measured infrared optical cross sections for a variety of chemical and biological aerosol simulants

Mv

K

1

CbncbrtmtTdvmhcarsf

Pa

2

4

easured infrared optical cross sections for aariety of chemical and biological aerosol simulants

ristan P. Gurton, David Ligon, and Rachid Dahmani

We conducted a series of spectral extinction measurements on a variety of aerosolized chemical andbiological simulants over the spectral range 3–13 �m using conventional Fourier-transform IR �FTIR�aerosol spectroscopy. Samples consist of both aerosolized particulates and atomized liquids. Materialsconsidered include Bacillus subtilis endospores, lyophilized ovalbumin, polyethylene glycol, dimethicone�SF-96�, and three common background materials: kaolin clay �hydrated aluminum silicate�, Arizonaroad dust �primarily SiO2�, and diesel soot. Aerosol size distributions and mass density were measuredsimultaneously with the FTIR spectra. As a result, all optical parameters presented here are massnormalized, i.e., in square meters per gram. In an effort to establish the utility of using Mie theory topredict such parameters, we conducted a series of calculations. For materials in which the complexindices of refraction are known, e.g., silicone oil �SF-96� and kaolin, measured size distributions wereconvolved with Mie theory and the resultant spectral extinction calculated. Where there was goodagreement between measured and calculated extinction spectra, absorption, total scattering, and back-scatter were also calculated. © 2004 Optical Society of America

OCIS codes: 010.1100, 010.1110, 010.1290.

cptopcwtattiiadArcsmstAiistd

. Introduction

urrently there is a great effort underway withinoth the defense and the academic research commu-ity to develop an optical-based technique that isapable of detecting the presence of harmful airborneiological or chemical agents. To accomplish this,esearchers require a set of fundamental parametershat describe the electromagnetic interaction withicrometer-sized particles �i.e., extinction, absorp-

ion, total scatter, and backscatter coefficients�.his paper is an extension of prior research con-ucted as a basic parameters study designed to pro-ide optical cross sections for a variety of aerosolaterials used to simulate the presence of more-

armful agents.1 For many common naturally oc-urring aerosols, such parameters are readilyvailable.2,3 However, for materials that are de-ived from biological or chemical matter, measured initu optical parameters in the infrared �IR� are woe-ully lacking.

The authors are with the U.S. Army Research Laboratory, 2800owder Mill Road, Adelphi, Maryland 20783-1145. The e-mailddress for K. P. Gurton is [email protected] 15 January 2004, revised manuscript received 14 May

004; accepted 24 May 2004.0003-6935�04�234564-07$15.00�0© 2004 Optical Society of America

564 APPLIED OPTICS � Vol. 43, No. 23 � 10 August 2004

In addition, for those who want to calculate opticalross sections, a similar knowledge of the bulk com-lex indices of refraction and particle size distribu-ions is required. Often Mie theory �appropriatenly for perfect spheres� is applied to media in whicharticle geometry is quite complex. Validity of thesealculations is sometimes questionable, particularlyhen the aerosol particle dimensions are comparable

o the wavelength in question. Similarly, issuesrise with experimental techniques used to measurehe complex indices of refraction for particulate ma-erial.4 The preferred method used to derive indicesn the IR involves the compression of the materialsnto small quasi-transparent pellets. Transmissionnd near-normal reflectance measurements are con-ucted on these pellets over a given spectral range.

Kramers–Kronig relation is then applied to theeflection spectra, and resultant complex indices arealculated. However, large uncertainties arise inpectral regions where anomalous dispersion for theaterial is great �i.e., regions where the absorption is

trong�. It is often these regions of strong absorp-ion that are of greatest interest to the researcher.s a result, we attempt to address some of these

ssues of accuracy for certain calculated parametersn the IR, when the aerosols are not necessarilypherical and whose indices of refraction are in ques-ion. In practice, accurate knowledge of both sizeistribution and complex indices of refraction is usu-

aecb

tiBmgw

c�eacvw

a�am

2

AECHas

Bprapcwsij

pam

tsrp

lspgTdts3msrtgmatahdaamwAnapsc1

ca

lly difficult to acquire. In a report soon to follow wexamine to what degree such unknown quantitiesan be inverted, by use of only spectral extinction orackscatter.5The materials considered here can be broken into

hree distinct groups. The first group is character-zed as typical biological simulants, which includeacillus subtilis var. niger �BG� endospores �a com-on soil-based bacteria with a mass density of 1.45

�cc� and dry powdered ovalbumin �lyophilized egghite�.The second group we considered consisted of two

ommon chemical species, polydimethyl siloxaneSF-96 grade 50, mass density 0.963 g�cc� and poly-thylene glycol �PEG 200, mass density 1.08 g�cc�,nd represented our chemical simulants. These twohemicals were chosen because of their relatively lowapor pressure, which allowed good droplet formationith little or no evaporation.The final group consisted of several background

erosols often present in the atmosphere, kaolin clayhydrated aluminum silicate, mass density 2.68 g�cc�nd Arizona road dust �primarily SiO2 and Al2O3,ass density 2.65 g�cc�.

. Experiment

ll samples presented here were provided by thedgewood Research, Development and Engineeringenter, Aberdeen Proving Ground, Maryland.owever, many of these materials are quite commonnd can be acquired through conventional chemicaluppliers.Aerosol samples derived from solid material �e.g.,

G, kaolin, and ovalbumin� were milled to produce aarticle distribution within an approximate sizeange of 0.7– 10 �m in diameter. We effectivelyerosolized these powdered materials by injectingressurized dry air through a cylindrical nozzle thatontained a spiraling array of fine stainless-steelires. A vortex created within the nozzle effectively

eparated and dispersed the dry particles with min-mal agglomeration. Care was taken to properly ad-ust the air pressure to levels that resulted in good

Fig. 1. Measured size distributio

article separation while remaining low enough tovoid fragmentation of the particles below theirilled diameters.Liquid samples were dispersed with a pharmaceu-

ical nebulizer. These nebulizers are specifically de-igned to produce particle distributions within theespiratory range of 0.8–4.0 �m in diameter whenressurized to 10 psi �517 Torr�.We found that, depending on the viscosity of the

iquid, generated size distributions can be shifted tomaller diameters by slightly increasing the pumpressure. As a result both PEG 200 and SF-96rade 50 were atomized by a pressure of 20 psi �1034orr�, which produced a distribution with a modaliameter 0.82 �m. We measured particle size dis-ributions in situ using a commercial particle sizepectrometer, TSI Aerodynamic Particle Sizer Model321. The TSI device utilizes a time-of-flightethod in which the particle drag velocity is mea-

ured while it is moved within a known laminar flowate. As the particles transverse through the devicehey encounter an optical cavity that houses two trig-ering lasers. Aerodynamic diameters are deter-ined from differences between the particle velocity

nd the laminar flow rate. Resultant size distribu-ions for aerosolized particulates and liquid dropletsre shown in Fig. 1. Because the TSI instrumentas a time-of-flight method to predict an aerodynamiciameter, accurate values for the bulk density of theerosol particle are required. Unfortunately, forerosols that were highly amorphous or that containany different constituents, accurate bulk densitiesere unattainable �e.g., soot and Arizona road dust�.s a result, size distributions for those materials areot presented. However, antidotal evidence-basederosol setting rates and video microscopy images ofarticles captured on slides suggest distributionsimilar to what is seen for kaolin �i.e., a large submi-rometer component with a model diameter betweenand 5 �m�.The primary transmission measurements were

onducted in a 0.8-m3 aerosol chamber that providedn optical path length of 0.61 m. Dispersed aerosols

particulates and liquid droplets.
n for
10 August 2004 � Vol. 43, No. 23 � APPLIED OPTICS 4565

wrihtvkmt

rFsswagawwtpass

wvatwmssg

ttsdttlfctramaop

3

Itwgcmia

tcpaesss

4

ere gently drawn into the chamber by a small-areaecirculating fan. Continuous monitoring of humid-ty levels was conducted with a filtered dew-pointygrometer that was inserted through the walls ofhe chamber. Because dry air was used to flush thearious optical windows, water-vapor levels wereept to a minimum. Relative humidity levels forost runs were typically � 8%, which is representa-

ive of a fairly dry environment.We obtained IR transmission spectra using a high-

esolution �0.02 wave-number� Bomem DA2.02ourier-transform IR �FTIR� spectrometer. For thistudy, the spectrometer was operated in a transmis-ion mode �i.e., spectral attenuation was measuredith the aerosol chamber placed between the sourcend the interferometer�. A broadband IR Nernstlower was collimated with a ZnSe condensing lensssembly and projected through the aerosol chamberith two BaF2 transmission windows that were fittedith dry-air flushes. Transmitted light was coupled

o the interferometer with a gold-surfaced f�4 off-axisarabola. Because spectra derived from aerosolsre devoid of any fine structure often seen in vaporpectra, the FTIR spectrometer was operated at apectral resolution of 4.0 cm�1.Aerosol mass densities �grams per cubic meter�ere periodically measured during the run by con-entional dosimetric techniques.6 We collected theerosol mass samples by drawing known volumes ofhe air–particle mixture onto polycarbonate filters,ith a nominal pore size of 0.20 �m, for predeter-ined periods of time. Results from the dosimetric

ampling were then used to mass normalize the mea-ured spectral extinction �in square meters perram�.Reproducibility among measured extinction spec-

ra recorded over several days was good. We at-ribute this to our ability to generate nearly identicalize distributions between each run during aerosolissemination. Much of the uncertainty inherent inhe measurement arose during the mass normaliza-ion process. Typical time periods necessary to col-ect a measurable amount of aerosol mass rangedrom 20 to as much as 90 s, depending on the aerosoloncentration. These mass measurements werehen matched to a corresponding extinction spectraecorded during approximately the same period. As

result a slight variance arose between variousass-normalized spectra. The standard deviation

mong mass-normalized spectra recorded continu-usly as the aerosol ranged from �8% to �15%, de-ending on the sample considered.

. Results

nterferograms were recorded before, during, and af-er each aerosol dispersion. A Bartlett apodizationas applied to each interferogram before the back-round spectra were removed. Transmission wasonverted to extinction by a Beer’s law relation. Weass normalized the results by dividing the path-

ntegrated extinction �1�m� by the correspondingerosol density �in grams per cubic meter�.


For materials in which complex indices of refrac-ion were available from prior studies7 �Figs. 2–4�, weonducted Mie theory calculations using the appro-riate size distribution shown in Fig. 1. Wheregreement between the measured and the calculatedxtinction was reasonable, the calculated spectral ab-orption, total scatter, and backscatter are also pre-ented �Figs 5–7�. For ovalbumin, PEG 200, dieseloot, and Arizona road dust in which no such indices

Fig. 2. Complex indices of refraction for BG endospores.7

Fig. 3. Complex indices for liquid silicone SF-96 grade 50.7

wi

4

Wctwrlc

mp

irvatwmpattetadvsF�ts

emBvctoptOw

Ft

ere available, only the measured spectral extinctions presented �Fig. 8�.

. Comments

hen comparing measured extinction with Mie-alculated values �Figs. 5–7�, one should keep in mindhat all the complex indices used in the calculationsere gathered from prior studies and may not accu-

ately represent the materials used here. Neverthe-ess, we thought it informative to show suchomparisons because it is often the case that one

Fig. 4. Complex indices of refraction for kaolin clay.7

ig. 5. Measured �red curve� and calculated �thick blue curve� specheory predicted total scatter �thin blue curve�, absorption �orang

ust rely on previously published constants to com-ute various optical parameters.For the three materials presented with refractive

ndices �i.e., BG, kaolin, and SF-96 grade 50�, theefractive index for SF-96 grade 50 is least likely toary from one measurement to the next because it ishighly standardized chemical where production is

ightly monitored. In addition, because it is a liquid,e know with certainty that the nebulized particlesust be spherical and thus most appropriate for ap-

lication of Mie theory. As one can see in Fig. 6,greement between measured and calculated spec-ral extinction is quite reasonable. Agreement forhe location of the peaks and valleys for the spectralxtinction is quite good. However, agreement be-ween the relative magnitude between the two is nots good and is attributed to uncertainties in the sizeistribution �Fig. 1�. In general, the underpredictedalues for the Mie extinction is synonymous with alightly broader size distribution than can be seen inig. 1. Nevertheless, where agreement is acceptable

e.g., 3–8.5 and 10–12 �m�, Mie-computed absorp-ion, scatter, and backscatter should be fairly repre-entative of the true values.As for the BG samples, little is known about the

xact form and conditions in which the indices wereeasured by Querry.7 What is certain is that theG samples provided by the Edgewood Research, De-elopment and Engineering Center were grown,leaned, and milled from a completely different batchhan used by Querry nearly a decade prior. In lieuf this and given the fact that growth media used toroduce bacterial endospores can be quite variable,he agreement that can be seen in Fig. 5 is good.ne notable region of divergence occurs in the long-ave IR between 9 and 10 �m. Although difficult to

extinction for aerosolized BG endospores. Also shown are the Mieve�, and backscatter �green curve�.

trale cur


saosspots

tegrEsAMc

soc

tap8

st�robodt

Pdsshdwa

Fs

FM

4

ee in Fig. 5, the measured spectral extinction showsmodest peak at 9.1 �m, but the calculated spectra

btained by Querry’s indices show this peak quiteuppressed and shifted toward 10 �m. Although re-earchers have attributed this peak to absorption byhosphates or polysaccharides, we believe a portionf the increased extinction seen in the measured spec-ra between 9 and 10 �m may be due to a residualilicate used during the milling process.Perhaps the least certain complex index used was

hat of kaolin. It is well known that the optical prop-rties for kaolin are quite variable, depending on theeographic location where it is mined �i.e., kaolin isoutinely sourced from North America, China, andurope�. As a result, good agreement between mea-ured and calculated extinction was not expected.s can be seen in Fig. 7, the spectral features of theie-calculated extinction appears suppressed when

ompared with the measured values.Figure 8 shows the measured mass-normalized

pectral extinction for PEG 200, Arizona road dust,valbumin, and diesel soot. As mentioned above,omplex indices of refraction were not available for

ig. 6. Measured �thick red curve� and calculated �thick blue curvehown are the Mie theory predicted total scatter �thin blue curve�

ig. 7. Measured �thick red curve� and calculated �thick blue curie theory predicted total scatter �thin blue curve�, absorption �or


hese materials, and as a result no Mie calculationsre presented. However, we note several key as-ects for the various spectra that can be seen in Fig..First, the two noticeable features in the extinction

pectra for diesel soot that can be seen at 4.25 �m andhe fine structure that can be seen between 5 and 8m are due to gaseous absorption of CO2 and water,espectively, that were generated during the burningf diesel fuel and proved extremely difficult to removey conventional FTIR background subtraction meth-ds. Prior studies have shown this spectra to beevoid of any real structure, and it decreases mono-onically with increasing wavelength.8

Although the preparation and dissemination of theEG 200 liquid was relatively straightforward, weid notice during several runs the emergence oftrong water-vapor absorption lines when the liquidamples were left in the open for periods of severalours due to the fact that PEG 200 is extremely hy-roscopic. For this reason, care should be takenhen the PEG 200 chemical simulant is used forerosol research because water uptake for atomized

ctral extinction for nebulized liquid silicone SF-96 grade 50. Alsoorption �orange curve�, and backscatter �green curve�.

ectral extinction for aerosolized kaolin clay. Also shown are thecurve�, and backscatter �green curve�.

� spe, abs

ve� spange

sm

tptt

sstq

wzIdnms�ip

FT

aaopfatipHattlcmpttt

Fmi

Famw

amples is likely except in the most arid of environ-ents.The extinction spectrum for ovalbumin turned out

o be somewhat surprising. As one can see by com-aring ovalbumin spectra with the Mie-calculated ex-inction spectra for BG that can be seen in Fig. 5, thewo spectra are nearly identical �Fig. 9�.

Where cost and the availability of a good biologicalimulant are an issue, use ovalbumin powder mayuffice. This would be particularly useful for fieldests in which large quantities of a simulant are re-uired.The last material shown in Fig. 8 is that of a some-hat standardized background aerosol called Ari-

ona road dust obtained from Powder Technologync., Burnsville, Minnesota. Because Arizona roadust is well characterized, it has become a popularatural obscurant used in many radiative transfereasurements. The particular type used for this

tudy was sized to have a nominal diameter of 1.13m with a standard deviation of 0.385 �m. Chem-

cal analysis shows the following compounds as aercentage of mass: 76% SiO2, 15% Al2O3, 4%

Fig. 8. Measured spectral extinction for PEG 2

ig. 9. Comparison of the measured extinction spectra for ovalbu-in �lyophilized egg white� with a Mie calculation in which the

ndices of refraction for BG endospores were used.

e2O3, 4% Na2O, as well as traces of CaO, MgO, andiO2.Finally, as part of this study we were provided with

n unusual material called Cab-O-Sil. Cab-O-Sil issynthetic, amorphous, untreated fumed silicon di-

xide that is composed of fine submicrometer SiO2articles. Used commercially as a thickening agentor food and cosmetics, this silica aerogel is extremelymorphous �94% of its volume is air� and its some-imes used as a fluidizer to improve aerosol dissem-nation efficiencies. As a result, we were asked toroduce spectra for this possible residual compound.owever, difficulties with extreme agglomerationrose during the FTIR measurement. Various at-empts to disperse dry well-separated Cab-O-Sil par-icles failed. Resultant measured spectra showedittle wavelength dependence regardless of the con-entration �Fig. 10�. These flat spectra are synony-ous with the presence of large-sized parameter

articles when compared with the wavelength �i.e.,he extinction mechanism is primarily due to the par-icle’s geometric cross section rather than any diffrac-ion effects�. Efforts to measure the in situ aerosol

iesel soot,1 Arizona road dust, and ovalbumin.

ig. 10. Measured extinction for agglomerated Cab-O-Sil aerogelt various concentrations. The flat spectral extinction is synony-ous with large-sized parameter particles �compared with theavelength�.

00, d


descdd

Ciaatr

R1

2

3

4

5

6

7

8

4

ensity also failed because we were unable to collectnough measurable mass on the dosimetric filter sub-trates during the spectral runs. Nevertheless, onean see the obvious absorption between 8 and 10 �mue to SiO2, which is also seen in the Arizona roadust spectra.

We thank the Edgewood Chemical and Biologicalenter for their continued support and sponsorship

n the area of aerosol research. In particular, were extremely appreciative to Richard Vanderbeeknd Alan Samuels for their technical expertise andhe insightful conversations that aided us in ownesearch.

eferences. K. P. Gurton, D. Ligon, and R. Kvavilashvili, “Measured infra-

red spectral extinction for aerosolized Bacillus subtilis var. ni-ger endospores from 3 to 13 �m,” Appl. Opt. 40, 4443–4448�2001�.


. A. Deepak, Atmospheric Aerosols: Their Formation, OpticalProperties, and Effects �Spectrum, Hampton, Va., 1982�.

. H. Gerber and E. Hindman, Light Absorption by Aerosol Parti-cles �Spectrum, Hampton, Va., 1982�.

. D. W. Wieliczka and M. R. Querry, “Four techniques to measurecomplex refractive indices of liquids and solids at carbon dioxidelaser wavelengths in the infrared spectral region,” CRDEC-CR-062 �Chemical Research, Development, and Engineering Cen-ter, Aberdeen Proving Grounds, Md., 1990�.

. D. Ligon, J. Gillespie, and P. Pellegrino, “Aerosol propertiesfrom spectral extinction and backscatter estimated by an in-verse Monte Carlo method,” Appl. Opt. 39, 4402–4410 �2000�.

. C. Bruce, Measuring Aerosol Density using Nephelometry andDosimetry �Center for Atmospheric Sciences, Las Cruces,N.Mex., 1987�.

. M. R. Querry, “Optical constants of minerals and other materi-als from the millimeter to the ultraviolet,” CRDEC-CR88009�Chemical Research, Development, and Engineering Center,Aberdeen Proving Grounds, Md., 1987�.

. C. W. Bruce, K. P. Gurton, and T. F. Stromberg, “Trans-spectralabsorption and scattering of electromagnetic radiation by dieselsoot,” Appl. Opt. 30, 1537–1546 �1991�.

measured infrared optical cross sections for a variety of chemical and biological aerosol simulants

Documents