spectroscopic ultra-trace detection of nitroaromatic gas vapor on rationally designed...

Published: February 20, 2011

r 2011 American Chemical Society 2243 dx.doi.org/10.1021/ac103123r |Anal. Chem. 2011, 83, 2243–2249

ARTICLE

pubs.acs.org/ac

Spectroscopic Ultra-Trace Detection of Nitroaromatic Gas Vapor onRationally Designed Two-Dimensional Nanoparticle Cluster ArraysJing Wang, Linglu Yang, Svetlana Boriskina, Bo Yan, and Bj€orn M. Reinhard*

Department of Chemistry and The Photonics Center, Boston University, Boston, Massachusetts 02215, United States

Nitroaromatic explosives are components of several widelyused high explosives, and there is an acute need for reliable andfieldable sensors that facilitate a rapid detection of their sub-limated vapor at low concentrations in the gas phase.1 Caninesare currently the most sensitive and versatile method for fielddetection of explosives. The animals tire, however, easily andcannot cover substantial areas,2,3 creating a need for technologiesthat enable the detection of explosive vapors with comparable orbetter sensitivities but at lower cost and with longer dutycycles.4,5 Several practical considerations make photonic sensorsappealing for explosive trace detection; they require no samplepreparation and many peripherals (lasers, spectrometers, etc.) toimplement portable devices are commercially available at rela-tively low cost. One example of a photonic sensor for nitroaro-matic explosives that has already been deployed uses luminescentpolymers6-9 as sensing element. Electron deficient unsaturatednitroaromatic explosive such as 2,4,6-trinitrotoluene (TNT) andits fabrication byproduct 2,4-dinitrotoluene (DNT) quench thefluorescence of the polymers and are thus detected. Since theselectivity of the sensors arises from the chemical nature of thepolymers, luminescent polymers are necessarily limited in therange of potential threats that can be detected. Spectroscopicsensors that utilize a vibrational spectroscopy such as Ramancould enable the identification of a much wider range ofexplosives as well as other chemical and biological compoundsand hold the promise for a unified sensor platform capable ofdetecting diverse threats simultaneously on one chip. Spectro-scopic sensors have, however, not yet achieved sensitivities forexplosive vapors comparable to that of the luminescent polymers.

Raman cross-section can be greatly enhanced in the vicinityof nanostructured metal surfaces, which effectively localize theincident electromagnetic radiation in junctions and crevices.10-15

These nanostructures act not only as antennas for the incidentbut also for the re-radiated light and can thus amplify the Ramansignal by many orders of magnitude. Random nanostructuredgold surfaces, generated through electrochemical roughening orprecipitation of gold colloids, typically achieve ensemble-aver-aged enhancement factors of up to 106. Because of this strongsignal amplification and molecular specificity, surface enhancedRaman scattering (SERS)16 offers new opportunities in ultra-trace analytics,17-20 and is of particular interest for detecting lowconcentration of explosive vapors.21,22 While random nanostruc-tured metal surfaces can provide high ensemble averaged SERSenhancement factors, substrates for critical applications, whichdemand high sensing fidelities, have to fulfill additional require-ments. Ideal SERS substrates provide low on-chip and chip-to-chip signal fluctuations, to warrant a high signal reproducibilityand reliability. This requirement is not fulfilled by random noblemetal nanostructures, which are prone to significant fluctuationsin the SERS signal enhancement. Furthermore, ultra-trace sen-sing applications, in particular, would benefit from the ability tocontrol the near-field response of the SERS substrate in a rationalfashion. The latter could facilitate the optimization of the E-fieldenhancement at both the pump and the Raman wavelength ofinterest, which would contribute to maximize the SERS signalintensity for the analyte of interest.23

The need for reliable and tunable SERS substrates has moti-vated the development of engineered SERS substrates, whichprovide a reproducible signal amplification according to rationaldesign criteria.17,24-27 Nanoparticle cluster arrays (NCAs) are

Received: November 28, 2010Accepted: January 30, 2011

ABSTRACT: Nanoparticle cluster arrays (NCAs) are engi-neered two-dimensional plasmonic arrays that provide highsignal enhancements for critical sensing applications usingsurface enhanced Raman spectroscopy (SERS). In this workwe demonstrate that rationally designed NCAs are capable ofdetecting ultra-traces of 2,4-dinitrotoluene (DNT) vapor.NCAs functionalized with a thin film of an aqueous NaOHsolution facilitated the detection of DNT vapor at a concentra-tion of at least 10 ppt, even in the presence of an excess ofpotential interferents, including Diesel fuel, fertilizers, and pesticides. Both in the presence and in the absence of this complexbackground the SERS signal intensity of the NO2 stretching mode showed a continuous, concentration dependent response overthe entire monitored concentration range (10 ppt-100 ppb). The small size, superb sensitivity, and selectivity, as well as the fastresponse time of <5 min, make NCAs a valuable photonic sensor platform for ultra-trace nitroaromatic gas vapor detection withpotential applications in landmine removal and homeland security.

2244 dx.doi.org/10.1021/ac103123r |Anal. Chem. 2011, 83, 2243–2249

Analytical Chemistry ARTICLE

one class of engineered SERS substrates, which utilize the E-fieldfocusing effect in nanoparticle clusters of well-defined size, self-assembled at predefined locations, to generate high and repro-ducible SERS signal enhancements.28,29 The E-field enhancementgenerated through plasmon coupling30-33 between the particleswithin the individual clusters is further increased in NCAsthrough an additional level of field enhancement arising fromintercluster interactions. Because of their larger size, clusterscouple over longer length scales than the individual particles, andthe interplay of electromagnetic inter- and intracluster interac-tions creates a synergistic multiscale cascade field enhancementwithin the NCAs.28,29

NCAs have proven to be a reliable SERS platform in practice.They have provided strong SERS signals from small moleculesand entire bacterial cells drop-coated on the NCAs. In thismanuscript we expand the application space of NCAs anddemonstrate that chemically activated NCAs are capable ofdetecting ultra-traces of nitroaromatic vapors in the presenceor absence of common interferents.

’EXPERIMENTAL SECTION

Template Guided Assembly of NCAs. NCAs were as-sembled using a template guided self-assembly approach.28,29

In a first step standard e-beam lithography using a Zeiss Suprascanning electron microscope (SEM) equipped with an e-beamblanker was used to generate a pattern of binding sites in a 180nm thick poly(methyl methacrylate) resist (PMMA 950 A3)film, which was coated by a 10 nm thin gold layer. After gold

etching using gold etchant for 10 s and then development inmethyl isobutyl ketone/isopropanol (MIBK/IPA = 1/3) solventmix for 70 s, the exposed binding sites on the quarz substratewere charged positively by incubation with a 2 mg/mL aqueoussolution of polylysine (Mw = 15-30 kDa) for 1 h. The positivecharge on the binding sites facilitated an efficient binding ofnegatively charged gold nanoparticles onto the washed substratesin a subsequent incubationwith a gold nanoparticle solution (∼1�1010 particles/mL) in 30 mM NaCl, 10 mM phosphate buffer(pH 8.0) in a water saturated atmosphere overnight. The goldnanoparticles used in the template guided assembly process werefunctionalized with a self-assembledmonolayer of carboxylic acidterminated polyethylene glycols (PEGs): HSC11H22(OC2H4)6-OCH2COOH. After removal of the PMMA template throughincubation in 1-methyl-2-pyrrolidone for 8 min, the assemblednanoparticle clusters remained bound to the substrate andformed the NCAs. To remove the PEGs on the nanoparticlesurface, the assembled NCAs were oxygen plasma cleaned rightbefore use. Periodic NCAs are described by a set of threeindependent parameters: particle diameter (d), binding sitediameter (D), and intercluster separation (Λ) (Figure 1a).SERS Measurements. All SERS measurements were per-

formed on an upright Olympus BX51W1 microscope equippedwith a 300 mm focal length imaging spectrometer (AndorShamrock 303i) and a back-illuminated CCD (Andor Idus)optimized for the near-Infrared spectral range (peak quantumefficiency >90% at 785 nm. A 1200 lines/mm grating with ablazing wavelength of 750 nm was used. The excitation laser wasa 785 nm diode laser. After passing through a 785 nm laser line

Figure 1. (a) Schematics of a nanoparticle cluster array (NCA). The array morphology is defined by the cluster binding size (D), intercluster separation(Λ), and nanoparticle diameter (d). (b) Simulated frequency spectra of the electric field intensity enhancement (over the free space value) averaged overisolated nanoparticle cluster geometries shown in the inset (blue) and clusters arranged into random 4 � 4 NCA configurations (red). The intensityvalues are averaged over various angles of the in-plane light polarization (blue and red) and over different clusters within NCAs (red). Intensityenhancement on a single 80 nm Au nanoparticle is shown for comparison (navy). (c) Near-field intensity distribution in a typical NCA configuration atpump (785 nm) and (d) Raman-shifted (876.5 nm) wavelength.



filter (Semrock, LL01-785-25), the laser light was injected intothe objective using a dichroic (Semrock, LPD-785RU) andfocused into the sample plane by a 40� air objective(numerical aperture (NA) = 0.65). The laser spot size on thesample had a radius of 40 μm. The laser power in the sampleplane was 27.7 mW corresponding to an irradiance of 551W/cm2.Light scattered off the sample was collected by same objectiveand filtered by the dichroic and an additional 803 nm long-passfilter (Semrock, LP02-785-RS). The active area for recordingSERS spectra was limited by a slit in the entrance port of thespectrometer to 4.5 μm � 31 μm. Ten individual acquisitionswith a 2 s integration time were accumulated for each spectrum.Before each measurement, the SERS spectrum of the aqueousNaOH solution on a NCA (without DNT vapor) was taken asbackground signal, which was subtracted from all subsequentlyrecorded DNT spectra on this chip. For the quantitative analysisof the SERS signal intensity we evaluated the peak intensity of theNO2 stretching mode at 1336 cm

-1.Computational Electromagnetics. To estimate the Raman

enhancement factors provided by the NCAs, we simulated thenear-field intensity spectra of typical 3-5-particle clusters andnanocluster arrays by using the multiparticle generalized Mietheory (GMT) algorithms. GMT is a classical electromagneticcomputational technique that provides a rigorous semianalyticalsolution to the problems of wave scattering by arbitrary arrays ofspherical particles.34,35 To stay within the framework of theclassical electromagnetic theory, all simulated cluster and NCAgeometries had nearest interparticle separations no less than1 nm.30,36 The intensity spectra were obtained under theexcitation by a plane wave incident normally to the array plane,which corresponds to the experimental conditions. The spectrawere averaged over all possible in-plane polarizations of theincident field. All the simulations we performed for d = 80 nmgold spheres with experimentally obtained gold refractive indexvalues from Johnson and Christy37 immersed in the ambientmedium with the refractive index n = 1.44.

’RESULTS AND DISCUSSION

We devised an experimental strategy for the SERS detection ofDNT vapor that seeks to maximize sensitivity through combina-tion of two synergistic effects: (i) analyte enrichment throughreactive absorption onto the sensor, and (ii) signal amplificationof the absorbed analytes through optimizedNCA substrates. Ourexperimental approach to achieve an enrichment of the analyte inthe vicinity of the NCA SERS substrate utilizes the acid-basebehavior of DNT, which is efficiently deprotonated in basicbuffers to form an anion. The deprotonated nitroaromaticspecies can form a stabilized Meisenheimer38 complex with asodium cation in the solution. Since the anion has a highersolubility than the neutral species, whose equilibrium concentra-tion in water is determined by the Henry’s law constant of KH =9.27 � 10-8 atm 3m

33mol-1,39 the deprotonation reaction

facilitates an increased mass transfer into the basic aqueoussolution. A thin film of this solution applied on a NCA therefore“captures” gas-phase molecules and makes them amenable todetection via SERS. Sylvia et al.21 have demonstrated before thatwetting of random SERS substrates with a mist of 10 mMaqueous solution of NaOH results in a strong increase in thedetection sensitivity. Sylvia et al. were able to detect DNT vaporat concentrations as low as 5 ppb using electrochemicallyroughened gold surfaces. Our goal was to further boost the

detection sensitivity of the SERS sniffer by replacing the randomSERS substrates through rationally designed NCAs.

A systematic characterization of the influence of the nanopar-ticle diameter on the SERS signal enhancement provided byNCAs has indicated that d = 80 nm diameter nanoparticlesprovide higher ensemble averaged SERS signal enhancementsthan 40 or 60 nm diameter building blocks.29 Consequently,we chose d = 80 nm particles as building blocks for the NCAsused in this work. The average SERS enhancement factorsprovided by NCAs are, however, not determined by the size ofthe nanoparticle building blocks alone. Because of the distancedependent electromagnetic interactions between nanoparticleswithin the clusters and between entire clusters, the array mor-phology as determined by the intercluster separation (Λ) and thecluster size (D) need to be considered, as well (Figure 1a).

We simulated the average intensity enhancement as function ofwavelength for a (Λ = 100 nm; D = 200 nm; d = 80 nm) NCA,which can be reliably fabricated with high purity. Figure 1bcompares the average E-field intensity enhancement for the(Λ = 100 nm; D = 200 nm; d = 80 nm) NCA with the averagesignal enhancement of isolated clusters (shown in the inset) andan individual 80 nm nanoparticle. The intensity spectra forisolated nanoparticle clusters were first averaged over all possiblepolarizations of the field incident on each cluster, and then over sixpossible cluster configurations shown in the inset. The averageintensity spectra for the NCAs were obtained by consideringseven random NCA configurations of 4 � 4 clusters and byaveraging over 112 (16 � 7) individual clusters and over variouspolarizations of the electric field. The gain in E-field enhancementthat results from an incorporation of individual clusters into anNCAs is well illustrated by Figure 1b. The NCAs provide higherE-field enhancement than either the individual nanoparticles orthe isolated nanoparticle clusters because of synergistic electro-magnetic interactions on multiple length scales in the array.

The signal intensity of a specific molecular vibration withfrequency ω in a SERS spectrum depends on the product of thelocal field enhancement factors [|Eloc(ω)|/|E0 (ω)|]2 at thepump and emission wavelengths.11,40 Eloc refers thereby to thelocal field experienced by the scatters, and E0 is the field of theincident light. The E-field intensity of the (Λ = 100 nm;D = 200nm) NCA in Figure 1b peaks at approximately 775 nm andtherefore offers high E-field intensities at both the 785 nm pump(Iav = 1425) and the 1350 cm

-1 Stokes Raman wavelength (Iav =876). The predicted E-field intensities translate into an averageRaman enhancement of 1.25� 106 for theNO2 stretchingmode,predestining this NCA morphology for applications as a spectro-scopic DNT sensor. The spatial E-field distribution in the NCAplane at both the pump and emission wavelength is plotted inFigures 1c,d; the E-field is localized in crevices and junctionsbetween the particles, which create “hot-spots” that sustain highE-field intensities at both wavelengths. We fabricated the (Λ =100 nm; D = 200 nm; d = 80 nm) NCAs using the templateguided self-assembly approach described in the ExperimentalSection. Scanning electron microscope (SEM) images of aresulting NCA are shown in Figures 2a-c. At the highestmagnification the configurations of the individual clusters arewell recognizable and the similarity to the simulated clusterconfigurations included in Figure 1 becomes apparent.

The experimental procedure for using the fabricated NCAs forvapor measurements is illustrated in Figure 2d. First a thin film of10 mM NaOH solution was applied to the NCAs using anebulizer; after this activation step the NCAs were transferred



into a glass chamber containing DNT vapor with a partialpressure of pDNT. The DNT atmosphere was generated byincubating an aqueous solution of DNT of known concentration(cDNT) in a closed glass chamber at 25 �C overnight. Using theHenry’s law constant (KH) of DNT the resulting partial pressureof the analyte in the gas-phase was calculated as pDNT =KH 3 cDNT. Through variation of cDNT we could systematicallychange the concentration of DNT in the gas phase. Since theglass chamber needed to be briefly opened for inserting theplasmonic sensors, the actual gas phase concentrations duringthe measurements were even lower than the calculated

equilibrium concentrations. The latter serve as conservativeestimates of the DNT vapor concentration in the followingdiscussion.

The SERS spectra recorded with NaOH activated NCAsexhibit the characteristic NO2 stretching mode at 1336 cm-1

after exposure to DNT vapor (Figure 3a). The signal intensity ofthis band is expected to depend both on the exposure time of theNCA to the analyte and its vapor concentration. First, wetherefore characterized the time dependence of the SERS signalintensity at a constant DNT concentration. In Figure 3b we plotthe peak intensity of the NO2 stretching mode obtained withNCAs incubated in the sublimated vapor of solid DNT fordifferent periods of time. While in the first 1-5 min the SERSsignal intensity shows a strong, continuous increase, at longerincubation times the SERS signal intensity does not continue toincrease. Instead, we find that after 8 min of incubation the signalis already significantly reduced when compared with the mea-surement taken after 5 min, and the signal continues to decreasewith increasing incubation time until the signal has completelydisappeared after 15 min.

To test whether the observed signal decrease resulted from aninsufficient stability of theNCAs when incubated with theNaOHmist, we inspected the NCAs before and after exposure in theSEM. These experiments revealed that the particles remainedbound to the substrate during the sensing experiments, arguingagainst a systematic removal of the nanoparticles through theNaOH solution as reason for the signal deterioration. Instead, weattribute the decrease in signal intensity at longer incubationtimes to the drying of the liquid film on the plasmonic array.Indeed, we observed that the liquid film sprayed onto the NCAslowly evaporated as function of time. The evaporation of theliquid on the NCA leads to a decrease in the ambient dielectricconstant when the aqueous solution is replaced by air.41 Thiseffect, together with the crystallization of NaOH on the NCAchip upon evaporation of the solvent, can account for a decreasein signal intensity as a function of time. Other effects that canpotentially contribute to the observed decrease in SERS signalintensity are interadsorbate interactions, which have been ob-served to result in a decrease of SERS signal intensity at surfacecoverages as low as a few tenths of a monolayer.42

Since we obtained the maximumDNT signal intensity with anincubation time of 5 min, all measurements reported in the

Figure 2. SEM images of a nanoparticle cluster array (NCA) withparticle diameter d = 80 nm, cluster binding size D = 200 nm, andintercluster separationΛ = 100 nm. Size bars are (a) 400 nm, (b) 1 μm,and (c) 2 μm. (d) Sample preparation: the NCAs are wetted with anaqueous NaOH mist and exposed to DNT vapor at a defined concen-tration, which is controlled by the DNT concentration in the aqueoussolution contained in the glass chamber at 25 �C.

Figure 3. (a) SERS spectra of DNT vapor at a concentration of 100 ppb and potential interferents (diesel, water, pesticide, fertilizer vapor at 25 �C). Allspectra were recorded onNCAs containing an aqueousNaOH film. The characteristic NO2 stretchingmode, which peaks at 1336 cm-1, is marked in theDNT spectrum. (b) Raman intensity as function of incubation time for NaOH solution activated NCAs with DNT vapor (200 ppb).



following were performed with this acquisition time. We pointout that although the NCAs reach peak performance after 5 min,this does not exclude that, depending on the concentration, DNTvapor can be detected much earlier. At a concentration of 100ppb DNT, for example, the analyte provides sufficient SERSsignal intensity for detection at the first recorded time point afteran incubation time of 1 min. Overall, we conclude that the sensorresponse is instantaneous, but the signal intensity for a givenconcentration is time dependent. In this regard the NCAs behavesimilar to other state-of-the art trace vapor explosive sensors withdiffusion controlled analyte adsorption.7,43

In the next step, we set out to explore the sensitivity of NaOHactivated NCAs for DNT vapor. To that end we exposed NCAsto DNT vapor concentrations between 100 ppb and 10 ppt andrecorded the corresponding SERS spectra. Representative spec-tra for the investigated DNT concentrations are shown inFigure 4a; in Figure 4b we plot only the spectra of the threelowest investigated DNT concentrations together with a back-ground spectrum with an enlarged intensity scale. The character-istic 1336 cm-1 NO2 stretching mode is unambiguouslyidentified in all spectra, even at concentrations as low as10 ppt. This impressive detection threshold considerably exceedsthe sensitivity of canines for gas-phase nitroaromatic com-pounds2 and is comparable to the best sensitivities achieved byboth photonic1,8,22 and non-photonic43-45 based gas sensors.

We cannot assume that under our experimental conditions theDNT distribution between vapor phase and NaOH solutionreaches an equilibrium. However, a double logarithmic plot ofthe recorded 1336 cm-1 SERS signal as function of the DNTconcentration (blue squares in Figure 5) shows a continuouslinear increase of the SERS intensity as function of concentration.This power-law dependence of the SERS intensity on the DNTvapor concentration confirms a systematic response of the sensorand enables a quantification of the DNT vapor concentrationthrough the measured SERS signal intensity. The error bars inFigure 5 represent the fluctuations in the SERS signal obtainedunder otherwise identical conditions on 5 different NCAs. Thesystematic concentration dependent response observed with theNCAs confirms that these engineered substrates are reliableSERS sensors that provide a reproducible signal amplification.

Most practical applications do not require the sensing of thepure explosive vapor but, instead, necessitate the detection of the

analyte in a complex atmosphere that can contain backgroundmolecules in a much higher concentration than the analyte.Atmospheric background can interfere with the detection ofultra-traces of explosive vapors in multiple ways. First, inter-ferents with spectral features that lie in the same spectral range ascharacteristic spectral features of the analyte could “cover” theexplosive vapor signal. For nitro-aromatic compound this is lessof a concern, since naturally occurring nitrates, which could coverthe characteristic NO2 stretching mode, are rare. A secondpotential threat to the sensitivity of NCAs in gas vapor sensingis a covering of the active sensor surface through abundantinterferents that coadsorb on the NCAs. The resulting blockingof the sensor surface would necessarily result in a decreasedsensitivity and would significantly impact overall sensor perfor-mance. To evaluate the influence of potential interferents we setout to quantify the sensitivity of the NCAs for DNT in thepresence of Diesel fuel, fertilizer, and pesticide vapor. We choseDiesel fuel since it is a known interferent for explosive vaporsensing, for instance, in landmine detection applications.21,46 Wealso included commercial fertilizers (Easy Gardener #6528) andpesticides (SC Johnson DRK 94892) as ubiquitous agricultural

Figure 4. (a) SERS spectra of DNT measured on NaOH activated NCAs for concentrations between 100 ppb and 10 ppt. (b) Magnified view of the250, 50, and 10 ppt spectra; a background spectrum is included, as well. The characteristic NO2 stretching mode at 1336 cm-1 is detected forconcentrations down to 10 ppt.

Figure 5. Peak intensities of the 1336 cm-1 band without (blue) andwith (red) interferents (diesel, fertilizer, and pesticide) as function of theDNT concentration.



compounds. The SERS spectra of the pure interferent vaporsrecorded on NaOH activated NCAs at 25 �C are included inFigure 3a. None of the considered compounds contain spectralfeatures that overlapwith the characteristic 1336 cm-1NO2 stretch-ing mode of DNT. We also note in passing that water vapor,which is a significant concern in other spectroscopic analytics(e.g., infrared spectroscopy),47,48 has very small Raman crosssections and therefore does not interfere with explosive vaporsensing using NCAs.

To validate the influence of the potential interferentsthrough co-adsorption on the NCA surface, we performed aseparate series of experiments in which we exposed the mistedNCAs to different concentrations of DNT vapor in the pre-sence of the background vapors. Diesel fuel, fertilizer, andpesticide were placed into the glass chamber containing theaqueous solution of DNT in separate open Eppendorf tubes.The atmosphere in the sample container was again equilibratedin a water bath at 25 �C overnight before the NaOH activatedNCA was inserted. The NCA was then incubated in thisatmosphere for 5 min, after which a SERS spectrum of thesample was acquired. We evaluated the 1336 cm-1 peak signalintensity over the same concentration range evaluated beforefor the quantification of the sensor response in the absence ofbackground. The data obtained in the presence of Diesel,fertilizer, and pesticide background are included as filled, redcircles in Figure 5. The Raman intensity increases again linearlyas function of DNT concentration in a log-log plot, but theresponse curve is shifted to somewhat lower intensities thanobserved before without background. The signal intensitiesremain, however, sufficiently high to enable the detection of10 ppt of DNT (Figure 6). We rationalize the excellentsensitivity obtained in the presence of the background vaporsthrough the selective enrichment of DNT on the sensor surfacethrough the aqueous NaOH film applied on the NCAs.Furthermore, the aqueous film protects the NCAs from hydro-phobic components with high vapor pressures such as theDiesel fuel. The outstanding sensitivity demonstrated by theNCAS in the presence of an excess of background vapor makesthese engineered SERS substrates a viable sensor platform forthe selective detection of ultra-traces of nitroaromatic com-pounds in complex atmospheres.

’CONCLUSIONS

NCAs are engineered SERS substrates that generate highE-field enhancements through a synergistic interplay of electro-magnetic interactions on multiple (inter- and intracluster) lengthscales. In this manuscript we demonstrate that rationally designedNCAs enable the detection of extremely low concentrations ofnitroaromatic vapors. Ultra-trace sensitivity for DNTwas achievedby combining signal amplification through optimized NCAs withsample enrichment on the NCAs. DNT enrichment was accom-plished through application of a thin film of aqueous NaOH ontothe NCAs. The analyte efficiently absorbs into the NaOH solutionwhere it can be detected through SERS. This experimentalapproach facilitated DNT vapor detection at a concentration of10 ppt within a fewminutes. HigherDNT concentrations could bedetected even faster. SERS is a vibrational spectroscopy and assuch provides molecular specific information. For the nitroaro-matic analyte investigated in this work, the characteristic NO2

stretching mode provided a marker band, which enabled itsselective detection in the presence of an excess of Diesel fuel,commercial fertilizer, and pesticide vapor. With these backgroundvapors the total SERS signal intensities measured for DNT weresomewhat lower than without, but the NCAs still achieved theimpressive DNT detection threshold of 10 ppt. The NCA sensorsprovided a continuous, concentration dependent response over alarge concentration range both in the presence and in the absenceof background vapors. This behavior will facilitate quantitativeDNT vapor concentration measurements using NCAs in thefuture. One limitation of the current assay is the drying of theNaOH mist on the NCA chip, which prevents a continuousexplosive vapor monitoring. We anticipate that this problem canbe overcome, for instance, using a sensor design that keeps theNCA chip in a water vapor saturated atmosphere.

DNT, whose vapor pressure is approximately 20 times higherthan that of TNT, is contained in military grade TNT withconcentration of up to 1%.49 DNT is therefore a commonmarkercompound used in the detection of the high explosive TNT. TheNCAs’ capability to detect ultra-traces of DNT vapor in thepresence of common interferents for gas sensors makes theseengineered plasmonic materials a promising photonic sensorplatform for landmine or concealed explosive detection.

’AUTHOR INFORMATION

Corresponding Author*E-mail: [email protected].

’ACKNOWLEDGMENT

We acknowledge financial support from the National ScienceFoundation through Grants CBET-0853798 and CBET-0953121,and the Army Research Laboratory (cooperative agreementDAAD 19-00-2-0005).

’REFERENCES

(1) Toal, S. J.; Trogler, W. C. J. Mater. Chem. 2006, 16, 2871–2883.(2) Oxley, J. C.; Waggoner, L. P. Detection of Explosives by Dogs;

Elsevier: Amsterdam, The Netherlands, 2009.(3) Czarnik, A. W. Nature 1998, 394, 417–418.(4) Moore, D. S. Rev. Sci. Instrum. 2004, 75, 2499–2512.(5) Moore, D. S. Sensing and Imaging 2007, 8, 9–38.(6) Germain, M. E.; Knapp, M. J. Chem. Soc. Rev. 2009,

38, 2543–2555.

Figure 6. SERS spectrum of 10 ppt DNT vapor in the presence ofDiesel fuel, pesticide, and fertilizer vapor background at 25 �C. The NO2

stretching mode is marked by an arrow.



(7) Yang, J.-S.; Swager, T. M. J. Am. Chem. Soc. 1998,120, 5321–5322.(8) Rose, A.; Zhu, Z.; Madigan, C. F.; Swager, T. M.; Bulovic, V.

Nature 2005, 434, 876–879.(9) Thomas, S. W.; Amara, J. P.; Bjork, R. E.; Swager, T. M. Chem.

Commun. 2005, 4572–4574.(10) Haynes, C. L.; McFarland, A. D.; Van Duyne, R. P. Anal. Chem.

2005, 77, 338a–346a.(11) Moskovits, M. Rev. Mod. Phys. 1985, 57, 783–826.(12) Stockman, M. I.; Shalaev, V. M.; Moskovits, M.; Botet, R.;

George, T. F. Phys. Rev. B 1992, 46, 2821–2830.(13) Shalaev, V. M.; Botet, R.; Tsai, D. P.; Kovacs, J.; Moskovits, M.

Phys. A 1994, 207, 197–207.(14) Moskovits, M. J. Raman Spectrosc. 2005, 36, 485–496.(15) Jackson, J. B.; Halas, N. J. Proc. Natl. Acad. Sci. U.S.A. 2004,

101, 17930–17935.(16) van Duyne, R. P.; Jeanmaire, D. L. J. Electroanal. Chem. 1977,

84, 1–20.(17) Haynes, C. L.; Yonzon, C. R.; Zhang, X. Y.; Van Duyne, R. P.

J. Raman Spectrosc. 2005, 36, 471–484.(18) Anker, J. N.; Hall, W. P.; Lyandres, O.; Shah, N. C.; Zhao, J.;

Van Duyne, R. P. Nat. Mater. 2008, 7, 442–453.(19) Baker, G. A.; Moore, D. S. Anal. Bioanal. Chem. 2005,

382, 1751–1770.(20) Ko, H.; Singumaneni, S.; Tsukruk, V. Small 2008,

4, 1576–1599.(21) Sylvia, J. M.; Janni, J. A.; Spencer, K. M. Anal. Chem. 2000,

72, 5834–5840.(22) Wackerbarth, H.; Salb, C.; Gundrum, L.; Niederkruger, M.;

Christou, K.; Beushausen, V.; Viol, W. Appl. Opt. 2010, 49, 4367–4371.(23) McFarland, A. D.; Young, M. A.; Dieringer, J. A.; van Duyne,

R. P. J. Phys. Chem. B 2005, 109, 11279–11285.(24) Haynes, C. L.; Van Duyne, R. P. J. Phys. Chem. B 2003,

107, 7426–7433.(25) Yu, Q. M.; Guan, P.; Qin, D.; Golden, G.; Wallace, P. M. Nano

Lett. 2008, 8, 1923–1928.(26) Gunnarsson, L.; Bjerneld, E. J.; Xu, H.; Petronis, S.; Kasemo, B.;

Kall, M. Appl. Phys. Lett. 2001, 78, 802–804.(27) Wells, S. M.; Retterer, S. D.; Oran, J. M.; Sepaniak, M. J. ACS

Nano 2009, 3, 3845–3853.(28) Yan, B.; Thubagere, A.; Premasiri, R.; Ziegler, L.; Dal Negro, L.;

Reinhard, B. M. ACS Nano 2009, 3, 1190–1202.(29) Yang, L.; Yan, B.; Premasiri, R.W.; Ziegler, L. D.; Dal Negro, L.;

Reinhard, B. M. Adv. Funct. Mater. 2010, 20, 2619–2628.(30) Yang, L.; Wang, H.; Reinhard, B. M. J. Phys. Chem. C 2010,

114, 4901–4908.(31) Reinhard, B. M.; Siu, M.; Agarwal, H.; Alivisatos, A. P.;

Liphardt, J. Nano Lett. 2005, 5, 2246–2252.(32) Su, K. H.; Wei, Q. H.; Zhang, X.; Mock, J. J.; Smith, D. R.;

Schultz, S. Nano Lett. 2003, 3, 1087–1090.(33) Nordlander, P.; Oubre, C.; Prodan, E.; Li, K.; Stockman, M. I.

Nano Lett. 2004, 4, 899–903.(34) Xu, Y. L. Appl. Opt. 1995, 34, 4573–4588.(35) Gopinath, A.; Boriskina, S. V.; Ning-Ning, F.; Reinhard, B. M.;

Dal Negro, L. Nano Lett. 2008, 8, 2423–2431.(36) Zuloaga, J.; Prodan, E.; Nordlander, P. Nano Lett. 2009,

9, 887–891.(37) Johnson, P. B.; Christy, R. W. Phys. Rev. B 1972, 6, 4370.(38) Meisenheimer, J. Justus Liebigs Ann. Chem. 1902, 323,

205–246.(39) US Environmental Protection Agency; Fact Sheet, Correcting

the Henry's Law Constant for Soil Temperature; http://www.epa.gov/oswer/riskassessment/airmodel/pdf/factsheet.pdf. Accessed February14, 2011.(40) Le Ru, E. C.; Etchegoin, P. G. Chem. Phys. Lett. 2006,

423, 63–66.(41) Soper, S. A.; Kuwana, T. Appl. Spectrosc. 1989, 43, 1180–1187.

(42) Norrod, K. L.; Sudnik, L. M.; Rousell, D.; Rowlen, K. L. Appl.Spectrosc. 1997, 51, 994–1001.

(43) Engel, Y.; Elnathan, R.; Pevzner, A.; Davidi, G.; Faxer, E.;Patolsky, F. Angew. Chem., Int. Ed. 2010, 49, 6830–6835.

(44) Jaworski, J. W.; Raorane, D.; Huh, J. H. Majumdar Langmuir2008, 24, 4938–4943.

(45) Pinnaduwage, L. A.; Thundat, T.; Hawk, J. E.; Hedden, D. L.;Britt, P. F.; Houser, E. J.; Stepnowski, S.; McGill, R. A.; Bubb, D. Sens.Actuators, B 2004, 99, 223–229.

(46) Yang, X.; Du, X. X.; Shi, J.; Swanson, B. Talanta 2001,54, 439–445.

(47) Bauer, C.; Sharma, A. K.; Willer, U.; Burgmeier, J.; Braunsch-weig, B.; Schade, W.; Blaser, S.; Hvozdara, L.; Muller, A.; Holl, G. Appl.Phys. B: Laser Opt. 2008, 92, 327–333.

(48) Stahl, D. C.; Tilotta, D. C. Environ. Sci. Technol. 2001,35, 3507–3512.

(49) George, V.; Jenkins, T. F.; Leggart, D. C.; Cragin, J. H.; Phelan,J.; Oxley, J.; Pennigton, J. Detection and Remediation of Mines andMinelike Targets IV; SPIE: Bellingham, WA, 1999.

spectroscopic ultra-trace detection of nitroaromatic gas vapor on rationally designed...

Documents