylvia J Janni J Clauson S and Spencer KM ldquoSERS of whole cell bacteria and trace levels of biological moleculesrdquo SPIE 4577 182-192 (2001) 21 Shende C Inscore F Gift A Maksymiuk P Farquharson S ldquoRapid dipicolinic acid extraction from Bacillus spores detected by surface-enhanced Raman spectroscopyrdquo App Spectrosc 58 accepted 22 Lee Y and Farquharson S ldquoSERS Sample Vials Based on Sol-Gel Process for Trace Pesticide Analysisrdquo SPIE
4206 140-146 (2000) 23 Farquharson S and Lee Y ldquoTrace Drug Analysis by SERSrdquo SPIE 4200-16 (2000) 24 Lee Y Farquharson S and Rainey P M Surface-Enhanced Raman Sensor for Trace Chemical Detection in
Water SPIE 3857 76-84 (1999) 25 Lee Y Farquharson S Kwong H and Shahriari M ldquoSol-Gel Chemical Sensor for Surface-Enhanced Raman
Spectroscopyrdquo SPIE 3537 252-260 (1998) 26 Farquharson S A Gift P Maksymiuk F Inscore and W Smith ldquopH dependence of methyl phosphonic acid
dipicolinic acid and cyanide by surface-enhanced Raman spectroscopyrdquo SPIE 5269-19 (2003) 27 Lide DR Ed Handbook of Chemistry and Physics CRC Press 77th Ed (1996-7) 28 Sosa C RJ Bartlett K KuBulat and WB Person ldquoA theoretical study of harmonic vibrational frequencies and
infrared intensities of XCH2CH2SCH2CH2X and XCH2CH2SH (= H Cl)rdquo J Phys Chem 93 577-588 (1993) 29 Hameka HF and JO Jenson ldquoComputer-generated predictions of the structure and the IR and Raman spectra of
VXrdquo ERDEC-TR-065 May 1993
- Chemagents Appendicespdf
-
- SPIE2001-4575-Bioagent-SERSapdf
-
- Detection of bioagent signatures A comparison of electrolytic and metal-doped sol-gel surface-enhanced Raman media
- Stuart Farquharson Wayne Smith and Yuan Lee
-
- Real-Time Analyzers 87 Church Street East Hartford CT 06108
-
- ABSTRACT
- 4 CONCLUSIONS
- 5 ACKNOWLEDGEMENTS
- 6 REFERENCES
-
SPIE -2003-5269 120
starting solution was adjusted to 325 using dilute KOH Again 2 mL were added to a vial and the SER spectrum recorded At this pH a reasonably strong 760 cm-1 band was observed This process was repeated as spectra were recorded at pHs of 70 74 75 79 85 and 100 A total of 1 mL of KOH was added diluting the total concentration by 10 Next the pH of the starting solution was made acidic by adding dilute nitric acid dropwise This time spectra were recorded at pHs of 72 69 64 and 37 Figure 4 shows the SER spectra for representative pHs (spectra were left out to simplify the figure) while Figure 3 shows the 760 cm-1 peak intensities as a function of pH (The band intensities were adjusted to compensate for dilution effects caused by the addition of HNO3 and KOH then normalized to 0021 M for the most intense band observed at pH 37) It is clear from Figure 3 that the 760 cm-1 band follows the MPA- concentration as a function of pH and must be assigned to this anion No bands were observed that corresponded to MPA or MPA= The lack of an MPA SER spectrum may be due to the absence of an attraction between the neutral analyte and the electropositive silver surface The same reasoning suggests that a strong SER spectrum should be observed for MPA= but it is not and a satisfactory explanation has not been found
0000
0005
0010
0015
0020
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14pH
Con
cent
ratio
n [M
]
pK1 = 212 pK2 = 729
MPA- MPA=
MPA
Figure 4 SER spectra of 002M MPA as a function of pH Conditions 100 mW of 785 nm 36 scans (1 min) 8 cm-1 recorded 2 min after sample introduction pH 19 and 20 not apparent on this scale pH 69 and 74 near identical to 70 and 75 and not shown for clarity
00
02
04
06
08
10
12
14
16
18
0 5 10 15 20 25 30Measurement Number
Ram
an In
tens
ity (7
60 c
m-1
)
Figure 3 Concentration dependence of MPA MPA- and MPA= as a function of pH for a 002M sample Intensity of 760 cm-1
band from Figure 6 as a function of pH ( for increasing basic adjustment for increasing acidic adjustment error was measured at ~10 for pH 64)
Figure 5 SER spectra of 002M MPA at pH 64 measured around a vial at three heights (9 points per height) Conditions as in Figure 4 but 10-sec scans
Since these measurements involved the removal and replacement of the SER-active vial in the sample holder to remove and add sample variation in the intensity as a function of vial position was minimized by illuminating the exact same height along the vial wall But this does not account for variability of the SERS response of the sol-gel coating around the vial To analyze this effect a vial containing MPA at pH 64 was rotated at ~ 40o
intervals at the original height and 18rdquo above and below this value Figure 5 shows the intensity of the 760 cm-1
band for the 27 positions It was found that the average value was 137plusmn014 an RSD of 10 overall and 5 for each height An error bar is included in Figure 4 for the pH 64 measurement
SPIE -2003-5269 121
With the above analyses in mind a preliminary investigation of the SER spectral response for MPA (as MPA-) as a function of concentration was performed A single vial was used for these measurements beginning with 1 mgL followed by measurements of 10 100 and 1000 mgL In all cases the pH was ~7 and 3 positions around the vial were measured per concentration Since the 760 cm-1 band was not observed for concentrations of 1 or 10 mgL using 100 mW of 785 nm the laser power at the sample was raised to 200 mW beginning with the 10 mgL concentration Photo-degradation was largely avoided (and not observed) by exposing the sample for only 33 seconds per spectral acquisition Representative spectra for 10 100 and 1000 mgL are shown in Figure 6 while a plot of the 760 cm-1 band intensity as a function of concentration is shown in Figure 7 These values were also used to estimate limits of detection based on the signal-to-noise ratio (SN) of the 760 cm-1 band using 100 mW of 785 nm laser excitation and a 10-min acquisition time As summarized in the Figure 7 inset the lower the measured concentration the lower the predicted LOD Even if as estimated 210 microgL could be measured using the silver-doped sol-gel vials an improvement of a factor of 70 is still required to meet JSAWM goals of measuring 3 microgL in 10-minutes Similar to MPA DPA is a diprotic acid (pKa1 = 216 and pKa2 = 692) and variations in pH will effect the relative concentrations of DPA DPA- and DPA= and possibly the SER spectra and band intensities This could prove significant if an acid or base is used to denature anthrax spores with the goal of extracting and analyzing DPA The MPA pH study described above was mimicked for DPA except that the starting solution consisted of 20 mg of DPA in 20 mL HPLC grade water (60x10-3M MW = 1671) The initial solution had a pH of 245 which was made basic by dilute KOH to pHs of 355 433 487 559 1069 and 1166 SER spectra were recorded at each pH using 100 mW of 785 nm and a 44-sec acquisition time Next one drop of concentrated nitric acid was used to remake the solution acidic at a pH of 200 Again sequential pH measurements were performed at 383 510 735 and 822 The solution pH was made acidic a third time but to pH 219 171 then 135 Throughout this process no more than 20 drops of acid or base were added and therefore the concentration was diluted by no more than 10 Most of the spectral bands showed a minor decrease in intensity as a function of increasing pH values However the bands at 525 and 795 cm-1 showed the most dramatic changes which occurred at acid pH Figure 8 shows the SER spectra of DPA for the spectral region and pH range of interest The identity of the DPA species was determined by plotting the normalized peak intensities with the lowest value set to 0 and the highest to 0006 M as a function of pH and overlaying these values on a plot of the relative concentrations for DPA DPA- and DPA= as previously done for MPA (Figure 9) As can be seen the 525 cm-1 band clearly corresponds to DPA The correspondence of the 795 cm-1 band to this species is less clear as the band retains intensity until pH of 55 This can be attributed to contribution to the overlapping band at 810 cm-1 which does not change as a function of pH The fact that most bands are observed at all pHs suggest that the primary interaction with silver is through the ring nitrogen This is supported by the fact that the most intense band occurs at 1008 cm-1 attributed to a symmetric ring breathing mode and that this interaction has been characterized for pyridine in numerous papers27
0
1
2
3
4
5
6
7
0 200 400 600 800 1000 1200
MPA Concentration (mgL)
760
Ban
d In
tens
ity
Figure 6 SER spectra of MPA in water at A) 1000 B) 100 and C) 10 mgL Conditions pH of 7 silver-doped sol-gel coated vial 200 mW of 785 nm 33-sec 8 cm-1 resolution
Figure 7 Plot of SER intensity of 760 cm-1 band of MPA as a function of concentration using 200 mW of 785 nm Inset table includes average intensity LOD standard deviation and percent deviation for each concentration but for 100 mW and 10 min
conc (mgL) LOD ave stddev dev10 021 002 1021100 072 010 14301000 312 040 1280
A
B C
SPIE -2003-5269 122
The ability to detect dipicolinic acid (DPA) by SERS is demonstrated in Figure 10 Even at 1 mgL the primary bands are visible The intensity of the dominant ring breathing mode observed at 1008 cm-1 for DPA in water (pH = 25 to 55) is plotted as a function of concentration in Figure 11 These values were also used to estimate limits of detection based on the SN of the 1008 cm-1 band using 100 mW of 785 nm laser excitation and a 10-min acquisition time Again the lower the measured concentration the lower the predicted LOD (see Figure 11 inset) and detection of 160 microgL is possible Gastrointestinal anthrax requires significant more spores than inhalation anthrax28 and a limit of detection might be placed at 1 million spores in 1 liter of water or 10 microgL Since each spore contains ~10 CaDPA by weight29 a goal for DPA might be 1 microgL indicating that the present measurements must be improved by nearly two orders of magnitude Due to the increased hazards of handling HCN gas KCN salt was used for these experiments Nevertheless all sample preparations were performed in a chemical hood KCN completely dissolves in water but its conjugate acid HCN is formed and has a Ka of 615x10-1030 viz
HCN CN- + H+ pKa = 921 Reaction 4
Figure 8 SER spectra of 1 mgmL DPA as a function of pH Conditions 100 mW 785 nm 100 scans (44-sec)
Figure 9 Concentration dependence of DPA DPA- and DPA= as a function of pH for a 0006M sample Intensity of 525 () and 795 (diams) cm-1 bands from Fig 8 as a function of pH
pH
135 171 219 383
0
0001
0002
0003
0004
0005
0006
0007
0 2 4 6 8 10 12 14pH
Con
cent
ratio
n [M
]
DPADPA-DPA=795525
DPA DPA=DPA-
pK2 = 692pK1 = 216
795 525
0
05
1
15
2
25
0 200 400 600 800 1000 1200
DPA Concentration (mgL)
1008
Ban
d In
tens
ity
conc (mgL) lod-10min-100mw1 017
10 016100 103
1000 355
Figure 10 SER spectra of DPA in water at A) 1000 B) 100 C) 10 and D) 1 mgL Conditions pH of 25-55 silver-doped sol-gel coated vial 175 mW of 785 nm 1-min 8 cm-1 D) has been multiplied by x10 to make bands visible
A
B C
D
Figure 11 Plot of SER intensity of 1008 cm-1 band of DPA as a function of concentration using 175 mW of 785 nm Inset table includes LOD in mgL for each concentration but for 100 mW and 10 min
SPIE -2003-5269 123
Consequently the cyanide concentration must be determined for each initial KCN concentration Specifically the samples prepared with concentrations of 01 1 10 100 and 1000 mgL of KCN produced CN- concentrations of 63x10-3 033 69 89 and 964 mgL at pHs of 816 90 967 102 and 107 respectively The pH dependence for the HCN and CN- concentrations are shown in Figure 12 Thus as the amount of KCN added to the solution decreases so does the pH of the solution (becomes less basic) and according to Reaction 4 the relative amount of CN- to HCN also decreases For example in the preparation of a 01 mgL solution of KCN the pH is shifted from 7 for pure water to only 816 and only 63 of the starting material becomes CN- or 63x10-3 mgL In comparison for a solution of 1000 mgL the pH is shifted from 7 to 107 and 96 of the starting material becomes CN- This is significant since the cyanide ion is better able to adsorb onto the silver particles and become SERS active SER spectra of 10 100 and 1000 mgL of KCN samples are shown in Figure 13 and a corresponding plot of the 2100 cm-1 band intensity as a function of concentration is shown in Figure 14
000102030405060708091011
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14pH
Con
cent
ratio
n [m
gm
L]
pKa = 921
CN -HCN
The SER spectra of cyanide are dominated by the single CN stretch at 2100 cm-1 which occurs in normal Raman spectra of solutions at 2080 cm-1 However a low frequency mode occurs at 135 cm-1 due to a Ag-CN stretch (not shown) indicative of a strong surface interaction It is also observed that as the concentration decreases the CN stretch shifts to 2140 cm-1 This shift has been attribute to the formation of a tetrahedral Ag(CN)3
2- surface structure31 as well
Figure 14 Concentration dependence of KCN SERS measured under conditions in Fig 11 Concentrations are 1 01 and 001 mgml Intensities are measured for the CN stretch at 2100 cm-1 Inset table includes LOD in mgL for each concentration in Figs 13 and 15 but for 100 mW and 10 min
Figure 13 SER spectra of KCN in water at A) 1000 B) 100 and C) 10 mgL Conditions pHs of 107 102 and 97 silver-doped sol-gel coated vial 100 mW of 785 nm 1-min 8 cm-1 C) has been multiplied by x10 to make band visible
Figure 12 Concentration dependence of HCN and CN- as a function of pH for a 1 mgmL sample Calculated intensity of 2100 () cm-1 band for a 1 mgml sample at pHs of 816 90 967 102 and 107
A
B
C
0
20
40
60
80
100
120
140
0 200 400 600 800 1000 1200CN Concentration (mgL)
2100
Ban
d In
tens
ity
Figure 15 SER spectra of KCN in water at A) 10 B) 1 and C) 01 mgL Conditions pHs of 97 12 and 12 gold-doped sol-gel coated vial 100 mW of 785 nm 1-min 8 cm-1 C) has been multiplied by x10 to make band visible
A
B
C
conc (mgL) lod-10min-100mw Condition01 001 Au-pH 121 007 Au-pH 12
69 003 Au-pH 9769 016 Ag-pH 9789 022 Ag-pH 102946 113 Ag-pH 107
SPIE -2003-5269 124
as to CN adsorbed to two different surface sites32 Alternatively the 2140 cm-1 band could be attributed to HCN since this species dominates at lower concentrations However it is unlikely that this species would be attracted to the electropositive silver surface Further both peaks should be present at pHs between 85 and 105 but this is not observed It has also been suggested that at concentrations near and above monolayer coverage the CN- species is forced to adsorb end-on due to crowding and at lower concentrations the molecule can reorient to lie flat33 This suggests that the 2100 and 2140 cm-1 bands correspond to the end-on and flat orientations respectively As Figures 13 and 14 show the intensity of the CN stretch for the 89 mgL sample is nearly as intense as the 964 mgmL band This suggests that the Raman signal for the flat orientation is more enhanced However more extensive measurements are required to verify this point Since resent research has suggested that cyanide may be more effectively detected on gold measurements of KCN solutions were also performed using gold-doped sol-gel vials Preliminary measurements are shown in Figure 15 for samples prepared from 01 1 and 10 mgL KCN Since the pHs are 816 90 967 the resultant CN- concentrations are 63x10-3 033 and 69 mgL Initially only the highest concentration was observed and the signal intensity was significantly better than the equivalent concentration measured using silver In an effort to shift Reaction 4 to the left transforming HCN to CN- (Le Chatelierrsquos principle) KOH was added to the lower concentration samples producing solutions with pH 12 This effectively forces all of the cyanide in solution to be CN- or 01 and 10 mgL respectively More importantly the CN stretch is now observed in the SER spectra The band appears at 2125 cm-1 as has been previously reported for gold12 As calculated for MPA and DPA LODs can be estimated from this data For the three concentrations of cyanide on silver the LODs are 016 to 11 mgL for 100 mW of 785 nm laser excitation and a 10-min acquisition time For gold pH adjusted the LODs are10 to 70 microgL an improvement of more than 10 times silver Nevertheless either substrate is sufficient to meet the JSAWM goals of measuring 3 mgL in 10-minutes as the requirements form cyanide are much less stringent than the nerve agents
4 CONCLUSIONS Here we examined the surface-enhanced Raman spectral response for methyl phosphonic acid dipicolinic acid and cyanide as a function of pH It was determined that the most prominent peak at 760 cm-1 reaches a maximum intensity between pH 3 and 7 and corresponds to the MPA- species Neither the MPA nor MPA= species appear to generate a SER spectrum and consequently no spectra were observed below pH 2 or above pH 8 In this study we also found that higher laser powers could cause photodegradation of MPA signified by the exponential growth of a band at 1050 cm-1 which is tentatively assigned to phosphorous acid Unlike MPA DPA was observed at all pHs This is attributed to the dominant interaction of the pyridine functional group with silver Minor spectral changes were observed at acid pHs and were assigned to neutral DPA Like MPA SER spectra of cyanide were pH dependent No spectra were observed for the HCN species while CN- was best observed at pHs more basic than 8 Preliminary concentration studies for the three analytes allowed estimating limits of detection for MPA DPA and CN using 100 mW of 785 nm and a 10-min acquisition time of 210 165 and 70 microgL respectively Although the latter value suggests that cyanide can be measured at sufficiently low concentrations to meet JSAWM goal improvements by 100 to 200 times are required for MPA and DPA It is clear from this study that pH of the sample is important and must be taken into account when developing concentration calibration curves as well as fieldable analyzers Future work will include tailoring the sol-gel with specific functional groups to dictate sample pH and thereby optimize sensitivity
ACKNOWLEDGMENTS The authors are grateful for the support of the US Army (DAAD13-02-C-0015 Joint Service Agent Water Monitor program) The authors would also like to thank Dr Steve Christensen of the US Army SBCCOM for helpful discussions and Mr Chetan Shende of Real-Time Analyzers for assistance in development of the gold-doped sol-gels
REFERENCES 1 Erickson B Analytical Chemistry News amp Features June 1 397A (1998)
SPIE -2003-5269 125
2 Johnston RL Hoefler CM Fargo JC and Moberley B AT-ONSITE 5-8 (1994) 3 Black RM RJ Clarke RW Read and MT Reid J Chromatography 662 301-321 (1994) 4 USA Army Center for Health Promotion and Preventive Medicine (USACHPPM) Technical Guideline 230A 5 Jensen JL NC Wong W R Loerop SPIE 4775-03 (2001) 6 Hoenig SL Handbook of Chemical Warfare and Terrorism Greenwood Press Wesport CT (2002) 7 Munro NB SS Talmage GD Griffin LC Waters AP Watson JF King and V Hauschild Env Health
Persp 107 933-974 (1999) 8 Holstege CP Kirk M Sidell FR Crit Care Clin13 923-42 (1997) 9 Farquharson S P Maksymiuk K Ong and S Christesen SPIE 4577 166-173 (2001) 10 Lee Y and S Farquharson SPIE 4378 21-26 (2001) 11 Spencer KM J Sylvia S Clauson and J Janni SPIE 4577 158-165 (2001) 12 Tessier P S Christesen K Ong E Clemente A Lenhoff E Kaler and O Velev Applied Spectroscopy 56
1524-1530 (2002) 13 Farquharson S WW Smith S Elliott and JF Sperry SPIE 3533 207-214 (1998) 14 Farquharson S WW Smith S Elliott and JF Sperry SPIE 3855110-116 (1999) 15 Farquharson S WW Smith YH Lee S Elliott and JF Sperry SPIE 4575 62-72 (2002) 16 Guzelian AA J Sylvia J Janni S Clauson and KM Spencer SPIE 4577 182-192 (2001) 17 Shende C F Inscore A Gift P Maksymiuk and S Farquharson in press 18 Weaver MJ S Farquharson and MA Tadayyoni J Chem Phys 82 4867-4874 (1985) 19 Alak AM and T Vo-Dinh Analytical Chemistry 59 2149-2153 (1987) 20 Lee Y and S Farquharson SPIE 4206 140-146 (2000) 21 Farquharson S and Y Lee SPIE 4200-16 (2000) 22 Lee Y S Farquharson and P M Rainey SPIE 3857 76-84 (1999) 23 Lee Y S Farquharson H Kwong and M Shahriari SPIE 3537 252-260 (1998) 24 Wang J M Pumera G Collins and A Mulchandani Analytical Chemistry 74 6121-6125 (2002) 25 Dou X YM Jung Z-Q Cao and Y Ozaki Applied Spectroscopy 53 1440-1447 (1999) 26 Data supplied by S Christesen and K Ewing 27 Kerker M and B Thompson Eds SPIE MS 10 (1990) 28 Inglesby TV DA Henderson JG Bartlett JAMA 287 2236 (2002) 29 FW Janssen AJ Lund and LE Anderson Science 127 26 (1958) 30 Lide DR Ed Handbook of Chemistry and Physics CRC Press 77th Ed (1996-7) 31 Billmann J G Kovacs and A Otto Surf Sci 92 153 (1980) 32 Murray CA and S Bodoff Phys Rev B 32 671 (1985) 33 Kellogg D and J Pemberton J Phys Chem 91 1120 (1987)
SPIE-2003-5269 16
Chemical agent detection by surface-enhanced Raman spectroscopy
Stuart Farquharson Alan Gift Paul Maksymiuk Frank Inscore and Wayne Smith
Real-Time Analyzers 87 Church Street East Hartford CT 06108
Kevin Morrisey and Steven D Christesen US Army SBCCOM Aberdeen Proving Ground MD 21010
ABSTRACT
In the past decade the Unites States and its allies have been challenged by a different kind of warfare exemplified by the terrorist attacks of September 11 2001 Although suicide bombings are the most often used form of terror military personnel must consider a wide range of attack scenarios Among these is the intentional poisoning of water supplies to obstruct military operations in Afghanistan and Iraq To counter such attacks the military is developing portable analyzers that can identify and quantify potential chemical agents in water supplies at microgram per liter concentrations within 10 minutes To aid this effort we have been investigating the value of a surface-enhanced Raman spectroscopy based portable analyzer In particular we have been developing silver-doped sol-gels to generate SER spectra of chemical agents and their hydrolysis products Here we present SER spectra of several chemical agents measured in a generic tap water Repeat measurements were performed to establish statistical error associated with SERS obtained using the sol-gel coated vials Keywords Chemical agents chemical agent detection SERS Raman spectroscopy
1 INTRODUCTION
In the autumn of 2001 terrorism within US borders became a sobering reality While extensive efforts are being implemented to secure the homeland US and Coalition forces in Afghanistan and Iraq are constantly faced with terrorist attacks In addition to car-bombings the military has identified several non-traditional attack scenarios including poisoning of water supplies by chemical warfare agents (CWAs) To counter this threat the Department of Defense has been investigating numerous approaches to detect such attacks under the auspices of the Joint Service Agent Water Monitor (JSAWM) program1 The JSAWM must be capable of identifying the chemical agents with no false-positives and quantifying the agents at microgram per liter concentrations within 10 minutes2 This includes the analysis of drinking water supplies distribution and storage systems Currently colorimetric paper is used to detect agents on-site while gas chromatography combined with mass spectrometry (GCMS) is used in mobile support laboratories However both methods have severe drawbacks The paper changes color in response to contact with many chemicals besides CWAs causing a high incidence of false positives3 while GCMS although very chemically specific requires up to an hour to perform and regular re-calibration456 Vibrational spectroscopy has also been investigated as a potential method of rapidly detecting CWAs7-
11 as early as 1985 in the case of infrared spectroscopy8 and 1988 in the case of Raman spectroscopy11 Again however these techniques also have limitations when it comes to measuring trace poisons in water Infrared spectra would be dominated by the very strong absorption of water which would obscure absorptions by most other chemicals present Whereas Raman spectroscopy is simply not a very sensitive technique and detection limits are typically grams per liter Surface-enhanced Raman spectroscopy (SERS) offers several advantages over conventional vibrational methods and that may provide the necessary sensitivity required for detecting trace quantities of chemical agents in water SERS employs the interaction of surface plasmon modes of metal particles with target analytes to increase scattering efficiency by as much as 1 million times12 In 1987 the potential of this approach to measure CWAs was demonstrated by measuring a series of pesticides13 Several of these organophosphonates have chemical structures similar to CWAs in particular P=O functional groups In the past few years we and others have further explored the ability of SERS to detect CWAs14-17 and even bioagents 18-21 We have been employing silver-doped sol-gels to promote the SER effect
SPIE-2003-5269 17
in these studies The porous silica network of the sol-gel matrix offers a unique environment for immobilizing and stabilizing SER-active metal particles22-25 The silver-doped sol-gels have been coated on the internal walls of standard glass vials to allow rapid SER analysis of small sample quantities without preparation We have measured over 100 chemicals with enhancements of 104 to 106 demonstrated reversible measurements in a flowing system reproducible measurements from vial-to-vial and measurements in multiple solvents including water21-25 Previously we used these vials to perform preliminary measurements of cyanide (CN) and methylphosphonic acid (MPA) Most of the nerve agents form MPA during hydrolysis while Tabun forms CN a chemical agent in its own right In another paper including in these proceedings we examined the limits of detection (LOD) for MPA by measuring a series of concentrations down to 50 mgL and estimated a limit of detection of 100 microgL26 These measurements provide encouragement in that SERS may satisfy the needs of the JSAWM To further establish the viability of SERS in particular silver-doped sol-gels here we present analysis of cyanide mustard and VX in tap water The measurements performed at the US Armyrsquos Edgewood Chemical Biological Center Aberdeen MD also included numerous repeat measurements to establish reproducibility
2 EXPERIMENTAL 2a General All chemicals including potassium cyanide 2-chloroethylethyl sulfide and those used to prepare the silver-doped and gold-doped sol-gel coated vials were obtained and used as received from Sigma-Aldrich All samples were prepared in a chemical hood using HPLC grade water unless otherwise noted (Fischer Scientific Fair Lawn NJ) for SERS measurements Once prepared the samples were transferred into the silver-doped sol-gel vials (Real-Time Analyzers Inc East Hartford CT) The vials were coated in a manner similar to that previously reported by adding ammonium hydroxide to a solution of silver nitrate tetramethyl orthosilicate (TMOS) and methanol22 The two precursor solutions were prepared mixed and transferred to 2-mL glass vials dried and heated After sol-gel formation the incorporated metal ions were reduced with dilute sodium borohydride (1mgmL) which is followed by a water wash to remove residual reducing agent After the resultant analyte solution was introduced the SER-active vial was fixed horizontally to an XY positioning stage (Conix Research Springfield OR) just inside the focal point of an f07 aspheric lens The lens focused the beam into the sample and collected the scattered radiation back along the same axis A dichroic filter (Omega Optical Brattleborough VT) was used to reflect the excitation laser to the lens and pass the Raman scattered radiation collected by the lens An f2 achromat was used to collimate the laser beam exiting a 200 microm core diameter source fiber optic while a second f2 achromat was used to focus the scattered radiation into a 365 microm fiber optic (Spectran Avon CT) A short pass filter was placed in the excitation beam path to block the silicon Raman scattering generated in the source fiber from reflecting off sampling optics and reaching the detector A long pass filter was placed in the collection beam path to block the sample Rayleigh scattering from reaching the detector A 785 nm diode laser (Process Instruments Inc model 785-600 Salt Lake City UT) was used to deliver 100 to 150 mW of power to the sample A Fourier transform Raman spectrometer (Real-Time Analyzers model IRA-785 East Hartford CT) and a silicon photo-avalanche detector (Perkin Elmer model C30902S Stamford CT) were used to acquire the SER spectra
2b Edgewood Chemical Biological Center The surface-enhanced Raman spectral measurements at the US Armyrsquos Chemical Biological center presented here were all performed on September 12 2003 To expedite measurements a plate was machined to hold up to 12 SER-active sample vials (Figure 1) The plate fit a standard XY plate reader that could be programmed Pure KCN bis-(2-chloroethyl)sulfide (distilled mustard gas HD) and ethyl S-2-diisopropylamino ethyl methylphosphonothioate (VX) were obtained on-site and used to prepare 1 gL tap water solutions in a chemical hood with appropriate safety equipment Simulated tap water was prepared by adding 10 chemicals most often found in tap water at appropriate concentrations to distilled water (Table 1) SER measurements were also performed in a chemical hood For added safety the FT-Raman instrument was placed outside the laboratory and 30 foot fiber optic and electrical cables were used to allow remote SERS measurements and plate manipulation For each experiment 1gL samples were prepared and added to 9 individual vials which were then loaded on the plate In some cases a tenth vial was included as a blank
Table 1 Generic Tap Water Composition
Compound mgL NaHCO3 100 CaSO4 27 MgSO4bull7H2O 67 NaNO3 1 Fulvic Acid 1 K2HPO4 07 KH2PO4 03 (NH4)2HSO4 001 NaCl 001 FeSO4 0001 pH 76-78
SPIE-2003-5269 18
A software program was written that allowed selecting the sequence that the vials were measured the number of positions along the length of the vials to measure (1 to 5) and the number of scans to co-add During sample analysis the program displayed the vial being analyzed the point being analyzed and the spectrum as it was being acquired Once all the data was collected a second software program was written to rapidly analyze the data The spectra collected for all the vials on a plate could be loaded at one time and then the spectra for each point could be displayed simultaneously or separately The user could then select the Raman peak to analyze in terms of peak height or area This was accomplished by selecting points on either side of the peak to define a baseline of zero The peak height or area could then be computed for all of the spectra loaded and then exported to a spreadsheet for statistical analysis
Figure 1 A) Vial Holder 6 slots to hold 2 vials each end-to-end B) Measurement Configuration Program user selects vials to measure sequence number of points per vial (1 to 5) and number of scans per point C) Spectral Acquisition Program shows spectrum being collected which vial and position D) Spectra Analysis Program user selects spectra to analyze by plate vial and point (s) as well as two wavenumbers defining the peak and the baseline to subtract The image is of 5 repeat measurements of 10 mgL KCN in generic tap water 16 sec each 100 mW of 785 nm
3 RESULTS AND DISCUSSION Raman and surface-enhanced Raman spectra were obtained for potassium cyanide bis-(2-chloroethyl)sulfide and ethyl S-2-diisopropylamino ethyl methylphosphonothioate representing three classes of chemical agents cyanides mustards and nerve agents respectively Spectra were also obtained for 2-chloroethyl ethyl sulfide (CEES) a structural analogue to HD which was included in the study to aid in assigning spectral bands KCN salt was used for cyanide experiments to avoid the increased hazards of handling HCN gas KCN completely dissolves in water forming its conjugate acid HCN according to its Ka of 615x10-1027 and at a concentration of 1 mgmL results in a pH 107 solution This is significant in that only CN- appears to interact sufficiently with silver to produce a SER spectrum and no spectral signal is observed below pH 726 Figure 2 shows the SER and normal Raman spectra for KCN The SERS spectrum of 1mgml KCN in water shows a single intense somewhat broad feature at
A B
C D
SPIE-2003-5269 19
2100 cm-1 assigned to the single CequivN stretch The band is much sharper in the normal Raman spectra of the solid KCN salt at 2074 cm-1 This band does broaden and shift to 2080 cm-1 in solution (not shown) However the observed SERS frequency is attributed to interaction with silver and low frequency mode at 135 cm-1 attributed to a Ag-CN stretch (not shown) supports this conclusion
cm-1 band to a C-S stretch but the authors concede that it is in fact more likely a C-Cl stretch28 It appears that the most intense Raman bands at 648 692 and 747 cm-1 shift to 620 660 and 720 cm-1 in the SER spectra and are tentatively assigned as above The width of these bands suggests that they overlap underlying spectral features Additional bands in the Raman spectra occur at 972 1034 1049 1263 1286 1423 1442 2865 2935 and 2960 cm-1 Corresponding bands occur in the SER spectra at 964 1015 1054 1286 1410 1447 2865 and 2935 cm-1 Most of these bands are associated with alkane modes specifically the bands at approximately 1040 cm-1 to a C-C stretch 1290 cm-1 to a CH2 in-phase twist 1440 cm-1 to a CH2 wag 2865 cm-1 to a symmetric CH2 stretch and 2965 cm-1 to an asymmetric CH2 stretch The Raman and SER spectra of sulfur mustard were measured at the Edgewood center (Figure 4) Both spectra are largely similar to CEES The C-Cl and C-S bands in the Raman spectrum of HD now occur at 640 655 700 739 and 760 cm-1 and are more resolved possibly due to the increased molecular symmetry Theoretical calculations indicate that the first three bands are due to C-Cl stretching modes and the latter two to C-S stretching modes28 Only the C-Cl bands maintain significant intensity in the SER spectra occurring at 624 and 643 cm-1 which is attributed to the
Prior to measurements of HD CEES was examined by Raman and SER spectroscopy (Figure 3) CEES also known as half-mustard is essentially identical to HD except one of the chlorine end atoms is replaced by a hydrogen atom Again although not as toxic as HD CEES is a blister agent and dilute aqueous samples were prepared using appropriate safety equipment Both the Raman and SER spectra of CEES are similar and dominated by bands between 600 and 800 cm-1 These are associated with C-Cl and C-S stretching modes which are tentatively assigned to 648 and 747 cm-1 in the Raman spectra respectively The shoulder at 630 cm-1 the overlapped band at 660 cm-1 and the strong band at 692 cm-1 could also be due to these modes or their asymmetric counterparts It is worth noting that theoretical calculations assign the 692
Figure 2 A) SER and B) NR spectra of KCN Conditions A) 1 mgml in tap water 100 mW of 785 nm at sample 1-min acquisition time B) solid 300 mW of 785 nm 5-min All spectra are 8 cm-1 resolution
A
B
Figure 3 A) SER and B) NR spectra of CEES Conditions A) 1 vv (10 mgml) in MeOH 100 mW of 785 nm 1-min acquisition time B) neat 300 mW of 785 nm 5-min
A
B
Cl-CH2-CH2-S-CH2-CH3
A
B
Figure 4 A) SER and B) NR of HD Conditions A) 1mgml in tap water B) pure both 100 mW of 785 nm 1-min
Cl-CH2-CH2-S-CH2-CH2-Cl
SPIE-2003-5269 20
expected strong interaction between chlorine and silver and adds support to the assignment of this band to a C-Cl stretch Weaker overlapping bands occur at 670 692 and 724 cm-1 the latter possibly due to C-S stretching modes Again the alkane modes are apparent in the normal Raman spectra of HD but only a broad feature at 1300 to 1450 cm-1 suggests CH2 contributions in the SER spectrum Although the observed bands in the VX spectrum have not been assigned (Figure 4) a computer generated Raman spectrum29 predicts many of the same features with surprising accuracy and are used here Two intense bands at 460 and 530 cm-1 closely match predicted bands at 463 and 546 cm-1 assigned to a CH3-P=O bend and a PO2CS wag Three highly overlapped bands occur at 694 745 and 771 cm-1 matching predicted bands at 713 730 and 760 cm-1 The first
Table 2 Measured SER peak heights for the CN stretch at 2100 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
Figure 5 A) SERS and B) NR spectra of VX Conditions A) 1 vv (10 mgml) in MeOH B) pure sample both 100 mW of 785 nm 1-min acquisition time
A
B
two have been assigned to a C-S stretch and CH2bend respectively while the latter has been attributed to either a P-C stretch or an O-C-C stretch Although the 745 cm-1 band may alternatively be assigned to a C-S stretch based on the previous measurements of CEES and HD The relatively intense bands at 890 1106 1218 1445 and 1465 cm-1 also match predicted bands at 880 1108 1216 1440 and 1464 cm-1 that are assigned to a C-C stretch CH3 rock N-C3 stretch various C-H3 bends and C-H bends respectively Both the computer generated and the measured spectra contain numerous other less intense bands One is worth mentioning A unique band appears at 370 cm-1 that is predicted at 368 cm-1 and corresponds to an O-P=O bend The surface-enhanced Raman spectrum of VX is also rich with spectral features It has the unique low frequency band at 370 cm-1 as well as a second band at 380 cm-1 that is assigned tothe S-P-O bend predicted in the normal Raman spectrum at 388 cm-1 Based on the measured and predicted normal Raman spectra the following SERS assignments are given 460 cm-1 to the CH3-P=O bend 544 cm-1 to the PO2CS wag 738 cm-1 to a C-S stretch (based on arguments above) 890 cm-1
to a C-C stretch 1101 cm-1 to a CH3 rock and 1456
cm-1 to a C-H bend The ability of SERS to measure chemical agents in water containing real-world chemical interferents was tested by using the generic tap water described in Table 1 The ability to reproduce measurements was accomplished by preparing three separate water stock solutions which were used to prepare three sample solutions each of 1 mgmL KCN in the generic tap water The samples defined as vials 1-9 were then measured at 4 points per vial in a semi-random fashion such that errors associated with stock solution preparation and errors associated with instrument drift could be identified No trends were apparent that signified such systematic errors Each spectrum collected consisted of 20 averaged scans taking 16 seconds at 8 cm-1 resolution The laser power at the sample was measured periodically during the day and it ranged from 102 to 105 mW spacccc
spot Vial 1 Vial 2 Vial 41 13754 13004 128192 13519 12692 12909
1 3 13541 12721 126394 13462 12648 12651
ave 13569 12766 12755Vial 6 Vial 3 Vial 5
1 11586 14068 10757 2 11236 14402 115122 3 11376 14546 11559
4 10894 11763 11214ave 11273 13695 11261
Vial 7 Vial8 Vial 91 11198 15804 11023
2 11243 15716 88753 3 11562 15216 11218
4 11655 15084 11077ave 11415 15455 10548
CN AVG STDEV ERR1 mgmL pk ht 12526 1560 1245
stock solution
SPIE-2003-5269 21
The CN measurements consisting of 36 data points produced an average height of 125 for the 2100 cm-1 peak with a standard deviation of 156 or 125 (Table 2) The HD and VX measurements were performed precisely the same way (Tables 3 and 4) For HD the 624 cm-1 peak was used for analysis and it had an average height of 53 with a standard deviation of 068 or 129 while for VX the 544 cm-1 peak was used for analysis and it had an average height of 1051 with a standard deviation of 308 or 293 The greater error in the VX measurements can be somewhat attributed to Vial 4 which produced lower SER signal intensities But removing this vial from the data set changes the standard deviation to 233 only a modest improvement
spot Vial 1 Vial 2 Vial 41 525 609 4682 45 675 484
1 3 527 644 5094 559 807 669
ave 51525 68375 5325Vial 6 Vial 3 Vial 5
1 521 574 575 2 536 53 4492 3 509 508 451
4 565 594 379ave 53275 5515 4635
Vial 7 Vial 8 Vial 91 457 589 413
2 583 497 4273 3 544 505 4
4 528 5 509ave 528 52275 43725
HD AVG STDEV ERR1 mgmL pk ht 530 068 1291
stock solution
spot Vial 1 Vial 2 Vial 41 1464 1034 4672 1485 99 61
1 3 1491 989 5684 1041 777 553
ave 137025 9475 5495Vial 6 Vial 3 Vial 5
1 1058 942 1293 2 697 121 965
2 3 727 1206 11124 689 1254 746
ave 79275 1153 1029Vial 7 Vial 8 Vial 9
1 1127 783 163 2 1358 812 1614
3 3 1371 908 15124 1097 875 1397
ave 123825 8445 153825
VX AVG STDEV ERR1 mgmL pk ht 1051 308 2925
stock solution
4 CONCLUSIONS In this paper we examined the ability of surface-enhanced Raman spectroscopy to reproducibly measure CN HD and VX in tap water without chemical interference Both normal and surface-enhanced Raman spectra were examined to select unique bands suitable to identify and quantify these chemical agents For SER measurements the 2100 cm-1 C-N stretch was used for CN the 624 cm-1 C-Cl stretch was used for HD and the 544 cm-1 PO2CS wag was used for VX It was determined that 1 mgmL samples of each of these chemicals measured 36 times in glass vials coated with a silver-doped sol-gel reproduced measurements with standard deviations of 125 129 and 293 It was further found that the 10 chemicals added to simulate generic tap water did not interfere with or alter the SER spectra It should be noted that the concentrations used in this study were considerably greater than those required by the JSAWM program Current work involves improving SER sensitivity and designing sampling systems with better reproducibility This includes the development of fractal silver and gold structures within the sol-gel matrix and the development of chemically selective sol-gels
ACKNOWLEDGMENTS The authors are grateful for the support of the US Army (DAAD13-02-C-0015 Joint Service Agent Water Monitor program) The authors would also like to thank Janet Jensen Ronald Crosier and Kristina Gonser for helpful discussions
Table 4 Measured SER peak heights for the VX band at 544 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
Table 3 Measured SER peak heights for the HD band at 624 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
SPIE-2003-5269 22
REFERENCES 1 Jensen JL NC Wong W R Loerop SPIE 4775-03 (2001) 2 JSAWM Requirements at wwwsbccomapgeaarmymilRDAecbcrtPRODSERJSAWMjsawmhtml 3 Erickson B ldquoThe Chemical Weapons Convention Redefines Analytical Challengerdquo Anal Chem 70 397A (1998) 4 Johnston RL Hoefler CM Fargo JC and Moberley B ldquoThe Defense Nuclear Agencyrsquos
ChemicalBiochemical Weapons Agreements Verification Technology Research Development Test and Evaluation Program and its Requirements for On-Site Analysisrdquo AT-ONSITE 5-8 (1994)
5 Black RM Clarke RJ Read RW and Reid MT ldquoApplication of GC-MS and GC-tandem MS to the analysis of chemical warfare samples found to contain residues of the nerve agent sarin sulphur mustard and their degradation productsrdquo J Chrom 662 301-321 (1994)
6 USA Army Center for Health Promotion and Preventive Medicine (USACHPPM) Technical Guideline 230A 7 Hoffland LD Piffath RJ Bouck JB rdquoSpectral signatures of chemical agents and simulantsrdquo Optical
Engineering 24 982-984 (1985) 8 Braue EHJ Pannella MGrdquoCIRCLE CELL FT-IR Analysis of Chemical Warfare Agents in Aqueous Solutionsrdquo
App Spectrosc 44 1513-1520 (1990) 9 Tseng C-H Mann CK and Vickers TJ ldquoDetermination of Organics on Metal Surfaces by Raman
Spectroscopyrdquo App Spectrosc 47 1767-1771 (1993) 10 Staff Reporter ldquoUS Army Tests Lidar to Detect Biological Toxinsrdquo Photonic Spectra pg 50 December 1998 11 Christesen SD Raman cross sections of chemical agents and simulants App Spectrosc 42 318-321 (1988) 12 Weaver MJ Farquharson S Tadayyoni MA ldquoSurface-enhancement factors for Raman scattering at silver electrodesrdquo J Chem Phys 82 4867-4874 (1985) 13 Alak AM and Vo-Dinh T ldquoSERS of Organophosphorous Chemical Agentsrdquo Anal Chem 59 2149-2153 (1987) 14 Lee Y Farquharson S ldquoRapid chemical agent identification by SERSrdquo SPIE 4378 21-26 (2001) 15 Farquharson S Maksymiuk P Ong K Christesen S ldquoChemical agent identification by surface-enhanced Raman spectroscopyrdquo SPIE 4577 166-173 (2001) 16 Spencer KM Sylvia J Clauson S and Janni J ldquoSurface Enhanced Raman as a Water Monitor for Warfare
Agents in Waterrdquo SPIE 4577 158-165 (2001) 17 Tessier P Christesen S Ong K Clemente E Lenhoff A Kaler E Velev O ldquoOn-line spectroscopic characterization of sodium cyanide with nanostructured gold surface-enhanced Raman spectroscopy substratesrdquo
App Spectrosc 56 1524-1530 (2002) 18 Farquharson S W W Smith S Elliott and J F Sperry Rapid biological agent identification by surface- enhanced Raman spectroscopy SPIE 3855110-116 (1999) 19 Farquharson S WW Smith YH Lee S Elliott and J F Sperry Detection of biological signatures A
comparison of electrolytic and metal-doped sol-gel surface-enhanced Raman media SPIE 4575 62-72 (2002) 20 Guzelian AA Sylvia J Janni J Clauson S and Spencer KM ldquoSERS of whole cell bacteria and trace levels of biological moleculesrdquo SPIE 4577 182-192 (2001) 21 Shende C Inscore F Gift A Maksymiuk P Farquharson S ldquoRapid dipicolinic acid extraction from Bacillus spores detected by surface-enhanced Raman spectroscopyrdquo App Spectrosc 58 accepted 22 Lee Y and Farquharson S ldquoSERS Sample Vials Based on Sol-Gel Process for Trace Pesticide Analysisrdquo SPIE
4206 140-146 (2000) 23 Farquharson S and Lee Y ldquoTrace Drug Analysis by SERSrdquo SPIE 4200-16 (2000) 24 Lee Y Farquharson S and Rainey P M Surface-Enhanced Raman Sensor for Trace Chemical Detection in
Water SPIE 3857 76-84 (1999) 25 Lee Y Farquharson S Kwong H and Shahriari M ldquoSol-Gel Chemical Sensor for Surface-Enhanced Raman
Spectroscopyrdquo SPIE 3537 252-260 (1998) 26 Farquharson S A Gift P Maksymiuk F Inscore and W Smith ldquopH dependence of methyl phosphonic acid
dipicolinic acid and cyanide by surface-enhanced Raman spectroscopyrdquo SPIE 5269-19 (2003) 27 Lide DR Ed Handbook of Chemistry and Physics CRC Press 77th Ed (1996-7) 28 Sosa C RJ Bartlett K KuBulat and WB Person ldquoA theoretical study of harmonic vibrational frequencies and
infrared intensities of XCH2CH2SCH2CH2X and XCH2CH2SH (= H Cl)rdquo J Phys Chem 93 577-588 (1993) 29 Hameka HF and JO Jenson ldquoComputer-generated predictions of the structure and the IR and Raman spectra of
VXrdquo ERDEC-TR-065 May 1993
- Chemagents Appendicespdf
-
- SPIE2001-4575-Bioagent-SERSapdf
-
- Detection of bioagent signatures A comparison of electrolytic and metal-doped sol-gel surface-enhanced Raman media
- Stuart Farquharson Wayne Smith and Yuan Lee
-
- Real-Time Analyzers 87 Church Street East Hartford CT 06108
-
- ABSTRACT
- 4 CONCLUSIONS
- 5 ACKNOWLEDGEMENTS
- 6 REFERENCES
-
SPIE -2003-5269 121
With the above analyses in mind a preliminary investigation of the SER spectral response for MPA (as MPA-) as a function of concentration was performed A single vial was used for these measurements beginning with 1 mgL followed by measurements of 10 100 and 1000 mgL In all cases the pH was ~7 and 3 positions around the vial were measured per concentration Since the 760 cm-1 band was not observed for concentrations of 1 or 10 mgL using 100 mW of 785 nm the laser power at the sample was raised to 200 mW beginning with the 10 mgL concentration Photo-degradation was largely avoided (and not observed) by exposing the sample for only 33 seconds per spectral acquisition Representative spectra for 10 100 and 1000 mgL are shown in Figure 6 while a plot of the 760 cm-1 band intensity as a function of concentration is shown in Figure 7 These values were also used to estimate limits of detection based on the signal-to-noise ratio (SN) of the 760 cm-1 band using 100 mW of 785 nm laser excitation and a 10-min acquisition time As summarized in the Figure 7 inset the lower the measured concentration the lower the predicted LOD Even if as estimated 210 microgL could be measured using the silver-doped sol-gel vials an improvement of a factor of 70 is still required to meet JSAWM goals of measuring 3 microgL in 10-minutes Similar to MPA DPA is a diprotic acid (pKa1 = 216 and pKa2 = 692) and variations in pH will effect the relative concentrations of DPA DPA- and DPA= and possibly the SER spectra and band intensities This could prove significant if an acid or base is used to denature anthrax spores with the goal of extracting and analyzing DPA The MPA pH study described above was mimicked for DPA except that the starting solution consisted of 20 mg of DPA in 20 mL HPLC grade water (60x10-3M MW = 1671) The initial solution had a pH of 245 which was made basic by dilute KOH to pHs of 355 433 487 559 1069 and 1166 SER spectra were recorded at each pH using 100 mW of 785 nm and a 44-sec acquisition time Next one drop of concentrated nitric acid was used to remake the solution acidic at a pH of 200 Again sequential pH measurements were performed at 383 510 735 and 822 The solution pH was made acidic a third time but to pH 219 171 then 135 Throughout this process no more than 20 drops of acid or base were added and therefore the concentration was diluted by no more than 10 Most of the spectral bands showed a minor decrease in intensity as a function of increasing pH values However the bands at 525 and 795 cm-1 showed the most dramatic changes which occurred at acid pH Figure 8 shows the SER spectra of DPA for the spectral region and pH range of interest The identity of the DPA species was determined by plotting the normalized peak intensities with the lowest value set to 0 and the highest to 0006 M as a function of pH and overlaying these values on a plot of the relative concentrations for DPA DPA- and DPA= as previously done for MPA (Figure 9) As can be seen the 525 cm-1 band clearly corresponds to DPA The correspondence of the 795 cm-1 band to this species is less clear as the band retains intensity until pH of 55 This can be attributed to contribution to the overlapping band at 810 cm-1 which does not change as a function of pH The fact that most bands are observed at all pHs suggest that the primary interaction with silver is through the ring nitrogen This is supported by the fact that the most intense band occurs at 1008 cm-1 attributed to a symmetric ring breathing mode and that this interaction has been characterized for pyridine in numerous papers27
0
1
2
3
4
5
6
7
0 200 400 600 800 1000 1200
MPA Concentration (mgL)
760
Ban
d In
tens
ity
Figure 6 SER spectra of MPA in water at A) 1000 B) 100 and C) 10 mgL Conditions pH of 7 silver-doped sol-gel coated vial 200 mW of 785 nm 33-sec 8 cm-1 resolution
Figure 7 Plot of SER intensity of 760 cm-1 band of MPA as a function of concentration using 200 mW of 785 nm Inset table includes average intensity LOD standard deviation and percent deviation for each concentration but for 100 mW and 10 min
conc (mgL) LOD ave stddev dev10 021 002 1021100 072 010 14301000 312 040 1280
A
B C
SPIE -2003-5269 122
The ability to detect dipicolinic acid (DPA) by SERS is demonstrated in Figure 10 Even at 1 mgL the primary bands are visible The intensity of the dominant ring breathing mode observed at 1008 cm-1 for DPA in water (pH = 25 to 55) is plotted as a function of concentration in Figure 11 These values were also used to estimate limits of detection based on the SN of the 1008 cm-1 band using 100 mW of 785 nm laser excitation and a 10-min acquisition time Again the lower the measured concentration the lower the predicted LOD (see Figure 11 inset) and detection of 160 microgL is possible Gastrointestinal anthrax requires significant more spores than inhalation anthrax28 and a limit of detection might be placed at 1 million spores in 1 liter of water or 10 microgL Since each spore contains ~10 CaDPA by weight29 a goal for DPA might be 1 microgL indicating that the present measurements must be improved by nearly two orders of magnitude Due to the increased hazards of handling HCN gas KCN salt was used for these experiments Nevertheless all sample preparations were performed in a chemical hood KCN completely dissolves in water but its conjugate acid HCN is formed and has a Ka of 615x10-1030 viz
HCN CN- + H+ pKa = 921 Reaction 4
Figure 8 SER spectra of 1 mgmL DPA as a function of pH Conditions 100 mW 785 nm 100 scans (44-sec)
Figure 9 Concentration dependence of DPA DPA- and DPA= as a function of pH for a 0006M sample Intensity of 525 () and 795 (diams) cm-1 bands from Fig 8 as a function of pH
pH
135 171 219 383
0
0001
0002
0003
0004
0005
0006
0007
0 2 4 6 8 10 12 14pH
Con
cent
ratio
n [M
]
DPADPA-DPA=795525
DPA DPA=DPA-
pK2 = 692pK1 = 216
795 525
0
05
1
15
2
25
0 200 400 600 800 1000 1200
DPA Concentration (mgL)
1008
Ban
d In
tens
ity
conc (mgL) lod-10min-100mw1 017
10 016100 103
1000 355
Figure 10 SER spectra of DPA in water at A) 1000 B) 100 C) 10 and D) 1 mgL Conditions pH of 25-55 silver-doped sol-gel coated vial 175 mW of 785 nm 1-min 8 cm-1 D) has been multiplied by x10 to make bands visible
A
B C
D
Figure 11 Plot of SER intensity of 1008 cm-1 band of DPA as a function of concentration using 175 mW of 785 nm Inset table includes LOD in mgL for each concentration but for 100 mW and 10 min
SPIE -2003-5269 123
Consequently the cyanide concentration must be determined for each initial KCN concentration Specifically the samples prepared with concentrations of 01 1 10 100 and 1000 mgL of KCN produced CN- concentrations of 63x10-3 033 69 89 and 964 mgL at pHs of 816 90 967 102 and 107 respectively The pH dependence for the HCN and CN- concentrations are shown in Figure 12 Thus as the amount of KCN added to the solution decreases so does the pH of the solution (becomes less basic) and according to Reaction 4 the relative amount of CN- to HCN also decreases For example in the preparation of a 01 mgL solution of KCN the pH is shifted from 7 for pure water to only 816 and only 63 of the starting material becomes CN- or 63x10-3 mgL In comparison for a solution of 1000 mgL the pH is shifted from 7 to 107 and 96 of the starting material becomes CN- This is significant since the cyanide ion is better able to adsorb onto the silver particles and become SERS active SER spectra of 10 100 and 1000 mgL of KCN samples are shown in Figure 13 and a corresponding plot of the 2100 cm-1 band intensity as a function of concentration is shown in Figure 14
000102030405060708091011
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14pH
Con
cent
ratio
n [m
gm
L]
pKa = 921
CN -HCN
The SER spectra of cyanide are dominated by the single CN stretch at 2100 cm-1 which occurs in normal Raman spectra of solutions at 2080 cm-1 However a low frequency mode occurs at 135 cm-1 due to a Ag-CN stretch (not shown) indicative of a strong surface interaction It is also observed that as the concentration decreases the CN stretch shifts to 2140 cm-1 This shift has been attribute to the formation of a tetrahedral Ag(CN)3
2- surface structure31 as well
Figure 14 Concentration dependence of KCN SERS measured under conditions in Fig 11 Concentrations are 1 01 and 001 mgml Intensities are measured for the CN stretch at 2100 cm-1 Inset table includes LOD in mgL for each concentration in Figs 13 and 15 but for 100 mW and 10 min
Figure 13 SER spectra of KCN in water at A) 1000 B) 100 and C) 10 mgL Conditions pHs of 107 102 and 97 silver-doped sol-gel coated vial 100 mW of 785 nm 1-min 8 cm-1 C) has been multiplied by x10 to make band visible
Figure 12 Concentration dependence of HCN and CN- as a function of pH for a 1 mgmL sample Calculated intensity of 2100 () cm-1 band for a 1 mgml sample at pHs of 816 90 967 102 and 107
A
B
C
0
20
40
60
80
100
120
140
0 200 400 600 800 1000 1200CN Concentration (mgL)
2100
Ban
d In
tens
ity
Figure 15 SER spectra of KCN in water at A) 10 B) 1 and C) 01 mgL Conditions pHs of 97 12 and 12 gold-doped sol-gel coated vial 100 mW of 785 nm 1-min 8 cm-1 C) has been multiplied by x10 to make band visible
A
B
C
conc (mgL) lod-10min-100mw Condition01 001 Au-pH 121 007 Au-pH 12
69 003 Au-pH 9769 016 Ag-pH 9789 022 Ag-pH 102946 113 Ag-pH 107
SPIE -2003-5269 124
as to CN adsorbed to two different surface sites32 Alternatively the 2140 cm-1 band could be attributed to HCN since this species dominates at lower concentrations However it is unlikely that this species would be attracted to the electropositive silver surface Further both peaks should be present at pHs between 85 and 105 but this is not observed It has also been suggested that at concentrations near and above monolayer coverage the CN- species is forced to adsorb end-on due to crowding and at lower concentrations the molecule can reorient to lie flat33 This suggests that the 2100 and 2140 cm-1 bands correspond to the end-on and flat orientations respectively As Figures 13 and 14 show the intensity of the CN stretch for the 89 mgL sample is nearly as intense as the 964 mgmL band This suggests that the Raman signal for the flat orientation is more enhanced However more extensive measurements are required to verify this point Since resent research has suggested that cyanide may be more effectively detected on gold measurements of KCN solutions were also performed using gold-doped sol-gel vials Preliminary measurements are shown in Figure 15 for samples prepared from 01 1 and 10 mgL KCN Since the pHs are 816 90 967 the resultant CN- concentrations are 63x10-3 033 and 69 mgL Initially only the highest concentration was observed and the signal intensity was significantly better than the equivalent concentration measured using silver In an effort to shift Reaction 4 to the left transforming HCN to CN- (Le Chatelierrsquos principle) KOH was added to the lower concentration samples producing solutions with pH 12 This effectively forces all of the cyanide in solution to be CN- or 01 and 10 mgL respectively More importantly the CN stretch is now observed in the SER spectra The band appears at 2125 cm-1 as has been previously reported for gold12 As calculated for MPA and DPA LODs can be estimated from this data For the three concentrations of cyanide on silver the LODs are 016 to 11 mgL for 100 mW of 785 nm laser excitation and a 10-min acquisition time For gold pH adjusted the LODs are10 to 70 microgL an improvement of more than 10 times silver Nevertheless either substrate is sufficient to meet the JSAWM goals of measuring 3 mgL in 10-minutes as the requirements form cyanide are much less stringent than the nerve agents
4 CONCLUSIONS Here we examined the surface-enhanced Raman spectral response for methyl phosphonic acid dipicolinic acid and cyanide as a function of pH It was determined that the most prominent peak at 760 cm-1 reaches a maximum intensity between pH 3 and 7 and corresponds to the MPA- species Neither the MPA nor MPA= species appear to generate a SER spectrum and consequently no spectra were observed below pH 2 or above pH 8 In this study we also found that higher laser powers could cause photodegradation of MPA signified by the exponential growth of a band at 1050 cm-1 which is tentatively assigned to phosphorous acid Unlike MPA DPA was observed at all pHs This is attributed to the dominant interaction of the pyridine functional group with silver Minor spectral changes were observed at acid pHs and were assigned to neutral DPA Like MPA SER spectra of cyanide were pH dependent No spectra were observed for the HCN species while CN- was best observed at pHs more basic than 8 Preliminary concentration studies for the three analytes allowed estimating limits of detection for MPA DPA and CN using 100 mW of 785 nm and a 10-min acquisition time of 210 165 and 70 microgL respectively Although the latter value suggests that cyanide can be measured at sufficiently low concentrations to meet JSAWM goal improvements by 100 to 200 times are required for MPA and DPA It is clear from this study that pH of the sample is important and must be taken into account when developing concentration calibration curves as well as fieldable analyzers Future work will include tailoring the sol-gel with specific functional groups to dictate sample pH and thereby optimize sensitivity
ACKNOWLEDGMENTS The authors are grateful for the support of the US Army (DAAD13-02-C-0015 Joint Service Agent Water Monitor program) The authors would also like to thank Dr Steve Christensen of the US Army SBCCOM for helpful discussions and Mr Chetan Shende of Real-Time Analyzers for assistance in development of the gold-doped sol-gels
REFERENCES 1 Erickson B Analytical Chemistry News amp Features June 1 397A (1998)
SPIE -2003-5269 125
2 Johnston RL Hoefler CM Fargo JC and Moberley B AT-ONSITE 5-8 (1994) 3 Black RM RJ Clarke RW Read and MT Reid J Chromatography 662 301-321 (1994) 4 USA Army Center for Health Promotion and Preventive Medicine (USACHPPM) Technical Guideline 230A 5 Jensen JL NC Wong W R Loerop SPIE 4775-03 (2001) 6 Hoenig SL Handbook of Chemical Warfare and Terrorism Greenwood Press Wesport CT (2002) 7 Munro NB SS Talmage GD Griffin LC Waters AP Watson JF King and V Hauschild Env Health
Persp 107 933-974 (1999) 8 Holstege CP Kirk M Sidell FR Crit Care Clin13 923-42 (1997) 9 Farquharson S P Maksymiuk K Ong and S Christesen SPIE 4577 166-173 (2001) 10 Lee Y and S Farquharson SPIE 4378 21-26 (2001) 11 Spencer KM J Sylvia S Clauson and J Janni SPIE 4577 158-165 (2001) 12 Tessier P S Christesen K Ong E Clemente A Lenhoff E Kaler and O Velev Applied Spectroscopy 56
1524-1530 (2002) 13 Farquharson S WW Smith S Elliott and JF Sperry SPIE 3533 207-214 (1998) 14 Farquharson S WW Smith S Elliott and JF Sperry SPIE 3855110-116 (1999) 15 Farquharson S WW Smith YH Lee S Elliott and JF Sperry SPIE 4575 62-72 (2002) 16 Guzelian AA J Sylvia J Janni S Clauson and KM Spencer SPIE 4577 182-192 (2001) 17 Shende C F Inscore A Gift P Maksymiuk and S Farquharson in press 18 Weaver MJ S Farquharson and MA Tadayyoni J Chem Phys 82 4867-4874 (1985) 19 Alak AM and T Vo-Dinh Analytical Chemistry 59 2149-2153 (1987) 20 Lee Y and S Farquharson SPIE 4206 140-146 (2000) 21 Farquharson S and Y Lee SPIE 4200-16 (2000) 22 Lee Y S Farquharson and P M Rainey SPIE 3857 76-84 (1999) 23 Lee Y S Farquharson H Kwong and M Shahriari SPIE 3537 252-260 (1998) 24 Wang J M Pumera G Collins and A Mulchandani Analytical Chemistry 74 6121-6125 (2002) 25 Dou X YM Jung Z-Q Cao and Y Ozaki Applied Spectroscopy 53 1440-1447 (1999) 26 Data supplied by S Christesen and K Ewing 27 Kerker M and B Thompson Eds SPIE MS 10 (1990) 28 Inglesby TV DA Henderson JG Bartlett JAMA 287 2236 (2002) 29 FW Janssen AJ Lund and LE Anderson Science 127 26 (1958) 30 Lide DR Ed Handbook of Chemistry and Physics CRC Press 77th Ed (1996-7) 31 Billmann J G Kovacs and A Otto Surf Sci 92 153 (1980) 32 Murray CA and S Bodoff Phys Rev B 32 671 (1985) 33 Kellogg D and J Pemberton J Phys Chem 91 1120 (1987)
SPIE-2003-5269 16
Chemical agent detection by surface-enhanced Raman spectroscopy
Stuart Farquharson Alan Gift Paul Maksymiuk Frank Inscore and Wayne Smith
Real-Time Analyzers 87 Church Street East Hartford CT 06108
Kevin Morrisey and Steven D Christesen US Army SBCCOM Aberdeen Proving Ground MD 21010
ABSTRACT
In the past decade the Unites States and its allies have been challenged by a different kind of warfare exemplified by the terrorist attacks of September 11 2001 Although suicide bombings are the most often used form of terror military personnel must consider a wide range of attack scenarios Among these is the intentional poisoning of water supplies to obstruct military operations in Afghanistan and Iraq To counter such attacks the military is developing portable analyzers that can identify and quantify potential chemical agents in water supplies at microgram per liter concentrations within 10 minutes To aid this effort we have been investigating the value of a surface-enhanced Raman spectroscopy based portable analyzer In particular we have been developing silver-doped sol-gels to generate SER spectra of chemical agents and their hydrolysis products Here we present SER spectra of several chemical agents measured in a generic tap water Repeat measurements were performed to establish statistical error associated with SERS obtained using the sol-gel coated vials Keywords Chemical agents chemical agent detection SERS Raman spectroscopy
1 INTRODUCTION
In the autumn of 2001 terrorism within US borders became a sobering reality While extensive efforts are being implemented to secure the homeland US and Coalition forces in Afghanistan and Iraq are constantly faced with terrorist attacks In addition to car-bombings the military has identified several non-traditional attack scenarios including poisoning of water supplies by chemical warfare agents (CWAs) To counter this threat the Department of Defense has been investigating numerous approaches to detect such attacks under the auspices of the Joint Service Agent Water Monitor (JSAWM) program1 The JSAWM must be capable of identifying the chemical agents with no false-positives and quantifying the agents at microgram per liter concentrations within 10 minutes2 This includes the analysis of drinking water supplies distribution and storage systems Currently colorimetric paper is used to detect agents on-site while gas chromatography combined with mass spectrometry (GCMS) is used in mobile support laboratories However both methods have severe drawbacks The paper changes color in response to contact with many chemicals besides CWAs causing a high incidence of false positives3 while GCMS although very chemically specific requires up to an hour to perform and regular re-calibration456 Vibrational spectroscopy has also been investigated as a potential method of rapidly detecting CWAs7-
11 as early as 1985 in the case of infrared spectroscopy8 and 1988 in the case of Raman spectroscopy11 Again however these techniques also have limitations when it comes to measuring trace poisons in water Infrared spectra would be dominated by the very strong absorption of water which would obscure absorptions by most other chemicals present Whereas Raman spectroscopy is simply not a very sensitive technique and detection limits are typically grams per liter Surface-enhanced Raman spectroscopy (SERS) offers several advantages over conventional vibrational methods and that may provide the necessary sensitivity required for detecting trace quantities of chemical agents in water SERS employs the interaction of surface plasmon modes of metal particles with target analytes to increase scattering efficiency by as much as 1 million times12 In 1987 the potential of this approach to measure CWAs was demonstrated by measuring a series of pesticides13 Several of these organophosphonates have chemical structures similar to CWAs in particular P=O functional groups In the past few years we and others have further explored the ability of SERS to detect CWAs14-17 and even bioagents 18-21 We have been employing silver-doped sol-gels to promote the SER effect
SPIE-2003-5269 17
in these studies The porous silica network of the sol-gel matrix offers a unique environment for immobilizing and stabilizing SER-active metal particles22-25 The silver-doped sol-gels have been coated on the internal walls of standard glass vials to allow rapid SER analysis of small sample quantities without preparation We have measured over 100 chemicals with enhancements of 104 to 106 demonstrated reversible measurements in a flowing system reproducible measurements from vial-to-vial and measurements in multiple solvents including water21-25 Previously we used these vials to perform preliminary measurements of cyanide (CN) and methylphosphonic acid (MPA) Most of the nerve agents form MPA during hydrolysis while Tabun forms CN a chemical agent in its own right In another paper including in these proceedings we examined the limits of detection (LOD) for MPA by measuring a series of concentrations down to 50 mgL and estimated a limit of detection of 100 microgL26 These measurements provide encouragement in that SERS may satisfy the needs of the JSAWM To further establish the viability of SERS in particular silver-doped sol-gels here we present analysis of cyanide mustard and VX in tap water The measurements performed at the US Armyrsquos Edgewood Chemical Biological Center Aberdeen MD also included numerous repeat measurements to establish reproducibility
2 EXPERIMENTAL 2a General All chemicals including potassium cyanide 2-chloroethylethyl sulfide and those used to prepare the silver-doped and gold-doped sol-gel coated vials were obtained and used as received from Sigma-Aldrich All samples were prepared in a chemical hood using HPLC grade water unless otherwise noted (Fischer Scientific Fair Lawn NJ) for SERS measurements Once prepared the samples were transferred into the silver-doped sol-gel vials (Real-Time Analyzers Inc East Hartford CT) The vials were coated in a manner similar to that previously reported by adding ammonium hydroxide to a solution of silver nitrate tetramethyl orthosilicate (TMOS) and methanol22 The two precursor solutions were prepared mixed and transferred to 2-mL glass vials dried and heated After sol-gel formation the incorporated metal ions were reduced with dilute sodium borohydride (1mgmL) which is followed by a water wash to remove residual reducing agent After the resultant analyte solution was introduced the SER-active vial was fixed horizontally to an XY positioning stage (Conix Research Springfield OR) just inside the focal point of an f07 aspheric lens The lens focused the beam into the sample and collected the scattered radiation back along the same axis A dichroic filter (Omega Optical Brattleborough VT) was used to reflect the excitation laser to the lens and pass the Raman scattered radiation collected by the lens An f2 achromat was used to collimate the laser beam exiting a 200 microm core diameter source fiber optic while a second f2 achromat was used to focus the scattered radiation into a 365 microm fiber optic (Spectran Avon CT) A short pass filter was placed in the excitation beam path to block the silicon Raman scattering generated in the source fiber from reflecting off sampling optics and reaching the detector A long pass filter was placed in the collection beam path to block the sample Rayleigh scattering from reaching the detector A 785 nm diode laser (Process Instruments Inc model 785-600 Salt Lake City UT) was used to deliver 100 to 150 mW of power to the sample A Fourier transform Raman spectrometer (Real-Time Analyzers model IRA-785 East Hartford CT) and a silicon photo-avalanche detector (Perkin Elmer model C30902S Stamford CT) were used to acquire the SER spectra
2b Edgewood Chemical Biological Center The surface-enhanced Raman spectral measurements at the US Armyrsquos Chemical Biological center presented here were all performed on September 12 2003 To expedite measurements a plate was machined to hold up to 12 SER-active sample vials (Figure 1) The plate fit a standard XY plate reader that could be programmed Pure KCN bis-(2-chloroethyl)sulfide (distilled mustard gas HD) and ethyl S-2-diisopropylamino ethyl methylphosphonothioate (VX) were obtained on-site and used to prepare 1 gL tap water solutions in a chemical hood with appropriate safety equipment Simulated tap water was prepared by adding 10 chemicals most often found in tap water at appropriate concentrations to distilled water (Table 1) SER measurements were also performed in a chemical hood For added safety the FT-Raman instrument was placed outside the laboratory and 30 foot fiber optic and electrical cables were used to allow remote SERS measurements and plate manipulation For each experiment 1gL samples were prepared and added to 9 individual vials which were then loaded on the plate In some cases a tenth vial was included as a blank
Table 1 Generic Tap Water Composition
Compound mgL NaHCO3 100 CaSO4 27 MgSO4bull7H2O 67 NaNO3 1 Fulvic Acid 1 K2HPO4 07 KH2PO4 03 (NH4)2HSO4 001 NaCl 001 FeSO4 0001 pH 76-78
SPIE-2003-5269 18
A software program was written that allowed selecting the sequence that the vials were measured the number of positions along the length of the vials to measure (1 to 5) and the number of scans to co-add During sample analysis the program displayed the vial being analyzed the point being analyzed and the spectrum as it was being acquired Once all the data was collected a second software program was written to rapidly analyze the data The spectra collected for all the vials on a plate could be loaded at one time and then the spectra for each point could be displayed simultaneously or separately The user could then select the Raman peak to analyze in terms of peak height or area This was accomplished by selecting points on either side of the peak to define a baseline of zero The peak height or area could then be computed for all of the spectra loaded and then exported to a spreadsheet for statistical analysis
Figure 1 A) Vial Holder 6 slots to hold 2 vials each end-to-end B) Measurement Configuration Program user selects vials to measure sequence number of points per vial (1 to 5) and number of scans per point C) Spectral Acquisition Program shows spectrum being collected which vial and position D) Spectra Analysis Program user selects spectra to analyze by plate vial and point (s) as well as two wavenumbers defining the peak and the baseline to subtract The image is of 5 repeat measurements of 10 mgL KCN in generic tap water 16 sec each 100 mW of 785 nm
3 RESULTS AND DISCUSSION Raman and surface-enhanced Raman spectra were obtained for potassium cyanide bis-(2-chloroethyl)sulfide and ethyl S-2-diisopropylamino ethyl methylphosphonothioate representing three classes of chemical agents cyanides mustards and nerve agents respectively Spectra were also obtained for 2-chloroethyl ethyl sulfide (CEES) a structural analogue to HD which was included in the study to aid in assigning spectral bands KCN salt was used for cyanide experiments to avoid the increased hazards of handling HCN gas KCN completely dissolves in water forming its conjugate acid HCN according to its Ka of 615x10-1027 and at a concentration of 1 mgmL results in a pH 107 solution This is significant in that only CN- appears to interact sufficiently with silver to produce a SER spectrum and no spectral signal is observed below pH 726 Figure 2 shows the SER and normal Raman spectra for KCN The SERS spectrum of 1mgml KCN in water shows a single intense somewhat broad feature at
A B
C D
SPIE-2003-5269 19
2100 cm-1 assigned to the single CequivN stretch The band is much sharper in the normal Raman spectra of the solid KCN salt at 2074 cm-1 This band does broaden and shift to 2080 cm-1 in solution (not shown) However the observed SERS frequency is attributed to interaction with silver and low frequency mode at 135 cm-1 attributed to a Ag-CN stretch (not shown) supports this conclusion
cm-1 band to a C-S stretch but the authors concede that it is in fact more likely a C-Cl stretch28 It appears that the most intense Raman bands at 648 692 and 747 cm-1 shift to 620 660 and 720 cm-1 in the SER spectra and are tentatively assigned as above The width of these bands suggests that they overlap underlying spectral features Additional bands in the Raman spectra occur at 972 1034 1049 1263 1286 1423 1442 2865 2935 and 2960 cm-1 Corresponding bands occur in the SER spectra at 964 1015 1054 1286 1410 1447 2865 and 2935 cm-1 Most of these bands are associated with alkane modes specifically the bands at approximately 1040 cm-1 to a C-C stretch 1290 cm-1 to a CH2 in-phase twist 1440 cm-1 to a CH2 wag 2865 cm-1 to a symmetric CH2 stretch and 2965 cm-1 to an asymmetric CH2 stretch The Raman and SER spectra of sulfur mustard were measured at the Edgewood center (Figure 4) Both spectra are largely similar to CEES The C-Cl and C-S bands in the Raman spectrum of HD now occur at 640 655 700 739 and 760 cm-1 and are more resolved possibly due to the increased molecular symmetry Theoretical calculations indicate that the first three bands are due to C-Cl stretching modes and the latter two to C-S stretching modes28 Only the C-Cl bands maintain significant intensity in the SER spectra occurring at 624 and 643 cm-1 which is attributed to the
Prior to measurements of HD CEES was examined by Raman and SER spectroscopy (Figure 3) CEES also known as half-mustard is essentially identical to HD except one of the chlorine end atoms is replaced by a hydrogen atom Again although not as toxic as HD CEES is a blister agent and dilute aqueous samples were prepared using appropriate safety equipment Both the Raman and SER spectra of CEES are similar and dominated by bands between 600 and 800 cm-1 These are associated with C-Cl and C-S stretching modes which are tentatively assigned to 648 and 747 cm-1 in the Raman spectra respectively The shoulder at 630 cm-1 the overlapped band at 660 cm-1 and the strong band at 692 cm-1 could also be due to these modes or their asymmetric counterparts It is worth noting that theoretical calculations assign the 692
Figure 2 A) SER and B) NR spectra of KCN Conditions A) 1 mgml in tap water 100 mW of 785 nm at sample 1-min acquisition time B) solid 300 mW of 785 nm 5-min All spectra are 8 cm-1 resolution
A
B
Figure 3 A) SER and B) NR spectra of CEES Conditions A) 1 vv (10 mgml) in MeOH 100 mW of 785 nm 1-min acquisition time B) neat 300 mW of 785 nm 5-min
A
B
Cl-CH2-CH2-S-CH2-CH3
A
B
Figure 4 A) SER and B) NR of HD Conditions A) 1mgml in tap water B) pure both 100 mW of 785 nm 1-min
Cl-CH2-CH2-S-CH2-CH2-Cl
SPIE-2003-5269 20
expected strong interaction between chlorine and silver and adds support to the assignment of this band to a C-Cl stretch Weaker overlapping bands occur at 670 692 and 724 cm-1 the latter possibly due to C-S stretching modes Again the alkane modes are apparent in the normal Raman spectra of HD but only a broad feature at 1300 to 1450 cm-1 suggests CH2 contributions in the SER spectrum Although the observed bands in the VX spectrum have not been assigned (Figure 4) a computer generated Raman spectrum29 predicts many of the same features with surprising accuracy and are used here Two intense bands at 460 and 530 cm-1 closely match predicted bands at 463 and 546 cm-1 assigned to a CH3-P=O bend and a PO2CS wag Three highly overlapped bands occur at 694 745 and 771 cm-1 matching predicted bands at 713 730 and 760 cm-1 The first
Table 2 Measured SER peak heights for the CN stretch at 2100 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
Figure 5 A) SERS and B) NR spectra of VX Conditions A) 1 vv (10 mgml) in MeOH B) pure sample both 100 mW of 785 nm 1-min acquisition time
A
B
two have been assigned to a C-S stretch and CH2bend respectively while the latter has been attributed to either a P-C stretch or an O-C-C stretch Although the 745 cm-1 band may alternatively be assigned to a C-S stretch based on the previous measurements of CEES and HD The relatively intense bands at 890 1106 1218 1445 and 1465 cm-1 also match predicted bands at 880 1108 1216 1440 and 1464 cm-1 that are assigned to a C-C stretch CH3 rock N-C3 stretch various C-H3 bends and C-H bends respectively Both the computer generated and the measured spectra contain numerous other less intense bands One is worth mentioning A unique band appears at 370 cm-1 that is predicted at 368 cm-1 and corresponds to an O-P=O bend The surface-enhanced Raman spectrum of VX is also rich with spectral features It has the unique low frequency band at 370 cm-1 as well as a second band at 380 cm-1 that is assigned tothe S-P-O bend predicted in the normal Raman spectrum at 388 cm-1 Based on the measured and predicted normal Raman spectra the following SERS assignments are given 460 cm-1 to the CH3-P=O bend 544 cm-1 to the PO2CS wag 738 cm-1 to a C-S stretch (based on arguments above) 890 cm-1
to a C-C stretch 1101 cm-1 to a CH3 rock and 1456
cm-1 to a C-H bend The ability of SERS to measure chemical agents in water containing real-world chemical interferents was tested by using the generic tap water described in Table 1 The ability to reproduce measurements was accomplished by preparing three separate water stock solutions which were used to prepare three sample solutions each of 1 mgmL KCN in the generic tap water The samples defined as vials 1-9 were then measured at 4 points per vial in a semi-random fashion such that errors associated with stock solution preparation and errors associated with instrument drift could be identified No trends were apparent that signified such systematic errors Each spectrum collected consisted of 20 averaged scans taking 16 seconds at 8 cm-1 resolution The laser power at the sample was measured periodically during the day and it ranged from 102 to 105 mW spacccc
spot Vial 1 Vial 2 Vial 41 13754 13004 128192 13519 12692 12909
1 3 13541 12721 126394 13462 12648 12651
ave 13569 12766 12755Vial 6 Vial 3 Vial 5
1 11586 14068 10757 2 11236 14402 115122 3 11376 14546 11559
4 10894 11763 11214ave 11273 13695 11261
Vial 7 Vial8 Vial 91 11198 15804 11023
2 11243 15716 88753 3 11562 15216 11218
4 11655 15084 11077ave 11415 15455 10548
CN AVG STDEV ERR1 mgmL pk ht 12526 1560 1245
stock solution
SPIE-2003-5269 21
The CN measurements consisting of 36 data points produced an average height of 125 for the 2100 cm-1 peak with a standard deviation of 156 or 125 (Table 2) The HD and VX measurements were performed precisely the same way (Tables 3 and 4) For HD the 624 cm-1 peak was used for analysis and it had an average height of 53 with a standard deviation of 068 or 129 while for VX the 544 cm-1 peak was used for analysis and it had an average height of 1051 with a standard deviation of 308 or 293 The greater error in the VX measurements can be somewhat attributed to Vial 4 which produced lower SER signal intensities But removing this vial from the data set changes the standard deviation to 233 only a modest improvement
spot Vial 1 Vial 2 Vial 41 525 609 4682 45 675 484
1 3 527 644 5094 559 807 669
ave 51525 68375 5325Vial 6 Vial 3 Vial 5
1 521 574 575 2 536 53 4492 3 509 508 451
4 565 594 379ave 53275 5515 4635
Vial 7 Vial 8 Vial 91 457 589 413
2 583 497 4273 3 544 505 4
4 528 5 509ave 528 52275 43725
HD AVG STDEV ERR1 mgmL pk ht 530 068 1291
stock solution
spot Vial 1 Vial 2 Vial 41 1464 1034 4672 1485 99 61
1 3 1491 989 5684 1041 777 553
ave 137025 9475 5495Vial 6 Vial 3 Vial 5
1 1058 942 1293 2 697 121 965
2 3 727 1206 11124 689 1254 746
ave 79275 1153 1029Vial 7 Vial 8 Vial 9
1 1127 783 163 2 1358 812 1614
3 3 1371 908 15124 1097 875 1397
ave 123825 8445 153825
VX AVG STDEV ERR1 mgmL pk ht 1051 308 2925
stock solution
4 CONCLUSIONS In this paper we examined the ability of surface-enhanced Raman spectroscopy to reproducibly measure CN HD and VX in tap water without chemical interference Both normal and surface-enhanced Raman spectra were examined to select unique bands suitable to identify and quantify these chemical agents For SER measurements the 2100 cm-1 C-N stretch was used for CN the 624 cm-1 C-Cl stretch was used for HD and the 544 cm-1 PO2CS wag was used for VX It was determined that 1 mgmL samples of each of these chemicals measured 36 times in glass vials coated with a silver-doped sol-gel reproduced measurements with standard deviations of 125 129 and 293 It was further found that the 10 chemicals added to simulate generic tap water did not interfere with or alter the SER spectra It should be noted that the concentrations used in this study were considerably greater than those required by the JSAWM program Current work involves improving SER sensitivity and designing sampling systems with better reproducibility This includes the development of fractal silver and gold structures within the sol-gel matrix and the development of chemically selective sol-gels
ACKNOWLEDGMENTS The authors are grateful for the support of the US Army (DAAD13-02-C-0015 Joint Service Agent Water Monitor program) The authors would also like to thank Janet Jensen Ronald Crosier and Kristina Gonser for helpful discussions
Table 4 Measured SER peak heights for the VX band at 544 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
Table 3 Measured SER peak heights for the HD band at 624 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
SPIE-2003-5269 22
REFERENCES 1 Jensen JL NC Wong W R Loerop SPIE 4775-03 (2001) 2 JSAWM Requirements at wwwsbccomapgeaarmymilRDAecbcrtPRODSERJSAWMjsawmhtml 3 Erickson B ldquoThe Chemical Weapons Convention Redefines Analytical Challengerdquo Anal Chem 70 397A (1998) 4 Johnston RL Hoefler CM Fargo JC and Moberley B ldquoThe Defense Nuclear Agencyrsquos
ChemicalBiochemical Weapons Agreements Verification Technology Research Development Test and Evaluation Program and its Requirements for On-Site Analysisrdquo AT-ONSITE 5-8 (1994)
5 Black RM Clarke RJ Read RW and Reid MT ldquoApplication of GC-MS and GC-tandem MS to the analysis of chemical warfare samples found to contain residues of the nerve agent sarin sulphur mustard and their degradation productsrdquo J Chrom 662 301-321 (1994)
6 USA Army Center for Health Promotion and Preventive Medicine (USACHPPM) Technical Guideline 230A 7 Hoffland LD Piffath RJ Bouck JB rdquoSpectral signatures of chemical agents and simulantsrdquo Optical
Engineering 24 982-984 (1985) 8 Braue EHJ Pannella MGrdquoCIRCLE CELL FT-IR Analysis of Chemical Warfare Agents in Aqueous Solutionsrdquo
App Spectrosc 44 1513-1520 (1990) 9 Tseng C-H Mann CK and Vickers TJ ldquoDetermination of Organics on Metal Surfaces by Raman
Spectroscopyrdquo App Spectrosc 47 1767-1771 (1993) 10 Staff Reporter ldquoUS Army Tests Lidar to Detect Biological Toxinsrdquo Photonic Spectra pg 50 December 1998 11 Christesen SD Raman cross sections of chemical agents and simulants App Spectrosc 42 318-321 (1988) 12 Weaver MJ Farquharson S Tadayyoni MA ldquoSurface-enhancement factors for Raman scattering at silver electrodesrdquo J Chem Phys 82 4867-4874 (1985) 13 Alak AM and Vo-Dinh T ldquoSERS of Organophosphorous Chemical Agentsrdquo Anal Chem 59 2149-2153 (1987) 14 Lee Y Farquharson S ldquoRapid chemical agent identification by SERSrdquo SPIE 4378 21-26 (2001) 15 Farquharson S Maksymiuk P Ong K Christesen S ldquoChemical agent identification by surface-enhanced Raman spectroscopyrdquo SPIE 4577 166-173 (2001) 16 Spencer KM Sylvia J Clauson S and Janni J ldquoSurface Enhanced Raman as a Water Monitor for Warfare
Agents in Waterrdquo SPIE 4577 158-165 (2001) 17 Tessier P Christesen S Ong K Clemente E Lenhoff A Kaler E Velev O ldquoOn-line spectroscopic characterization of sodium cyanide with nanostructured gold surface-enhanced Raman spectroscopy substratesrdquo
App Spectrosc 56 1524-1530 (2002) 18 Farquharson S W W Smith S Elliott and J F Sperry Rapid biological agent identification by surface- enhanced Raman spectroscopy SPIE 3855110-116 (1999) 19 Farquharson S WW Smith YH Lee S Elliott and J F Sperry Detection of biological signatures A
comparison of electrolytic and metal-doped sol-gel surface-enhanced Raman media SPIE 4575 62-72 (2002) 20 Guzelian AA Sylvia J Janni J Clauson S and Spencer KM ldquoSERS of whole cell bacteria and trace levels of biological moleculesrdquo SPIE 4577 182-192 (2001) 21 Shende C Inscore F Gift A Maksymiuk P Farquharson S ldquoRapid dipicolinic acid extraction from Bacillus spores detected by surface-enhanced Raman spectroscopyrdquo App Spectrosc 58 accepted 22 Lee Y and Farquharson S ldquoSERS Sample Vials Based on Sol-Gel Process for Trace Pesticide Analysisrdquo SPIE
4206 140-146 (2000) 23 Farquharson S and Lee Y ldquoTrace Drug Analysis by SERSrdquo SPIE 4200-16 (2000) 24 Lee Y Farquharson S and Rainey P M Surface-Enhanced Raman Sensor for Trace Chemical Detection in
Water SPIE 3857 76-84 (1999) 25 Lee Y Farquharson S Kwong H and Shahriari M ldquoSol-Gel Chemical Sensor for Surface-Enhanced Raman
Spectroscopyrdquo SPIE 3537 252-260 (1998) 26 Farquharson S A Gift P Maksymiuk F Inscore and W Smith ldquopH dependence of methyl phosphonic acid
dipicolinic acid and cyanide by surface-enhanced Raman spectroscopyrdquo SPIE 5269-19 (2003) 27 Lide DR Ed Handbook of Chemistry and Physics CRC Press 77th Ed (1996-7) 28 Sosa C RJ Bartlett K KuBulat and WB Person ldquoA theoretical study of harmonic vibrational frequencies and
infrared intensities of XCH2CH2SCH2CH2X and XCH2CH2SH (= H Cl)rdquo J Phys Chem 93 577-588 (1993) 29 Hameka HF and JO Jenson ldquoComputer-generated predictions of the structure and the IR and Raman spectra of
VXrdquo ERDEC-TR-065 May 1993
- Chemagents Appendicespdf
-
- SPIE2001-4575-Bioagent-SERSapdf
-
- Detection of bioagent signatures A comparison of electrolytic and metal-doped sol-gel surface-enhanced Raman media
- Stuart Farquharson Wayne Smith and Yuan Lee
-
- Real-Time Analyzers 87 Church Street East Hartford CT 06108
-
- ABSTRACT
- 4 CONCLUSIONS
- 5 ACKNOWLEDGEMENTS
- 6 REFERENCES
-
SPIE -2003-5269 122
The ability to detect dipicolinic acid (DPA) by SERS is demonstrated in Figure 10 Even at 1 mgL the primary bands are visible The intensity of the dominant ring breathing mode observed at 1008 cm-1 for DPA in water (pH = 25 to 55) is plotted as a function of concentration in Figure 11 These values were also used to estimate limits of detection based on the SN of the 1008 cm-1 band using 100 mW of 785 nm laser excitation and a 10-min acquisition time Again the lower the measured concentration the lower the predicted LOD (see Figure 11 inset) and detection of 160 microgL is possible Gastrointestinal anthrax requires significant more spores than inhalation anthrax28 and a limit of detection might be placed at 1 million spores in 1 liter of water or 10 microgL Since each spore contains ~10 CaDPA by weight29 a goal for DPA might be 1 microgL indicating that the present measurements must be improved by nearly two orders of magnitude Due to the increased hazards of handling HCN gas KCN salt was used for these experiments Nevertheless all sample preparations were performed in a chemical hood KCN completely dissolves in water but its conjugate acid HCN is formed and has a Ka of 615x10-1030 viz
HCN CN- + H+ pKa = 921 Reaction 4
Figure 8 SER spectra of 1 mgmL DPA as a function of pH Conditions 100 mW 785 nm 100 scans (44-sec)
Figure 9 Concentration dependence of DPA DPA- and DPA= as a function of pH for a 0006M sample Intensity of 525 () and 795 (diams) cm-1 bands from Fig 8 as a function of pH
pH
135 171 219 383
0
0001
0002
0003
0004
0005
0006
0007
0 2 4 6 8 10 12 14pH
Con
cent
ratio
n [M
]
DPADPA-DPA=795525
DPA DPA=DPA-
pK2 = 692pK1 = 216
795 525
0
05
1
15
2
25
0 200 400 600 800 1000 1200
DPA Concentration (mgL)
1008
Ban
d In
tens
ity
conc (mgL) lod-10min-100mw1 017
10 016100 103
1000 355
Figure 10 SER spectra of DPA in water at A) 1000 B) 100 C) 10 and D) 1 mgL Conditions pH of 25-55 silver-doped sol-gel coated vial 175 mW of 785 nm 1-min 8 cm-1 D) has been multiplied by x10 to make bands visible
A
B C
D
Figure 11 Plot of SER intensity of 1008 cm-1 band of DPA as a function of concentration using 175 mW of 785 nm Inset table includes LOD in mgL for each concentration but for 100 mW and 10 min
SPIE -2003-5269 123
Consequently the cyanide concentration must be determined for each initial KCN concentration Specifically the samples prepared with concentrations of 01 1 10 100 and 1000 mgL of KCN produced CN- concentrations of 63x10-3 033 69 89 and 964 mgL at pHs of 816 90 967 102 and 107 respectively The pH dependence for the HCN and CN- concentrations are shown in Figure 12 Thus as the amount of KCN added to the solution decreases so does the pH of the solution (becomes less basic) and according to Reaction 4 the relative amount of CN- to HCN also decreases For example in the preparation of a 01 mgL solution of KCN the pH is shifted from 7 for pure water to only 816 and only 63 of the starting material becomes CN- or 63x10-3 mgL In comparison for a solution of 1000 mgL the pH is shifted from 7 to 107 and 96 of the starting material becomes CN- This is significant since the cyanide ion is better able to adsorb onto the silver particles and become SERS active SER spectra of 10 100 and 1000 mgL of KCN samples are shown in Figure 13 and a corresponding plot of the 2100 cm-1 band intensity as a function of concentration is shown in Figure 14
000102030405060708091011
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14pH
Con
cent
ratio
n [m
gm
L]
pKa = 921
CN -HCN
The SER spectra of cyanide are dominated by the single CN stretch at 2100 cm-1 which occurs in normal Raman spectra of solutions at 2080 cm-1 However a low frequency mode occurs at 135 cm-1 due to a Ag-CN stretch (not shown) indicative of a strong surface interaction It is also observed that as the concentration decreases the CN stretch shifts to 2140 cm-1 This shift has been attribute to the formation of a tetrahedral Ag(CN)3
2- surface structure31 as well
Figure 14 Concentration dependence of KCN SERS measured under conditions in Fig 11 Concentrations are 1 01 and 001 mgml Intensities are measured for the CN stretch at 2100 cm-1 Inset table includes LOD in mgL for each concentration in Figs 13 and 15 but for 100 mW and 10 min
Figure 13 SER spectra of KCN in water at A) 1000 B) 100 and C) 10 mgL Conditions pHs of 107 102 and 97 silver-doped sol-gel coated vial 100 mW of 785 nm 1-min 8 cm-1 C) has been multiplied by x10 to make band visible
Figure 12 Concentration dependence of HCN and CN- as a function of pH for a 1 mgmL sample Calculated intensity of 2100 () cm-1 band for a 1 mgml sample at pHs of 816 90 967 102 and 107
A
B
C
0
20
40
60
80
100
120
140
0 200 400 600 800 1000 1200CN Concentration (mgL)
2100
Ban
d In
tens
ity
Figure 15 SER spectra of KCN in water at A) 10 B) 1 and C) 01 mgL Conditions pHs of 97 12 and 12 gold-doped sol-gel coated vial 100 mW of 785 nm 1-min 8 cm-1 C) has been multiplied by x10 to make band visible
A
B
C
conc (mgL) lod-10min-100mw Condition01 001 Au-pH 121 007 Au-pH 12
69 003 Au-pH 9769 016 Ag-pH 9789 022 Ag-pH 102946 113 Ag-pH 107
SPIE -2003-5269 124
as to CN adsorbed to two different surface sites32 Alternatively the 2140 cm-1 band could be attributed to HCN since this species dominates at lower concentrations However it is unlikely that this species would be attracted to the electropositive silver surface Further both peaks should be present at pHs between 85 and 105 but this is not observed It has also been suggested that at concentrations near and above monolayer coverage the CN- species is forced to adsorb end-on due to crowding and at lower concentrations the molecule can reorient to lie flat33 This suggests that the 2100 and 2140 cm-1 bands correspond to the end-on and flat orientations respectively As Figures 13 and 14 show the intensity of the CN stretch for the 89 mgL sample is nearly as intense as the 964 mgmL band This suggests that the Raman signal for the flat orientation is more enhanced However more extensive measurements are required to verify this point Since resent research has suggested that cyanide may be more effectively detected on gold measurements of KCN solutions were also performed using gold-doped sol-gel vials Preliminary measurements are shown in Figure 15 for samples prepared from 01 1 and 10 mgL KCN Since the pHs are 816 90 967 the resultant CN- concentrations are 63x10-3 033 and 69 mgL Initially only the highest concentration was observed and the signal intensity was significantly better than the equivalent concentration measured using silver In an effort to shift Reaction 4 to the left transforming HCN to CN- (Le Chatelierrsquos principle) KOH was added to the lower concentration samples producing solutions with pH 12 This effectively forces all of the cyanide in solution to be CN- or 01 and 10 mgL respectively More importantly the CN stretch is now observed in the SER spectra The band appears at 2125 cm-1 as has been previously reported for gold12 As calculated for MPA and DPA LODs can be estimated from this data For the three concentrations of cyanide on silver the LODs are 016 to 11 mgL for 100 mW of 785 nm laser excitation and a 10-min acquisition time For gold pH adjusted the LODs are10 to 70 microgL an improvement of more than 10 times silver Nevertheless either substrate is sufficient to meet the JSAWM goals of measuring 3 mgL in 10-minutes as the requirements form cyanide are much less stringent than the nerve agents
4 CONCLUSIONS Here we examined the surface-enhanced Raman spectral response for methyl phosphonic acid dipicolinic acid and cyanide as a function of pH It was determined that the most prominent peak at 760 cm-1 reaches a maximum intensity between pH 3 and 7 and corresponds to the MPA- species Neither the MPA nor MPA= species appear to generate a SER spectrum and consequently no spectra were observed below pH 2 or above pH 8 In this study we also found that higher laser powers could cause photodegradation of MPA signified by the exponential growth of a band at 1050 cm-1 which is tentatively assigned to phosphorous acid Unlike MPA DPA was observed at all pHs This is attributed to the dominant interaction of the pyridine functional group with silver Minor spectral changes were observed at acid pHs and were assigned to neutral DPA Like MPA SER spectra of cyanide were pH dependent No spectra were observed for the HCN species while CN- was best observed at pHs more basic than 8 Preliminary concentration studies for the three analytes allowed estimating limits of detection for MPA DPA and CN using 100 mW of 785 nm and a 10-min acquisition time of 210 165 and 70 microgL respectively Although the latter value suggests that cyanide can be measured at sufficiently low concentrations to meet JSAWM goal improvements by 100 to 200 times are required for MPA and DPA It is clear from this study that pH of the sample is important and must be taken into account when developing concentration calibration curves as well as fieldable analyzers Future work will include tailoring the sol-gel with specific functional groups to dictate sample pH and thereby optimize sensitivity
ACKNOWLEDGMENTS The authors are grateful for the support of the US Army (DAAD13-02-C-0015 Joint Service Agent Water Monitor program) The authors would also like to thank Dr Steve Christensen of the US Army SBCCOM for helpful discussions and Mr Chetan Shende of Real-Time Analyzers for assistance in development of the gold-doped sol-gels
REFERENCES 1 Erickson B Analytical Chemistry News amp Features June 1 397A (1998)
SPIE -2003-5269 125
2 Johnston RL Hoefler CM Fargo JC and Moberley B AT-ONSITE 5-8 (1994) 3 Black RM RJ Clarke RW Read and MT Reid J Chromatography 662 301-321 (1994) 4 USA Army Center for Health Promotion and Preventive Medicine (USACHPPM) Technical Guideline 230A 5 Jensen JL NC Wong W R Loerop SPIE 4775-03 (2001) 6 Hoenig SL Handbook of Chemical Warfare and Terrorism Greenwood Press Wesport CT (2002) 7 Munro NB SS Talmage GD Griffin LC Waters AP Watson JF King and V Hauschild Env Health
Persp 107 933-974 (1999) 8 Holstege CP Kirk M Sidell FR Crit Care Clin13 923-42 (1997) 9 Farquharson S P Maksymiuk K Ong and S Christesen SPIE 4577 166-173 (2001) 10 Lee Y and S Farquharson SPIE 4378 21-26 (2001) 11 Spencer KM J Sylvia S Clauson and J Janni SPIE 4577 158-165 (2001) 12 Tessier P S Christesen K Ong E Clemente A Lenhoff E Kaler and O Velev Applied Spectroscopy 56
1524-1530 (2002) 13 Farquharson S WW Smith S Elliott and JF Sperry SPIE 3533 207-214 (1998) 14 Farquharson S WW Smith S Elliott and JF Sperry SPIE 3855110-116 (1999) 15 Farquharson S WW Smith YH Lee S Elliott and JF Sperry SPIE 4575 62-72 (2002) 16 Guzelian AA J Sylvia J Janni S Clauson and KM Spencer SPIE 4577 182-192 (2001) 17 Shende C F Inscore A Gift P Maksymiuk and S Farquharson in press 18 Weaver MJ S Farquharson and MA Tadayyoni J Chem Phys 82 4867-4874 (1985) 19 Alak AM and T Vo-Dinh Analytical Chemistry 59 2149-2153 (1987) 20 Lee Y and S Farquharson SPIE 4206 140-146 (2000) 21 Farquharson S and Y Lee SPIE 4200-16 (2000) 22 Lee Y S Farquharson and P M Rainey SPIE 3857 76-84 (1999) 23 Lee Y S Farquharson H Kwong and M Shahriari SPIE 3537 252-260 (1998) 24 Wang J M Pumera G Collins and A Mulchandani Analytical Chemistry 74 6121-6125 (2002) 25 Dou X YM Jung Z-Q Cao and Y Ozaki Applied Spectroscopy 53 1440-1447 (1999) 26 Data supplied by S Christesen and K Ewing 27 Kerker M and B Thompson Eds SPIE MS 10 (1990) 28 Inglesby TV DA Henderson JG Bartlett JAMA 287 2236 (2002) 29 FW Janssen AJ Lund and LE Anderson Science 127 26 (1958) 30 Lide DR Ed Handbook of Chemistry and Physics CRC Press 77th Ed (1996-7) 31 Billmann J G Kovacs and A Otto Surf Sci 92 153 (1980) 32 Murray CA and S Bodoff Phys Rev B 32 671 (1985) 33 Kellogg D and J Pemberton J Phys Chem 91 1120 (1987)
SPIE-2003-5269 16
Chemical agent detection by surface-enhanced Raman spectroscopy
Stuart Farquharson Alan Gift Paul Maksymiuk Frank Inscore and Wayne Smith
Real-Time Analyzers 87 Church Street East Hartford CT 06108
Kevin Morrisey and Steven D Christesen US Army SBCCOM Aberdeen Proving Ground MD 21010
ABSTRACT
In the past decade the Unites States and its allies have been challenged by a different kind of warfare exemplified by the terrorist attacks of September 11 2001 Although suicide bombings are the most often used form of terror military personnel must consider a wide range of attack scenarios Among these is the intentional poisoning of water supplies to obstruct military operations in Afghanistan and Iraq To counter such attacks the military is developing portable analyzers that can identify and quantify potential chemical agents in water supplies at microgram per liter concentrations within 10 minutes To aid this effort we have been investigating the value of a surface-enhanced Raman spectroscopy based portable analyzer In particular we have been developing silver-doped sol-gels to generate SER spectra of chemical agents and their hydrolysis products Here we present SER spectra of several chemical agents measured in a generic tap water Repeat measurements were performed to establish statistical error associated with SERS obtained using the sol-gel coated vials Keywords Chemical agents chemical agent detection SERS Raman spectroscopy
1 INTRODUCTION
In the autumn of 2001 terrorism within US borders became a sobering reality While extensive efforts are being implemented to secure the homeland US and Coalition forces in Afghanistan and Iraq are constantly faced with terrorist attacks In addition to car-bombings the military has identified several non-traditional attack scenarios including poisoning of water supplies by chemical warfare agents (CWAs) To counter this threat the Department of Defense has been investigating numerous approaches to detect such attacks under the auspices of the Joint Service Agent Water Monitor (JSAWM) program1 The JSAWM must be capable of identifying the chemical agents with no false-positives and quantifying the agents at microgram per liter concentrations within 10 minutes2 This includes the analysis of drinking water supplies distribution and storage systems Currently colorimetric paper is used to detect agents on-site while gas chromatography combined with mass spectrometry (GCMS) is used in mobile support laboratories However both methods have severe drawbacks The paper changes color in response to contact with many chemicals besides CWAs causing a high incidence of false positives3 while GCMS although very chemically specific requires up to an hour to perform and regular re-calibration456 Vibrational spectroscopy has also been investigated as a potential method of rapidly detecting CWAs7-
11 as early as 1985 in the case of infrared spectroscopy8 and 1988 in the case of Raman spectroscopy11 Again however these techniques also have limitations when it comes to measuring trace poisons in water Infrared spectra would be dominated by the very strong absorption of water which would obscure absorptions by most other chemicals present Whereas Raman spectroscopy is simply not a very sensitive technique and detection limits are typically grams per liter Surface-enhanced Raman spectroscopy (SERS) offers several advantages over conventional vibrational methods and that may provide the necessary sensitivity required for detecting trace quantities of chemical agents in water SERS employs the interaction of surface plasmon modes of metal particles with target analytes to increase scattering efficiency by as much as 1 million times12 In 1987 the potential of this approach to measure CWAs was demonstrated by measuring a series of pesticides13 Several of these organophosphonates have chemical structures similar to CWAs in particular P=O functional groups In the past few years we and others have further explored the ability of SERS to detect CWAs14-17 and even bioagents 18-21 We have been employing silver-doped sol-gels to promote the SER effect
SPIE-2003-5269 17
in these studies The porous silica network of the sol-gel matrix offers a unique environment for immobilizing and stabilizing SER-active metal particles22-25 The silver-doped sol-gels have been coated on the internal walls of standard glass vials to allow rapid SER analysis of small sample quantities without preparation We have measured over 100 chemicals with enhancements of 104 to 106 demonstrated reversible measurements in a flowing system reproducible measurements from vial-to-vial and measurements in multiple solvents including water21-25 Previously we used these vials to perform preliminary measurements of cyanide (CN) and methylphosphonic acid (MPA) Most of the nerve agents form MPA during hydrolysis while Tabun forms CN a chemical agent in its own right In another paper including in these proceedings we examined the limits of detection (LOD) for MPA by measuring a series of concentrations down to 50 mgL and estimated a limit of detection of 100 microgL26 These measurements provide encouragement in that SERS may satisfy the needs of the JSAWM To further establish the viability of SERS in particular silver-doped sol-gels here we present analysis of cyanide mustard and VX in tap water The measurements performed at the US Armyrsquos Edgewood Chemical Biological Center Aberdeen MD also included numerous repeat measurements to establish reproducibility
2 EXPERIMENTAL 2a General All chemicals including potassium cyanide 2-chloroethylethyl sulfide and those used to prepare the silver-doped and gold-doped sol-gel coated vials were obtained and used as received from Sigma-Aldrich All samples were prepared in a chemical hood using HPLC grade water unless otherwise noted (Fischer Scientific Fair Lawn NJ) for SERS measurements Once prepared the samples were transferred into the silver-doped sol-gel vials (Real-Time Analyzers Inc East Hartford CT) The vials were coated in a manner similar to that previously reported by adding ammonium hydroxide to a solution of silver nitrate tetramethyl orthosilicate (TMOS) and methanol22 The two precursor solutions were prepared mixed and transferred to 2-mL glass vials dried and heated After sol-gel formation the incorporated metal ions were reduced with dilute sodium borohydride (1mgmL) which is followed by a water wash to remove residual reducing agent After the resultant analyte solution was introduced the SER-active vial was fixed horizontally to an XY positioning stage (Conix Research Springfield OR) just inside the focal point of an f07 aspheric lens The lens focused the beam into the sample and collected the scattered radiation back along the same axis A dichroic filter (Omega Optical Brattleborough VT) was used to reflect the excitation laser to the lens and pass the Raman scattered radiation collected by the lens An f2 achromat was used to collimate the laser beam exiting a 200 microm core diameter source fiber optic while a second f2 achromat was used to focus the scattered radiation into a 365 microm fiber optic (Spectran Avon CT) A short pass filter was placed in the excitation beam path to block the silicon Raman scattering generated in the source fiber from reflecting off sampling optics and reaching the detector A long pass filter was placed in the collection beam path to block the sample Rayleigh scattering from reaching the detector A 785 nm diode laser (Process Instruments Inc model 785-600 Salt Lake City UT) was used to deliver 100 to 150 mW of power to the sample A Fourier transform Raman spectrometer (Real-Time Analyzers model IRA-785 East Hartford CT) and a silicon photo-avalanche detector (Perkin Elmer model C30902S Stamford CT) were used to acquire the SER spectra
2b Edgewood Chemical Biological Center The surface-enhanced Raman spectral measurements at the US Armyrsquos Chemical Biological center presented here were all performed on September 12 2003 To expedite measurements a plate was machined to hold up to 12 SER-active sample vials (Figure 1) The plate fit a standard XY plate reader that could be programmed Pure KCN bis-(2-chloroethyl)sulfide (distilled mustard gas HD) and ethyl S-2-diisopropylamino ethyl methylphosphonothioate (VX) were obtained on-site and used to prepare 1 gL tap water solutions in a chemical hood with appropriate safety equipment Simulated tap water was prepared by adding 10 chemicals most often found in tap water at appropriate concentrations to distilled water (Table 1) SER measurements were also performed in a chemical hood For added safety the FT-Raman instrument was placed outside the laboratory and 30 foot fiber optic and electrical cables were used to allow remote SERS measurements and plate manipulation For each experiment 1gL samples were prepared and added to 9 individual vials which were then loaded on the plate In some cases a tenth vial was included as a blank
Table 1 Generic Tap Water Composition
Compound mgL NaHCO3 100 CaSO4 27 MgSO4bull7H2O 67 NaNO3 1 Fulvic Acid 1 K2HPO4 07 KH2PO4 03 (NH4)2HSO4 001 NaCl 001 FeSO4 0001 pH 76-78
SPIE-2003-5269 18
A software program was written that allowed selecting the sequence that the vials were measured the number of positions along the length of the vials to measure (1 to 5) and the number of scans to co-add During sample analysis the program displayed the vial being analyzed the point being analyzed and the spectrum as it was being acquired Once all the data was collected a second software program was written to rapidly analyze the data The spectra collected for all the vials on a plate could be loaded at one time and then the spectra for each point could be displayed simultaneously or separately The user could then select the Raman peak to analyze in terms of peak height or area This was accomplished by selecting points on either side of the peak to define a baseline of zero The peak height or area could then be computed for all of the spectra loaded and then exported to a spreadsheet for statistical analysis
Figure 1 A) Vial Holder 6 slots to hold 2 vials each end-to-end B) Measurement Configuration Program user selects vials to measure sequence number of points per vial (1 to 5) and number of scans per point C) Spectral Acquisition Program shows spectrum being collected which vial and position D) Spectra Analysis Program user selects spectra to analyze by plate vial and point (s) as well as two wavenumbers defining the peak and the baseline to subtract The image is of 5 repeat measurements of 10 mgL KCN in generic tap water 16 sec each 100 mW of 785 nm
3 RESULTS AND DISCUSSION Raman and surface-enhanced Raman spectra were obtained for potassium cyanide bis-(2-chloroethyl)sulfide and ethyl S-2-diisopropylamino ethyl methylphosphonothioate representing three classes of chemical agents cyanides mustards and nerve agents respectively Spectra were also obtained for 2-chloroethyl ethyl sulfide (CEES) a structural analogue to HD which was included in the study to aid in assigning spectral bands KCN salt was used for cyanide experiments to avoid the increased hazards of handling HCN gas KCN completely dissolves in water forming its conjugate acid HCN according to its Ka of 615x10-1027 and at a concentration of 1 mgmL results in a pH 107 solution This is significant in that only CN- appears to interact sufficiently with silver to produce a SER spectrum and no spectral signal is observed below pH 726 Figure 2 shows the SER and normal Raman spectra for KCN The SERS spectrum of 1mgml KCN in water shows a single intense somewhat broad feature at
A B
C D
SPIE-2003-5269 19
2100 cm-1 assigned to the single CequivN stretch The band is much sharper in the normal Raman spectra of the solid KCN salt at 2074 cm-1 This band does broaden and shift to 2080 cm-1 in solution (not shown) However the observed SERS frequency is attributed to interaction with silver and low frequency mode at 135 cm-1 attributed to a Ag-CN stretch (not shown) supports this conclusion
cm-1 band to a C-S stretch but the authors concede that it is in fact more likely a C-Cl stretch28 It appears that the most intense Raman bands at 648 692 and 747 cm-1 shift to 620 660 and 720 cm-1 in the SER spectra and are tentatively assigned as above The width of these bands suggests that they overlap underlying spectral features Additional bands in the Raman spectra occur at 972 1034 1049 1263 1286 1423 1442 2865 2935 and 2960 cm-1 Corresponding bands occur in the SER spectra at 964 1015 1054 1286 1410 1447 2865 and 2935 cm-1 Most of these bands are associated with alkane modes specifically the bands at approximately 1040 cm-1 to a C-C stretch 1290 cm-1 to a CH2 in-phase twist 1440 cm-1 to a CH2 wag 2865 cm-1 to a symmetric CH2 stretch and 2965 cm-1 to an asymmetric CH2 stretch The Raman and SER spectra of sulfur mustard were measured at the Edgewood center (Figure 4) Both spectra are largely similar to CEES The C-Cl and C-S bands in the Raman spectrum of HD now occur at 640 655 700 739 and 760 cm-1 and are more resolved possibly due to the increased molecular symmetry Theoretical calculations indicate that the first three bands are due to C-Cl stretching modes and the latter two to C-S stretching modes28 Only the C-Cl bands maintain significant intensity in the SER spectra occurring at 624 and 643 cm-1 which is attributed to the
Prior to measurements of HD CEES was examined by Raman and SER spectroscopy (Figure 3) CEES also known as half-mustard is essentially identical to HD except one of the chlorine end atoms is replaced by a hydrogen atom Again although not as toxic as HD CEES is a blister agent and dilute aqueous samples were prepared using appropriate safety equipment Both the Raman and SER spectra of CEES are similar and dominated by bands between 600 and 800 cm-1 These are associated with C-Cl and C-S stretching modes which are tentatively assigned to 648 and 747 cm-1 in the Raman spectra respectively The shoulder at 630 cm-1 the overlapped band at 660 cm-1 and the strong band at 692 cm-1 could also be due to these modes or their asymmetric counterparts It is worth noting that theoretical calculations assign the 692
Figure 2 A) SER and B) NR spectra of KCN Conditions A) 1 mgml in tap water 100 mW of 785 nm at sample 1-min acquisition time B) solid 300 mW of 785 nm 5-min All spectra are 8 cm-1 resolution
A
B
Figure 3 A) SER and B) NR spectra of CEES Conditions A) 1 vv (10 mgml) in MeOH 100 mW of 785 nm 1-min acquisition time B) neat 300 mW of 785 nm 5-min
A
B
Cl-CH2-CH2-S-CH2-CH3
A
B
Figure 4 A) SER and B) NR of HD Conditions A) 1mgml in tap water B) pure both 100 mW of 785 nm 1-min
Cl-CH2-CH2-S-CH2-CH2-Cl
SPIE-2003-5269 20
expected strong interaction between chlorine and silver and adds support to the assignment of this band to a C-Cl stretch Weaker overlapping bands occur at 670 692 and 724 cm-1 the latter possibly due to C-S stretching modes Again the alkane modes are apparent in the normal Raman spectra of HD but only a broad feature at 1300 to 1450 cm-1 suggests CH2 contributions in the SER spectrum Although the observed bands in the VX spectrum have not been assigned (Figure 4) a computer generated Raman spectrum29 predicts many of the same features with surprising accuracy and are used here Two intense bands at 460 and 530 cm-1 closely match predicted bands at 463 and 546 cm-1 assigned to a CH3-P=O bend and a PO2CS wag Three highly overlapped bands occur at 694 745 and 771 cm-1 matching predicted bands at 713 730 and 760 cm-1 The first
Table 2 Measured SER peak heights for the CN stretch at 2100 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
Figure 5 A) SERS and B) NR spectra of VX Conditions A) 1 vv (10 mgml) in MeOH B) pure sample both 100 mW of 785 nm 1-min acquisition time
A
B
two have been assigned to a C-S stretch and CH2bend respectively while the latter has been attributed to either a P-C stretch or an O-C-C stretch Although the 745 cm-1 band may alternatively be assigned to a C-S stretch based on the previous measurements of CEES and HD The relatively intense bands at 890 1106 1218 1445 and 1465 cm-1 also match predicted bands at 880 1108 1216 1440 and 1464 cm-1 that are assigned to a C-C stretch CH3 rock N-C3 stretch various C-H3 bends and C-H bends respectively Both the computer generated and the measured spectra contain numerous other less intense bands One is worth mentioning A unique band appears at 370 cm-1 that is predicted at 368 cm-1 and corresponds to an O-P=O bend The surface-enhanced Raman spectrum of VX is also rich with spectral features It has the unique low frequency band at 370 cm-1 as well as a second band at 380 cm-1 that is assigned tothe S-P-O bend predicted in the normal Raman spectrum at 388 cm-1 Based on the measured and predicted normal Raman spectra the following SERS assignments are given 460 cm-1 to the CH3-P=O bend 544 cm-1 to the PO2CS wag 738 cm-1 to a C-S stretch (based on arguments above) 890 cm-1
to a C-C stretch 1101 cm-1 to a CH3 rock and 1456
cm-1 to a C-H bend The ability of SERS to measure chemical agents in water containing real-world chemical interferents was tested by using the generic tap water described in Table 1 The ability to reproduce measurements was accomplished by preparing three separate water stock solutions which were used to prepare three sample solutions each of 1 mgmL KCN in the generic tap water The samples defined as vials 1-9 were then measured at 4 points per vial in a semi-random fashion such that errors associated with stock solution preparation and errors associated with instrument drift could be identified No trends were apparent that signified such systematic errors Each spectrum collected consisted of 20 averaged scans taking 16 seconds at 8 cm-1 resolution The laser power at the sample was measured periodically during the day and it ranged from 102 to 105 mW spacccc
spot Vial 1 Vial 2 Vial 41 13754 13004 128192 13519 12692 12909
1 3 13541 12721 126394 13462 12648 12651
ave 13569 12766 12755Vial 6 Vial 3 Vial 5
1 11586 14068 10757 2 11236 14402 115122 3 11376 14546 11559
4 10894 11763 11214ave 11273 13695 11261
Vial 7 Vial8 Vial 91 11198 15804 11023
2 11243 15716 88753 3 11562 15216 11218
4 11655 15084 11077ave 11415 15455 10548
CN AVG STDEV ERR1 mgmL pk ht 12526 1560 1245
stock solution
SPIE-2003-5269 21
The CN measurements consisting of 36 data points produced an average height of 125 for the 2100 cm-1 peak with a standard deviation of 156 or 125 (Table 2) The HD and VX measurements were performed precisely the same way (Tables 3 and 4) For HD the 624 cm-1 peak was used for analysis and it had an average height of 53 with a standard deviation of 068 or 129 while for VX the 544 cm-1 peak was used for analysis and it had an average height of 1051 with a standard deviation of 308 or 293 The greater error in the VX measurements can be somewhat attributed to Vial 4 which produced lower SER signal intensities But removing this vial from the data set changes the standard deviation to 233 only a modest improvement
spot Vial 1 Vial 2 Vial 41 525 609 4682 45 675 484
1 3 527 644 5094 559 807 669
ave 51525 68375 5325Vial 6 Vial 3 Vial 5
1 521 574 575 2 536 53 4492 3 509 508 451
4 565 594 379ave 53275 5515 4635
Vial 7 Vial 8 Vial 91 457 589 413
2 583 497 4273 3 544 505 4
4 528 5 509ave 528 52275 43725
HD AVG STDEV ERR1 mgmL pk ht 530 068 1291
stock solution
spot Vial 1 Vial 2 Vial 41 1464 1034 4672 1485 99 61
1 3 1491 989 5684 1041 777 553
ave 137025 9475 5495Vial 6 Vial 3 Vial 5
1 1058 942 1293 2 697 121 965
2 3 727 1206 11124 689 1254 746
ave 79275 1153 1029Vial 7 Vial 8 Vial 9
1 1127 783 163 2 1358 812 1614
3 3 1371 908 15124 1097 875 1397
ave 123825 8445 153825
VX AVG STDEV ERR1 mgmL pk ht 1051 308 2925
stock solution
4 CONCLUSIONS In this paper we examined the ability of surface-enhanced Raman spectroscopy to reproducibly measure CN HD and VX in tap water without chemical interference Both normal and surface-enhanced Raman spectra were examined to select unique bands suitable to identify and quantify these chemical agents For SER measurements the 2100 cm-1 C-N stretch was used for CN the 624 cm-1 C-Cl stretch was used for HD and the 544 cm-1 PO2CS wag was used for VX It was determined that 1 mgmL samples of each of these chemicals measured 36 times in glass vials coated with a silver-doped sol-gel reproduced measurements with standard deviations of 125 129 and 293 It was further found that the 10 chemicals added to simulate generic tap water did not interfere with or alter the SER spectra It should be noted that the concentrations used in this study were considerably greater than those required by the JSAWM program Current work involves improving SER sensitivity and designing sampling systems with better reproducibility This includes the development of fractal silver and gold structures within the sol-gel matrix and the development of chemically selective sol-gels
ACKNOWLEDGMENTS The authors are grateful for the support of the US Army (DAAD13-02-C-0015 Joint Service Agent Water Monitor program) The authors would also like to thank Janet Jensen Ronald Crosier and Kristina Gonser for helpful discussions
Table 4 Measured SER peak heights for the VX band at 544 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
Table 3 Measured SER peak heights for the HD band at 624 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
SPIE-2003-5269 22
REFERENCES 1 Jensen JL NC Wong W R Loerop SPIE 4775-03 (2001) 2 JSAWM Requirements at wwwsbccomapgeaarmymilRDAecbcrtPRODSERJSAWMjsawmhtml 3 Erickson B ldquoThe Chemical Weapons Convention Redefines Analytical Challengerdquo Anal Chem 70 397A (1998) 4 Johnston RL Hoefler CM Fargo JC and Moberley B ldquoThe Defense Nuclear Agencyrsquos
ChemicalBiochemical Weapons Agreements Verification Technology Research Development Test and Evaluation Program and its Requirements for On-Site Analysisrdquo AT-ONSITE 5-8 (1994)
5 Black RM Clarke RJ Read RW and Reid MT ldquoApplication of GC-MS and GC-tandem MS to the analysis of chemical warfare samples found to contain residues of the nerve agent sarin sulphur mustard and their degradation productsrdquo J Chrom 662 301-321 (1994)
6 USA Army Center for Health Promotion and Preventive Medicine (USACHPPM) Technical Guideline 230A 7 Hoffland LD Piffath RJ Bouck JB rdquoSpectral signatures of chemical agents and simulantsrdquo Optical
Engineering 24 982-984 (1985) 8 Braue EHJ Pannella MGrdquoCIRCLE CELL FT-IR Analysis of Chemical Warfare Agents in Aqueous Solutionsrdquo
App Spectrosc 44 1513-1520 (1990) 9 Tseng C-H Mann CK and Vickers TJ ldquoDetermination of Organics on Metal Surfaces by Raman
Spectroscopyrdquo App Spectrosc 47 1767-1771 (1993) 10 Staff Reporter ldquoUS Army Tests Lidar to Detect Biological Toxinsrdquo Photonic Spectra pg 50 December 1998 11 Christesen SD Raman cross sections of chemical agents and simulants App Spectrosc 42 318-321 (1988) 12 Weaver MJ Farquharson S Tadayyoni MA ldquoSurface-enhancement factors for Raman scattering at silver electrodesrdquo J Chem Phys 82 4867-4874 (1985) 13 Alak AM and Vo-Dinh T ldquoSERS of Organophosphorous Chemical Agentsrdquo Anal Chem 59 2149-2153 (1987) 14 Lee Y Farquharson S ldquoRapid chemical agent identification by SERSrdquo SPIE 4378 21-26 (2001) 15 Farquharson S Maksymiuk P Ong K Christesen S ldquoChemical agent identification by surface-enhanced Raman spectroscopyrdquo SPIE 4577 166-173 (2001) 16 Spencer KM Sylvia J Clauson S and Janni J ldquoSurface Enhanced Raman as a Water Monitor for Warfare
Agents in Waterrdquo SPIE 4577 158-165 (2001) 17 Tessier P Christesen S Ong K Clemente E Lenhoff A Kaler E Velev O ldquoOn-line spectroscopic characterization of sodium cyanide with nanostructured gold surface-enhanced Raman spectroscopy substratesrdquo
App Spectrosc 56 1524-1530 (2002) 18 Farquharson S W W Smith S Elliott and J F Sperry Rapid biological agent identification by surface- enhanced Raman spectroscopy SPIE 3855110-116 (1999) 19 Farquharson S WW Smith YH Lee S Elliott and J F Sperry Detection of biological signatures A
comparison of electrolytic and metal-doped sol-gel surface-enhanced Raman media SPIE 4575 62-72 (2002) 20 Guzelian AA Sylvia J Janni J Clauson S and Spencer KM ldquoSERS of whole cell bacteria and trace levels of biological moleculesrdquo SPIE 4577 182-192 (2001) 21 Shende C Inscore F Gift A Maksymiuk P Farquharson S ldquoRapid dipicolinic acid extraction from Bacillus spores detected by surface-enhanced Raman spectroscopyrdquo App Spectrosc 58 accepted 22 Lee Y and Farquharson S ldquoSERS Sample Vials Based on Sol-Gel Process for Trace Pesticide Analysisrdquo SPIE
4206 140-146 (2000) 23 Farquharson S and Lee Y ldquoTrace Drug Analysis by SERSrdquo SPIE 4200-16 (2000) 24 Lee Y Farquharson S and Rainey P M Surface-Enhanced Raman Sensor for Trace Chemical Detection in
Water SPIE 3857 76-84 (1999) 25 Lee Y Farquharson S Kwong H and Shahriari M ldquoSol-Gel Chemical Sensor for Surface-Enhanced Raman
Spectroscopyrdquo SPIE 3537 252-260 (1998) 26 Farquharson S A Gift P Maksymiuk F Inscore and W Smith ldquopH dependence of methyl phosphonic acid
dipicolinic acid and cyanide by surface-enhanced Raman spectroscopyrdquo SPIE 5269-19 (2003) 27 Lide DR Ed Handbook of Chemistry and Physics CRC Press 77th Ed (1996-7) 28 Sosa C RJ Bartlett K KuBulat and WB Person ldquoA theoretical study of harmonic vibrational frequencies and
infrared intensities of XCH2CH2SCH2CH2X and XCH2CH2SH (= H Cl)rdquo J Phys Chem 93 577-588 (1993) 29 Hameka HF and JO Jenson ldquoComputer-generated predictions of the structure and the IR and Raman spectra of
VXrdquo ERDEC-TR-065 May 1993
- Chemagents Appendicespdf
-
- SPIE2001-4575-Bioagent-SERSapdf
-
- Detection of bioagent signatures A comparison of electrolytic and metal-doped sol-gel surface-enhanced Raman media
- Stuart Farquharson Wayne Smith and Yuan Lee
-
- Real-Time Analyzers 87 Church Street East Hartford CT 06108
-
- ABSTRACT
- 4 CONCLUSIONS
- 5 ACKNOWLEDGEMENTS
- 6 REFERENCES
-
SPIE -2003-5269 123
Consequently the cyanide concentration must be determined for each initial KCN concentration Specifically the samples prepared with concentrations of 01 1 10 100 and 1000 mgL of KCN produced CN- concentrations of 63x10-3 033 69 89 and 964 mgL at pHs of 816 90 967 102 and 107 respectively The pH dependence for the HCN and CN- concentrations are shown in Figure 12 Thus as the amount of KCN added to the solution decreases so does the pH of the solution (becomes less basic) and according to Reaction 4 the relative amount of CN- to HCN also decreases For example in the preparation of a 01 mgL solution of KCN the pH is shifted from 7 for pure water to only 816 and only 63 of the starting material becomes CN- or 63x10-3 mgL In comparison for a solution of 1000 mgL the pH is shifted from 7 to 107 and 96 of the starting material becomes CN- This is significant since the cyanide ion is better able to adsorb onto the silver particles and become SERS active SER spectra of 10 100 and 1000 mgL of KCN samples are shown in Figure 13 and a corresponding plot of the 2100 cm-1 band intensity as a function of concentration is shown in Figure 14
000102030405060708091011
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14pH
Con
cent
ratio
n [m
gm
L]
pKa = 921
CN -HCN
The SER spectra of cyanide are dominated by the single CN stretch at 2100 cm-1 which occurs in normal Raman spectra of solutions at 2080 cm-1 However a low frequency mode occurs at 135 cm-1 due to a Ag-CN stretch (not shown) indicative of a strong surface interaction It is also observed that as the concentration decreases the CN stretch shifts to 2140 cm-1 This shift has been attribute to the formation of a tetrahedral Ag(CN)3
2- surface structure31 as well
Figure 14 Concentration dependence of KCN SERS measured under conditions in Fig 11 Concentrations are 1 01 and 001 mgml Intensities are measured for the CN stretch at 2100 cm-1 Inset table includes LOD in mgL for each concentration in Figs 13 and 15 but for 100 mW and 10 min
Figure 13 SER spectra of KCN in water at A) 1000 B) 100 and C) 10 mgL Conditions pHs of 107 102 and 97 silver-doped sol-gel coated vial 100 mW of 785 nm 1-min 8 cm-1 C) has been multiplied by x10 to make band visible
Figure 12 Concentration dependence of HCN and CN- as a function of pH for a 1 mgmL sample Calculated intensity of 2100 () cm-1 band for a 1 mgml sample at pHs of 816 90 967 102 and 107
A
B
C
0
20
40
60
80
100
120
140
0 200 400 600 800 1000 1200CN Concentration (mgL)
2100
Ban
d In
tens
ity
Figure 15 SER spectra of KCN in water at A) 10 B) 1 and C) 01 mgL Conditions pHs of 97 12 and 12 gold-doped sol-gel coated vial 100 mW of 785 nm 1-min 8 cm-1 C) has been multiplied by x10 to make band visible
A
B
C
conc (mgL) lod-10min-100mw Condition01 001 Au-pH 121 007 Au-pH 12
69 003 Au-pH 9769 016 Ag-pH 9789 022 Ag-pH 102946 113 Ag-pH 107
SPIE -2003-5269 124
as to CN adsorbed to two different surface sites32 Alternatively the 2140 cm-1 band could be attributed to HCN since this species dominates at lower concentrations However it is unlikely that this species would be attracted to the electropositive silver surface Further both peaks should be present at pHs between 85 and 105 but this is not observed It has also been suggested that at concentrations near and above monolayer coverage the CN- species is forced to adsorb end-on due to crowding and at lower concentrations the molecule can reorient to lie flat33 This suggests that the 2100 and 2140 cm-1 bands correspond to the end-on and flat orientations respectively As Figures 13 and 14 show the intensity of the CN stretch for the 89 mgL sample is nearly as intense as the 964 mgmL band This suggests that the Raman signal for the flat orientation is more enhanced However more extensive measurements are required to verify this point Since resent research has suggested that cyanide may be more effectively detected on gold measurements of KCN solutions were also performed using gold-doped sol-gel vials Preliminary measurements are shown in Figure 15 for samples prepared from 01 1 and 10 mgL KCN Since the pHs are 816 90 967 the resultant CN- concentrations are 63x10-3 033 and 69 mgL Initially only the highest concentration was observed and the signal intensity was significantly better than the equivalent concentration measured using silver In an effort to shift Reaction 4 to the left transforming HCN to CN- (Le Chatelierrsquos principle) KOH was added to the lower concentration samples producing solutions with pH 12 This effectively forces all of the cyanide in solution to be CN- or 01 and 10 mgL respectively More importantly the CN stretch is now observed in the SER spectra The band appears at 2125 cm-1 as has been previously reported for gold12 As calculated for MPA and DPA LODs can be estimated from this data For the three concentrations of cyanide on silver the LODs are 016 to 11 mgL for 100 mW of 785 nm laser excitation and a 10-min acquisition time For gold pH adjusted the LODs are10 to 70 microgL an improvement of more than 10 times silver Nevertheless either substrate is sufficient to meet the JSAWM goals of measuring 3 mgL in 10-minutes as the requirements form cyanide are much less stringent than the nerve agents
4 CONCLUSIONS Here we examined the surface-enhanced Raman spectral response for methyl phosphonic acid dipicolinic acid and cyanide as a function of pH It was determined that the most prominent peak at 760 cm-1 reaches a maximum intensity between pH 3 and 7 and corresponds to the MPA- species Neither the MPA nor MPA= species appear to generate a SER spectrum and consequently no spectra were observed below pH 2 or above pH 8 In this study we also found that higher laser powers could cause photodegradation of MPA signified by the exponential growth of a band at 1050 cm-1 which is tentatively assigned to phosphorous acid Unlike MPA DPA was observed at all pHs This is attributed to the dominant interaction of the pyridine functional group with silver Minor spectral changes were observed at acid pHs and were assigned to neutral DPA Like MPA SER spectra of cyanide were pH dependent No spectra were observed for the HCN species while CN- was best observed at pHs more basic than 8 Preliminary concentration studies for the three analytes allowed estimating limits of detection for MPA DPA and CN using 100 mW of 785 nm and a 10-min acquisition time of 210 165 and 70 microgL respectively Although the latter value suggests that cyanide can be measured at sufficiently low concentrations to meet JSAWM goal improvements by 100 to 200 times are required for MPA and DPA It is clear from this study that pH of the sample is important and must be taken into account when developing concentration calibration curves as well as fieldable analyzers Future work will include tailoring the sol-gel with specific functional groups to dictate sample pH and thereby optimize sensitivity
ACKNOWLEDGMENTS The authors are grateful for the support of the US Army (DAAD13-02-C-0015 Joint Service Agent Water Monitor program) The authors would also like to thank Dr Steve Christensen of the US Army SBCCOM for helpful discussions and Mr Chetan Shende of Real-Time Analyzers for assistance in development of the gold-doped sol-gels
REFERENCES 1 Erickson B Analytical Chemistry News amp Features June 1 397A (1998)
SPIE -2003-5269 125
2 Johnston RL Hoefler CM Fargo JC and Moberley B AT-ONSITE 5-8 (1994) 3 Black RM RJ Clarke RW Read and MT Reid J Chromatography 662 301-321 (1994) 4 USA Army Center for Health Promotion and Preventive Medicine (USACHPPM) Technical Guideline 230A 5 Jensen JL NC Wong W R Loerop SPIE 4775-03 (2001) 6 Hoenig SL Handbook of Chemical Warfare and Terrorism Greenwood Press Wesport CT (2002) 7 Munro NB SS Talmage GD Griffin LC Waters AP Watson JF King and V Hauschild Env Health
Persp 107 933-974 (1999) 8 Holstege CP Kirk M Sidell FR Crit Care Clin13 923-42 (1997) 9 Farquharson S P Maksymiuk K Ong and S Christesen SPIE 4577 166-173 (2001) 10 Lee Y and S Farquharson SPIE 4378 21-26 (2001) 11 Spencer KM J Sylvia S Clauson and J Janni SPIE 4577 158-165 (2001) 12 Tessier P S Christesen K Ong E Clemente A Lenhoff E Kaler and O Velev Applied Spectroscopy 56
1524-1530 (2002) 13 Farquharson S WW Smith S Elliott and JF Sperry SPIE 3533 207-214 (1998) 14 Farquharson S WW Smith S Elliott and JF Sperry SPIE 3855110-116 (1999) 15 Farquharson S WW Smith YH Lee S Elliott and JF Sperry SPIE 4575 62-72 (2002) 16 Guzelian AA J Sylvia J Janni S Clauson and KM Spencer SPIE 4577 182-192 (2001) 17 Shende C F Inscore A Gift P Maksymiuk and S Farquharson in press 18 Weaver MJ S Farquharson and MA Tadayyoni J Chem Phys 82 4867-4874 (1985) 19 Alak AM and T Vo-Dinh Analytical Chemistry 59 2149-2153 (1987) 20 Lee Y and S Farquharson SPIE 4206 140-146 (2000) 21 Farquharson S and Y Lee SPIE 4200-16 (2000) 22 Lee Y S Farquharson and P M Rainey SPIE 3857 76-84 (1999) 23 Lee Y S Farquharson H Kwong and M Shahriari SPIE 3537 252-260 (1998) 24 Wang J M Pumera G Collins and A Mulchandani Analytical Chemistry 74 6121-6125 (2002) 25 Dou X YM Jung Z-Q Cao and Y Ozaki Applied Spectroscopy 53 1440-1447 (1999) 26 Data supplied by S Christesen and K Ewing 27 Kerker M and B Thompson Eds SPIE MS 10 (1990) 28 Inglesby TV DA Henderson JG Bartlett JAMA 287 2236 (2002) 29 FW Janssen AJ Lund and LE Anderson Science 127 26 (1958) 30 Lide DR Ed Handbook of Chemistry and Physics CRC Press 77th Ed (1996-7) 31 Billmann J G Kovacs and A Otto Surf Sci 92 153 (1980) 32 Murray CA and S Bodoff Phys Rev B 32 671 (1985) 33 Kellogg D and J Pemberton J Phys Chem 91 1120 (1987)
SPIE-2003-5269 16
Chemical agent detection by surface-enhanced Raman spectroscopy
Stuart Farquharson Alan Gift Paul Maksymiuk Frank Inscore and Wayne Smith
Real-Time Analyzers 87 Church Street East Hartford CT 06108
Kevin Morrisey and Steven D Christesen US Army SBCCOM Aberdeen Proving Ground MD 21010
ABSTRACT
In the past decade the Unites States and its allies have been challenged by a different kind of warfare exemplified by the terrorist attacks of September 11 2001 Although suicide bombings are the most often used form of terror military personnel must consider a wide range of attack scenarios Among these is the intentional poisoning of water supplies to obstruct military operations in Afghanistan and Iraq To counter such attacks the military is developing portable analyzers that can identify and quantify potential chemical agents in water supplies at microgram per liter concentrations within 10 minutes To aid this effort we have been investigating the value of a surface-enhanced Raman spectroscopy based portable analyzer In particular we have been developing silver-doped sol-gels to generate SER spectra of chemical agents and their hydrolysis products Here we present SER spectra of several chemical agents measured in a generic tap water Repeat measurements were performed to establish statistical error associated with SERS obtained using the sol-gel coated vials Keywords Chemical agents chemical agent detection SERS Raman spectroscopy
1 INTRODUCTION
In the autumn of 2001 terrorism within US borders became a sobering reality While extensive efforts are being implemented to secure the homeland US and Coalition forces in Afghanistan and Iraq are constantly faced with terrorist attacks In addition to car-bombings the military has identified several non-traditional attack scenarios including poisoning of water supplies by chemical warfare agents (CWAs) To counter this threat the Department of Defense has been investigating numerous approaches to detect such attacks under the auspices of the Joint Service Agent Water Monitor (JSAWM) program1 The JSAWM must be capable of identifying the chemical agents with no false-positives and quantifying the agents at microgram per liter concentrations within 10 minutes2 This includes the analysis of drinking water supplies distribution and storage systems Currently colorimetric paper is used to detect agents on-site while gas chromatography combined with mass spectrometry (GCMS) is used in mobile support laboratories However both methods have severe drawbacks The paper changes color in response to contact with many chemicals besides CWAs causing a high incidence of false positives3 while GCMS although very chemically specific requires up to an hour to perform and regular re-calibration456 Vibrational spectroscopy has also been investigated as a potential method of rapidly detecting CWAs7-
11 as early as 1985 in the case of infrared spectroscopy8 and 1988 in the case of Raman spectroscopy11 Again however these techniques also have limitations when it comes to measuring trace poisons in water Infrared spectra would be dominated by the very strong absorption of water which would obscure absorptions by most other chemicals present Whereas Raman spectroscopy is simply not a very sensitive technique and detection limits are typically grams per liter Surface-enhanced Raman spectroscopy (SERS) offers several advantages over conventional vibrational methods and that may provide the necessary sensitivity required for detecting trace quantities of chemical agents in water SERS employs the interaction of surface plasmon modes of metal particles with target analytes to increase scattering efficiency by as much as 1 million times12 In 1987 the potential of this approach to measure CWAs was demonstrated by measuring a series of pesticides13 Several of these organophosphonates have chemical structures similar to CWAs in particular P=O functional groups In the past few years we and others have further explored the ability of SERS to detect CWAs14-17 and even bioagents 18-21 We have been employing silver-doped sol-gels to promote the SER effect
SPIE-2003-5269 17
in these studies The porous silica network of the sol-gel matrix offers a unique environment for immobilizing and stabilizing SER-active metal particles22-25 The silver-doped sol-gels have been coated on the internal walls of standard glass vials to allow rapid SER analysis of small sample quantities without preparation We have measured over 100 chemicals with enhancements of 104 to 106 demonstrated reversible measurements in a flowing system reproducible measurements from vial-to-vial and measurements in multiple solvents including water21-25 Previously we used these vials to perform preliminary measurements of cyanide (CN) and methylphosphonic acid (MPA) Most of the nerve agents form MPA during hydrolysis while Tabun forms CN a chemical agent in its own right In another paper including in these proceedings we examined the limits of detection (LOD) for MPA by measuring a series of concentrations down to 50 mgL and estimated a limit of detection of 100 microgL26 These measurements provide encouragement in that SERS may satisfy the needs of the JSAWM To further establish the viability of SERS in particular silver-doped sol-gels here we present analysis of cyanide mustard and VX in tap water The measurements performed at the US Armyrsquos Edgewood Chemical Biological Center Aberdeen MD also included numerous repeat measurements to establish reproducibility
2 EXPERIMENTAL 2a General All chemicals including potassium cyanide 2-chloroethylethyl sulfide and those used to prepare the silver-doped and gold-doped sol-gel coated vials were obtained and used as received from Sigma-Aldrich All samples were prepared in a chemical hood using HPLC grade water unless otherwise noted (Fischer Scientific Fair Lawn NJ) for SERS measurements Once prepared the samples were transferred into the silver-doped sol-gel vials (Real-Time Analyzers Inc East Hartford CT) The vials were coated in a manner similar to that previously reported by adding ammonium hydroxide to a solution of silver nitrate tetramethyl orthosilicate (TMOS) and methanol22 The two precursor solutions were prepared mixed and transferred to 2-mL glass vials dried and heated After sol-gel formation the incorporated metal ions were reduced with dilute sodium borohydride (1mgmL) which is followed by a water wash to remove residual reducing agent After the resultant analyte solution was introduced the SER-active vial was fixed horizontally to an XY positioning stage (Conix Research Springfield OR) just inside the focal point of an f07 aspheric lens The lens focused the beam into the sample and collected the scattered radiation back along the same axis A dichroic filter (Omega Optical Brattleborough VT) was used to reflect the excitation laser to the lens and pass the Raman scattered radiation collected by the lens An f2 achromat was used to collimate the laser beam exiting a 200 microm core diameter source fiber optic while a second f2 achromat was used to focus the scattered radiation into a 365 microm fiber optic (Spectran Avon CT) A short pass filter was placed in the excitation beam path to block the silicon Raman scattering generated in the source fiber from reflecting off sampling optics and reaching the detector A long pass filter was placed in the collection beam path to block the sample Rayleigh scattering from reaching the detector A 785 nm diode laser (Process Instruments Inc model 785-600 Salt Lake City UT) was used to deliver 100 to 150 mW of power to the sample A Fourier transform Raman spectrometer (Real-Time Analyzers model IRA-785 East Hartford CT) and a silicon photo-avalanche detector (Perkin Elmer model C30902S Stamford CT) were used to acquire the SER spectra
2b Edgewood Chemical Biological Center The surface-enhanced Raman spectral measurements at the US Armyrsquos Chemical Biological center presented here were all performed on September 12 2003 To expedite measurements a plate was machined to hold up to 12 SER-active sample vials (Figure 1) The plate fit a standard XY plate reader that could be programmed Pure KCN bis-(2-chloroethyl)sulfide (distilled mustard gas HD) and ethyl S-2-diisopropylamino ethyl methylphosphonothioate (VX) were obtained on-site and used to prepare 1 gL tap water solutions in a chemical hood with appropriate safety equipment Simulated tap water was prepared by adding 10 chemicals most often found in tap water at appropriate concentrations to distilled water (Table 1) SER measurements were also performed in a chemical hood For added safety the FT-Raman instrument was placed outside the laboratory and 30 foot fiber optic and electrical cables were used to allow remote SERS measurements and plate manipulation For each experiment 1gL samples were prepared and added to 9 individual vials which were then loaded on the plate In some cases a tenth vial was included as a blank
Table 1 Generic Tap Water Composition
Compound mgL NaHCO3 100 CaSO4 27 MgSO4bull7H2O 67 NaNO3 1 Fulvic Acid 1 K2HPO4 07 KH2PO4 03 (NH4)2HSO4 001 NaCl 001 FeSO4 0001 pH 76-78
SPIE-2003-5269 18
A software program was written that allowed selecting the sequence that the vials were measured the number of positions along the length of the vials to measure (1 to 5) and the number of scans to co-add During sample analysis the program displayed the vial being analyzed the point being analyzed and the spectrum as it was being acquired Once all the data was collected a second software program was written to rapidly analyze the data The spectra collected for all the vials on a plate could be loaded at one time and then the spectra for each point could be displayed simultaneously or separately The user could then select the Raman peak to analyze in terms of peak height or area This was accomplished by selecting points on either side of the peak to define a baseline of zero The peak height or area could then be computed for all of the spectra loaded and then exported to a spreadsheet for statistical analysis
Figure 1 A) Vial Holder 6 slots to hold 2 vials each end-to-end B) Measurement Configuration Program user selects vials to measure sequence number of points per vial (1 to 5) and number of scans per point C) Spectral Acquisition Program shows spectrum being collected which vial and position D) Spectra Analysis Program user selects spectra to analyze by plate vial and point (s) as well as two wavenumbers defining the peak and the baseline to subtract The image is of 5 repeat measurements of 10 mgL KCN in generic tap water 16 sec each 100 mW of 785 nm
3 RESULTS AND DISCUSSION Raman and surface-enhanced Raman spectra were obtained for potassium cyanide bis-(2-chloroethyl)sulfide and ethyl S-2-diisopropylamino ethyl methylphosphonothioate representing three classes of chemical agents cyanides mustards and nerve agents respectively Spectra were also obtained for 2-chloroethyl ethyl sulfide (CEES) a structural analogue to HD which was included in the study to aid in assigning spectral bands KCN salt was used for cyanide experiments to avoid the increased hazards of handling HCN gas KCN completely dissolves in water forming its conjugate acid HCN according to its Ka of 615x10-1027 and at a concentration of 1 mgmL results in a pH 107 solution This is significant in that only CN- appears to interact sufficiently with silver to produce a SER spectrum and no spectral signal is observed below pH 726 Figure 2 shows the SER and normal Raman spectra for KCN The SERS spectrum of 1mgml KCN in water shows a single intense somewhat broad feature at
A B
C D
SPIE-2003-5269 19
2100 cm-1 assigned to the single CequivN stretch The band is much sharper in the normal Raman spectra of the solid KCN salt at 2074 cm-1 This band does broaden and shift to 2080 cm-1 in solution (not shown) However the observed SERS frequency is attributed to interaction with silver and low frequency mode at 135 cm-1 attributed to a Ag-CN stretch (not shown) supports this conclusion
cm-1 band to a C-S stretch but the authors concede that it is in fact more likely a C-Cl stretch28 It appears that the most intense Raman bands at 648 692 and 747 cm-1 shift to 620 660 and 720 cm-1 in the SER spectra and are tentatively assigned as above The width of these bands suggests that they overlap underlying spectral features Additional bands in the Raman spectra occur at 972 1034 1049 1263 1286 1423 1442 2865 2935 and 2960 cm-1 Corresponding bands occur in the SER spectra at 964 1015 1054 1286 1410 1447 2865 and 2935 cm-1 Most of these bands are associated with alkane modes specifically the bands at approximately 1040 cm-1 to a C-C stretch 1290 cm-1 to a CH2 in-phase twist 1440 cm-1 to a CH2 wag 2865 cm-1 to a symmetric CH2 stretch and 2965 cm-1 to an asymmetric CH2 stretch The Raman and SER spectra of sulfur mustard were measured at the Edgewood center (Figure 4) Both spectra are largely similar to CEES The C-Cl and C-S bands in the Raman spectrum of HD now occur at 640 655 700 739 and 760 cm-1 and are more resolved possibly due to the increased molecular symmetry Theoretical calculations indicate that the first three bands are due to C-Cl stretching modes and the latter two to C-S stretching modes28 Only the C-Cl bands maintain significant intensity in the SER spectra occurring at 624 and 643 cm-1 which is attributed to the
Prior to measurements of HD CEES was examined by Raman and SER spectroscopy (Figure 3) CEES also known as half-mustard is essentially identical to HD except one of the chlorine end atoms is replaced by a hydrogen atom Again although not as toxic as HD CEES is a blister agent and dilute aqueous samples were prepared using appropriate safety equipment Both the Raman and SER spectra of CEES are similar and dominated by bands between 600 and 800 cm-1 These are associated with C-Cl and C-S stretching modes which are tentatively assigned to 648 and 747 cm-1 in the Raman spectra respectively The shoulder at 630 cm-1 the overlapped band at 660 cm-1 and the strong band at 692 cm-1 could also be due to these modes or their asymmetric counterparts It is worth noting that theoretical calculations assign the 692
Figure 2 A) SER and B) NR spectra of KCN Conditions A) 1 mgml in tap water 100 mW of 785 nm at sample 1-min acquisition time B) solid 300 mW of 785 nm 5-min All spectra are 8 cm-1 resolution
A
B
Figure 3 A) SER and B) NR spectra of CEES Conditions A) 1 vv (10 mgml) in MeOH 100 mW of 785 nm 1-min acquisition time B) neat 300 mW of 785 nm 5-min
A
B
Cl-CH2-CH2-S-CH2-CH3
A
B
Figure 4 A) SER and B) NR of HD Conditions A) 1mgml in tap water B) pure both 100 mW of 785 nm 1-min
Cl-CH2-CH2-S-CH2-CH2-Cl
SPIE-2003-5269 20
expected strong interaction between chlorine and silver and adds support to the assignment of this band to a C-Cl stretch Weaker overlapping bands occur at 670 692 and 724 cm-1 the latter possibly due to C-S stretching modes Again the alkane modes are apparent in the normal Raman spectra of HD but only a broad feature at 1300 to 1450 cm-1 suggests CH2 contributions in the SER spectrum Although the observed bands in the VX spectrum have not been assigned (Figure 4) a computer generated Raman spectrum29 predicts many of the same features with surprising accuracy and are used here Two intense bands at 460 and 530 cm-1 closely match predicted bands at 463 and 546 cm-1 assigned to a CH3-P=O bend and a PO2CS wag Three highly overlapped bands occur at 694 745 and 771 cm-1 matching predicted bands at 713 730 and 760 cm-1 The first
Table 2 Measured SER peak heights for the CN stretch at 2100 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
Figure 5 A) SERS and B) NR spectra of VX Conditions A) 1 vv (10 mgml) in MeOH B) pure sample both 100 mW of 785 nm 1-min acquisition time
A
B
two have been assigned to a C-S stretch and CH2bend respectively while the latter has been attributed to either a P-C stretch or an O-C-C stretch Although the 745 cm-1 band may alternatively be assigned to a C-S stretch based on the previous measurements of CEES and HD The relatively intense bands at 890 1106 1218 1445 and 1465 cm-1 also match predicted bands at 880 1108 1216 1440 and 1464 cm-1 that are assigned to a C-C stretch CH3 rock N-C3 stretch various C-H3 bends and C-H bends respectively Both the computer generated and the measured spectra contain numerous other less intense bands One is worth mentioning A unique band appears at 370 cm-1 that is predicted at 368 cm-1 and corresponds to an O-P=O bend The surface-enhanced Raman spectrum of VX is also rich with spectral features It has the unique low frequency band at 370 cm-1 as well as a second band at 380 cm-1 that is assigned tothe S-P-O bend predicted in the normal Raman spectrum at 388 cm-1 Based on the measured and predicted normal Raman spectra the following SERS assignments are given 460 cm-1 to the CH3-P=O bend 544 cm-1 to the PO2CS wag 738 cm-1 to a C-S stretch (based on arguments above) 890 cm-1
to a C-C stretch 1101 cm-1 to a CH3 rock and 1456
cm-1 to a C-H bend The ability of SERS to measure chemical agents in water containing real-world chemical interferents was tested by using the generic tap water described in Table 1 The ability to reproduce measurements was accomplished by preparing three separate water stock solutions which were used to prepare three sample solutions each of 1 mgmL KCN in the generic tap water The samples defined as vials 1-9 were then measured at 4 points per vial in a semi-random fashion such that errors associated with stock solution preparation and errors associated with instrument drift could be identified No trends were apparent that signified such systematic errors Each spectrum collected consisted of 20 averaged scans taking 16 seconds at 8 cm-1 resolution The laser power at the sample was measured periodically during the day and it ranged from 102 to 105 mW spacccc
spot Vial 1 Vial 2 Vial 41 13754 13004 128192 13519 12692 12909
1 3 13541 12721 126394 13462 12648 12651
ave 13569 12766 12755Vial 6 Vial 3 Vial 5
1 11586 14068 10757 2 11236 14402 115122 3 11376 14546 11559
4 10894 11763 11214ave 11273 13695 11261
Vial 7 Vial8 Vial 91 11198 15804 11023
2 11243 15716 88753 3 11562 15216 11218
4 11655 15084 11077ave 11415 15455 10548
CN AVG STDEV ERR1 mgmL pk ht 12526 1560 1245
stock solution
SPIE-2003-5269 21
The CN measurements consisting of 36 data points produced an average height of 125 for the 2100 cm-1 peak with a standard deviation of 156 or 125 (Table 2) The HD and VX measurements were performed precisely the same way (Tables 3 and 4) For HD the 624 cm-1 peak was used for analysis and it had an average height of 53 with a standard deviation of 068 or 129 while for VX the 544 cm-1 peak was used for analysis and it had an average height of 1051 with a standard deviation of 308 or 293 The greater error in the VX measurements can be somewhat attributed to Vial 4 which produced lower SER signal intensities But removing this vial from the data set changes the standard deviation to 233 only a modest improvement
spot Vial 1 Vial 2 Vial 41 525 609 4682 45 675 484
1 3 527 644 5094 559 807 669
ave 51525 68375 5325Vial 6 Vial 3 Vial 5
1 521 574 575 2 536 53 4492 3 509 508 451
4 565 594 379ave 53275 5515 4635
Vial 7 Vial 8 Vial 91 457 589 413
2 583 497 4273 3 544 505 4
4 528 5 509ave 528 52275 43725
HD AVG STDEV ERR1 mgmL pk ht 530 068 1291
stock solution
spot Vial 1 Vial 2 Vial 41 1464 1034 4672 1485 99 61
1 3 1491 989 5684 1041 777 553
ave 137025 9475 5495Vial 6 Vial 3 Vial 5
1 1058 942 1293 2 697 121 965
2 3 727 1206 11124 689 1254 746
ave 79275 1153 1029Vial 7 Vial 8 Vial 9
1 1127 783 163 2 1358 812 1614
3 3 1371 908 15124 1097 875 1397
ave 123825 8445 153825
VX AVG STDEV ERR1 mgmL pk ht 1051 308 2925
stock solution
4 CONCLUSIONS In this paper we examined the ability of surface-enhanced Raman spectroscopy to reproducibly measure CN HD and VX in tap water without chemical interference Both normal and surface-enhanced Raman spectra were examined to select unique bands suitable to identify and quantify these chemical agents For SER measurements the 2100 cm-1 C-N stretch was used for CN the 624 cm-1 C-Cl stretch was used for HD and the 544 cm-1 PO2CS wag was used for VX It was determined that 1 mgmL samples of each of these chemicals measured 36 times in glass vials coated with a silver-doped sol-gel reproduced measurements with standard deviations of 125 129 and 293 It was further found that the 10 chemicals added to simulate generic tap water did not interfere with or alter the SER spectra It should be noted that the concentrations used in this study were considerably greater than those required by the JSAWM program Current work involves improving SER sensitivity and designing sampling systems with better reproducibility This includes the development of fractal silver and gold structures within the sol-gel matrix and the development of chemically selective sol-gels
ACKNOWLEDGMENTS The authors are grateful for the support of the US Army (DAAD13-02-C-0015 Joint Service Agent Water Monitor program) The authors would also like to thank Janet Jensen Ronald Crosier and Kristina Gonser for helpful discussions
Table 4 Measured SER peak heights for the VX band at 544 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
Table 3 Measured SER peak heights for the HD band at 624 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
SPIE-2003-5269 22
REFERENCES 1 Jensen JL NC Wong W R Loerop SPIE 4775-03 (2001) 2 JSAWM Requirements at wwwsbccomapgeaarmymilRDAecbcrtPRODSERJSAWMjsawmhtml 3 Erickson B ldquoThe Chemical Weapons Convention Redefines Analytical Challengerdquo Anal Chem 70 397A (1998) 4 Johnston RL Hoefler CM Fargo JC and Moberley B ldquoThe Defense Nuclear Agencyrsquos
ChemicalBiochemical Weapons Agreements Verification Technology Research Development Test and Evaluation Program and its Requirements for On-Site Analysisrdquo AT-ONSITE 5-8 (1994)
5 Black RM Clarke RJ Read RW and Reid MT ldquoApplication of GC-MS and GC-tandem MS to the analysis of chemical warfare samples found to contain residues of the nerve agent sarin sulphur mustard and their degradation productsrdquo J Chrom 662 301-321 (1994)
6 USA Army Center for Health Promotion and Preventive Medicine (USACHPPM) Technical Guideline 230A 7 Hoffland LD Piffath RJ Bouck JB rdquoSpectral signatures of chemical agents and simulantsrdquo Optical
Engineering 24 982-984 (1985) 8 Braue EHJ Pannella MGrdquoCIRCLE CELL FT-IR Analysis of Chemical Warfare Agents in Aqueous Solutionsrdquo
App Spectrosc 44 1513-1520 (1990) 9 Tseng C-H Mann CK and Vickers TJ ldquoDetermination of Organics on Metal Surfaces by Raman
Spectroscopyrdquo App Spectrosc 47 1767-1771 (1993) 10 Staff Reporter ldquoUS Army Tests Lidar to Detect Biological Toxinsrdquo Photonic Spectra pg 50 December 1998 11 Christesen SD Raman cross sections of chemical agents and simulants App Spectrosc 42 318-321 (1988) 12 Weaver MJ Farquharson S Tadayyoni MA ldquoSurface-enhancement factors for Raman scattering at silver electrodesrdquo J Chem Phys 82 4867-4874 (1985) 13 Alak AM and Vo-Dinh T ldquoSERS of Organophosphorous Chemical Agentsrdquo Anal Chem 59 2149-2153 (1987) 14 Lee Y Farquharson S ldquoRapid chemical agent identification by SERSrdquo SPIE 4378 21-26 (2001) 15 Farquharson S Maksymiuk P Ong K Christesen S ldquoChemical agent identification by surface-enhanced Raman spectroscopyrdquo SPIE 4577 166-173 (2001) 16 Spencer KM Sylvia J Clauson S and Janni J ldquoSurface Enhanced Raman as a Water Monitor for Warfare
Agents in Waterrdquo SPIE 4577 158-165 (2001) 17 Tessier P Christesen S Ong K Clemente E Lenhoff A Kaler E Velev O ldquoOn-line spectroscopic characterization of sodium cyanide with nanostructured gold surface-enhanced Raman spectroscopy substratesrdquo
App Spectrosc 56 1524-1530 (2002) 18 Farquharson S W W Smith S Elliott and J F Sperry Rapid biological agent identification by surface- enhanced Raman spectroscopy SPIE 3855110-116 (1999) 19 Farquharson S WW Smith YH Lee S Elliott and J F Sperry Detection of biological signatures A
comparison of electrolytic and metal-doped sol-gel surface-enhanced Raman media SPIE 4575 62-72 (2002) 20 Guzelian AA Sylvia J Janni J Clauson S and Spencer KM ldquoSERS of whole cell bacteria and trace levels of biological moleculesrdquo SPIE 4577 182-192 (2001) 21 Shende C Inscore F Gift A Maksymiuk P Farquharson S ldquoRapid dipicolinic acid extraction from Bacillus spores detected by surface-enhanced Raman spectroscopyrdquo App Spectrosc 58 accepted 22 Lee Y and Farquharson S ldquoSERS Sample Vials Based on Sol-Gel Process for Trace Pesticide Analysisrdquo SPIE
4206 140-146 (2000) 23 Farquharson S and Lee Y ldquoTrace Drug Analysis by SERSrdquo SPIE 4200-16 (2000) 24 Lee Y Farquharson S and Rainey P M Surface-Enhanced Raman Sensor for Trace Chemical Detection in
Water SPIE 3857 76-84 (1999) 25 Lee Y Farquharson S Kwong H and Shahriari M ldquoSol-Gel Chemical Sensor for Surface-Enhanced Raman
Spectroscopyrdquo SPIE 3537 252-260 (1998) 26 Farquharson S A Gift P Maksymiuk F Inscore and W Smith ldquopH dependence of methyl phosphonic acid
dipicolinic acid and cyanide by surface-enhanced Raman spectroscopyrdquo SPIE 5269-19 (2003) 27 Lide DR Ed Handbook of Chemistry and Physics CRC Press 77th Ed (1996-7) 28 Sosa C RJ Bartlett K KuBulat and WB Person ldquoA theoretical study of harmonic vibrational frequencies and
infrared intensities of XCH2CH2SCH2CH2X and XCH2CH2SH (= H Cl)rdquo J Phys Chem 93 577-588 (1993) 29 Hameka HF and JO Jenson ldquoComputer-generated predictions of the structure and the IR and Raman spectra of
VXrdquo ERDEC-TR-065 May 1993
- Chemagents Appendicespdf
-
- SPIE2001-4575-Bioagent-SERSapdf
-
- Detection of bioagent signatures A comparison of electrolytic and metal-doped sol-gel surface-enhanced Raman media
- Stuart Farquharson Wayne Smith and Yuan Lee
-
- Real-Time Analyzers 87 Church Street East Hartford CT 06108
-
- ABSTRACT
- 4 CONCLUSIONS
- 5 ACKNOWLEDGEMENTS
- 6 REFERENCES
-
SPIE -2003-5269 124
as to CN adsorbed to two different surface sites32 Alternatively the 2140 cm-1 band could be attributed to HCN since this species dominates at lower concentrations However it is unlikely that this species would be attracted to the electropositive silver surface Further both peaks should be present at pHs between 85 and 105 but this is not observed It has also been suggested that at concentrations near and above monolayer coverage the CN- species is forced to adsorb end-on due to crowding and at lower concentrations the molecule can reorient to lie flat33 This suggests that the 2100 and 2140 cm-1 bands correspond to the end-on and flat orientations respectively As Figures 13 and 14 show the intensity of the CN stretch for the 89 mgL sample is nearly as intense as the 964 mgmL band This suggests that the Raman signal for the flat orientation is more enhanced However more extensive measurements are required to verify this point Since resent research has suggested that cyanide may be more effectively detected on gold measurements of KCN solutions were also performed using gold-doped sol-gel vials Preliminary measurements are shown in Figure 15 for samples prepared from 01 1 and 10 mgL KCN Since the pHs are 816 90 967 the resultant CN- concentrations are 63x10-3 033 and 69 mgL Initially only the highest concentration was observed and the signal intensity was significantly better than the equivalent concentration measured using silver In an effort to shift Reaction 4 to the left transforming HCN to CN- (Le Chatelierrsquos principle) KOH was added to the lower concentration samples producing solutions with pH 12 This effectively forces all of the cyanide in solution to be CN- or 01 and 10 mgL respectively More importantly the CN stretch is now observed in the SER spectra The band appears at 2125 cm-1 as has been previously reported for gold12 As calculated for MPA and DPA LODs can be estimated from this data For the three concentrations of cyanide on silver the LODs are 016 to 11 mgL for 100 mW of 785 nm laser excitation and a 10-min acquisition time For gold pH adjusted the LODs are10 to 70 microgL an improvement of more than 10 times silver Nevertheless either substrate is sufficient to meet the JSAWM goals of measuring 3 mgL in 10-minutes as the requirements form cyanide are much less stringent than the nerve agents
4 CONCLUSIONS Here we examined the surface-enhanced Raman spectral response for methyl phosphonic acid dipicolinic acid and cyanide as a function of pH It was determined that the most prominent peak at 760 cm-1 reaches a maximum intensity between pH 3 and 7 and corresponds to the MPA- species Neither the MPA nor MPA= species appear to generate a SER spectrum and consequently no spectra were observed below pH 2 or above pH 8 In this study we also found that higher laser powers could cause photodegradation of MPA signified by the exponential growth of a band at 1050 cm-1 which is tentatively assigned to phosphorous acid Unlike MPA DPA was observed at all pHs This is attributed to the dominant interaction of the pyridine functional group with silver Minor spectral changes were observed at acid pHs and were assigned to neutral DPA Like MPA SER spectra of cyanide were pH dependent No spectra were observed for the HCN species while CN- was best observed at pHs more basic than 8 Preliminary concentration studies for the three analytes allowed estimating limits of detection for MPA DPA and CN using 100 mW of 785 nm and a 10-min acquisition time of 210 165 and 70 microgL respectively Although the latter value suggests that cyanide can be measured at sufficiently low concentrations to meet JSAWM goal improvements by 100 to 200 times are required for MPA and DPA It is clear from this study that pH of the sample is important and must be taken into account when developing concentration calibration curves as well as fieldable analyzers Future work will include tailoring the sol-gel with specific functional groups to dictate sample pH and thereby optimize sensitivity
ACKNOWLEDGMENTS The authors are grateful for the support of the US Army (DAAD13-02-C-0015 Joint Service Agent Water Monitor program) The authors would also like to thank Dr Steve Christensen of the US Army SBCCOM for helpful discussions and Mr Chetan Shende of Real-Time Analyzers for assistance in development of the gold-doped sol-gels
REFERENCES 1 Erickson B Analytical Chemistry News amp Features June 1 397A (1998)
SPIE -2003-5269 125
2 Johnston RL Hoefler CM Fargo JC and Moberley B AT-ONSITE 5-8 (1994) 3 Black RM RJ Clarke RW Read and MT Reid J Chromatography 662 301-321 (1994) 4 USA Army Center for Health Promotion and Preventive Medicine (USACHPPM) Technical Guideline 230A 5 Jensen JL NC Wong W R Loerop SPIE 4775-03 (2001) 6 Hoenig SL Handbook of Chemical Warfare and Terrorism Greenwood Press Wesport CT (2002) 7 Munro NB SS Talmage GD Griffin LC Waters AP Watson JF King and V Hauschild Env Health
Persp 107 933-974 (1999) 8 Holstege CP Kirk M Sidell FR Crit Care Clin13 923-42 (1997) 9 Farquharson S P Maksymiuk K Ong and S Christesen SPIE 4577 166-173 (2001) 10 Lee Y and S Farquharson SPIE 4378 21-26 (2001) 11 Spencer KM J Sylvia S Clauson and J Janni SPIE 4577 158-165 (2001) 12 Tessier P S Christesen K Ong E Clemente A Lenhoff E Kaler and O Velev Applied Spectroscopy 56
1524-1530 (2002) 13 Farquharson S WW Smith S Elliott and JF Sperry SPIE 3533 207-214 (1998) 14 Farquharson S WW Smith S Elliott and JF Sperry SPIE 3855110-116 (1999) 15 Farquharson S WW Smith YH Lee S Elliott and JF Sperry SPIE 4575 62-72 (2002) 16 Guzelian AA J Sylvia J Janni S Clauson and KM Spencer SPIE 4577 182-192 (2001) 17 Shende C F Inscore A Gift P Maksymiuk and S Farquharson in press 18 Weaver MJ S Farquharson and MA Tadayyoni J Chem Phys 82 4867-4874 (1985) 19 Alak AM and T Vo-Dinh Analytical Chemistry 59 2149-2153 (1987) 20 Lee Y and S Farquharson SPIE 4206 140-146 (2000) 21 Farquharson S and Y Lee SPIE 4200-16 (2000) 22 Lee Y S Farquharson and P M Rainey SPIE 3857 76-84 (1999) 23 Lee Y S Farquharson H Kwong and M Shahriari SPIE 3537 252-260 (1998) 24 Wang J M Pumera G Collins and A Mulchandani Analytical Chemistry 74 6121-6125 (2002) 25 Dou X YM Jung Z-Q Cao and Y Ozaki Applied Spectroscopy 53 1440-1447 (1999) 26 Data supplied by S Christesen and K Ewing 27 Kerker M and B Thompson Eds SPIE MS 10 (1990) 28 Inglesby TV DA Henderson JG Bartlett JAMA 287 2236 (2002) 29 FW Janssen AJ Lund and LE Anderson Science 127 26 (1958) 30 Lide DR Ed Handbook of Chemistry and Physics CRC Press 77th Ed (1996-7) 31 Billmann J G Kovacs and A Otto Surf Sci 92 153 (1980) 32 Murray CA and S Bodoff Phys Rev B 32 671 (1985) 33 Kellogg D and J Pemberton J Phys Chem 91 1120 (1987)
SPIE-2003-5269 16
Chemical agent detection by surface-enhanced Raman spectroscopy
Stuart Farquharson Alan Gift Paul Maksymiuk Frank Inscore and Wayne Smith
Real-Time Analyzers 87 Church Street East Hartford CT 06108
Kevin Morrisey and Steven D Christesen US Army SBCCOM Aberdeen Proving Ground MD 21010
ABSTRACT
In the past decade the Unites States and its allies have been challenged by a different kind of warfare exemplified by the terrorist attacks of September 11 2001 Although suicide bombings are the most often used form of terror military personnel must consider a wide range of attack scenarios Among these is the intentional poisoning of water supplies to obstruct military operations in Afghanistan and Iraq To counter such attacks the military is developing portable analyzers that can identify and quantify potential chemical agents in water supplies at microgram per liter concentrations within 10 minutes To aid this effort we have been investigating the value of a surface-enhanced Raman spectroscopy based portable analyzer In particular we have been developing silver-doped sol-gels to generate SER spectra of chemical agents and their hydrolysis products Here we present SER spectra of several chemical agents measured in a generic tap water Repeat measurements were performed to establish statistical error associated with SERS obtained using the sol-gel coated vials Keywords Chemical agents chemical agent detection SERS Raman spectroscopy
1 INTRODUCTION
In the autumn of 2001 terrorism within US borders became a sobering reality While extensive efforts are being implemented to secure the homeland US and Coalition forces in Afghanistan and Iraq are constantly faced with terrorist attacks In addition to car-bombings the military has identified several non-traditional attack scenarios including poisoning of water supplies by chemical warfare agents (CWAs) To counter this threat the Department of Defense has been investigating numerous approaches to detect such attacks under the auspices of the Joint Service Agent Water Monitor (JSAWM) program1 The JSAWM must be capable of identifying the chemical agents with no false-positives and quantifying the agents at microgram per liter concentrations within 10 minutes2 This includes the analysis of drinking water supplies distribution and storage systems Currently colorimetric paper is used to detect agents on-site while gas chromatography combined with mass spectrometry (GCMS) is used in mobile support laboratories However both methods have severe drawbacks The paper changes color in response to contact with many chemicals besides CWAs causing a high incidence of false positives3 while GCMS although very chemically specific requires up to an hour to perform and regular re-calibration456 Vibrational spectroscopy has also been investigated as a potential method of rapidly detecting CWAs7-
11 as early as 1985 in the case of infrared spectroscopy8 and 1988 in the case of Raman spectroscopy11 Again however these techniques also have limitations when it comes to measuring trace poisons in water Infrared spectra would be dominated by the very strong absorption of water which would obscure absorptions by most other chemicals present Whereas Raman spectroscopy is simply not a very sensitive technique and detection limits are typically grams per liter Surface-enhanced Raman spectroscopy (SERS) offers several advantages over conventional vibrational methods and that may provide the necessary sensitivity required for detecting trace quantities of chemical agents in water SERS employs the interaction of surface plasmon modes of metal particles with target analytes to increase scattering efficiency by as much as 1 million times12 In 1987 the potential of this approach to measure CWAs was demonstrated by measuring a series of pesticides13 Several of these organophosphonates have chemical structures similar to CWAs in particular P=O functional groups In the past few years we and others have further explored the ability of SERS to detect CWAs14-17 and even bioagents 18-21 We have been employing silver-doped sol-gels to promote the SER effect
SPIE-2003-5269 17
in these studies The porous silica network of the sol-gel matrix offers a unique environment for immobilizing and stabilizing SER-active metal particles22-25 The silver-doped sol-gels have been coated on the internal walls of standard glass vials to allow rapid SER analysis of small sample quantities without preparation We have measured over 100 chemicals with enhancements of 104 to 106 demonstrated reversible measurements in a flowing system reproducible measurements from vial-to-vial and measurements in multiple solvents including water21-25 Previously we used these vials to perform preliminary measurements of cyanide (CN) and methylphosphonic acid (MPA) Most of the nerve agents form MPA during hydrolysis while Tabun forms CN a chemical agent in its own right In another paper including in these proceedings we examined the limits of detection (LOD) for MPA by measuring a series of concentrations down to 50 mgL and estimated a limit of detection of 100 microgL26 These measurements provide encouragement in that SERS may satisfy the needs of the JSAWM To further establish the viability of SERS in particular silver-doped sol-gels here we present analysis of cyanide mustard and VX in tap water The measurements performed at the US Armyrsquos Edgewood Chemical Biological Center Aberdeen MD also included numerous repeat measurements to establish reproducibility
2 EXPERIMENTAL 2a General All chemicals including potassium cyanide 2-chloroethylethyl sulfide and those used to prepare the silver-doped and gold-doped sol-gel coated vials were obtained and used as received from Sigma-Aldrich All samples were prepared in a chemical hood using HPLC grade water unless otherwise noted (Fischer Scientific Fair Lawn NJ) for SERS measurements Once prepared the samples were transferred into the silver-doped sol-gel vials (Real-Time Analyzers Inc East Hartford CT) The vials were coated in a manner similar to that previously reported by adding ammonium hydroxide to a solution of silver nitrate tetramethyl orthosilicate (TMOS) and methanol22 The two precursor solutions were prepared mixed and transferred to 2-mL glass vials dried and heated After sol-gel formation the incorporated metal ions were reduced with dilute sodium borohydride (1mgmL) which is followed by a water wash to remove residual reducing agent After the resultant analyte solution was introduced the SER-active vial was fixed horizontally to an XY positioning stage (Conix Research Springfield OR) just inside the focal point of an f07 aspheric lens The lens focused the beam into the sample and collected the scattered radiation back along the same axis A dichroic filter (Omega Optical Brattleborough VT) was used to reflect the excitation laser to the lens and pass the Raman scattered radiation collected by the lens An f2 achromat was used to collimate the laser beam exiting a 200 microm core diameter source fiber optic while a second f2 achromat was used to focus the scattered radiation into a 365 microm fiber optic (Spectran Avon CT) A short pass filter was placed in the excitation beam path to block the silicon Raman scattering generated in the source fiber from reflecting off sampling optics and reaching the detector A long pass filter was placed in the collection beam path to block the sample Rayleigh scattering from reaching the detector A 785 nm diode laser (Process Instruments Inc model 785-600 Salt Lake City UT) was used to deliver 100 to 150 mW of power to the sample A Fourier transform Raman spectrometer (Real-Time Analyzers model IRA-785 East Hartford CT) and a silicon photo-avalanche detector (Perkin Elmer model C30902S Stamford CT) were used to acquire the SER spectra
2b Edgewood Chemical Biological Center The surface-enhanced Raman spectral measurements at the US Armyrsquos Chemical Biological center presented here were all performed on September 12 2003 To expedite measurements a plate was machined to hold up to 12 SER-active sample vials (Figure 1) The plate fit a standard XY plate reader that could be programmed Pure KCN bis-(2-chloroethyl)sulfide (distilled mustard gas HD) and ethyl S-2-diisopropylamino ethyl methylphosphonothioate (VX) were obtained on-site and used to prepare 1 gL tap water solutions in a chemical hood with appropriate safety equipment Simulated tap water was prepared by adding 10 chemicals most often found in tap water at appropriate concentrations to distilled water (Table 1) SER measurements were also performed in a chemical hood For added safety the FT-Raman instrument was placed outside the laboratory and 30 foot fiber optic and electrical cables were used to allow remote SERS measurements and plate manipulation For each experiment 1gL samples were prepared and added to 9 individual vials which were then loaded on the plate In some cases a tenth vial was included as a blank
Table 1 Generic Tap Water Composition
Compound mgL NaHCO3 100 CaSO4 27 MgSO4bull7H2O 67 NaNO3 1 Fulvic Acid 1 K2HPO4 07 KH2PO4 03 (NH4)2HSO4 001 NaCl 001 FeSO4 0001 pH 76-78
SPIE-2003-5269 18
A software program was written that allowed selecting the sequence that the vials were measured the number of positions along the length of the vials to measure (1 to 5) and the number of scans to co-add During sample analysis the program displayed the vial being analyzed the point being analyzed and the spectrum as it was being acquired Once all the data was collected a second software program was written to rapidly analyze the data The spectra collected for all the vials on a plate could be loaded at one time and then the spectra for each point could be displayed simultaneously or separately The user could then select the Raman peak to analyze in terms of peak height or area This was accomplished by selecting points on either side of the peak to define a baseline of zero The peak height or area could then be computed for all of the spectra loaded and then exported to a spreadsheet for statistical analysis
Figure 1 A) Vial Holder 6 slots to hold 2 vials each end-to-end B) Measurement Configuration Program user selects vials to measure sequence number of points per vial (1 to 5) and number of scans per point C) Spectral Acquisition Program shows spectrum being collected which vial and position D) Spectra Analysis Program user selects spectra to analyze by plate vial and point (s) as well as two wavenumbers defining the peak and the baseline to subtract The image is of 5 repeat measurements of 10 mgL KCN in generic tap water 16 sec each 100 mW of 785 nm
3 RESULTS AND DISCUSSION Raman and surface-enhanced Raman spectra were obtained for potassium cyanide bis-(2-chloroethyl)sulfide and ethyl S-2-diisopropylamino ethyl methylphosphonothioate representing three classes of chemical agents cyanides mustards and nerve agents respectively Spectra were also obtained for 2-chloroethyl ethyl sulfide (CEES) a structural analogue to HD which was included in the study to aid in assigning spectral bands KCN salt was used for cyanide experiments to avoid the increased hazards of handling HCN gas KCN completely dissolves in water forming its conjugate acid HCN according to its Ka of 615x10-1027 and at a concentration of 1 mgmL results in a pH 107 solution This is significant in that only CN- appears to interact sufficiently with silver to produce a SER spectrum and no spectral signal is observed below pH 726 Figure 2 shows the SER and normal Raman spectra for KCN The SERS spectrum of 1mgml KCN in water shows a single intense somewhat broad feature at
A B
C D
SPIE-2003-5269 19
2100 cm-1 assigned to the single CequivN stretch The band is much sharper in the normal Raman spectra of the solid KCN salt at 2074 cm-1 This band does broaden and shift to 2080 cm-1 in solution (not shown) However the observed SERS frequency is attributed to interaction with silver and low frequency mode at 135 cm-1 attributed to a Ag-CN stretch (not shown) supports this conclusion
cm-1 band to a C-S stretch but the authors concede that it is in fact more likely a C-Cl stretch28 It appears that the most intense Raman bands at 648 692 and 747 cm-1 shift to 620 660 and 720 cm-1 in the SER spectra and are tentatively assigned as above The width of these bands suggests that they overlap underlying spectral features Additional bands in the Raman spectra occur at 972 1034 1049 1263 1286 1423 1442 2865 2935 and 2960 cm-1 Corresponding bands occur in the SER spectra at 964 1015 1054 1286 1410 1447 2865 and 2935 cm-1 Most of these bands are associated with alkane modes specifically the bands at approximately 1040 cm-1 to a C-C stretch 1290 cm-1 to a CH2 in-phase twist 1440 cm-1 to a CH2 wag 2865 cm-1 to a symmetric CH2 stretch and 2965 cm-1 to an asymmetric CH2 stretch The Raman and SER spectra of sulfur mustard were measured at the Edgewood center (Figure 4) Both spectra are largely similar to CEES The C-Cl and C-S bands in the Raman spectrum of HD now occur at 640 655 700 739 and 760 cm-1 and are more resolved possibly due to the increased molecular symmetry Theoretical calculations indicate that the first three bands are due to C-Cl stretching modes and the latter two to C-S stretching modes28 Only the C-Cl bands maintain significant intensity in the SER spectra occurring at 624 and 643 cm-1 which is attributed to the
Prior to measurements of HD CEES was examined by Raman and SER spectroscopy (Figure 3) CEES also known as half-mustard is essentially identical to HD except one of the chlorine end atoms is replaced by a hydrogen atom Again although not as toxic as HD CEES is a blister agent and dilute aqueous samples were prepared using appropriate safety equipment Both the Raman and SER spectra of CEES are similar and dominated by bands between 600 and 800 cm-1 These are associated with C-Cl and C-S stretching modes which are tentatively assigned to 648 and 747 cm-1 in the Raman spectra respectively The shoulder at 630 cm-1 the overlapped band at 660 cm-1 and the strong band at 692 cm-1 could also be due to these modes or their asymmetric counterparts It is worth noting that theoretical calculations assign the 692
Figure 2 A) SER and B) NR spectra of KCN Conditions A) 1 mgml in tap water 100 mW of 785 nm at sample 1-min acquisition time B) solid 300 mW of 785 nm 5-min All spectra are 8 cm-1 resolution
A
B
Figure 3 A) SER and B) NR spectra of CEES Conditions A) 1 vv (10 mgml) in MeOH 100 mW of 785 nm 1-min acquisition time B) neat 300 mW of 785 nm 5-min
A
B
Cl-CH2-CH2-S-CH2-CH3
A
B
Figure 4 A) SER and B) NR of HD Conditions A) 1mgml in tap water B) pure both 100 mW of 785 nm 1-min
Cl-CH2-CH2-S-CH2-CH2-Cl
SPIE-2003-5269 20
expected strong interaction between chlorine and silver and adds support to the assignment of this band to a C-Cl stretch Weaker overlapping bands occur at 670 692 and 724 cm-1 the latter possibly due to C-S stretching modes Again the alkane modes are apparent in the normal Raman spectra of HD but only a broad feature at 1300 to 1450 cm-1 suggests CH2 contributions in the SER spectrum Although the observed bands in the VX spectrum have not been assigned (Figure 4) a computer generated Raman spectrum29 predicts many of the same features with surprising accuracy and are used here Two intense bands at 460 and 530 cm-1 closely match predicted bands at 463 and 546 cm-1 assigned to a CH3-P=O bend and a PO2CS wag Three highly overlapped bands occur at 694 745 and 771 cm-1 matching predicted bands at 713 730 and 760 cm-1 The first
Table 2 Measured SER peak heights for the CN stretch at 2100 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
Figure 5 A) SERS and B) NR spectra of VX Conditions A) 1 vv (10 mgml) in MeOH B) pure sample both 100 mW of 785 nm 1-min acquisition time
A
B
two have been assigned to a C-S stretch and CH2bend respectively while the latter has been attributed to either a P-C stretch or an O-C-C stretch Although the 745 cm-1 band may alternatively be assigned to a C-S stretch based on the previous measurements of CEES and HD The relatively intense bands at 890 1106 1218 1445 and 1465 cm-1 also match predicted bands at 880 1108 1216 1440 and 1464 cm-1 that are assigned to a C-C stretch CH3 rock N-C3 stretch various C-H3 bends and C-H bends respectively Both the computer generated and the measured spectra contain numerous other less intense bands One is worth mentioning A unique band appears at 370 cm-1 that is predicted at 368 cm-1 and corresponds to an O-P=O bend The surface-enhanced Raman spectrum of VX is also rich with spectral features It has the unique low frequency band at 370 cm-1 as well as a second band at 380 cm-1 that is assigned tothe S-P-O bend predicted in the normal Raman spectrum at 388 cm-1 Based on the measured and predicted normal Raman spectra the following SERS assignments are given 460 cm-1 to the CH3-P=O bend 544 cm-1 to the PO2CS wag 738 cm-1 to a C-S stretch (based on arguments above) 890 cm-1
to a C-C stretch 1101 cm-1 to a CH3 rock and 1456
cm-1 to a C-H bend The ability of SERS to measure chemical agents in water containing real-world chemical interferents was tested by using the generic tap water described in Table 1 The ability to reproduce measurements was accomplished by preparing three separate water stock solutions which were used to prepare three sample solutions each of 1 mgmL KCN in the generic tap water The samples defined as vials 1-9 were then measured at 4 points per vial in a semi-random fashion such that errors associated with stock solution preparation and errors associated with instrument drift could be identified No trends were apparent that signified such systematic errors Each spectrum collected consisted of 20 averaged scans taking 16 seconds at 8 cm-1 resolution The laser power at the sample was measured periodically during the day and it ranged from 102 to 105 mW spacccc
spot Vial 1 Vial 2 Vial 41 13754 13004 128192 13519 12692 12909
1 3 13541 12721 126394 13462 12648 12651
ave 13569 12766 12755Vial 6 Vial 3 Vial 5
1 11586 14068 10757 2 11236 14402 115122 3 11376 14546 11559
4 10894 11763 11214ave 11273 13695 11261
Vial 7 Vial8 Vial 91 11198 15804 11023
2 11243 15716 88753 3 11562 15216 11218
4 11655 15084 11077ave 11415 15455 10548
CN AVG STDEV ERR1 mgmL pk ht 12526 1560 1245
stock solution
SPIE-2003-5269 21
The CN measurements consisting of 36 data points produced an average height of 125 for the 2100 cm-1 peak with a standard deviation of 156 or 125 (Table 2) The HD and VX measurements were performed precisely the same way (Tables 3 and 4) For HD the 624 cm-1 peak was used for analysis and it had an average height of 53 with a standard deviation of 068 or 129 while for VX the 544 cm-1 peak was used for analysis and it had an average height of 1051 with a standard deviation of 308 or 293 The greater error in the VX measurements can be somewhat attributed to Vial 4 which produced lower SER signal intensities But removing this vial from the data set changes the standard deviation to 233 only a modest improvement
spot Vial 1 Vial 2 Vial 41 525 609 4682 45 675 484
1 3 527 644 5094 559 807 669
ave 51525 68375 5325Vial 6 Vial 3 Vial 5
1 521 574 575 2 536 53 4492 3 509 508 451
4 565 594 379ave 53275 5515 4635
Vial 7 Vial 8 Vial 91 457 589 413
2 583 497 4273 3 544 505 4
4 528 5 509ave 528 52275 43725
HD AVG STDEV ERR1 mgmL pk ht 530 068 1291
stock solution
spot Vial 1 Vial 2 Vial 41 1464 1034 4672 1485 99 61
1 3 1491 989 5684 1041 777 553
ave 137025 9475 5495Vial 6 Vial 3 Vial 5
1 1058 942 1293 2 697 121 965
2 3 727 1206 11124 689 1254 746
ave 79275 1153 1029Vial 7 Vial 8 Vial 9
1 1127 783 163 2 1358 812 1614
3 3 1371 908 15124 1097 875 1397
ave 123825 8445 153825
VX AVG STDEV ERR1 mgmL pk ht 1051 308 2925
stock solution
4 CONCLUSIONS In this paper we examined the ability of surface-enhanced Raman spectroscopy to reproducibly measure CN HD and VX in tap water without chemical interference Both normal and surface-enhanced Raman spectra were examined to select unique bands suitable to identify and quantify these chemical agents For SER measurements the 2100 cm-1 C-N stretch was used for CN the 624 cm-1 C-Cl stretch was used for HD and the 544 cm-1 PO2CS wag was used for VX It was determined that 1 mgmL samples of each of these chemicals measured 36 times in glass vials coated with a silver-doped sol-gel reproduced measurements with standard deviations of 125 129 and 293 It was further found that the 10 chemicals added to simulate generic tap water did not interfere with or alter the SER spectra It should be noted that the concentrations used in this study were considerably greater than those required by the JSAWM program Current work involves improving SER sensitivity and designing sampling systems with better reproducibility This includes the development of fractal silver and gold structures within the sol-gel matrix and the development of chemically selective sol-gels
ACKNOWLEDGMENTS The authors are grateful for the support of the US Army (DAAD13-02-C-0015 Joint Service Agent Water Monitor program) The authors would also like to thank Janet Jensen Ronald Crosier and Kristina Gonser for helpful discussions
Table 4 Measured SER peak heights for the VX band at 544 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
Table 3 Measured SER peak heights for the HD band at 624 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
SPIE-2003-5269 22
REFERENCES 1 Jensen JL NC Wong W R Loerop SPIE 4775-03 (2001) 2 JSAWM Requirements at wwwsbccomapgeaarmymilRDAecbcrtPRODSERJSAWMjsawmhtml 3 Erickson B ldquoThe Chemical Weapons Convention Redefines Analytical Challengerdquo Anal Chem 70 397A (1998) 4 Johnston RL Hoefler CM Fargo JC and Moberley B ldquoThe Defense Nuclear Agencyrsquos
ChemicalBiochemical Weapons Agreements Verification Technology Research Development Test and Evaluation Program and its Requirements for On-Site Analysisrdquo AT-ONSITE 5-8 (1994)
5 Black RM Clarke RJ Read RW and Reid MT ldquoApplication of GC-MS and GC-tandem MS to the analysis of chemical warfare samples found to contain residues of the nerve agent sarin sulphur mustard and their degradation productsrdquo J Chrom 662 301-321 (1994)
6 USA Army Center for Health Promotion and Preventive Medicine (USACHPPM) Technical Guideline 230A 7 Hoffland LD Piffath RJ Bouck JB rdquoSpectral signatures of chemical agents and simulantsrdquo Optical
Engineering 24 982-984 (1985) 8 Braue EHJ Pannella MGrdquoCIRCLE CELL FT-IR Analysis of Chemical Warfare Agents in Aqueous Solutionsrdquo
App Spectrosc 44 1513-1520 (1990) 9 Tseng C-H Mann CK and Vickers TJ ldquoDetermination of Organics on Metal Surfaces by Raman
Spectroscopyrdquo App Spectrosc 47 1767-1771 (1993) 10 Staff Reporter ldquoUS Army Tests Lidar to Detect Biological Toxinsrdquo Photonic Spectra pg 50 December 1998 11 Christesen SD Raman cross sections of chemical agents and simulants App Spectrosc 42 318-321 (1988) 12 Weaver MJ Farquharson S Tadayyoni MA ldquoSurface-enhancement factors for Raman scattering at silver electrodesrdquo J Chem Phys 82 4867-4874 (1985) 13 Alak AM and Vo-Dinh T ldquoSERS of Organophosphorous Chemical Agentsrdquo Anal Chem 59 2149-2153 (1987) 14 Lee Y Farquharson S ldquoRapid chemical agent identification by SERSrdquo SPIE 4378 21-26 (2001) 15 Farquharson S Maksymiuk P Ong K Christesen S ldquoChemical agent identification by surface-enhanced Raman spectroscopyrdquo SPIE 4577 166-173 (2001) 16 Spencer KM Sylvia J Clauson S and Janni J ldquoSurface Enhanced Raman as a Water Monitor for Warfare
Agents in Waterrdquo SPIE 4577 158-165 (2001) 17 Tessier P Christesen S Ong K Clemente E Lenhoff A Kaler E Velev O ldquoOn-line spectroscopic characterization of sodium cyanide with nanostructured gold surface-enhanced Raman spectroscopy substratesrdquo
App Spectrosc 56 1524-1530 (2002) 18 Farquharson S W W Smith S Elliott and J F Sperry Rapid biological agent identification by surface- enhanced Raman spectroscopy SPIE 3855110-116 (1999) 19 Farquharson S WW Smith YH Lee S Elliott and J F Sperry Detection of biological signatures A
comparison of electrolytic and metal-doped sol-gel surface-enhanced Raman media SPIE 4575 62-72 (2002) 20 Guzelian AA Sylvia J Janni J Clauson S and Spencer KM ldquoSERS of whole cell bacteria and trace levels of biological moleculesrdquo SPIE 4577 182-192 (2001) 21 Shende C Inscore F Gift A Maksymiuk P Farquharson S ldquoRapid dipicolinic acid extraction from Bacillus spores detected by surface-enhanced Raman spectroscopyrdquo App Spectrosc 58 accepted 22 Lee Y and Farquharson S ldquoSERS Sample Vials Based on Sol-Gel Process for Trace Pesticide Analysisrdquo SPIE
4206 140-146 (2000) 23 Farquharson S and Lee Y ldquoTrace Drug Analysis by SERSrdquo SPIE 4200-16 (2000) 24 Lee Y Farquharson S and Rainey P M Surface-Enhanced Raman Sensor for Trace Chemical Detection in
Water SPIE 3857 76-84 (1999) 25 Lee Y Farquharson S Kwong H and Shahriari M ldquoSol-Gel Chemical Sensor for Surface-Enhanced Raman
Spectroscopyrdquo SPIE 3537 252-260 (1998) 26 Farquharson S A Gift P Maksymiuk F Inscore and W Smith ldquopH dependence of methyl phosphonic acid
dipicolinic acid and cyanide by surface-enhanced Raman spectroscopyrdquo SPIE 5269-19 (2003) 27 Lide DR Ed Handbook of Chemistry and Physics CRC Press 77th Ed (1996-7) 28 Sosa C RJ Bartlett K KuBulat and WB Person ldquoA theoretical study of harmonic vibrational frequencies and
infrared intensities of XCH2CH2SCH2CH2X and XCH2CH2SH (= H Cl)rdquo J Phys Chem 93 577-588 (1993) 29 Hameka HF and JO Jenson ldquoComputer-generated predictions of the structure and the IR and Raman spectra of
VXrdquo ERDEC-TR-065 May 1993
- Chemagents Appendicespdf
-
- SPIE2001-4575-Bioagent-SERSapdf
-
- Detection of bioagent signatures A comparison of electrolytic and metal-doped sol-gel surface-enhanced Raman media
- Stuart Farquharson Wayne Smith and Yuan Lee
-
- Real-Time Analyzers 87 Church Street East Hartford CT 06108
-
- ABSTRACT
- 4 CONCLUSIONS
- 5 ACKNOWLEDGEMENTS
- 6 REFERENCES
-
SPIE -2003-5269 125
2 Johnston RL Hoefler CM Fargo JC and Moberley B AT-ONSITE 5-8 (1994) 3 Black RM RJ Clarke RW Read and MT Reid J Chromatography 662 301-321 (1994) 4 USA Army Center for Health Promotion and Preventive Medicine (USACHPPM) Technical Guideline 230A 5 Jensen JL NC Wong W R Loerop SPIE 4775-03 (2001) 6 Hoenig SL Handbook of Chemical Warfare and Terrorism Greenwood Press Wesport CT (2002) 7 Munro NB SS Talmage GD Griffin LC Waters AP Watson JF King and V Hauschild Env Health
Persp 107 933-974 (1999) 8 Holstege CP Kirk M Sidell FR Crit Care Clin13 923-42 (1997) 9 Farquharson S P Maksymiuk K Ong and S Christesen SPIE 4577 166-173 (2001) 10 Lee Y and S Farquharson SPIE 4378 21-26 (2001) 11 Spencer KM J Sylvia S Clauson and J Janni SPIE 4577 158-165 (2001) 12 Tessier P S Christesen K Ong E Clemente A Lenhoff E Kaler and O Velev Applied Spectroscopy 56
1524-1530 (2002) 13 Farquharson S WW Smith S Elliott and JF Sperry SPIE 3533 207-214 (1998) 14 Farquharson S WW Smith S Elliott and JF Sperry SPIE 3855110-116 (1999) 15 Farquharson S WW Smith YH Lee S Elliott and JF Sperry SPIE 4575 62-72 (2002) 16 Guzelian AA J Sylvia J Janni S Clauson and KM Spencer SPIE 4577 182-192 (2001) 17 Shende C F Inscore A Gift P Maksymiuk and S Farquharson in press 18 Weaver MJ S Farquharson and MA Tadayyoni J Chem Phys 82 4867-4874 (1985) 19 Alak AM and T Vo-Dinh Analytical Chemistry 59 2149-2153 (1987) 20 Lee Y and S Farquharson SPIE 4206 140-146 (2000) 21 Farquharson S and Y Lee SPIE 4200-16 (2000) 22 Lee Y S Farquharson and P M Rainey SPIE 3857 76-84 (1999) 23 Lee Y S Farquharson H Kwong and M Shahriari SPIE 3537 252-260 (1998) 24 Wang J M Pumera G Collins and A Mulchandani Analytical Chemistry 74 6121-6125 (2002) 25 Dou X YM Jung Z-Q Cao and Y Ozaki Applied Spectroscopy 53 1440-1447 (1999) 26 Data supplied by S Christesen and K Ewing 27 Kerker M and B Thompson Eds SPIE MS 10 (1990) 28 Inglesby TV DA Henderson JG Bartlett JAMA 287 2236 (2002) 29 FW Janssen AJ Lund and LE Anderson Science 127 26 (1958) 30 Lide DR Ed Handbook of Chemistry and Physics CRC Press 77th Ed (1996-7) 31 Billmann J G Kovacs and A Otto Surf Sci 92 153 (1980) 32 Murray CA and S Bodoff Phys Rev B 32 671 (1985) 33 Kellogg D and J Pemberton J Phys Chem 91 1120 (1987)
SPIE-2003-5269 16
Chemical agent detection by surface-enhanced Raman spectroscopy
Stuart Farquharson Alan Gift Paul Maksymiuk Frank Inscore and Wayne Smith
Real-Time Analyzers 87 Church Street East Hartford CT 06108
Kevin Morrisey and Steven D Christesen US Army SBCCOM Aberdeen Proving Ground MD 21010
ABSTRACT
In the past decade the Unites States and its allies have been challenged by a different kind of warfare exemplified by the terrorist attacks of September 11 2001 Although suicide bombings are the most often used form of terror military personnel must consider a wide range of attack scenarios Among these is the intentional poisoning of water supplies to obstruct military operations in Afghanistan and Iraq To counter such attacks the military is developing portable analyzers that can identify and quantify potential chemical agents in water supplies at microgram per liter concentrations within 10 minutes To aid this effort we have been investigating the value of a surface-enhanced Raman spectroscopy based portable analyzer In particular we have been developing silver-doped sol-gels to generate SER spectra of chemical agents and their hydrolysis products Here we present SER spectra of several chemical agents measured in a generic tap water Repeat measurements were performed to establish statistical error associated with SERS obtained using the sol-gel coated vials Keywords Chemical agents chemical agent detection SERS Raman spectroscopy
1 INTRODUCTION
In the autumn of 2001 terrorism within US borders became a sobering reality While extensive efforts are being implemented to secure the homeland US and Coalition forces in Afghanistan and Iraq are constantly faced with terrorist attacks In addition to car-bombings the military has identified several non-traditional attack scenarios including poisoning of water supplies by chemical warfare agents (CWAs) To counter this threat the Department of Defense has been investigating numerous approaches to detect such attacks under the auspices of the Joint Service Agent Water Monitor (JSAWM) program1 The JSAWM must be capable of identifying the chemical agents with no false-positives and quantifying the agents at microgram per liter concentrations within 10 minutes2 This includes the analysis of drinking water supplies distribution and storage systems Currently colorimetric paper is used to detect agents on-site while gas chromatography combined with mass spectrometry (GCMS) is used in mobile support laboratories However both methods have severe drawbacks The paper changes color in response to contact with many chemicals besides CWAs causing a high incidence of false positives3 while GCMS although very chemically specific requires up to an hour to perform and regular re-calibration456 Vibrational spectroscopy has also been investigated as a potential method of rapidly detecting CWAs7-
11 as early as 1985 in the case of infrared spectroscopy8 and 1988 in the case of Raman spectroscopy11 Again however these techniques also have limitations when it comes to measuring trace poisons in water Infrared spectra would be dominated by the very strong absorption of water which would obscure absorptions by most other chemicals present Whereas Raman spectroscopy is simply not a very sensitive technique and detection limits are typically grams per liter Surface-enhanced Raman spectroscopy (SERS) offers several advantages over conventional vibrational methods and that may provide the necessary sensitivity required for detecting trace quantities of chemical agents in water SERS employs the interaction of surface plasmon modes of metal particles with target analytes to increase scattering efficiency by as much as 1 million times12 In 1987 the potential of this approach to measure CWAs was demonstrated by measuring a series of pesticides13 Several of these organophosphonates have chemical structures similar to CWAs in particular P=O functional groups In the past few years we and others have further explored the ability of SERS to detect CWAs14-17 and even bioagents 18-21 We have been employing silver-doped sol-gels to promote the SER effect
SPIE-2003-5269 17
in these studies The porous silica network of the sol-gel matrix offers a unique environment for immobilizing and stabilizing SER-active metal particles22-25 The silver-doped sol-gels have been coated on the internal walls of standard glass vials to allow rapid SER analysis of small sample quantities without preparation We have measured over 100 chemicals with enhancements of 104 to 106 demonstrated reversible measurements in a flowing system reproducible measurements from vial-to-vial and measurements in multiple solvents including water21-25 Previously we used these vials to perform preliminary measurements of cyanide (CN) and methylphosphonic acid (MPA) Most of the nerve agents form MPA during hydrolysis while Tabun forms CN a chemical agent in its own right In another paper including in these proceedings we examined the limits of detection (LOD) for MPA by measuring a series of concentrations down to 50 mgL and estimated a limit of detection of 100 microgL26 These measurements provide encouragement in that SERS may satisfy the needs of the JSAWM To further establish the viability of SERS in particular silver-doped sol-gels here we present analysis of cyanide mustard and VX in tap water The measurements performed at the US Armyrsquos Edgewood Chemical Biological Center Aberdeen MD also included numerous repeat measurements to establish reproducibility
2 EXPERIMENTAL 2a General All chemicals including potassium cyanide 2-chloroethylethyl sulfide and those used to prepare the silver-doped and gold-doped sol-gel coated vials were obtained and used as received from Sigma-Aldrich All samples were prepared in a chemical hood using HPLC grade water unless otherwise noted (Fischer Scientific Fair Lawn NJ) for SERS measurements Once prepared the samples were transferred into the silver-doped sol-gel vials (Real-Time Analyzers Inc East Hartford CT) The vials were coated in a manner similar to that previously reported by adding ammonium hydroxide to a solution of silver nitrate tetramethyl orthosilicate (TMOS) and methanol22 The two precursor solutions were prepared mixed and transferred to 2-mL glass vials dried and heated After sol-gel formation the incorporated metal ions were reduced with dilute sodium borohydride (1mgmL) which is followed by a water wash to remove residual reducing agent After the resultant analyte solution was introduced the SER-active vial was fixed horizontally to an XY positioning stage (Conix Research Springfield OR) just inside the focal point of an f07 aspheric lens The lens focused the beam into the sample and collected the scattered radiation back along the same axis A dichroic filter (Omega Optical Brattleborough VT) was used to reflect the excitation laser to the lens and pass the Raman scattered radiation collected by the lens An f2 achromat was used to collimate the laser beam exiting a 200 microm core diameter source fiber optic while a second f2 achromat was used to focus the scattered radiation into a 365 microm fiber optic (Spectran Avon CT) A short pass filter was placed in the excitation beam path to block the silicon Raman scattering generated in the source fiber from reflecting off sampling optics and reaching the detector A long pass filter was placed in the collection beam path to block the sample Rayleigh scattering from reaching the detector A 785 nm diode laser (Process Instruments Inc model 785-600 Salt Lake City UT) was used to deliver 100 to 150 mW of power to the sample A Fourier transform Raman spectrometer (Real-Time Analyzers model IRA-785 East Hartford CT) and a silicon photo-avalanche detector (Perkin Elmer model C30902S Stamford CT) were used to acquire the SER spectra
2b Edgewood Chemical Biological Center The surface-enhanced Raman spectral measurements at the US Armyrsquos Chemical Biological center presented here were all performed on September 12 2003 To expedite measurements a plate was machined to hold up to 12 SER-active sample vials (Figure 1) The plate fit a standard XY plate reader that could be programmed Pure KCN bis-(2-chloroethyl)sulfide (distilled mustard gas HD) and ethyl S-2-diisopropylamino ethyl methylphosphonothioate (VX) were obtained on-site and used to prepare 1 gL tap water solutions in a chemical hood with appropriate safety equipment Simulated tap water was prepared by adding 10 chemicals most often found in tap water at appropriate concentrations to distilled water (Table 1) SER measurements were also performed in a chemical hood For added safety the FT-Raman instrument was placed outside the laboratory and 30 foot fiber optic and electrical cables were used to allow remote SERS measurements and plate manipulation For each experiment 1gL samples were prepared and added to 9 individual vials which were then loaded on the plate In some cases a tenth vial was included as a blank
Table 1 Generic Tap Water Composition
Compound mgL NaHCO3 100 CaSO4 27 MgSO4bull7H2O 67 NaNO3 1 Fulvic Acid 1 K2HPO4 07 KH2PO4 03 (NH4)2HSO4 001 NaCl 001 FeSO4 0001 pH 76-78
SPIE-2003-5269 18
A software program was written that allowed selecting the sequence that the vials were measured the number of positions along the length of the vials to measure (1 to 5) and the number of scans to co-add During sample analysis the program displayed the vial being analyzed the point being analyzed and the spectrum as it was being acquired Once all the data was collected a second software program was written to rapidly analyze the data The spectra collected for all the vials on a plate could be loaded at one time and then the spectra for each point could be displayed simultaneously or separately The user could then select the Raman peak to analyze in terms of peak height or area This was accomplished by selecting points on either side of the peak to define a baseline of zero The peak height or area could then be computed for all of the spectra loaded and then exported to a spreadsheet for statistical analysis
Figure 1 A) Vial Holder 6 slots to hold 2 vials each end-to-end B) Measurement Configuration Program user selects vials to measure sequence number of points per vial (1 to 5) and number of scans per point C) Spectral Acquisition Program shows spectrum being collected which vial and position D) Spectra Analysis Program user selects spectra to analyze by plate vial and point (s) as well as two wavenumbers defining the peak and the baseline to subtract The image is of 5 repeat measurements of 10 mgL KCN in generic tap water 16 sec each 100 mW of 785 nm
3 RESULTS AND DISCUSSION Raman and surface-enhanced Raman spectra were obtained for potassium cyanide bis-(2-chloroethyl)sulfide and ethyl S-2-diisopropylamino ethyl methylphosphonothioate representing three classes of chemical agents cyanides mustards and nerve agents respectively Spectra were also obtained for 2-chloroethyl ethyl sulfide (CEES) a structural analogue to HD which was included in the study to aid in assigning spectral bands KCN salt was used for cyanide experiments to avoid the increased hazards of handling HCN gas KCN completely dissolves in water forming its conjugate acid HCN according to its Ka of 615x10-1027 and at a concentration of 1 mgmL results in a pH 107 solution This is significant in that only CN- appears to interact sufficiently with silver to produce a SER spectrum and no spectral signal is observed below pH 726 Figure 2 shows the SER and normal Raman spectra for KCN The SERS spectrum of 1mgml KCN in water shows a single intense somewhat broad feature at
A B
C D
SPIE-2003-5269 19
2100 cm-1 assigned to the single CequivN stretch The band is much sharper in the normal Raman spectra of the solid KCN salt at 2074 cm-1 This band does broaden and shift to 2080 cm-1 in solution (not shown) However the observed SERS frequency is attributed to interaction with silver and low frequency mode at 135 cm-1 attributed to a Ag-CN stretch (not shown) supports this conclusion
cm-1 band to a C-S stretch but the authors concede that it is in fact more likely a C-Cl stretch28 It appears that the most intense Raman bands at 648 692 and 747 cm-1 shift to 620 660 and 720 cm-1 in the SER spectra and are tentatively assigned as above The width of these bands suggests that they overlap underlying spectral features Additional bands in the Raman spectra occur at 972 1034 1049 1263 1286 1423 1442 2865 2935 and 2960 cm-1 Corresponding bands occur in the SER spectra at 964 1015 1054 1286 1410 1447 2865 and 2935 cm-1 Most of these bands are associated with alkane modes specifically the bands at approximately 1040 cm-1 to a C-C stretch 1290 cm-1 to a CH2 in-phase twist 1440 cm-1 to a CH2 wag 2865 cm-1 to a symmetric CH2 stretch and 2965 cm-1 to an asymmetric CH2 stretch The Raman and SER spectra of sulfur mustard were measured at the Edgewood center (Figure 4) Both spectra are largely similar to CEES The C-Cl and C-S bands in the Raman spectrum of HD now occur at 640 655 700 739 and 760 cm-1 and are more resolved possibly due to the increased molecular symmetry Theoretical calculations indicate that the first three bands are due to C-Cl stretching modes and the latter two to C-S stretching modes28 Only the C-Cl bands maintain significant intensity in the SER spectra occurring at 624 and 643 cm-1 which is attributed to the
Prior to measurements of HD CEES was examined by Raman and SER spectroscopy (Figure 3) CEES also known as half-mustard is essentially identical to HD except one of the chlorine end atoms is replaced by a hydrogen atom Again although not as toxic as HD CEES is a blister agent and dilute aqueous samples were prepared using appropriate safety equipment Both the Raman and SER spectra of CEES are similar and dominated by bands between 600 and 800 cm-1 These are associated with C-Cl and C-S stretching modes which are tentatively assigned to 648 and 747 cm-1 in the Raman spectra respectively The shoulder at 630 cm-1 the overlapped band at 660 cm-1 and the strong band at 692 cm-1 could also be due to these modes or their asymmetric counterparts It is worth noting that theoretical calculations assign the 692
Figure 2 A) SER and B) NR spectra of KCN Conditions A) 1 mgml in tap water 100 mW of 785 nm at sample 1-min acquisition time B) solid 300 mW of 785 nm 5-min All spectra are 8 cm-1 resolution
A
B
Figure 3 A) SER and B) NR spectra of CEES Conditions A) 1 vv (10 mgml) in MeOH 100 mW of 785 nm 1-min acquisition time B) neat 300 mW of 785 nm 5-min
A
B
Cl-CH2-CH2-S-CH2-CH3
A
B
Figure 4 A) SER and B) NR of HD Conditions A) 1mgml in tap water B) pure both 100 mW of 785 nm 1-min
Cl-CH2-CH2-S-CH2-CH2-Cl
SPIE-2003-5269 20
expected strong interaction between chlorine and silver and adds support to the assignment of this band to a C-Cl stretch Weaker overlapping bands occur at 670 692 and 724 cm-1 the latter possibly due to C-S stretching modes Again the alkane modes are apparent in the normal Raman spectra of HD but only a broad feature at 1300 to 1450 cm-1 suggests CH2 contributions in the SER spectrum Although the observed bands in the VX spectrum have not been assigned (Figure 4) a computer generated Raman spectrum29 predicts many of the same features with surprising accuracy and are used here Two intense bands at 460 and 530 cm-1 closely match predicted bands at 463 and 546 cm-1 assigned to a CH3-P=O bend and a PO2CS wag Three highly overlapped bands occur at 694 745 and 771 cm-1 matching predicted bands at 713 730 and 760 cm-1 The first
Table 2 Measured SER peak heights for the CN stretch at 2100 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
Figure 5 A) SERS and B) NR spectra of VX Conditions A) 1 vv (10 mgml) in MeOH B) pure sample both 100 mW of 785 nm 1-min acquisition time
A
B
two have been assigned to a C-S stretch and CH2bend respectively while the latter has been attributed to either a P-C stretch or an O-C-C stretch Although the 745 cm-1 band may alternatively be assigned to a C-S stretch based on the previous measurements of CEES and HD The relatively intense bands at 890 1106 1218 1445 and 1465 cm-1 also match predicted bands at 880 1108 1216 1440 and 1464 cm-1 that are assigned to a C-C stretch CH3 rock N-C3 stretch various C-H3 bends and C-H bends respectively Both the computer generated and the measured spectra contain numerous other less intense bands One is worth mentioning A unique band appears at 370 cm-1 that is predicted at 368 cm-1 and corresponds to an O-P=O bend The surface-enhanced Raman spectrum of VX is also rich with spectral features It has the unique low frequency band at 370 cm-1 as well as a second band at 380 cm-1 that is assigned tothe S-P-O bend predicted in the normal Raman spectrum at 388 cm-1 Based on the measured and predicted normal Raman spectra the following SERS assignments are given 460 cm-1 to the CH3-P=O bend 544 cm-1 to the PO2CS wag 738 cm-1 to a C-S stretch (based on arguments above) 890 cm-1
to a C-C stretch 1101 cm-1 to a CH3 rock and 1456
cm-1 to a C-H bend The ability of SERS to measure chemical agents in water containing real-world chemical interferents was tested by using the generic tap water described in Table 1 The ability to reproduce measurements was accomplished by preparing three separate water stock solutions which were used to prepare three sample solutions each of 1 mgmL KCN in the generic tap water The samples defined as vials 1-9 were then measured at 4 points per vial in a semi-random fashion such that errors associated with stock solution preparation and errors associated with instrument drift could be identified No trends were apparent that signified such systematic errors Each spectrum collected consisted of 20 averaged scans taking 16 seconds at 8 cm-1 resolution The laser power at the sample was measured periodically during the day and it ranged from 102 to 105 mW spacccc
spot Vial 1 Vial 2 Vial 41 13754 13004 128192 13519 12692 12909
1 3 13541 12721 126394 13462 12648 12651
ave 13569 12766 12755Vial 6 Vial 3 Vial 5
1 11586 14068 10757 2 11236 14402 115122 3 11376 14546 11559
4 10894 11763 11214ave 11273 13695 11261
Vial 7 Vial8 Vial 91 11198 15804 11023
2 11243 15716 88753 3 11562 15216 11218
4 11655 15084 11077ave 11415 15455 10548
CN AVG STDEV ERR1 mgmL pk ht 12526 1560 1245
stock solution
SPIE-2003-5269 21
The CN measurements consisting of 36 data points produced an average height of 125 for the 2100 cm-1 peak with a standard deviation of 156 or 125 (Table 2) The HD and VX measurements were performed precisely the same way (Tables 3 and 4) For HD the 624 cm-1 peak was used for analysis and it had an average height of 53 with a standard deviation of 068 or 129 while for VX the 544 cm-1 peak was used for analysis and it had an average height of 1051 with a standard deviation of 308 or 293 The greater error in the VX measurements can be somewhat attributed to Vial 4 which produced lower SER signal intensities But removing this vial from the data set changes the standard deviation to 233 only a modest improvement
spot Vial 1 Vial 2 Vial 41 525 609 4682 45 675 484
1 3 527 644 5094 559 807 669
ave 51525 68375 5325Vial 6 Vial 3 Vial 5
1 521 574 575 2 536 53 4492 3 509 508 451
4 565 594 379ave 53275 5515 4635
Vial 7 Vial 8 Vial 91 457 589 413
2 583 497 4273 3 544 505 4
4 528 5 509ave 528 52275 43725
HD AVG STDEV ERR1 mgmL pk ht 530 068 1291
stock solution
spot Vial 1 Vial 2 Vial 41 1464 1034 4672 1485 99 61
1 3 1491 989 5684 1041 777 553
ave 137025 9475 5495Vial 6 Vial 3 Vial 5
1 1058 942 1293 2 697 121 965
2 3 727 1206 11124 689 1254 746
ave 79275 1153 1029Vial 7 Vial 8 Vial 9
1 1127 783 163 2 1358 812 1614
3 3 1371 908 15124 1097 875 1397
ave 123825 8445 153825
VX AVG STDEV ERR1 mgmL pk ht 1051 308 2925
stock solution
4 CONCLUSIONS In this paper we examined the ability of surface-enhanced Raman spectroscopy to reproducibly measure CN HD and VX in tap water without chemical interference Both normal and surface-enhanced Raman spectra were examined to select unique bands suitable to identify and quantify these chemical agents For SER measurements the 2100 cm-1 C-N stretch was used for CN the 624 cm-1 C-Cl stretch was used for HD and the 544 cm-1 PO2CS wag was used for VX It was determined that 1 mgmL samples of each of these chemicals measured 36 times in glass vials coated with a silver-doped sol-gel reproduced measurements with standard deviations of 125 129 and 293 It was further found that the 10 chemicals added to simulate generic tap water did not interfere with or alter the SER spectra It should be noted that the concentrations used in this study were considerably greater than those required by the JSAWM program Current work involves improving SER sensitivity and designing sampling systems with better reproducibility This includes the development of fractal silver and gold structures within the sol-gel matrix and the development of chemically selective sol-gels
ACKNOWLEDGMENTS The authors are grateful for the support of the US Army (DAAD13-02-C-0015 Joint Service Agent Water Monitor program) The authors would also like to thank Janet Jensen Ronald Crosier and Kristina Gonser for helpful discussions
Table 4 Measured SER peak heights for the VX band at 544 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
Table 3 Measured SER peak heights for the HD band at 624 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
SPIE-2003-5269 22
REFERENCES 1 Jensen JL NC Wong W R Loerop SPIE 4775-03 (2001) 2 JSAWM Requirements at wwwsbccomapgeaarmymilRDAecbcrtPRODSERJSAWMjsawmhtml 3 Erickson B ldquoThe Chemical Weapons Convention Redefines Analytical Challengerdquo Anal Chem 70 397A (1998) 4 Johnston RL Hoefler CM Fargo JC and Moberley B ldquoThe Defense Nuclear Agencyrsquos
ChemicalBiochemical Weapons Agreements Verification Technology Research Development Test and Evaluation Program and its Requirements for On-Site Analysisrdquo AT-ONSITE 5-8 (1994)
5 Black RM Clarke RJ Read RW and Reid MT ldquoApplication of GC-MS and GC-tandem MS to the analysis of chemical warfare samples found to contain residues of the nerve agent sarin sulphur mustard and their degradation productsrdquo J Chrom 662 301-321 (1994)
6 USA Army Center for Health Promotion and Preventive Medicine (USACHPPM) Technical Guideline 230A 7 Hoffland LD Piffath RJ Bouck JB rdquoSpectral signatures of chemical agents and simulantsrdquo Optical
Engineering 24 982-984 (1985) 8 Braue EHJ Pannella MGrdquoCIRCLE CELL FT-IR Analysis of Chemical Warfare Agents in Aqueous Solutionsrdquo
App Spectrosc 44 1513-1520 (1990) 9 Tseng C-H Mann CK and Vickers TJ ldquoDetermination of Organics on Metal Surfaces by Raman
Spectroscopyrdquo App Spectrosc 47 1767-1771 (1993) 10 Staff Reporter ldquoUS Army Tests Lidar to Detect Biological Toxinsrdquo Photonic Spectra pg 50 December 1998 11 Christesen SD Raman cross sections of chemical agents and simulants App Spectrosc 42 318-321 (1988) 12 Weaver MJ Farquharson S Tadayyoni MA ldquoSurface-enhancement factors for Raman scattering at silver electrodesrdquo J Chem Phys 82 4867-4874 (1985) 13 Alak AM and Vo-Dinh T ldquoSERS of Organophosphorous Chemical Agentsrdquo Anal Chem 59 2149-2153 (1987) 14 Lee Y Farquharson S ldquoRapid chemical agent identification by SERSrdquo SPIE 4378 21-26 (2001) 15 Farquharson S Maksymiuk P Ong K Christesen S ldquoChemical agent identification by surface-enhanced Raman spectroscopyrdquo SPIE 4577 166-173 (2001) 16 Spencer KM Sylvia J Clauson S and Janni J ldquoSurface Enhanced Raman as a Water Monitor for Warfare
Agents in Waterrdquo SPIE 4577 158-165 (2001) 17 Tessier P Christesen S Ong K Clemente E Lenhoff A Kaler E Velev O ldquoOn-line spectroscopic characterization of sodium cyanide with nanostructured gold surface-enhanced Raman spectroscopy substratesrdquo
App Spectrosc 56 1524-1530 (2002) 18 Farquharson S W W Smith S Elliott and J F Sperry Rapid biological agent identification by surface- enhanced Raman spectroscopy SPIE 3855110-116 (1999) 19 Farquharson S WW Smith YH Lee S Elliott and J F Sperry Detection of biological signatures A
comparison of electrolytic and metal-doped sol-gel surface-enhanced Raman media SPIE 4575 62-72 (2002) 20 Guzelian AA Sylvia J Janni J Clauson S and Spencer KM ldquoSERS of whole cell bacteria and trace levels of biological moleculesrdquo SPIE 4577 182-192 (2001) 21 Shende C Inscore F Gift A Maksymiuk P Farquharson S ldquoRapid dipicolinic acid extraction from Bacillus spores detected by surface-enhanced Raman spectroscopyrdquo App Spectrosc 58 accepted 22 Lee Y and Farquharson S ldquoSERS Sample Vials Based on Sol-Gel Process for Trace Pesticide Analysisrdquo SPIE
4206 140-146 (2000) 23 Farquharson S and Lee Y ldquoTrace Drug Analysis by SERSrdquo SPIE 4200-16 (2000) 24 Lee Y Farquharson S and Rainey P M Surface-Enhanced Raman Sensor for Trace Chemical Detection in
Water SPIE 3857 76-84 (1999) 25 Lee Y Farquharson S Kwong H and Shahriari M ldquoSol-Gel Chemical Sensor for Surface-Enhanced Raman
Spectroscopyrdquo SPIE 3537 252-260 (1998) 26 Farquharson S A Gift P Maksymiuk F Inscore and W Smith ldquopH dependence of methyl phosphonic acid
dipicolinic acid and cyanide by surface-enhanced Raman spectroscopyrdquo SPIE 5269-19 (2003) 27 Lide DR Ed Handbook of Chemistry and Physics CRC Press 77th Ed (1996-7) 28 Sosa C RJ Bartlett K KuBulat and WB Person ldquoA theoretical study of harmonic vibrational frequencies and
infrared intensities of XCH2CH2SCH2CH2X and XCH2CH2SH (= H Cl)rdquo J Phys Chem 93 577-588 (1993) 29 Hameka HF and JO Jenson ldquoComputer-generated predictions of the structure and the IR and Raman spectra of
VXrdquo ERDEC-TR-065 May 1993
- Chemagents Appendicespdf
-
- SPIE2001-4575-Bioagent-SERSapdf
-
- Detection of bioagent signatures A comparison of electrolytic and metal-doped sol-gel surface-enhanced Raman media
- Stuart Farquharson Wayne Smith and Yuan Lee
-
- Real-Time Analyzers 87 Church Street East Hartford CT 06108
-
- ABSTRACT
- 4 CONCLUSIONS
- 5 ACKNOWLEDGEMENTS
- 6 REFERENCES
-
SPIE-2003-5269 16
Chemical agent detection by surface-enhanced Raman spectroscopy
Stuart Farquharson Alan Gift Paul Maksymiuk Frank Inscore and Wayne Smith
Real-Time Analyzers 87 Church Street East Hartford CT 06108
Kevin Morrisey and Steven D Christesen US Army SBCCOM Aberdeen Proving Ground MD 21010
ABSTRACT
In the past decade the Unites States and its allies have been challenged by a different kind of warfare exemplified by the terrorist attacks of September 11 2001 Although suicide bombings are the most often used form of terror military personnel must consider a wide range of attack scenarios Among these is the intentional poisoning of water supplies to obstruct military operations in Afghanistan and Iraq To counter such attacks the military is developing portable analyzers that can identify and quantify potential chemical agents in water supplies at microgram per liter concentrations within 10 minutes To aid this effort we have been investigating the value of a surface-enhanced Raman spectroscopy based portable analyzer In particular we have been developing silver-doped sol-gels to generate SER spectra of chemical agents and their hydrolysis products Here we present SER spectra of several chemical agents measured in a generic tap water Repeat measurements were performed to establish statistical error associated with SERS obtained using the sol-gel coated vials Keywords Chemical agents chemical agent detection SERS Raman spectroscopy
1 INTRODUCTION
In the autumn of 2001 terrorism within US borders became a sobering reality While extensive efforts are being implemented to secure the homeland US and Coalition forces in Afghanistan and Iraq are constantly faced with terrorist attacks In addition to car-bombings the military has identified several non-traditional attack scenarios including poisoning of water supplies by chemical warfare agents (CWAs) To counter this threat the Department of Defense has been investigating numerous approaches to detect such attacks under the auspices of the Joint Service Agent Water Monitor (JSAWM) program1 The JSAWM must be capable of identifying the chemical agents with no false-positives and quantifying the agents at microgram per liter concentrations within 10 minutes2 This includes the analysis of drinking water supplies distribution and storage systems Currently colorimetric paper is used to detect agents on-site while gas chromatography combined with mass spectrometry (GCMS) is used in mobile support laboratories However both methods have severe drawbacks The paper changes color in response to contact with many chemicals besides CWAs causing a high incidence of false positives3 while GCMS although very chemically specific requires up to an hour to perform and regular re-calibration456 Vibrational spectroscopy has also been investigated as a potential method of rapidly detecting CWAs7-
11 as early as 1985 in the case of infrared spectroscopy8 and 1988 in the case of Raman spectroscopy11 Again however these techniques also have limitations when it comes to measuring trace poisons in water Infrared spectra would be dominated by the very strong absorption of water which would obscure absorptions by most other chemicals present Whereas Raman spectroscopy is simply not a very sensitive technique and detection limits are typically grams per liter Surface-enhanced Raman spectroscopy (SERS) offers several advantages over conventional vibrational methods and that may provide the necessary sensitivity required for detecting trace quantities of chemical agents in water SERS employs the interaction of surface plasmon modes of metal particles with target analytes to increase scattering efficiency by as much as 1 million times12 In 1987 the potential of this approach to measure CWAs was demonstrated by measuring a series of pesticides13 Several of these organophosphonates have chemical structures similar to CWAs in particular P=O functional groups In the past few years we and others have further explored the ability of SERS to detect CWAs14-17 and even bioagents 18-21 We have been employing silver-doped sol-gels to promote the SER effect
SPIE-2003-5269 17
in these studies The porous silica network of the sol-gel matrix offers a unique environment for immobilizing and stabilizing SER-active metal particles22-25 The silver-doped sol-gels have been coated on the internal walls of standard glass vials to allow rapid SER analysis of small sample quantities without preparation We have measured over 100 chemicals with enhancements of 104 to 106 demonstrated reversible measurements in a flowing system reproducible measurements from vial-to-vial and measurements in multiple solvents including water21-25 Previously we used these vials to perform preliminary measurements of cyanide (CN) and methylphosphonic acid (MPA) Most of the nerve agents form MPA during hydrolysis while Tabun forms CN a chemical agent in its own right In another paper including in these proceedings we examined the limits of detection (LOD) for MPA by measuring a series of concentrations down to 50 mgL and estimated a limit of detection of 100 microgL26 These measurements provide encouragement in that SERS may satisfy the needs of the JSAWM To further establish the viability of SERS in particular silver-doped sol-gels here we present analysis of cyanide mustard and VX in tap water The measurements performed at the US Armyrsquos Edgewood Chemical Biological Center Aberdeen MD also included numerous repeat measurements to establish reproducibility
2 EXPERIMENTAL 2a General All chemicals including potassium cyanide 2-chloroethylethyl sulfide and those used to prepare the silver-doped and gold-doped sol-gel coated vials were obtained and used as received from Sigma-Aldrich All samples were prepared in a chemical hood using HPLC grade water unless otherwise noted (Fischer Scientific Fair Lawn NJ) for SERS measurements Once prepared the samples were transferred into the silver-doped sol-gel vials (Real-Time Analyzers Inc East Hartford CT) The vials were coated in a manner similar to that previously reported by adding ammonium hydroxide to a solution of silver nitrate tetramethyl orthosilicate (TMOS) and methanol22 The two precursor solutions were prepared mixed and transferred to 2-mL glass vials dried and heated After sol-gel formation the incorporated metal ions were reduced with dilute sodium borohydride (1mgmL) which is followed by a water wash to remove residual reducing agent After the resultant analyte solution was introduced the SER-active vial was fixed horizontally to an XY positioning stage (Conix Research Springfield OR) just inside the focal point of an f07 aspheric lens The lens focused the beam into the sample and collected the scattered radiation back along the same axis A dichroic filter (Omega Optical Brattleborough VT) was used to reflect the excitation laser to the lens and pass the Raman scattered radiation collected by the lens An f2 achromat was used to collimate the laser beam exiting a 200 microm core diameter source fiber optic while a second f2 achromat was used to focus the scattered radiation into a 365 microm fiber optic (Spectran Avon CT) A short pass filter was placed in the excitation beam path to block the silicon Raman scattering generated in the source fiber from reflecting off sampling optics and reaching the detector A long pass filter was placed in the collection beam path to block the sample Rayleigh scattering from reaching the detector A 785 nm diode laser (Process Instruments Inc model 785-600 Salt Lake City UT) was used to deliver 100 to 150 mW of power to the sample A Fourier transform Raman spectrometer (Real-Time Analyzers model IRA-785 East Hartford CT) and a silicon photo-avalanche detector (Perkin Elmer model C30902S Stamford CT) were used to acquire the SER spectra
2b Edgewood Chemical Biological Center The surface-enhanced Raman spectral measurements at the US Armyrsquos Chemical Biological center presented here were all performed on September 12 2003 To expedite measurements a plate was machined to hold up to 12 SER-active sample vials (Figure 1) The plate fit a standard XY plate reader that could be programmed Pure KCN bis-(2-chloroethyl)sulfide (distilled mustard gas HD) and ethyl S-2-diisopropylamino ethyl methylphosphonothioate (VX) were obtained on-site and used to prepare 1 gL tap water solutions in a chemical hood with appropriate safety equipment Simulated tap water was prepared by adding 10 chemicals most often found in tap water at appropriate concentrations to distilled water (Table 1) SER measurements were also performed in a chemical hood For added safety the FT-Raman instrument was placed outside the laboratory and 30 foot fiber optic and electrical cables were used to allow remote SERS measurements and plate manipulation For each experiment 1gL samples were prepared and added to 9 individual vials which were then loaded on the plate In some cases a tenth vial was included as a blank
Table 1 Generic Tap Water Composition
Compound mgL NaHCO3 100 CaSO4 27 MgSO4bull7H2O 67 NaNO3 1 Fulvic Acid 1 K2HPO4 07 KH2PO4 03 (NH4)2HSO4 001 NaCl 001 FeSO4 0001 pH 76-78
SPIE-2003-5269 18
A software program was written that allowed selecting the sequence that the vials were measured the number of positions along the length of the vials to measure (1 to 5) and the number of scans to co-add During sample analysis the program displayed the vial being analyzed the point being analyzed and the spectrum as it was being acquired Once all the data was collected a second software program was written to rapidly analyze the data The spectra collected for all the vials on a plate could be loaded at one time and then the spectra for each point could be displayed simultaneously or separately The user could then select the Raman peak to analyze in terms of peak height or area This was accomplished by selecting points on either side of the peak to define a baseline of zero The peak height or area could then be computed for all of the spectra loaded and then exported to a spreadsheet for statistical analysis
Figure 1 A) Vial Holder 6 slots to hold 2 vials each end-to-end B) Measurement Configuration Program user selects vials to measure sequence number of points per vial (1 to 5) and number of scans per point C) Spectral Acquisition Program shows spectrum being collected which vial and position D) Spectra Analysis Program user selects spectra to analyze by plate vial and point (s) as well as two wavenumbers defining the peak and the baseline to subtract The image is of 5 repeat measurements of 10 mgL KCN in generic tap water 16 sec each 100 mW of 785 nm
3 RESULTS AND DISCUSSION Raman and surface-enhanced Raman spectra were obtained for potassium cyanide bis-(2-chloroethyl)sulfide and ethyl S-2-diisopropylamino ethyl methylphosphonothioate representing three classes of chemical agents cyanides mustards and nerve agents respectively Spectra were also obtained for 2-chloroethyl ethyl sulfide (CEES) a structural analogue to HD which was included in the study to aid in assigning spectral bands KCN salt was used for cyanide experiments to avoid the increased hazards of handling HCN gas KCN completely dissolves in water forming its conjugate acid HCN according to its Ka of 615x10-1027 and at a concentration of 1 mgmL results in a pH 107 solution This is significant in that only CN- appears to interact sufficiently with silver to produce a SER spectrum and no spectral signal is observed below pH 726 Figure 2 shows the SER and normal Raman spectra for KCN The SERS spectrum of 1mgml KCN in water shows a single intense somewhat broad feature at
A B
C D
SPIE-2003-5269 19
2100 cm-1 assigned to the single CequivN stretch The band is much sharper in the normal Raman spectra of the solid KCN salt at 2074 cm-1 This band does broaden and shift to 2080 cm-1 in solution (not shown) However the observed SERS frequency is attributed to interaction with silver and low frequency mode at 135 cm-1 attributed to a Ag-CN stretch (not shown) supports this conclusion
cm-1 band to a C-S stretch but the authors concede that it is in fact more likely a C-Cl stretch28 It appears that the most intense Raman bands at 648 692 and 747 cm-1 shift to 620 660 and 720 cm-1 in the SER spectra and are tentatively assigned as above The width of these bands suggests that they overlap underlying spectral features Additional bands in the Raman spectra occur at 972 1034 1049 1263 1286 1423 1442 2865 2935 and 2960 cm-1 Corresponding bands occur in the SER spectra at 964 1015 1054 1286 1410 1447 2865 and 2935 cm-1 Most of these bands are associated with alkane modes specifically the bands at approximately 1040 cm-1 to a C-C stretch 1290 cm-1 to a CH2 in-phase twist 1440 cm-1 to a CH2 wag 2865 cm-1 to a symmetric CH2 stretch and 2965 cm-1 to an asymmetric CH2 stretch The Raman and SER spectra of sulfur mustard were measured at the Edgewood center (Figure 4) Both spectra are largely similar to CEES The C-Cl and C-S bands in the Raman spectrum of HD now occur at 640 655 700 739 and 760 cm-1 and are more resolved possibly due to the increased molecular symmetry Theoretical calculations indicate that the first three bands are due to C-Cl stretching modes and the latter two to C-S stretching modes28 Only the C-Cl bands maintain significant intensity in the SER spectra occurring at 624 and 643 cm-1 which is attributed to the
Prior to measurements of HD CEES was examined by Raman and SER spectroscopy (Figure 3) CEES also known as half-mustard is essentially identical to HD except one of the chlorine end atoms is replaced by a hydrogen atom Again although not as toxic as HD CEES is a blister agent and dilute aqueous samples were prepared using appropriate safety equipment Both the Raman and SER spectra of CEES are similar and dominated by bands between 600 and 800 cm-1 These are associated with C-Cl and C-S stretching modes which are tentatively assigned to 648 and 747 cm-1 in the Raman spectra respectively The shoulder at 630 cm-1 the overlapped band at 660 cm-1 and the strong band at 692 cm-1 could also be due to these modes or their asymmetric counterparts It is worth noting that theoretical calculations assign the 692
Figure 2 A) SER and B) NR spectra of KCN Conditions A) 1 mgml in tap water 100 mW of 785 nm at sample 1-min acquisition time B) solid 300 mW of 785 nm 5-min All spectra are 8 cm-1 resolution
A
B
Figure 3 A) SER and B) NR spectra of CEES Conditions A) 1 vv (10 mgml) in MeOH 100 mW of 785 nm 1-min acquisition time B) neat 300 mW of 785 nm 5-min
A
B
Cl-CH2-CH2-S-CH2-CH3
A
B
Figure 4 A) SER and B) NR of HD Conditions A) 1mgml in tap water B) pure both 100 mW of 785 nm 1-min
Cl-CH2-CH2-S-CH2-CH2-Cl
SPIE-2003-5269 20
expected strong interaction between chlorine and silver and adds support to the assignment of this band to a C-Cl stretch Weaker overlapping bands occur at 670 692 and 724 cm-1 the latter possibly due to C-S stretching modes Again the alkane modes are apparent in the normal Raman spectra of HD but only a broad feature at 1300 to 1450 cm-1 suggests CH2 contributions in the SER spectrum Although the observed bands in the VX spectrum have not been assigned (Figure 4) a computer generated Raman spectrum29 predicts many of the same features with surprising accuracy and are used here Two intense bands at 460 and 530 cm-1 closely match predicted bands at 463 and 546 cm-1 assigned to a CH3-P=O bend and a PO2CS wag Three highly overlapped bands occur at 694 745 and 771 cm-1 matching predicted bands at 713 730 and 760 cm-1 The first
Table 2 Measured SER peak heights for the CN stretch at 2100 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
Figure 5 A) SERS and B) NR spectra of VX Conditions A) 1 vv (10 mgml) in MeOH B) pure sample both 100 mW of 785 nm 1-min acquisition time
A
B
two have been assigned to a C-S stretch and CH2bend respectively while the latter has been attributed to either a P-C stretch or an O-C-C stretch Although the 745 cm-1 band may alternatively be assigned to a C-S stretch based on the previous measurements of CEES and HD The relatively intense bands at 890 1106 1218 1445 and 1465 cm-1 also match predicted bands at 880 1108 1216 1440 and 1464 cm-1 that are assigned to a C-C stretch CH3 rock N-C3 stretch various C-H3 bends and C-H bends respectively Both the computer generated and the measured spectra contain numerous other less intense bands One is worth mentioning A unique band appears at 370 cm-1 that is predicted at 368 cm-1 and corresponds to an O-P=O bend The surface-enhanced Raman spectrum of VX is also rich with spectral features It has the unique low frequency band at 370 cm-1 as well as a second band at 380 cm-1 that is assigned tothe S-P-O bend predicted in the normal Raman spectrum at 388 cm-1 Based on the measured and predicted normal Raman spectra the following SERS assignments are given 460 cm-1 to the CH3-P=O bend 544 cm-1 to the PO2CS wag 738 cm-1 to a C-S stretch (based on arguments above) 890 cm-1
to a C-C stretch 1101 cm-1 to a CH3 rock and 1456
cm-1 to a C-H bend The ability of SERS to measure chemical agents in water containing real-world chemical interferents was tested by using the generic tap water described in Table 1 The ability to reproduce measurements was accomplished by preparing three separate water stock solutions which were used to prepare three sample solutions each of 1 mgmL KCN in the generic tap water The samples defined as vials 1-9 were then measured at 4 points per vial in a semi-random fashion such that errors associated with stock solution preparation and errors associated with instrument drift could be identified No trends were apparent that signified such systematic errors Each spectrum collected consisted of 20 averaged scans taking 16 seconds at 8 cm-1 resolution The laser power at the sample was measured periodically during the day and it ranged from 102 to 105 mW spacccc
spot Vial 1 Vial 2 Vial 41 13754 13004 128192 13519 12692 12909
1 3 13541 12721 126394 13462 12648 12651
ave 13569 12766 12755Vial 6 Vial 3 Vial 5
1 11586 14068 10757 2 11236 14402 115122 3 11376 14546 11559
4 10894 11763 11214ave 11273 13695 11261
Vial 7 Vial8 Vial 91 11198 15804 11023
2 11243 15716 88753 3 11562 15216 11218
4 11655 15084 11077ave 11415 15455 10548
CN AVG STDEV ERR1 mgmL pk ht 12526 1560 1245
stock solution
SPIE-2003-5269 21
The CN measurements consisting of 36 data points produced an average height of 125 for the 2100 cm-1 peak with a standard deviation of 156 or 125 (Table 2) The HD and VX measurements were performed precisely the same way (Tables 3 and 4) For HD the 624 cm-1 peak was used for analysis and it had an average height of 53 with a standard deviation of 068 or 129 while for VX the 544 cm-1 peak was used for analysis and it had an average height of 1051 with a standard deviation of 308 or 293 The greater error in the VX measurements can be somewhat attributed to Vial 4 which produced lower SER signal intensities But removing this vial from the data set changes the standard deviation to 233 only a modest improvement
spot Vial 1 Vial 2 Vial 41 525 609 4682 45 675 484
1 3 527 644 5094 559 807 669
ave 51525 68375 5325Vial 6 Vial 3 Vial 5
1 521 574 575 2 536 53 4492 3 509 508 451
4 565 594 379ave 53275 5515 4635
Vial 7 Vial 8 Vial 91 457 589 413
2 583 497 4273 3 544 505 4
4 528 5 509ave 528 52275 43725
HD AVG STDEV ERR1 mgmL pk ht 530 068 1291
stock solution
spot Vial 1 Vial 2 Vial 41 1464 1034 4672 1485 99 61
1 3 1491 989 5684 1041 777 553
ave 137025 9475 5495Vial 6 Vial 3 Vial 5
1 1058 942 1293 2 697 121 965
2 3 727 1206 11124 689 1254 746
ave 79275 1153 1029Vial 7 Vial 8 Vial 9
1 1127 783 163 2 1358 812 1614
3 3 1371 908 15124 1097 875 1397
ave 123825 8445 153825
VX AVG STDEV ERR1 mgmL pk ht 1051 308 2925
stock solution
4 CONCLUSIONS In this paper we examined the ability of surface-enhanced Raman spectroscopy to reproducibly measure CN HD and VX in tap water without chemical interference Both normal and surface-enhanced Raman spectra were examined to select unique bands suitable to identify and quantify these chemical agents For SER measurements the 2100 cm-1 C-N stretch was used for CN the 624 cm-1 C-Cl stretch was used for HD and the 544 cm-1 PO2CS wag was used for VX It was determined that 1 mgmL samples of each of these chemicals measured 36 times in glass vials coated with a silver-doped sol-gel reproduced measurements with standard deviations of 125 129 and 293 It was further found that the 10 chemicals added to simulate generic tap water did not interfere with or alter the SER spectra It should be noted that the concentrations used in this study were considerably greater than those required by the JSAWM program Current work involves improving SER sensitivity and designing sampling systems with better reproducibility This includes the development of fractal silver and gold structures within the sol-gel matrix and the development of chemically selective sol-gels
ACKNOWLEDGMENTS The authors are grateful for the support of the US Army (DAAD13-02-C-0015 Joint Service Agent Water Monitor program) The authors would also like to thank Janet Jensen Ronald Crosier and Kristina Gonser for helpful discussions
Table 4 Measured SER peak heights for the VX band at 544 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
Table 3 Measured SER peak heights for the HD band at 624 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
SPIE-2003-5269 22
REFERENCES 1 Jensen JL NC Wong W R Loerop SPIE 4775-03 (2001) 2 JSAWM Requirements at wwwsbccomapgeaarmymilRDAecbcrtPRODSERJSAWMjsawmhtml 3 Erickson B ldquoThe Chemical Weapons Convention Redefines Analytical Challengerdquo Anal Chem 70 397A (1998) 4 Johnston RL Hoefler CM Fargo JC and Moberley B ldquoThe Defense Nuclear Agencyrsquos
ChemicalBiochemical Weapons Agreements Verification Technology Research Development Test and Evaluation Program and its Requirements for On-Site Analysisrdquo AT-ONSITE 5-8 (1994)
5 Black RM Clarke RJ Read RW and Reid MT ldquoApplication of GC-MS and GC-tandem MS to the analysis of chemical warfare samples found to contain residues of the nerve agent sarin sulphur mustard and their degradation productsrdquo J Chrom 662 301-321 (1994)
6 USA Army Center for Health Promotion and Preventive Medicine (USACHPPM) Technical Guideline 230A 7 Hoffland LD Piffath RJ Bouck JB rdquoSpectral signatures of chemical agents and simulantsrdquo Optical
Engineering 24 982-984 (1985) 8 Braue EHJ Pannella MGrdquoCIRCLE CELL FT-IR Analysis of Chemical Warfare Agents in Aqueous Solutionsrdquo
App Spectrosc 44 1513-1520 (1990) 9 Tseng C-H Mann CK and Vickers TJ ldquoDetermination of Organics on Metal Surfaces by Raman
Spectroscopyrdquo App Spectrosc 47 1767-1771 (1993) 10 Staff Reporter ldquoUS Army Tests Lidar to Detect Biological Toxinsrdquo Photonic Spectra pg 50 December 1998 11 Christesen SD Raman cross sections of chemical agents and simulants App Spectrosc 42 318-321 (1988) 12 Weaver MJ Farquharson S Tadayyoni MA ldquoSurface-enhancement factors for Raman scattering at silver electrodesrdquo J Chem Phys 82 4867-4874 (1985) 13 Alak AM and Vo-Dinh T ldquoSERS of Organophosphorous Chemical Agentsrdquo Anal Chem 59 2149-2153 (1987) 14 Lee Y Farquharson S ldquoRapid chemical agent identification by SERSrdquo SPIE 4378 21-26 (2001) 15 Farquharson S Maksymiuk P Ong K Christesen S ldquoChemical agent identification by surface-enhanced Raman spectroscopyrdquo SPIE 4577 166-173 (2001) 16 Spencer KM Sylvia J Clauson S and Janni J ldquoSurface Enhanced Raman as a Water Monitor for Warfare
Agents in Waterrdquo SPIE 4577 158-165 (2001) 17 Tessier P Christesen S Ong K Clemente E Lenhoff A Kaler E Velev O ldquoOn-line spectroscopic characterization of sodium cyanide with nanostructured gold surface-enhanced Raman spectroscopy substratesrdquo
App Spectrosc 56 1524-1530 (2002) 18 Farquharson S W W Smith S Elliott and J F Sperry Rapid biological agent identification by surface- enhanced Raman spectroscopy SPIE 3855110-116 (1999) 19 Farquharson S WW Smith YH Lee S Elliott and J F Sperry Detection of biological signatures A
comparison of electrolytic and metal-doped sol-gel surface-enhanced Raman media SPIE 4575 62-72 (2002) 20 Guzelian AA Sylvia J Janni J Clauson S and Spencer KM ldquoSERS of whole cell bacteria and trace levels of biological moleculesrdquo SPIE 4577 182-192 (2001) 21 Shende C Inscore F Gift A Maksymiuk P Farquharson S ldquoRapid dipicolinic acid extraction from Bacillus spores detected by surface-enhanced Raman spectroscopyrdquo App Spectrosc 58 accepted 22 Lee Y and Farquharson S ldquoSERS Sample Vials Based on Sol-Gel Process for Trace Pesticide Analysisrdquo SPIE
4206 140-146 (2000) 23 Farquharson S and Lee Y ldquoTrace Drug Analysis by SERSrdquo SPIE 4200-16 (2000) 24 Lee Y Farquharson S and Rainey P M Surface-Enhanced Raman Sensor for Trace Chemical Detection in
Water SPIE 3857 76-84 (1999) 25 Lee Y Farquharson S Kwong H and Shahriari M ldquoSol-Gel Chemical Sensor for Surface-Enhanced Raman
Spectroscopyrdquo SPIE 3537 252-260 (1998) 26 Farquharson S A Gift P Maksymiuk F Inscore and W Smith ldquopH dependence of methyl phosphonic acid
dipicolinic acid and cyanide by surface-enhanced Raman spectroscopyrdquo SPIE 5269-19 (2003) 27 Lide DR Ed Handbook of Chemistry and Physics CRC Press 77th Ed (1996-7) 28 Sosa C RJ Bartlett K KuBulat and WB Person ldquoA theoretical study of harmonic vibrational frequencies and
infrared intensities of XCH2CH2SCH2CH2X and XCH2CH2SH (= H Cl)rdquo J Phys Chem 93 577-588 (1993) 29 Hameka HF and JO Jenson ldquoComputer-generated predictions of the structure and the IR and Raman spectra of
VXrdquo ERDEC-TR-065 May 1993
- Chemagents Appendicespdf
-
- SPIE2001-4575-Bioagent-SERSapdf
-
- Detection of bioagent signatures A comparison of electrolytic and metal-doped sol-gel surface-enhanced Raman media
- Stuart Farquharson Wayne Smith and Yuan Lee
-
- Real-Time Analyzers 87 Church Street East Hartford CT 06108
-
- ABSTRACT
- 4 CONCLUSIONS
- 5 ACKNOWLEDGEMENTS
- 6 REFERENCES
-
SPIE-2003-5269 17
in these studies The porous silica network of the sol-gel matrix offers a unique environment for immobilizing and stabilizing SER-active metal particles22-25 The silver-doped sol-gels have been coated on the internal walls of standard glass vials to allow rapid SER analysis of small sample quantities without preparation We have measured over 100 chemicals with enhancements of 104 to 106 demonstrated reversible measurements in a flowing system reproducible measurements from vial-to-vial and measurements in multiple solvents including water21-25 Previously we used these vials to perform preliminary measurements of cyanide (CN) and methylphosphonic acid (MPA) Most of the nerve agents form MPA during hydrolysis while Tabun forms CN a chemical agent in its own right In another paper including in these proceedings we examined the limits of detection (LOD) for MPA by measuring a series of concentrations down to 50 mgL and estimated a limit of detection of 100 microgL26 These measurements provide encouragement in that SERS may satisfy the needs of the JSAWM To further establish the viability of SERS in particular silver-doped sol-gels here we present analysis of cyanide mustard and VX in tap water The measurements performed at the US Armyrsquos Edgewood Chemical Biological Center Aberdeen MD also included numerous repeat measurements to establish reproducibility
2 EXPERIMENTAL 2a General All chemicals including potassium cyanide 2-chloroethylethyl sulfide and those used to prepare the silver-doped and gold-doped sol-gel coated vials were obtained and used as received from Sigma-Aldrich All samples were prepared in a chemical hood using HPLC grade water unless otherwise noted (Fischer Scientific Fair Lawn NJ) for SERS measurements Once prepared the samples were transferred into the silver-doped sol-gel vials (Real-Time Analyzers Inc East Hartford CT) The vials were coated in a manner similar to that previously reported by adding ammonium hydroxide to a solution of silver nitrate tetramethyl orthosilicate (TMOS) and methanol22 The two precursor solutions were prepared mixed and transferred to 2-mL glass vials dried and heated After sol-gel formation the incorporated metal ions were reduced with dilute sodium borohydride (1mgmL) which is followed by a water wash to remove residual reducing agent After the resultant analyte solution was introduced the SER-active vial was fixed horizontally to an XY positioning stage (Conix Research Springfield OR) just inside the focal point of an f07 aspheric lens The lens focused the beam into the sample and collected the scattered radiation back along the same axis A dichroic filter (Omega Optical Brattleborough VT) was used to reflect the excitation laser to the lens and pass the Raman scattered radiation collected by the lens An f2 achromat was used to collimate the laser beam exiting a 200 microm core diameter source fiber optic while a second f2 achromat was used to focus the scattered radiation into a 365 microm fiber optic (Spectran Avon CT) A short pass filter was placed in the excitation beam path to block the silicon Raman scattering generated in the source fiber from reflecting off sampling optics and reaching the detector A long pass filter was placed in the collection beam path to block the sample Rayleigh scattering from reaching the detector A 785 nm diode laser (Process Instruments Inc model 785-600 Salt Lake City UT) was used to deliver 100 to 150 mW of power to the sample A Fourier transform Raman spectrometer (Real-Time Analyzers model IRA-785 East Hartford CT) and a silicon photo-avalanche detector (Perkin Elmer model C30902S Stamford CT) were used to acquire the SER spectra
2b Edgewood Chemical Biological Center The surface-enhanced Raman spectral measurements at the US Armyrsquos Chemical Biological center presented here were all performed on September 12 2003 To expedite measurements a plate was machined to hold up to 12 SER-active sample vials (Figure 1) The plate fit a standard XY plate reader that could be programmed Pure KCN bis-(2-chloroethyl)sulfide (distilled mustard gas HD) and ethyl S-2-diisopropylamino ethyl methylphosphonothioate (VX) were obtained on-site and used to prepare 1 gL tap water solutions in a chemical hood with appropriate safety equipment Simulated tap water was prepared by adding 10 chemicals most often found in tap water at appropriate concentrations to distilled water (Table 1) SER measurements were also performed in a chemical hood For added safety the FT-Raman instrument was placed outside the laboratory and 30 foot fiber optic and electrical cables were used to allow remote SERS measurements and plate manipulation For each experiment 1gL samples were prepared and added to 9 individual vials which were then loaded on the plate In some cases a tenth vial was included as a blank
Table 1 Generic Tap Water Composition
Compound mgL NaHCO3 100 CaSO4 27 MgSO4bull7H2O 67 NaNO3 1 Fulvic Acid 1 K2HPO4 07 KH2PO4 03 (NH4)2HSO4 001 NaCl 001 FeSO4 0001 pH 76-78
SPIE-2003-5269 18
A software program was written that allowed selecting the sequence that the vials were measured the number of positions along the length of the vials to measure (1 to 5) and the number of scans to co-add During sample analysis the program displayed the vial being analyzed the point being analyzed and the spectrum as it was being acquired Once all the data was collected a second software program was written to rapidly analyze the data The spectra collected for all the vials on a plate could be loaded at one time and then the spectra for each point could be displayed simultaneously or separately The user could then select the Raman peak to analyze in terms of peak height or area This was accomplished by selecting points on either side of the peak to define a baseline of zero The peak height or area could then be computed for all of the spectra loaded and then exported to a spreadsheet for statistical analysis
Figure 1 A) Vial Holder 6 slots to hold 2 vials each end-to-end B) Measurement Configuration Program user selects vials to measure sequence number of points per vial (1 to 5) and number of scans per point C) Spectral Acquisition Program shows spectrum being collected which vial and position D) Spectra Analysis Program user selects spectra to analyze by plate vial and point (s) as well as two wavenumbers defining the peak and the baseline to subtract The image is of 5 repeat measurements of 10 mgL KCN in generic tap water 16 sec each 100 mW of 785 nm
3 RESULTS AND DISCUSSION Raman and surface-enhanced Raman spectra were obtained for potassium cyanide bis-(2-chloroethyl)sulfide and ethyl S-2-diisopropylamino ethyl methylphosphonothioate representing three classes of chemical agents cyanides mustards and nerve agents respectively Spectra were also obtained for 2-chloroethyl ethyl sulfide (CEES) a structural analogue to HD which was included in the study to aid in assigning spectral bands KCN salt was used for cyanide experiments to avoid the increased hazards of handling HCN gas KCN completely dissolves in water forming its conjugate acid HCN according to its Ka of 615x10-1027 and at a concentration of 1 mgmL results in a pH 107 solution This is significant in that only CN- appears to interact sufficiently with silver to produce a SER spectrum and no spectral signal is observed below pH 726 Figure 2 shows the SER and normal Raman spectra for KCN The SERS spectrum of 1mgml KCN in water shows a single intense somewhat broad feature at
A B
C D
SPIE-2003-5269 19
2100 cm-1 assigned to the single CequivN stretch The band is much sharper in the normal Raman spectra of the solid KCN salt at 2074 cm-1 This band does broaden and shift to 2080 cm-1 in solution (not shown) However the observed SERS frequency is attributed to interaction with silver and low frequency mode at 135 cm-1 attributed to a Ag-CN stretch (not shown) supports this conclusion
cm-1 band to a C-S stretch but the authors concede that it is in fact more likely a C-Cl stretch28 It appears that the most intense Raman bands at 648 692 and 747 cm-1 shift to 620 660 and 720 cm-1 in the SER spectra and are tentatively assigned as above The width of these bands suggests that they overlap underlying spectral features Additional bands in the Raman spectra occur at 972 1034 1049 1263 1286 1423 1442 2865 2935 and 2960 cm-1 Corresponding bands occur in the SER spectra at 964 1015 1054 1286 1410 1447 2865 and 2935 cm-1 Most of these bands are associated with alkane modes specifically the bands at approximately 1040 cm-1 to a C-C stretch 1290 cm-1 to a CH2 in-phase twist 1440 cm-1 to a CH2 wag 2865 cm-1 to a symmetric CH2 stretch and 2965 cm-1 to an asymmetric CH2 stretch The Raman and SER spectra of sulfur mustard were measured at the Edgewood center (Figure 4) Both spectra are largely similar to CEES The C-Cl and C-S bands in the Raman spectrum of HD now occur at 640 655 700 739 and 760 cm-1 and are more resolved possibly due to the increased molecular symmetry Theoretical calculations indicate that the first three bands are due to C-Cl stretching modes and the latter two to C-S stretching modes28 Only the C-Cl bands maintain significant intensity in the SER spectra occurring at 624 and 643 cm-1 which is attributed to the
Prior to measurements of HD CEES was examined by Raman and SER spectroscopy (Figure 3) CEES also known as half-mustard is essentially identical to HD except one of the chlorine end atoms is replaced by a hydrogen atom Again although not as toxic as HD CEES is a blister agent and dilute aqueous samples were prepared using appropriate safety equipment Both the Raman and SER spectra of CEES are similar and dominated by bands between 600 and 800 cm-1 These are associated with C-Cl and C-S stretching modes which are tentatively assigned to 648 and 747 cm-1 in the Raman spectra respectively The shoulder at 630 cm-1 the overlapped band at 660 cm-1 and the strong band at 692 cm-1 could also be due to these modes or their asymmetric counterparts It is worth noting that theoretical calculations assign the 692
Figure 2 A) SER and B) NR spectra of KCN Conditions A) 1 mgml in tap water 100 mW of 785 nm at sample 1-min acquisition time B) solid 300 mW of 785 nm 5-min All spectra are 8 cm-1 resolution
A
B
Figure 3 A) SER and B) NR spectra of CEES Conditions A) 1 vv (10 mgml) in MeOH 100 mW of 785 nm 1-min acquisition time B) neat 300 mW of 785 nm 5-min
A
B
Cl-CH2-CH2-S-CH2-CH3
A
B
Figure 4 A) SER and B) NR of HD Conditions A) 1mgml in tap water B) pure both 100 mW of 785 nm 1-min
Cl-CH2-CH2-S-CH2-CH2-Cl
SPIE-2003-5269 20
expected strong interaction between chlorine and silver and adds support to the assignment of this band to a C-Cl stretch Weaker overlapping bands occur at 670 692 and 724 cm-1 the latter possibly due to C-S stretching modes Again the alkane modes are apparent in the normal Raman spectra of HD but only a broad feature at 1300 to 1450 cm-1 suggests CH2 contributions in the SER spectrum Although the observed bands in the VX spectrum have not been assigned (Figure 4) a computer generated Raman spectrum29 predicts many of the same features with surprising accuracy and are used here Two intense bands at 460 and 530 cm-1 closely match predicted bands at 463 and 546 cm-1 assigned to a CH3-P=O bend and a PO2CS wag Three highly overlapped bands occur at 694 745 and 771 cm-1 matching predicted bands at 713 730 and 760 cm-1 The first
Table 2 Measured SER peak heights for the CN stretch at 2100 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
Figure 5 A) SERS and B) NR spectra of VX Conditions A) 1 vv (10 mgml) in MeOH B) pure sample both 100 mW of 785 nm 1-min acquisition time
A
B
two have been assigned to a C-S stretch and CH2bend respectively while the latter has been attributed to either a P-C stretch or an O-C-C stretch Although the 745 cm-1 band may alternatively be assigned to a C-S stretch based on the previous measurements of CEES and HD The relatively intense bands at 890 1106 1218 1445 and 1465 cm-1 also match predicted bands at 880 1108 1216 1440 and 1464 cm-1 that are assigned to a C-C stretch CH3 rock N-C3 stretch various C-H3 bends and C-H bends respectively Both the computer generated and the measured spectra contain numerous other less intense bands One is worth mentioning A unique band appears at 370 cm-1 that is predicted at 368 cm-1 and corresponds to an O-P=O bend The surface-enhanced Raman spectrum of VX is also rich with spectral features It has the unique low frequency band at 370 cm-1 as well as a second band at 380 cm-1 that is assigned tothe S-P-O bend predicted in the normal Raman spectrum at 388 cm-1 Based on the measured and predicted normal Raman spectra the following SERS assignments are given 460 cm-1 to the CH3-P=O bend 544 cm-1 to the PO2CS wag 738 cm-1 to a C-S stretch (based on arguments above) 890 cm-1
to a C-C stretch 1101 cm-1 to a CH3 rock and 1456
cm-1 to a C-H bend The ability of SERS to measure chemical agents in water containing real-world chemical interferents was tested by using the generic tap water described in Table 1 The ability to reproduce measurements was accomplished by preparing three separate water stock solutions which were used to prepare three sample solutions each of 1 mgmL KCN in the generic tap water The samples defined as vials 1-9 were then measured at 4 points per vial in a semi-random fashion such that errors associated with stock solution preparation and errors associated with instrument drift could be identified No trends were apparent that signified such systematic errors Each spectrum collected consisted of 20 averaged scans taking 16 seconds at 8 cm-1 resolution The laser power at the sample was measured periodically during the day and it ranged from 102 to 105 mW spacccc
spot Vial 1 Vial 2 Vial 41 13754 13004 128192 13519 12692 12909
1 3 13541 12721 126394 13462 12648 12651
ave 13569 12766 12755Vial 6 Vial 3 Vial 5
1 11586 14068 10757 2 11236 14402 115122 3 11376 14546 11559
4 10894 11763 11214ave 11273 13695 11261
Vial 7 Vial8 Vial 91 11198 15804 11023
2 11243 15716 88753 3 11562 15216 11218
4 11655 15084 11077ave 11415 15455 10548
CN AVG STDEV ERR1 mgmL pk ht 12526 1560 1245
stock solution
SPIE-2003-5269 21
The CN measurements consisting of 36 data points produced an average height of 125 for the 2100 cm-1 peak with a standard deviation of 156 or 125 (Table 2) The HD and VX measurements were performed precisely the same way (Tables 3 and 4) For HD the 624 cm-1 peak was used for analysis and it had an average height of 53 with a standard deviation of 068 or 129 while for VX the 544 cm-1 peak was used for analysis and it had an average height of 1051 with a standard deviation of 308 or 293 The greater error in the VX measurements can be somewhat attributed to Vial 4 which produced lower SER signal intensities But removing this vial from the data set changes the standard deviation to 233 only a modest improvement
spot Vial 1 Vial 2 Vial 41 525 609 4682 45 675 484
1 3 527 644 5094 559 807 669
ave 51525 68375 5325Vial 6 Vial 3 Vial 5
1 521 574 575 2 536 53 4492 3 509 508 451
4 565 594 379ave 53275 5515 4635
Vial 7 Vial 8 Vial 91 457 589 413
2 583 497 4273 3 544 505 4
4 528 5 509ave 528 52275 43725
HD AVG STDEV ERR1 mgmL pk ht 530 068 1291
stock solution
spot Vial 1 Vial 2 Vial 41 1464 1034 4672 1485 99 61
1 3 1491 989 5684 1041 777 553
ave 137025 9475 5495Vial 6 Vial 3 Vial 5
1 1058 942 1293 2 697 121 965
2 3 727 1206 11124 689 1254 746
ave 79275 1153 1029Vial 7 Vial 8 Vial 9
1 1127 783 163 2 1358 812 1614
3 3 1371 908 15124 1097 875 1397
ave 123825 8445 153825
VX AVG STDEV ERR1 mgmL pk ht 1051 308 2925
stock solution
4 CONCLUSIONS In this paper we examined the ability of surface-enhanced Raman spectroscopy to reproducibly measure CN HD and VX in tap water without chemical interference Both normal and surface-enhanced Raman spectra were examined to select unique bands suitable to identify and quantify these chemical agents For SER measurements the 2100 cm-1 C-N stretch was used for CN the 624 cm-1 C-Cl stretch was used for HD and the 544 cm-1 PO2CS wag was used for VX It was determined that 1 mgmL samples of each of these chemicals measured 36 times in glass vials coated with a silver-doped sol-gel reproduced measurements with standard deviations of 125 129 and 293 It was further found that the 10 chemicals added to simulate generic tap water did not interfere with or alter the SER spectra It should be noted that the concentrations used in this study were considerably greater than those required by the JSAWM program Current work involves improving SER sensitivity and designing sampling systems with better reproducibility This includes the development of fractal silver and gold structures within the sol-gel matrix and the development of chemically selective sol-gels
ACKNOWLEDGMENTS The authors are grateful for the support of the US Army (DAAD13-02-C-0015 Joint Service Agent Water Monitor program) The authors would also like to thank Janet Jensen Ronald Crosier and Kristina Gonser for helpful discussions
Table 4 Measured SER peak heights for the VX band at 544 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
Table 3 Measured SER peak heights for the HD band at 624 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
SPIE-2003-5269 22
REFERENCES 1 Jensen JL NC Wong W R Loerop SPIE 4775-03 (2001) 2 JSAWM Requirements at wwwsbccomapgeaarmymilRDAecbcrtPRODSERJSAWMjsawmhtml 3 Erickson B ldquoThe Chemical Weapons Convention Redefines Analytical Challengerdquo Anal Chem 70 397A (1998) 4 Johnston RL Hoefler CM Fargo JC and Moberley B ldquoThe Defense Nuclear Agencyrsquos
ChemicalBiochemical Weapons Agreements Verification Technology Research Development Test and Evaluation Program and its Requirements for On-Site Analysisrdquo AT-ONSITE 5-8 (1994)
5 Black RM Clarke RJ Read RW and Reid MT ldquoApplication of GC-MS and GC-tandem MS to the analysis of chemical warfare samples found to contain residues of the nerve agent sarin sulphur mustard and their degradation productsrdquo J Chrom 662 301-321 (1994)
6 USA Army Center for Health Promotion and Preventive Medicine (USACHPPM) Technical Guideline 230A 7 Hoffland LD Piffath RJ Bouck JB rdquoSpectral signatures of chemical agents and simulantsrdquo Optical
Engineering 24 982-984 (1985) 8 Braue EHJ Pannella MGrdquoCIRCLE CELL FT-IR Analysis of Chemical Warfare Agents in Aqueous Solutionsrdquo
App Spectrosc 44 1513-1520 (1990) 9 Tseng C-H Mann CK and Vickers TJ ldquoDetermination of Organics on Metal Surfaces by Raman
Spectroscopyrdquo App Spectrosc 47 1767-1771 (1993) 10 Staff Reporter ldquoUS Army Tests Lidar to Detect Biological Toxinsrdquo Photonic Spectra pg 50 December 1998 11 Christesen SD Raman cross sections of chemical agents and simulants App Spectrosc 42 318-321 (1988) 12 Weaver MJ Farquharson S Tadayyoni MA ldquoSurface-enhancement factors for Raman scattering at silver electrodesrdquo J Chem Phys 82 4867-4874 (1985) 13 Alak AM and Vo-Dinh T ldquoSERS of Organophosphorous Chemical Agentsrdquo Anal Chem 59 2149-2153 (1987) 14 Lee Y Farquharson S ldquoRapid chemical agent identification by SERSrdquo SPIE 4378 21-26 (2001) 15 Farquharson S Maksymiuk P Ong K Christesen S ldquoChemical agent identification by surface-enhanced Raman spectroscopyrdquo SPIE 4577 166-173 (2001) 16 Spencer KM Sylvia J Clauson S and Janni J ldquoSurface Enhanced Raman as a Water Monitor for Warfare
Agents in Waterrdquo SPIE 4577 158-165 (2001) 17 Tessier P Christesen S Ong K Clemente E Lenhoff A Kaler E Velev O ldquoOn-line spectroscopic characterization of sodium cyanide with nanostructured gold surface-enhanced Raman spectroscopy substratesrdquo
App Spectrosc 56 1524-1530 (2002) 18 Farquharson S W W Smith S Elliott and J F Sperry Rapid biological agent identification by surface- enhanced Raman spectroscopy SPIE 3855110-116 (1999) 19 Farquharson S WW Smith YH Lee S Elliott and J F Sperry Detection of biological signatures A
comparison of electrolytic and metal-doped sol-gel surface-enhanced Raman media SPIE 4575 62-72 (2002) 20 Guzelian AA Sylvia J Janni J Clauson S and Spencer KM ldquoSERS of whole cell bacteria and trace levels of biological moleculesrdquo SPIE 4577 182-192 (2001) 21 Shende C Inscore F Gift A Maksymiuk P Farquharson S ldquoRapid dipicolinic acid extraction from Bacillus spores detected by surface-enhanced Raman spectroscopyrdquo App Spectrosc 58 accepted 22 Lee Y and Farquharson S ldquoSERS Sample Vials Based on Sol-Gel Process for Trace Pesticide Analysisrdquo SPIE
4206 140-146 (2000) 23 Farquharson S and Lee Y ldquoTrace Drug Analysis by SERSrdquo SPIE 4200-16 (2000) 24 Lee Y Farquharson S and Rainey P M Surface-Enhanced Raman Sensor for Trace Chemical Detection in
Water SPIE 3857 76-84 (1999) 25 Lee Y Farquharson S Kwong H and Shahriari M ldquoSol-Gel Chemical Sensor for Surface-Enhanced Raman
Spectroscopyrdquo SPIE 3537 252-260 (1998) 26 Farquharson S A Gift P Maksymiuk F Inscore and W Smith ldquopH dependence of methyl phosphonic acid
dipicolinic acid and cyanide by surface-enhanced Raman spectroscopyrdquo SPIE 5269-19 (2003) 27 Lide DR Ed Handbook of Chemistry and Physics CRC Press 77th Ed (1996-7) 28 Sosa C RJ Bartlett K KuBulat and WB Person ldquoA theoretical study of harmonic vibrational frequencies and
infrared intensities of XCH2CH2SCH2CH2X and XCH2CH2SH (= H Cl)rdquo J Phys Chem 93 577-588 (1993) 29 Hameka HF and JO Jenson ldquoComputer-generated predictions of the structure and the IR and Raman spectra of
VXrdquo ERDEC-TR-065 May 1993
- Chemagents Appendicespdf
-
- SPIE2001-4575-Bioagent-SERSapdf
-
- Detection of bioagent signatures A comparison of electrolytic and metal-doped sol-gel surface-enhanced Raman media
- Stuart Farquharson Wayne Smith and Yuan Lee
-
- Real-Time Analyzers 87 Church Street East Hartford CT 06108
-
- ABSTRACT
- 4 CONCLUSIONS
- 5 ACKNOWLEDGEMENTS
- 6 REFERENCES
-
SPIE-2003-5269 18
A software program was written that allowed selecting the sequence that the vials were measured the number of positions along the length of the vials to measure (1 to 5) and the number of scans to co-add During sample analysis the program displayed the vial being analyzed the point being analyzed and the spectrum as it was being acquired Once all the data was collected a second software program was written to rapidly analyze the data The spectra collected for all the vials on a plate could be loaded at one time and then the spectra for each point could be displayed simultaneously or separately The user could then select the Raman peak to analyze in terms of peak height or area This was accomplished by selecting points on either side of the peak to define a baseline of zero The peak height or area could then be computed for all of the spectra loaded and then exported to a spreadsheet for statistical analysis
Figure 1 A) Vial Holder 6 slots to hold 2 vials each end-to-end B) Measurement Configuration Program user selects vials to measure sequence number of points per vial (1 to 5) and number of scans per point C) Spectral Acquisition Program shows spectrum being collected which vial and position D) Spectra Analysis Program user selects spectra to analyze by plate vial and point (s) as well as two wavenumbers defining the peak and the baseline to subtract The image is of 5 repeat measurements of 10 mgL KCN in generic tap water 16 sec each 100 mW of 785 nm
3 RESULTS AND DISCUSSION Raman and surface-enhanced Raman spectra were obtained for potassium cyanide bis-(2-chloroethyl)sulfide and ethyl S-2-diisopropylamino ethyl methylphosphonothioate representing three classes of chemical agents cyanides mustards and nerve agents respectively Spectra were also obtained for 2-chloroethyl ethyl sulfide (CEES) a structural analogue to HD which was included in the study to aid in assigning spectral bands KCN salt was used for cyanide experiments to avoid the increased hazards of handling HCN gas KCN completely dissolves in water forming its conjugate acid HCN according to its Ka of 615x10-1027 and at a concentration of 1 mgmL results in a pH 107 solution This is significant in that only CN- appears to interact sufficiently with silver to produce a SER spectrum and no spectral signal is observed below pH 726 Figure 2 shows the SER and normal Raman spectra for KCN The SERS spectrum of 1mgml KCN in water shows a single intense somewhat broad feature at
A B
C D
SPIE-2003-5269 19
2100 cm-1 assigned to the single CequivN stretch The band is much sharper in the normal Raman spectra of the solid KCN salt at 2074 cm-1 This band does broaden and shift to 2080 cm-1 in solution (not shown) However the observed SERS frequency is attributed to interaction with silver and low frequency mode at 135 cm-1 attributed to a Ag-CN stretch (not shown) supports this conclusion
cm-1 band to a C-S stretch but the authors concede that it is in fact more likely a C-Cl stretch28 It appears that the most intense Raman bands at 648 692 and 747 cm-1 shift to 620 660 and 720 cm-1 in the SER spectra and are tentatively assigned as above The width of these bands suggests that they overlap underlying spectral features Additional bands in the Raman spectra occur at 972 1034 1049 1263 1286 1423 1442 2865 2935 and 2960 cm-1 Corresponding bands occur in the SER spectra at 964 1015 1054 1286 1410 1447 2865 and 2935 cm-1 Most of these bands are associated with alkane modes specifically the bands at approximately 1040 cm-1 to a C-C stretch 1290 cm-1 to a CH2 in-phase twist 1440 cm-1 to a CH2 wag 2865 cm-1 to a symmetric CH2 stretch and 2965 cm-1 to an asymmetric CH2 stretch The Raman and SER spectra of sulfur mustard were measured at the Edgewood center (Figure 4) Both spectra are largely similar to CEES The C-Cl and C-S bands in the Raman spectrum of HD now occur at 640 655 700 739 and 760 cm-1 and are more resolved possibly due to the increased molecular symmetry Theoretical calculations indicate that the first three bands are due to C-Cl stretching modes and the latter two to C-S stretching modes28 Only the C-Cl bands maintain significant intensity in the SER spectra occurring at 624 and 643 cm-1 which is attributed to the
Prior to measurements of HD CEES was examined by Raman and SER spectroscopy (Figure 3) CEES also known as half-mustard is essentially identical to HD except one of the chlorine end atoms is replaced by a hydrogen atom Again although not as toxic as HD CEES is a blister agent and dilute aqueous samples were prepared using appropriate safety equipment Both the Raman and SER spectra of CEES are similar and dominated by bands between 600 and 800 cm-1 These are associated with C-Cl and C-S stretching modes which are tentatively assigned to 648 and 747 cm-1 in the Raman spectra respectively The shoulder at 630 cm-1 the overlapped band at 660 cm-1 and the strong band at 692 cm-1 could also be due to these modes or their asymmetric counterparts It is worth noting that theoretical calculations assign the 692
Figure 2 A) SER and B) NR spectra of KCN Conditions A) 1 mgml in tap water 100 mW of 785 nm at sample 1-min acquisition time B) solid 300 mW of 785 nm 5-min All spectra are 8 cm-1 resolution
A
B
Figure 3 A) SER and B) NR spectra of CEES Conditions A) 1 vv (10 mgml) in MeOH 100 mW of 785 nm 1-min acquisition time B) neat 300 mW of 785 nm 5-min
A
B
Cl-CH2-CH2-S-CH2-CH3
A
B
Figure 4 A) SER and B) NR of HD Conditions A) 1mgml in tap water B) pure both 100 mW of 785 nm 1-min
Cl-CH2-CH2-S-CH2-CH2-Cl
SPIE-2003-5269 20
expected strong interaction between chlorine and silver and adds support to the assignment of this band to a C-Cl stretch Weaker overlapping bands occur at 670 692 and 724 cm-1 the latter possibly due to C-S stretching modes Again the alkane modes are apparent in the normal Raman spectra of HD but only a broad feature at 1300 to 1450 cm-1 suggests CH2 contributions in the SER spectrum Although the observed bands in the VX spectrum have not been assigned (Figure 4) a computer generated Raman spectrum29 predicts many of the same features with surprising accuracy and are used here Two intense bands at 460 and 530 cm-1 closely match predicted bands at 463 and 546 cm-1 assigned to a CH3-P=O bend and a PO2CS wag Three highly overlapped bands occur at 694 745 and 771 cm-1 matching predicted bands at 713 730 and 760 cm-1 The first
Table 2 Measured SER peak heights for the CN stretch at 2100 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
Figure 5 A) SERS and B) NR spectra of VX Conditions A) 1 vv (10 mgml) in MeOH B) pure sample both 100 mW of 785 nm 1-min acquisition time
A
B
two have been assigned to a C-S stretch and CH2bend respectively while the latter has been attributed to either a P-C stretch or an O-C-C stretch Although the 745 cm-1 band may alternatively be assigned to a C-S stretch based on the previous measurements of CEES and HD The relatively intense bands at 890 1106 1218 1445 and 1465 cm-1 also match predicted bands at 880 1108 1216 1440 and 1464 cm-1 that are assigned to a C-C stretch CH3 rock N-C3 stretch various C-H3 bends and C-H bends respectively Both the computer generated and the measured spectra contain numerous other less intense bands One is worth mentioning A unique band appears at 370 cm-1 that is predicted at 368 cm-1 and corresponds to an O-P=O bend The surface-enhanced Raman spectrum of VX is also rich with spectral features It has the unique low frequency band at 370 cm-1 as well as a second band at 380 cm-1 that is assigned tothe S-P-O bend predicted in the normal Raman spectrum at 388 cm-1 Based on the measured and predicted normal Raman spectra the following SERS assignments are given 460 cm-1 to the CH3-P=O bend 544 cm-1 to the PO2CS wag 738 cm-1 to a C-S stretch (based on arguments above) 890 cm-1
to a C-C stretch 1101 cm-1 to a CH3 rock and 1456
cm-1 to a C-H bend The ability of SERS to measure chemical agents in water containing real-world chemical interferents was tested by using the generic tap water described in Table 1 The ability to reproduce measurements was accomplished by preparing three separate water stock solutions which were used to prepare three sample solutions each of 1 mgmL KCN in the generic tap water The samples defined as vials 1-9 were then measured at 4 points per vial in a semi-random fashion such that errors associated with stock solution preparation and errors associated with instrument drift could be identified No trends were apparent that signified such systematic errors Each spectrum collected consisted of 20 averaged scans taking 16 seconds at 8 cm-1 resolution The laser power at the sample was measured periodically during the day and it ranged from 102 to 105 mW spacccc
spot Vial 1 Vial 2 Vial 41 13754 13004 128192 13519 12692 12909
1 3 13541 12721 126394 13462 12648 12651
ave 13569 12766 12755Vial 6 Vial 3 Vial 5
1 11586 14068 10757 2 11236 14402 115122 3 11376 14546 11559
4 10894 11763 11214ave 11273 13695 11261
Vial 7 Vial8 Vial 91 11198 15804 11023
2 11243 15716 88753 3 11562 15216 11218
4 11655 15084 11077ave 11415 15455 10548
CN AVG STDEV ERR1 mgmL pk ht 12526 1560 1245
stock solution
SPIE-2003-5269 21
The CN measurements consisting of 36 data points produced an average height of 125 for the 2100 cm-1 peak with a standard deviation of 156 or 125 (Table 2) The HD and VX measurements were performed precisely the same way (Tables 3 and 4) For HD the 624 cm-1 peak was used for analysis and it had an average height of 53 with a standard deviation of 068 or 129 while for VX the 544 cm-1 peak was used for analysis and it had an average height of 1051 with a standard deviation of 308 or 293 The greater error in the VX measurements can be somewhat attributed to Vial 4 which produced lower SER signal intensities But removing this vial from the data set changes the standard deviation to 233 only a modest improvement
spot Vial 1 Vial 2 Vial 41 525 609 4682 45 675 484
1 3 527 644 5094 559 807 669
ave 51525 68375 5325Vial 6 Vial 3 Vial 5
1 521 574 575 2 536 53 4492 3 509 508 451
4 565 594 379ave 53275 5515 4635
Vial 7 Vial 8 Vial 91 457 589 413
2 583 497 4273 3 544 505 4
4 528 5 509ave 528 52275 43725
HD AVG STDEV ERR1 mgmL pk ht 530 068 1291
stock solution
spot Vial 1 Vial 2 Vial 41 1464 1034 4672 1485 99 61
1 3 1491 989 5684 1041 777 553
ave 137025 9475 5495Vial 6 Vial 3 Vial 5
1 1058 942 1293 2 697 121 965
2 3 727 1206 11124 689 1254 746
ave 79275 1153 1029Vial 7 Vial 8 Vial 9
1 1127 783 163 2 1358 812 1614
3 3 1371 908 15124 1097 875 1397
ave 123825 8445 153825
VX AVG STDEV ERR1 mgmL pk ht 1051 308 2925
stock solution
4 CONCLUSIONS In this paper we examined the ability of surface-enhanced Raman spectroscopy to reproducibly measure CN HD and VX in tap water without chemical interference Both normal and surface-enhanced Raman spectra were examined to select unique bands suitable to identify and quantify these chemical agents For SER measurements the 2100 cm-1 C-N stretch was used for CN the 624 cm-1 C-Cl stretch was used for HD and the 544 cm-1 PO2CS wag was used for VX It was determined that 1 mgmL samples of each of these chemicals measured 36 times in glass vials coated with a silver-doped sol-gel reproduced measurements with standard deviations of 125 129 and 293 It was further found that the 10 chemicals added to simulate generic tap water did not interfere with or alter the SER spectra It should be noted that the concentrations used in this study were considerably greater than those required by the JSAWM program Current work involves improving SER sensitivity and designing sampling systems with better reproducibility This includes the development of fractal silver and gold structures within the sol-gel matrix and the development of chemically selective sol-gels
ACKNOWLEDGMENTS The authors are grateful for the support of the US Army (DAAD13-02-C-0015 Joint Service Agent Water Monitor program) The authors would also like to thank Janet Jensen Ronald Crosier and Kristina Gonser for helpful discussions
Table 4 Measured SER peak heights for the VX band at 544 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
Table 3 Measured SER peak heights for the HD band at 624 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
SPIE-2003-5269 22
REFERENCES 1 Jensen JL NC Wong W R Loerop SPIE 4775-03 (2001) 2 JSAWM Requirements at wwwsbccomapgeaarmymilRDAecbcrtPRODSERJSAWMjsawmhtml 3 Erickson B ldquoThe Chemical Weapons Convention Redefines Analytical Challengerdquo Anal Chem 70 397A (1998) 4 Johnston RL Hoefler CM Fargo JC and Moberley B ldquoThe Defense Nuclear Agencyrsquos
ChemicalBiochemical Weapons Agreements Verification Technology Research Development Test and Evaluation Program and its Requirements for On-Site Analysisrdquo AT-ONSITE 5-8 (1994)
5 Black RM Clarke RJ Read RW and Reid MT ldquoApplication of GC-MS and GC-tandem MS to the analysis of chemical warfare samples found to contain residues of the nerve agent sarin sulphur mustard and their degradation productsrdquo J Chrom 662 301-321 (1994)
6 USA Army Center for Health Promotion and Preventive Medicine (USACHPPM) Technical Guideline 230A 7 Hoffland LD Piffath RJ Bouck JB rdquoSpectral signatures of chemical agents and simulantsrdquo Optical
Engineering 24 982-984 (1985) 8 Braue EHJ Pannella MGrdquoCIRCLE CELL FT-IR Analysis of Chemical Warfare Agents in Aqueous Solutionsrdquo
App Spectrosc 44 1513-1520 (1990) 9 Tseng C-H Mann CK and Vickers TJ ldquoDetermination of Organics on Metal Surfaces by Raman
Spectroscopyrdquo App Spectrosc 47 1767-1771 (1993) 10 Staff Reporter ldquoUS Army Tests Lidar to Detect Biological Toxinsrdquo Photonic Spectra pg 50 December 1998 11 Christesen SD Raman cross sections of chemical agents and simulants App Spectrosc 42 318-321 (1988) 12 Weaver MJ Farquharson S Tadayyoni MA ldquoSurface-enhancement factors for Raman scattering at silver electrodesrdquo J Chem Phys 82 4867-4874 (1985) 13 Alak AM and Vo-Dinh T ldquoSERS of Organophosphorous Chemical Agentsrdquo Anal Chem 59 2149-2153 (1987) 14 Lee Y Farquharson S ldquoRapid chemical agent identification by SERSrdquo SPIE 4378 21-26 (2001) 15 Farquharson S Maksymiuk P Ong K Christesen S ldquoChemical agent identification by surface-enhanced Raman spectroscopyrdquo SPIE 4577 166-173 (2001) 16 Spencer KM Sylvia J Clauson S and Janni J ldquoSurface Enhanced Raman as a Water Monitor for Warfare
Agents in Waterrdquo SPIE 4577 158-165 (2001) 17 Tessier P Christesen S Ong K Clemente E Lenhoff A Kaler E Velev O ldquoOn-line spectroscopic characterization of sodium cyanide with nanostructured gold surface-enhanced Raman spectroscopy substratesrdquo
App Spectrosc 56 1524-1530 (2002) 18 Farquharson S W W Smith S Elliott and J F Sperry Rapid biological agent identification by surface- enhanced Raman spectroscopy SPIE 3855110-116 (1999) 19 Farquharson S WW Smith YH Lee S Elliott and J F Sperry Detection of biological signatures A
comparison of electrolytic and metal-doped sol-gel surface-enhanced Raman media SPIE 4575 62-72 (2002) 20 Guzelian AA Sylvia J Janni J Clauson S and Spencer KM ldquoSERS of whole cell bacteria and trace levels of biological moleculesrdquo SPIE 4577 182-192 (2001) 21 Shende C Inscore F Gift A Maksymiuk P Farquharson S ldquoRapid dipicolinic acid extraction from Bacillus spores detected by surface-enhanced Raman spectroscopyrdquo App Spectrosc 58 accepted 22 Lee Y and Farquharson S ldquoSERS Sample Vials Based on Sol-Gel Process for Trace Pesticide Analysisrdquo SPIE
4206 140-146 (2000) 23 Farquharson S and Lee Y ldquoTrace Drug Analysis by SERSrdquo SPIE 4200-16 (2000) 24 Lee Y Farquharson S and Rainey P M Surface-Enhanced Raman Sensor for Trace Chemical Detection in
Water SPIE 3857 76-84 (1999) 25 Lee Y Farquharson S Kwong H and Shahriari M ldquoSol-Gel Chemical Sensor for Surface-Enhanced Raman
Spectroscopyrdquo SPIE 3537 252-260 (1998) 26 Farquharson S A Gift P Maksymiuk F Inscore and W Smith ldquopH dependence of methyl phosphonic acid
dipicolinic acid and cyanide by surface-enhanced Raman spectroscopyrdquo SPIE 5269-19 (2003) 27 Lide DR Ed Handbook of Chemistry and Physics CRC Press 77th Ed (1996-7) 28 Sosa C RJ Bartlett K KuBulat and WB Person ldquoA theoretical study of harmonic vibrational frequencies and
infrared intensities of XCH2CH2SCH2CH2X and XCH2CH2SH (= H Cl)rdquo J Phys Chem 93 577-588 (1993) 29 Hameka HF and JO Jenson ldquoComputer-generated predictions of the structure and the IR and Raman spectra of
VXrdquo ERDEC-TR-065 May 1993
- Chemagents Appendicespdf
-
- SPIE2001-4575-Bioagent-SERSapdf
-
- Detection of bioagent signatures A comparison of electrolytic and metal-doped sol-gel surface-enhanced Raman media
- Stuart Farquharson Wayne Smith and Yuan Lee
-
- Real-Time Analyzers 87 Church Street East Hartford CT 06108
-
- ABSTRACT
- 4 CONCLUSIONS
- 5 ACKNOWLEDGEMENTS
- 6 REFERENCES
-
SPIE-2003-5269 19
2100 cm-1 assigned to the single CequivN stretch The band is much sharper in the normal Raman spectra of the solid KCN salt at 2074 cm-1 This band does broaden and shift to 2080 cm-1 in solution (not shown) However the observed SERS frequency is attributed to interaction with silver and low frequency mode at 135 cm-1 attributed to a Ag-CN stretch (not shown) supports this conclusion
cm-1 band to a C-S stretch but the authors concede that it is in fact more likely a C-Cl stretch28 It appears that the most intense Raman bands at 648 692 and 747 cm-1 shift to 620 660 and 720 cm-1 in the SER spectra and are tentatively assigned as above The width of these bands suggests that they overlap underlying spectral features Additional bands in the Raman spectra occur at 972 1034 1049 1263 1286 1423 1442 2865 2935 and 2960 cm-1 Corresponding bands occur in the SER spectra at 964 1015 1054 1286 1410 1447 2865 and 2935 cm-1 Most of these bands are associated with alkane modes specifically the bands at approximately 1040 cm-1 to a C-C stretch 1290 cm-1 to a CH2 in-phase twist 1440 cm-1 to a CH2 wag 2865 cm-1 to a symmetric CH2 stretch and 2965 cm-1 to an asymmetric CH2 stretch The Raman and SER spectra of sulfur mustard were measured at the Edgewood center (Figure 4) Both spectra are largely similar to CEES The C-Cl and C-S bands in the Raman spectrum of HD now occur at 640 655 700 739 and 760 cm-1 and are more resolved possibly due to the increased molecular symmetry Theoretical calculations indicate that the first three bands are due to C-Cl stretching modes and the latter two to C-S stretching modes28 Only the C-Cl bands maintain significant intensity in the SER spectra occurring at 624 and 643 cm-1 which is attributed to the
Prior to measurements of HD CEES was examined by Raman and SER spectroscopy (Figure 3) CEES also known as half-mustard is essentially identical to HD except one of the chlorine end atoms is replaced by a hydrogen atom Again although not as toxic as HD CEES is a blister agent and dilute aqueous samples were prepared using appropriate safety equipment Both the Raman and SER spectra of CEES are similar and dominated by bands between 600 and 800 cm-1 These are associated with C-Cl and C-S stretching modes which are tentatively assigned to 648 and 747 cm-1 in the Raman spectra respectively The shoulder at 630 cm-1 the overlapped band at 660 cm-1 and the strong band at 692 cm-1 could also be due to these modes or their asymmetric counterparts It is worth noting that theoretical calculations assign the 692
Figure 2 A) SER and B) NR spectra of KCN Conditions A) 1 mgml in tap water 100 mW of 785 nm at sample 1-min acquisition time B) solid 300 mW of 785 nm 5-min All spectra are 8 cm-1 resolution
A
B
Figure 3 A) SER and B) NR spectra of CEES Conditions A) 1 vv (10 mgml) in MeOH 100 mW of 785 nm 1-min acquisition time B) neat 300 mW of 785 nm 5-min
A
B
Cl-CH2-CH2-S-CH2-CH3
A
B
Figure 4 A) SER and B) NR of HD Conditions A) 1mgml in tap water B) pure both 100 mW of 785 nm 1-min
Cl-CH2-CH2-S-CH2-CH2-Cl
SPIE-2003-5269 20
expected strong interaction between chlorine and silver and adds support to the assignment of this band to a C-Cl stretch Weaker overlapping bands occur at 670 692 and 724 cm-1 the latter possibly due to C-S stretching modes Again the alkane modes are apparent in the normal Raman spectra of HD but only a broad feature at 1300 to 1450 cm-1 suggests CH2 contributions in the SER spectrum Although the observed bands in the VX spectrum have not been assigned (Figure 4) a computer generated Raman spectrum29 predicts many of the same features with surprising accuracy and are used here Two intense bands at 460 and 530 cm-1 closely match predicted bands at 463 and 546 cm-1 assigned to a CH3-P=O bend and a PO2CS wag Three highly overlapped bands occur at 694 745 and 771 cm-1 matching predicted bands at 713 730 and 760 cm-1 The first
Table 2 Measured SER peak heights for the CN stretch at 2100 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
Figure 5 A) SERS and B) NR spectra of VX Conditions A) 1 vv (10 mgml) in MeOH B) pure sample both 100 mW of 785 nm 1-min acquisition time
A
B
two have been assigned to a C-S stretch and CH2bend respectively while the latter has been attributed to either a P-C stretch or an O-C-C stretch Although the 745 cm-1 band may alternatively be assigned to a C-S stretch based on the previous measurements of CEES and HD The relatively intense bands at 890 1106 1218 1445 and 1465 cm-1 also match predicted bands at 880 1108 1216 1440 and 1464 cm-1 that are assigned to a C-C stretch CH3 rock N-C3 stretch various C-H3 bends and C-H bends respectively Both the computer generated and the measured spectra contain numerous other less intense bands One is worth mentioning A unique band appears at 370 cm-1 that is predicted at 368 cm-1 and corresponds to an O-P=O bend The surface-enhanced Raman spectrum of VX is also rich with spectral features It has the unique low frequency band at 370 cm-1 as well as a second band at 380 cm-1 that is assigned tothe S-P-O bend predicted in the normal Raman spectrum at 388 cm-1 Based on the measured and predicted normal Raman spectra the following SERS assignments are given 460 cm-1 to the CH3-P=O bend 544 cm-1 to the PO2CS wag 738 cm-1 to a C-S stretch (based on arguments above) 890 cm-1
to a C-C stretch 1101 cm-1 to a CH3 rock and 1456
cm-1 to a C-H bend The ability of SERS to measure chemical agents in water containing real-world chemical interferents was tested by using the generic tap water described in Table 1 The ability to reproduce measurements was accomplished by preparing three separate water stock solutions which were used to prepare three sample solutions each of 1 mgmL KCN in the generic tap water The samples defined as vials 1-9 were then measured at 4 points per vial in a semi-random fashion such that errors associated with stock solution preparation and errors associated with instrument drift could be identified No trends were apparent that signified such systematic errors Each spectrum collected consisted of 20 averaged scans taking 16 seconds at 8 cm-1 resolution The laser power at the sample was measured periodically during the day and it ranged from 102 to 105 mW spacccc
spot Vial 1 Vial 2 Vial 41 13754 13004 128192 13519 12692 12909
1 3 13541 12721 126394 13462 12648 12651
ave 13569 12766 12755Vial 6 Vial 3 Vial 5
1 11586 14068 10757 2 11236 14402 115122 3 11376 14546 11559
4 10894 11763 11214ave 11273 13695 11261
Vial 7 Vial8 Vial 91 11198 15804 11023
2 11243 15716 88753 3 11562 15216 11218
4 11655 15084 11077ave 11415 15455 10548
CN AVG STDEV ERR1 mgmL pk ht 12526 1560 1245
stock solution
SPIE-2003-5269 21
The CN measurements consisting of 36 data points produced an average height of 125 for the 2100 cm-1 peak with a standard deviation of 156 or 125 (Table 2) The HD and VX measurements were performed precisely the same way (Tables 3 and 4) For HD the 624 cm-1 peak was used for analysis and it had an average height of 53 with a standard deviation of 068 or 129 while for VX the 544 cm-1 peak was used for analysis and it had an average height of 1051 with a standard deviation of 308 or 293 The greater error in the VX measurements can be somewhat attributed to Vial 4 which produced lower SER signal intensities But removing this vial from the data set changes the standard deviation to 233 only a modest improvement
spot Vial 1 Vial 2 Vial 41 525 609 4682 45 675 484
1 3 527 644 5094 559 807 669
ave 51525 68375 5325Vial 6 Vial 3 Vial 5
1 521 574 575 2 536 53 4492 3 509 508 451
4 565 594 379ave 53275 5515 4635
Vial 7 Vial 8 Vial 91 457 589 413
2 583 497 4273 3 544 505 4
4 528 5 509ave 528 52275 43725
HD AVG STDEV ERR1 mgmL pk ht 530 068 1291
stock solution
spot Vial 1 Vial 2 Vial 41 1464 1034 4672 1485 99 61
1 3 1491 989 5684 1041 777 553
ave 137025 9475 5495Vial 6 Vial 3 Vial 5
1 1058 942 1293 2 697 121 965
2 3 727 1206 11124 689 1254 746
ave 79275 1153 1029Vial 7 Vial 8 Vial 9
1 1127 783 163 2 1358 812 1614
3 3 1371 908 15124 1097 875 1397
ave 123825 8445 153825
VX AVG STDEV ERR1 mgmL pk ht 1051 308 2925
stock solution
4 CONCLUSIONS In this paper we examined the ability of surface-enhanced Raman spectroscopy to reproducibly measure CN HD and VX in tap water without chemical interference Both normal and surface-enhanced Raman spectra were examined to select unique bands suitable to identify and quantify these chemical agents For SER measurements the 2100 cm-1 C-N stretch was used for CN the 624 cm-1 C-Cl stretch was used for HD and the 544 cm-1 PO2CS wag was used for VX It was determined that 1 mgmL samples of each of these chemicals measured 36 times in glass vials coated with a silver-doped sol-gel reproduced measurements with standard deviations of 125 129 and 293 It was further found that the 10 chemicals added to simulate generic tap water did not interfere with or alter the SER spectra It should be noted that the concentrations used in this study were considerably greater than those required by the JSAWM program Current work involves improving SER sensitivity and designing sampling systems with better reproducibility This includes the development of fractal silver and gold structures within the sol-gel matrix and the development of chemically selective sol-gels
ACKNOWLEDGMENTS The authors are grateful for the support of the US Army (DAAD13-02-C-0015 Joint Service Agent Water Monitor program) The authors would also like to thank Janet Jensen Ronald Crosier and Kristina Gonser for helpful discussions
Table 4 Measured SER peak heights for the VX band at 544 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
Table 3 Measured SER peak heights for the HD band at 624 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
SPIE-2003-5269 22
REFERENCES 1 Jensen JL NC Wong W R Loerop SPIE 4775-03 (2001) 2 JSAWM Requirements at wwwsbccomapgeaarmymilRDAecbcrtPRODSERJSAWMjsawmhtml 3 Erickson B ldquoThe Chemical Weapons Convention Redefines Analytical Challengerdquo Anal Chem 70 397A (1998) 4 Johnston RL Hoefler CM Fargo JC and Moberley B ldquoThe Defense Nuclear Agencyrsquos
ChemicalBiochemical Weapons Agreements Verification Technology Research Development Test and Evaluation Program and its Requirements for On-Site Analysisrdquo AT-ONSITE 5-8 (1994)
5 Black RM Clarke RJ Read RW and Reid MT ldquoApplication of GC-MS and GC-tandem MS to the analysis of chemical warfare samples found to contain residues of the nerve agent sarin sulphur mustard and their degradation productsrdquo J Chrom 662 301-321 (1994)
6 USA Army Center for Health Promotion and Preventive Medicine (USACHPPM) Technical Guideline 230A 7 Hoffland LD Piffath RJ Bouck JB rdquoSpectral signatures of chemical agents and simulantsrdquo Optical
Engineering 24 982-984 (1985) 8 Braue EHJ Pannella MGrdquoCIRCLE CELL FT-IR Analysis of Chemical Warfare Agents in Aqueous Solutionsrdquo
App Spectrosc 44 1513-1520 (1990) 9 Tseng C-H Mann CK and Vickers TJ ldquoDetermination of Organics on Metal Surfaces by Raman
Spectroscopyrdquo App Spectrosc 47 1767-1771 (1993) 10 Staff Reporter ldquoUS Army Tests Lidar to Detect Biological Toxinsrdquo Photonic Spectra pg 50 December 1998 11 Christesen SD Raman cross sections of chemical agents and simulants App Spectrosc 42 318-321 (1988) 12 Weaver MJ Farquharson S Tadayyoni MA ldquoSurface-enhancement factors for Raman scattering at silver electrodesrdquo J Chem Phys 82 4867-4874 (1985) 13 Alak AM and Vo-Dinh T ldquoSERS of Organophosphorous Chemical Agentsrdquo Anal Chem 59 2149-2153 (1987) 14 Lee Y Farquharson S ldquoRapid chemical agent identification by SERSrdquo SPIE 4378 21-26 (2001) 15 Farquharson S Maksymiuk P Ong K Christesen S ldquoChemical agent identification by surface-enhanced Raman spectroscopyrdquo SPIE 4577 166-173 (2001) 16 Spencer KM Sylvia J Clauson S and Janni J ldquoSurface Enhanced Raman as a Water Monitor for Warfare
Agents in Waterrdquo SPIE 4577 158-165 (2001) 17 Tessier P Christesen S Ong K Clemente E Lenhoff A Kaler E Velev O ldquoOn-line spectroscopic characterization of sodium cyanide with nanostructured gold surface-enhanced Raman spectroscopy substratesrdquo
App Spectrosc 56 1524-1530 (2002) 18 Farquharson S W W Smith S Elliott and J F Sperry Rapid biological agent identification by surface- enhanced Raman spectroscopy SPIE 3855110-116 (1999) 19 Farquharson S WW Smith YH Lee S Elliott and J F Sperry Detection of biological signatures A
comparison of electrolytic and metal-doped sol-gel surface-enhanced Raman media SPIE 4575 62-72 (2002) 20 Guzelian AA Sylvia J Janni J Clauson S and Spencer KM ldquoSERS of whole cell bacteria and trace levels of biological moleculesrdquo SPIE 4577 182-192 (2001) 21 Shende C Inscore F Gift A Maksymiuk P Farquharson S ldquoRapid dipicolinic acid extraction from Bacillus spores detected by surface-enhanced Raman spectroscopyrdquo App Spectrosc 58 accepted 22 Lee Y and Farquharson S ldquoSERS Sample Vials Based on Sol-Gel Process for Trace Pesticide Analysisrdquo SPIE
4206 140-146 (2000) 23 Farquharson S and Lee Y ldquoTrace Drug Analysis by SERSrdquo SPIE 4200-16 (2000) 24 Lee Y Farquharson S and Rainey P M Surface-Enhanced Raman Sensor for Trace Chemical Detection in
Water SPIE 3857 76-84 (1999) 25 Lee Y Farquharson S Kwong H and Shahriari M ldquoSol-Gel Chemical Sensor for Surface-Enhanced Raman
Spectroscopyrdquo SPIE 3537 252-260 (1998) 26 Farquharson S A Gift P Maksymiuk F Inscore and W Smith ldquopH dependence of methyl phosphonic acid
dipicolinic acid and cyanide by surface-enhanced Raman spectroscopyrdquo SPIE 5269-19 (2003) 27 Lide DR Ed Handbook of Chemistry and Physics CRC Press 77th Ed (1996-7) 28 Sosa C RJ Bartlett K KuBulat and WB Person ldquoA theoretical study of harmonic vibrational frequencies and
infrared intensities of XCH2CH2SCH2CH2X and XCH2CH2SH (= H Cl)rdquo J Phys Chem 93 577-588 (1993) 29 Hameka HF and JO Jenson ldquoComputer-generated predictions of the structure and the IR and Raman spectra of
VXrdquo ERDEC-TR-065 May 1993
- Chemagents Appendicespdf
-
- SPIE2001-4575-Bioagent-SERSapdf
-
- Detection of bioagent signatures A comparison of electrolytic and metal-doped sol-gel surface-enhanced Raman media
- Stuart Farquharson Wayne Smith and Yuan Lee
-
- Real-Time Analyzers 87 Church Street East Hartford CT 06108
-
- ABSTRACT
- 4 CONCLUSIONS
- 5 ACKNOWLEDGEMENTS
- 6 REFERENCES
-
SPIE-2003-5269 20
expected strong interaction between chlorine and silver and adds support to the assignment of this band to a C-Cl stretch Weaker overlapping bands occur at 670 692 and 724 cm-1 the latter possibly due to C-S stretching modes Again the alkane modes are apparent in the normal Raman spectra of HD but only a broad feature at 1300 to 1450 cm-1 suggests CH2 contributions in the SER spectrum Although the observed bands in the VX spectrum have not been assigned (Figure 4) a computer generated Raman spectrum29 predicts many of the same features with surprising accuracy and are used here Two intense bands at 460 and 530 cm-1 closely match predicted bands at 463 and 546 cm-1 assigned to a CH3-P=O bend and a PO2CS wag Three highly overlapped bands occur at 694 745 and 771 cm-1 matching predicted bands at 713 730 and 760 cm-1 The first
Table 2 Measured SER peak heights for the CN stretch at 2100 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
Figure 5 A) SERS and B) NR spectra of VX Conditions A) 1 vv (10 mgml) in MeOH B) pure sample both 100 mW of 785 nm 1-min acquisition time
A
B
two have been assigned to a C-S stretch and CH2bend respectively while the latter has been attributed to either a P-C stretch or an O-C-C stretch Although the 745 cm-1 band may alternatively be assigned to a C-S stretch based on the previous measurements of CEES and HD The relatively intense bands at 890 1106 1218 1445 and 1465 cm-1 also match predicted bands at 880 1108 1216 1440 and 1464 cm-1 that are assigned to a C-C stretch CH3 rock N-C3 stretch various C-H3 bends and C-H bends respectively Both the computer generated and the measured spectra contain numerous other less intense bands One is worth mentioning A unique band appears at 370 cm-1 that is predicted at 368 cm-1 and corresponds to an O-P=O bend The surface-enhanced Raman spectrum of VX is also rich with spectral features It has the unique low frequency band at 370 cm-1 as well as a second band at 380 cm-1 that is assigned tothe S-P-O bend predicted in the normal Raman spectrum at 388 cm-1 Based on the measured and predicted normal Raman spectra the following SERS assignments are given 460 cm-1 to the CH3-P=O bend 544 cm-1 to the PO2CS wag 738 cm-1 to a C-S stretch (based on arguments above) 890 cm-1
to a C-C stretch 1101 cm-1 to a CH3 rock and 1456
cm-1 to a C-H bend The ability of SERS to measure chemical agents in water containing real-world chemical interferents was tested by using the generic tap water described in Table 1 The ability to reproduce measurements was accomplished by preparing three separate water stock solutions which were used to prepare three sample solutions each of 1 mgmL KCN in the generic tap water The samples defined as vials 1-9 were then measured at 4 points per vial in a semi-random fashion such that errors associated with stock solution preparation and errors associated with instrument drift could be identified No trends were apparent that signified such systematic errors Each spectrum collected consisted of 20 averaged scans taking 16 seconds at 8 cm-1 resolution The laser power at the sample was measured periodically during the day and it ranged from 102 to 105 mW spacccc
spot Vial 1 Vial 2 Vial 41 13754 13004 128192 13519 12692 12909
1 3 13541 12721 126394 13462 12648 12651
ave 13569 12766 12755Vial 6 Vial 3 Vial 5
1 11586 14068 10757 2 11236 14402 115122 3 11376 14546 11559
4 10894 11763 11214ave 11273 13695 11261
Vial 7 Vial8 Vial 91 11198 15804 11023
2 11243 15716 88753 3 11562 15216 11218
4 11655 15084 11077ave 11415 15455 10548
CN AVG STDEV ERR1 mgmL pk ht 12526 1560 1245
stock solution
SPIE-2003-5269 21
The CN measurements consisting of 36 data points produced an average height of 125 for the 2100 cm-1 peak with a standard deviation of 156 or 125 (Table 2) The HD and VX measurements were performed precisely the same way (Tables 3 and 4) For HD the 624 cm-1 peak was used for analysis and it had an average height of 53 with a standard deviation of 068 or 129 while for VX the 544 cm-1 peak was used for analysis and it had an average height of 1051 with a standard deviation of 308 or 293 The greater error in the VX measurements can be somewhat attributed to Vial 4 which produced lower SER signal intensities But removing this vial from the data set changes the standard deviation to 233 only a modest improvement
spot Vial 1 Vial 2 Vial 41 525 609 4682 45 675 484
1 3 527 644 5094 559 807 669
ave 51525 68375 5325Vial 6 Vial 3 Vial 5
1 521 574 575 2 536 53 4492 3 509 508 451
4 565 594 379ave 53275 5515 4635
Vial 7 Vial 8 Vial 91 457 589 413
2 583 497 4273 3 544 505 4
4 528 5 509ave 528 52275 43725
HD AVG STDEV ERR1 mgmL pk ht 530 068 1291
stock solution
spot Vial 1 Vial 2 Vial 41 1464 1034 4672 1485 99 61
1 3 1491 989 5684 1041 777 553
ave 137025 9475 5495Vial 6 Vial 3 Vial 5
1 1058 942 1293 2 697 121 965
2 3 727 1206 11124 689 1254 746
ave 79275 1153 1029Vial 7 Vial 8 Vial 9
1 1127 783 163 2 1358 812 1614
3 3 1371 908 15124 1097 875 1397
ave 123825 8445 153825
VX AVG STDEV ERR1 mgmL pk ht 1051 308 2925
stock solution
4 CONCLUSIONS In this paper we examined the ability of surface-enhanced Raman spectroscopy to reproducibly measure CN HD and VX in tap water without chemical interference Both normal and surface-enhanced Raman spectra were examined to select unique bands suitable to identify and quantify these chemical agents For SER measurements the 2100 cm-1 C-N stretch was used for CN the 624 cm-1 C-Cl stretch was used for HD and the 544 cm-1 PO2CS wag was used for VX It was determined that 1 mgmL samples of each of these chemicals measured 36 times in glass vials coated with a silver-doped sol-gel reproduced measurements with standard deviations of 125 129 and 293 It was further found that the 10 chemicals added to simulate generic tap water did not interfere with or alter the SER spectra It should be noted that the concentrations used in this study were considerably greater than those required by the JSAWM program Current work involves improving SER sensitivity and designing sampling systems with better reproducibility This includes the development of fractal silver and gold structures within the sol-gel matrix and the development of chemically selective sol-gels
ACKNOWLEDGMENTS The authors are grateful for the support of the US Army (DAAD13-02-C-0015 Joint Service Agent Water Monitor program) The authors would also like to thank Janet Jensen Ronald Crosier and Kristina Gonser for helpful discussions
Table 4 Measured SER peak heights for the VX band at 544 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
Table 3 Measured SER peak heights for the HD band at 624 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
SPIE-2003-5269 22
REFERENCES 1 Jensen JL NC Wong W R Loerop SPIE 4775-03 (2001) 2 JSAWM Requirements at wwwsbccomapgeaarmymilRDAecbcrtPRODSERJSAWMjsawmhtml 3 Erickson B ldquoThe Chemical Weapons Convention Redefines Analytical Challengerdquo Anal Chem 70 397A (1998) 4 Johnston RL Hoefler CM Fargo JC and Moberley B ldquoThe Defense Nuclear Agencyrsquos
ChemicalBiochemical Weapons Agreements Verification Technology Research Development Test and Evaluation Program and its Requirements for On-Site Analysisrdquo AT-ONSITE 5-8 (1994)
5 Black RM Clarke RJ Read RW and Reid MT ldquoApplication of GC-MS and GC-tandem MS to the analysis of chemical warfare samples found to contain residues of the nerve agent sarin sulphur mustard and their degradation productsrdquo J Chrom 662 301-321 (1994)
6 USA Army Center for Health Promotion and Preventive Medicine (USACHPPM) Technical Guideline 230A 7 Hoffland LD Piffath RJ Bouck JB rdquoSpectral signatures of chemical agents and simulantsrdquo Optical
Engineering 24 982-984 (1985) 8 Braue EHJ Pannella MGrdquoCIRCLE CELL FT-IR Analysis of Chemical Warfare Agents in Aqueous Solutionsrdquo
App Spectrosc 44 1513-1520 (1990) 9 Tseng C-H Mann CK and Vickers TJ ldquoDetermination of Organics on Metal Surfaces by Raman
Spectroscopyrdquo App Spectrosc 47 1767-1771 (1993) 10 Staff Reporter ldquoUS Army Tests Lidar to Detect Biological Toxinsrdquo Photonic Spectra pg 50 December 1998 11 Christesen SD Raman cross sections of chemical agents and simulants App Spectrosc 42 318-321 (1988) 12 Weaver MJ Farquharson S Tadayyoni MA ldquoSurface-enhancement factors for Raman scattering at silver electrodesrdquo J Chem Phys 82 4867-4874 (1985) 13 Alak AM and Vo-Dinh T ldquoSERS of Organophosphorous Chemical Agentsrdquo Anal Chem 59 2149-2153 (1987) 14 Lee Y Farquharson S ldquoRapid chemical agent identification by SERSrdquo SPIE 4378 21-26 (2001) 15 Farquharson S Maksymiuk P Ong K Christesen S ldquoChemical agent identification by surface-enhanced Raman spectroscopyrdquo SPIE 4577 166-173 (2001) 16 Spencer KM Sylvia J Clauson S and Janni J ldquoSurface Enhanced Raman as a Water Monitor for Warfare
Agents in Waterrdquo SPIE 4577 158-165 (2001) 17 Tessier P Christesen S Ong K Clemente E Lenhoff A Kaler E Velev O ldquoOn-line spectroscopic characterization of sodium cyanide with nanostructured gold surface-enhanced Raman spectroscopy substratesrdquo
App Spectrosc 56 1524-1530 (2002) 18 Farquharson S W W Smith S Elliott and J F Sperry Rapid biological agent identification by surface- enhanced Raman spectroscopy SPIE 3855110-116 (1999) 19 Farquharson S WW Smith YH Lee S Elliott and J F Sperry Detection of biological signatures A
comparison of electrolytic and metal-doped sol-gel surface-enhanced Raman media SPIE 4575 62-72 (2002) 20 Guzelian AA Sylvia J Janni J Clauson S and Spencer KM ldquoSERS of whole cell bacteria and trace levels of biological moleculesrdquo SPIE 4577 182-192 (2001) 21 Shende C Inscore F Gift A Maksymiuk P Farquharson S ldquoRapid dipicolinic acid extraction from Bacillus spores detected by surface-enhanced Raman spectroscopyrdquo App Spectrosc 58 accepted 22 Lee Y and Farquharson S ldquoSERS Sample Vials Based on Sol-Gel Process for Trace Pesticide Analysisrdquo SPIE
4206 140-146 (2000) 23 Farquharson S and Lee Y ldquoTrace Drug Analysis by SERSrdquo SPIE 4200-16 (2000) 24 Lee Y Farquharson S and Rainey P M Surface-Enhanced Raman Sensor for Trace Chemical Detection in
Water SPIE 3857 76-84 (1999) 25 Lee Y Farquharson S Kwong H and Shahriari M ldquoSol-Gel Chemical Sensor for Surface-Enhanced Raman
Spectroscopyrdquo SPIE 3537 252-260 (1998) 26 Farquharson S A Gift P Maksymiuk F Inscore and W Smith ldquopH dependence of methyl phosphonic acid
dipicolinic acid and cyanide by surface-enhanced Raman spectroscopyrdquo SPIE 5269-19 (2003) 27 Lide DR Ed Handbook of Chemistry and Physics CRC Press 77th Ed (1996-7) 28 Sosa C RJ Bartlett K KuBulat and WB Person ldquoA theoretical study of harmonic vibrational frequencies and
infrared intensities of XCH2CH2SCH2CH2X and XCH2CH2SH (= H Cl)rdquo J Phys Chem 93 577-588 (1993) 29 Hameka HF and JO Jenson ldquoComputer-generated predictions of the structure and the IR and Raman spectra of
VXrdquo ERDEC-TR-065 May 1993
- Chemagents Appendicespdf
-
- SPIE2001-4575-Bioagent-SERSapdf
-
- Detection of bioagent signatures A comparison of electrolytic and metal-doped sol-gel surface-enhanced Raman media
- Stuart Farquharson Wayne Smith and Yuan Lee
-
- Real-Time Analyzers 87 Church Street East Hartford CT 06108
-
- ABSTRACT
- 4 CONCLUSIONS
- 5 ACKNOWLEDGEMENTS
- 6 REFERENCES
-
SPIE-2003-5269 21
The CN measurements consisting of 36 data points produced an average height of 125 for the 2100 cm-1 peak with a standard deviation of 156 or 125 (Table 2) The HD and VX measurements were performed precisely the same way (Tables 3 and 4) For HD the 624 cm-1 peak was used for analysis and it had an average height of 53 with a standard deviation of 068 or 129 while for VX the 544 cm-1 peak was used for analysis and it had an average height of 1051 with a standard deviation of 308 or 293 The greater error in the VX measurements can be somewhat attributed to Vial 4 which produced lower SER signal intensities But removing this vial from the data set changes the standard deviation to 233 only a modest improvement
spot Vial 1 Vial 2 Vial 41 525 609 4682 45 675 484
1 3 527 644 5094 559 807 669
ave 51525 68375 5325Vial 6 Vial 3 Vial 5
1 521 574 575 2 536 53 4492 3 509 508 451
4 565 594 379ave 53275 5515 4635
Vial 7 Vial 8 Vial 91 457 589 413
2 583 497 4273 3 544 505 4
4 528 5 509ave 528 52275 43725
HD AVG STDEV ERR1 mgmL pk ht 530 068 1291
stock solution
spot Vial 1 Vial 2 Vial 41 1464 1034 4672 1485 99 61
1 3 1491 989 5684 1041 777 553
ave 137025 9475 5495Vial 6 Vial 3 Vial 5
1 1058 942 1293 2 697 121 965
2 3 727 1206 11124 689 1254 746
ave 79275 1153 1029Vial 7 Vial 8 Vial 9
1 1127 783 163 2 1358 812 1614
3 3 1371 908 15124 1097 875 1397
ave 123825 8445 153825
VX AVG STDEV ERR1 mgmL pk ht 1051 308 2925
stock solution
4 CONCLUSIONS In this paper we examined the ability of surface-enhanced Raman spectroscopy to reproducibly measure CN HD and VX in tap water without chemical interference Both normal and surface-enhanced Raman spectra were examined to select unique bands suitable to identify and quantify these chemical agents For SER measurements the 2100 cm-1 C-N stretch was used for CN the 624 cm-1 C-Cl stretch was used for HD and the 544 cm-1 PO2CS wag was used for VX It was determined that 1 mgmL samples of each of these chemicals measured 36 times in glass vials coated with a silver-doped sol-gel reproduced measurements with standard deviations of 125 129 and 293 It was further found that the 10 chemicals added to simulate generic tap water did not interfere with or alter the SER spectra It should be noted that the concentrations used in this study were considerably greater than those required by the JSAWM program Current work involves improving SER sensitivity and designing sampling systems with better reproducibility This includes the development of fractal silver and gold structures within the sol-gel matrix and the development of chemically selective sol-gels
ACKNOWLEDGMENTS The authors are grateful for the support of the US Army (DAAD13-02-C-0015 Joint Service Agent Water Monitor program) The authors would also like to thank Janet Jensen Ronald Crosier and Kristina Gonser for helpful discussions
Table 4 Measured SER peak heights for the VX band at 544 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
Table 3 Measured SER peak heights for the HD band at 624 cm-1 for 1 mgmL of sample in 3 stock solutions repeated 3 times and measured 4 times per vial
SPIE-2003-5269 22
REFERENCES 1 Jensen JL NC Wong W R Loerop SPIE 4775-03 (2001) 2 JSAWM Requirements at wwwsbccomapgeaarmymilRDAecbcrtPRODSERJSAWMjsawmhtml 3 Erickson B ldquoThe Chemical Weapons Convention Redefines Analytical Challengerdquo Anal Chem 70 397A (1998) 4 Johnston RL Hoefler CM Fargo JC and Moberley B ldquoThe Defense Nuclear Agencyrsquos
ChemicalBiochemical Weapons Agreements Verification Technology Research Development Test and Evaluation Program and its Requirements for On-Site Analysisrdquo AT-ONSITE 5-8 (1994)
5 Black RM Clarke RJ Read RW and Reid MT ldquoApplication of GC-MS and GC-tandem MS to the analysis of chemical warfare samples found to contain residues of the nerve agent sarin sulphur mustard and their degradation productsrdquo J Chrom 662 301-321 (1994)
6 USA Army Center for Health Promotion and Preventive Medicine (USACHPPM) Technical Guideline 230A 7 Hoffland LD Piffath RJ Bouck JB rdquoSpectral signatures of chemical agents and simulantsrdquo Optical
Engineering 24 982-984 (1985) 8 Braue EHJ Pannella MGrdquoCIRCLE CELL FT-IR Analysis of Chemical Warfare Agents in Aqueous Solutionsrdquo
App Spectrosc 44 1513-1520 (1990) 9 Tseng C-H Mann CK and Vickers TJ ldquoDetermination of Organics on Metal Surfaces by Raman
Spectroscopyrdquo App Spectrosc 47 1767-1771 (1993) 10 Staff Reporter ldquoUS Army Tests Lidar to Detect Biological Toxinsrdquo Photonic Spectra pg 50 December 1998 11 Christesen SD Raman cross sections of chemical agents and simulants App Spectrosc 42 318-321 (1988) 12 Weaver MJ Farquharson S Tadayyoni MA ldquoSurface-enhancement factors for Raman scattering at silver electrodesrdquo J Chem Phys 82 4867-4874 (1985) 13 Alak AM and Vo-Dinh T ldquoSERS of Organophosphorous Chemical Agentsrdquo Anal Chem 59 2149-2153 (1987) 14 Lee Y Farquharson S ldquoRapid chemical agent identification by SERSrdquo SPIE 4378 21-26 (2001) 15 Farquharson S Maksymiuk P Ong K Christesen S ldquoChemical agent identification by surface-enhanced Raman spectroscopyrdquo SPIE 4577 166-173 (2001) 16 Spencer KM Sylvia J Clauson S and Janni J ldquoSurface Enhanced Raman as a Water Monitor for Warfare
Agents in Waterrdquo SPIE 4577 158-165 (2001) 17 Tessier P Christesen S Ong K Clemente E Lenhoff A Kaler E Velev O ldquoOn-line spectroscopic characterization of sodium cyanide with nanostructured gold surface-enhanced Raman spectroscopy substratesrdquo
App Spectrosc 56 1524-1530 (2002) 18 Farquharson S W W Smith S Elliott and J F Sperry Rapid biological agent identification by surface- enhanced Raman spectroscopy SPIE 3855110-116 (1999) 19 Farquharson S WW Smith YH Lee S Elliott and J F Sperry Detection of biological signatures A
comparison of electrolytic and metal-doped sol-gel surface-enhanced Raman media SPIE 4575 62-72 (2002) 20 Guzelian AA Sylvia J Janni J Clauson S and Spencer KM ldquoSERS of whole cell bacteria and trace levels of biological moleculesrdquo SPIE 4577 182-192 (2001) 21 Shende C Inscore F Gift A Maksymiuk P Farquharson S ldquoRapid dipicolinic acid extraction from Bacillus spores detected by surface-enhanced Raman spectroscopyrdquo App Spectrosc 58 accepted 22 Lee Y and Farquharson S ldquoSERS Sample Vials Based on Sol-Gel Process for Trace Pesticide Analysisrdquo SPIE
4206 140-146 (2000) 23 Farquharson S and Lee Y ldquoTrace Drug Analysis by SERSrdquo SPIE 4200-16 (2000) 24 Lee Y Farquharson S and Rainey P M Surface-Enhanced Raman Sensor for Trace Chemical Detection in
Water SPIE 3857 76-84 (1999) 25 Lee Y Farquharson S Kwong H and Shahriari M ldquoSol-Gel Chemical Sensor for Surface-Enhanced Raman
Spectroscopyrdquo SPIE 3537 252-260 (1998) 26 Farquharson S A Gift P Maksymiuk F Inscore and W Smith ldquopH dependence of methyl phosphonic acid
dipicolinic acid and cyanide by surface-enhanced Raman spectroscopyrdquo SPIE 5269-19 (2003) 27 Lide DR Ed Handbook of Chemistry and Physics CRC Press 77th Ed (1996-7) 28 Sosa C RJ Bartlett K KuBulat and WB Person ldquoA theoretical study of harmonic vibrational frequencies and
infrared intensities of XCH2CH2SCH2CH2X and XCH2CH2SH (= H Cl)rdquo J Phys Chem 93 577-588 (1993) 29 Hameka HF and JO Jenson ldquoComputer-generated predictions of the structure and the IR and Raman spectra of
VXrdquo ERDEC-TR-065 May 1993
- Chemagents Appendicespdf
-
- SPIE2001-4575-Bioagent-SERSapdf
-
- Detection of bioagent signatures A comparison of electrolytic and metal-doped sol-gel surface-enhanced Raman media
- Stuart Farquharson Wayne Smith and Yuan Lee
-
- Real-Time Analyzers 87 Church Street East Hartford CT 06108
-
- ABSTRACT
- 4 CONCLUSIONS
- 5 ACKNOWLEDGEMENTS
- 6 REFERENCES
-
SPIE-2003-5269 22
REFERENCES 1 Jensen JL NC Wong W R Loerop SPIE 4775-03 (2001) 2 JSAWM Requirements at wwwsbccomapgeaarmymilRDAecbcrtPRODSERJSAWMjsawmhtml 3 Erickson B ldquoThe Chemical Weapons Convention Redefines Analytical Challengerdquo Anal Chem 70 397A (1998) 4 Johnston RL Hoefler CM Fargo JC and Moberley B ldquoThe Defense Nuclear Agencyrsquos
ChemicalBiochemical Weapons Agreements Verification Technology Research Development Test and Evaluation Program and its Requirements for On-Site Analysisrdquo AT-ONSITE 5-8 (1994)
5 Black RM Clarke RJ Read RW and Reid MT ldquoApplication of GC-MS and GC-tandem MS to the analysis of chemical warfare samples found to contain residues of the nerve agent sarin sulphur mustard and their degradation productsrdquo J Chrom 662 301-321 (1994)
6 USA Army Center for Health Promotion and Preventive Medicine (USACHPPM) Technical Guideline 230A 7 Hoffland LD Piffath RJ Bouck JB rdquoSpectral signatures of chemical agents and simulantsrdquo Optical
Engineering 24 982-984 (1985) 8 Braue EHJ Pannella MGrdquoCIRCLE CELL FT-IR Analysis of Chemical Warfare Agents in Aqueous Solutionsrdquo
App Spectrosc 44 1513-1520 (1990) 9 Tseng C-H Mann CK and Vickers TJ ldquoDetermination of Organics on Metal Surfaces by Raman
Spectroscopyrdquo App Spectrosc 47 1767-1771 (1993) 10 Staff Reporter ldquoUS Army Tests Lidar to Detect Biological Toxinsrdquo Photonic Spectra pg 50 December 1998 11 Christesen SD Raman cross sections of chemical agents and simulants App Spectrosc 42 318-321 (1988) 12 Weaver MJ Farquharson S Tadayyoni MA ldquoSurface-enhancement factors for Raman scattering at silver electrodesrdquo J Chem Phys 82 4867-4874 (1985) 13 Alak AM and Vo-Dinh T ldquoSERS of Organophosphorous Chemical Agentsrdquo Anal Chem 59 2149-2153 (1987) 14 Lee Y Farquharson S ldquoRapid chemical agent identification by SERSrdquo SPIE 4378 21-26 (2001) 15 Farquharson S Maksymiuk P Ong K Christesen S ldquoChemical agent identification by surface-enhanced Raman spectroscopyrdquo SPIE 4577 166-173 (2001) 16 Spencer KM Sylvia J Clauson S and Janni J ldquoSurface Enhanced Raman as a Water Monitor for Warfare
Agents in Waterrdquo SPIE 4577 158-165 (2001) 17 Tessier P Christesen S Ong K Clemente E Lenhoff A Kaler E Velev O ldquoOn-line spectroscopic characterization of sodium cyanide with nanostructured gold surface-enhanced Raman spectroscopy substratesrdquo
App Spectrosc 56 1524-1530 (2002) 18 Farquharson S W W Smith S Elliott and J F Sperry Rapid biological agent identification by surface- enhanced Raman spectroscopy SPIE 3855110-116 (1999) 19 Farquharson S WW Smith YH Lee S Elliott and J F Sperry Detection of biological signatures A
comparison of electrolytic and metal-doped sol-gel surface-enhanced Raman media SPIE 4575 62-72 (2002) 20 Guzelian AA Sylvia J Janni J Clauson S and Spencer KM ldquoSERS of whole cell bacteria and trace levels of biological moleculesrdquo SPIE 4577 182-192 (2001) 21 Shende C Inscore F Gift A Maksymiuk P Farquharson S ldquoRapid dipicolinic acid extraction from Bacillus spores detected by surface-enhanced Raman spectroscopyrdquo App Spectrosc 58 accepted 22 Lee Y and Farquharson S ldquoSERS Sample Vials Based on Sol-Gel Process for Trace Pesticide Analysisrdquo SPIE
4206 140-146 (2000) 23 Farquharson S and Lee Y ldquoTrace Drug Analysis by SERSrdquo SPIE 4200-16 (2000) 24 Lee Y Farquharson S and Rainey P M Surface-Enhanced Raman Sensor for Trace Chemical Detection in
Water SPIE 3857 76-84 (1999) 25 Lee Y Farquharson S Kwong H and Shahriari M ldquoSol-Gel Chemical Sensor for Surface-Enhanced Raman
Spectroscopyrdquo SPIE 3537 252-260 (1998) 26 Farquharson S A Gift P Maksymiuk F Inscore and W Smith ldquopH dependence of methyl phosphonic acid
dipicolinic acid and cyanide by surface-enhanced Raman spectroscopyrdquo SPIE 5269-19 (2003) 27 Lide DR Ed Handbook of Chemistry and Physics CRC Press 77th Ed (1996-7) 28 Sosa C RJ Bartlett K KuBulat and WB Person ldquoA theoretical study of harmonic vibrational frequencies and
infrared intensities of XCH2CH2SCH2CH2X and XCH2CH2SH (= H Cl)rdquo J Phys Chem 93 577-588 (1993) 29 Hameka HF and JO Jenson ldquoComputer-generated predictions of the structure and the IR and Raman spectra of
VXrdquo ERDEC-TR-065 May 1993
- Chemagents Appendicespdf
-
- SPIE2001-4575-Bioagent-SERSapdf
-
- Detection of bioagent signatures A comparison of electrolytic and metal-doped sol-gel surface-enhanced Raman media
- Stuart Farquharson Wayne Smith and Yuan Lee
-
- Real-Time Analyzers 87 Church Street East Hartford CT 06108
-
- ABSTRACT
- 4 CONCLUSIONS
- 5 ACKNOWLEDGEMENTS
- 6 REFERENCES
-