Trace detection of explosives with low vapor emissionsby laser surface photofragmentation–fragment detectionspectroscopy with an improved ionization probe

Jerry Cabalo and Rosario Sausa

Trace explosive residues are measured in real time by surface laser photofragmentation–fragmentdetection (SPF–FD) spectroscopy at ambient conditions. A 248�nm laser photofragments the targetresidue on a substrate, and a 226�nm laser ionizes the resulting NO fragment by resonance-enhancedmultiphoton ionization by means of its A–X �0, 0� transitions near 226 nm. We tested two probes onselected explosives and modeled their electric field in the presence of a substrate with an ion opticssimulation program. The limits of detection range from 1 to 15 ng�cm2 (signal-to-noise ratio of 3) at 1 atmand 298 K and depend on the electrode orientation and mechanism for NO formation.

OCIS codes: 300.0300, 140.3450, 300.6410, 300.6360, 280.3420.

1. Introduction

The sensitive detection of explosive compounds inreal time at ambient pressure and temperature is akey issue for law enforcement and airport security.Two laser spectroscopic techniques that are used todetect 2,4,6-trinitrotoluene (TNT) in the gas phaseare cavity ringdown spectroscopy1,2 and laserphotofragmentation–fragment detection (PF–FD)spectroscopy. The cavity ringdown-spectroscopy tech-nique probes the parent molecule, whereas thePF–FD technique probes the characteristic NO pho-tofragment by laser-induced fluorescence or resonance-enhanced multiphoton ionization (REMPI).7 The NOfragment is characteristic of the NO2 functionalgroup that is present in these molecules. Both tech-niques can detect TNT in the parts per million byvolume to parts per billion by volume range nearambient conditions. However, these techniques can-not detect other important energetic materials such

as hexahydro-1,3,5-hexanitro-1,3,5-triazine (RDX),octohydro-1,3,5,7-octonitro-1,3,5,7-octazocine (HMX),or hexanitrohexazaisowurtzitane (CL20) at ambientconditions because their vapor pressure concentra-tions are several of magnitudes less than that of TNTat 298 K and 1 atm. Structures of the energetic ma-terials are shown in Fig. 1.

Laser irradiation of the solid energetic materialswith ultraviolet (UV) light is a means of generatinggaseous molecules or fragments that can be detectedby mass spectrometry, optical spectroscopy, or both.Photothermal and photochemical processes produceatomic and molecular fragments from the surface.Tang and co-workers showed that the 226�nm irra-diation of RDX produces both positive and negativeion masses ranging from 15 to 176 amu.12 They as-signed the m�z � 30 peak to NO. Dickinson andco-workers observed small masses of both neutraland ionic species, including NO, as well as photoelec-trons when they irradiated RDX crystals with248�nm light at a fluence �5 mJ �cm2.13 As they in-creased the laser fluence, they observed an increasein the kinetic energy of the products. They observed aclear etching of the crystal at a laser fluence�90 mJ�cm2 and an intense peak attributable to NO.NO is formed in its long-lived, metastable, electronicstate (� � 50 �s) from the collisional reneutralizationof energetic NO�. At a laser fluence of �600 mJ�cm2,they observed a chemiluminescence at 385 nm fromelectronically excited CN.

The authors were with the U.S. Army Research Laboratory, MailStop AMSRL-WM-BD, Aberdeen Proving Ground, Maryland21005-5069 when this research was performed. The e-mail addressfor R. Sausa is sausa@arl.army.mil. The current address for J.Cabalo, a National Research Council Postdoctoral Associate, isU.S. Army Edgewood Chemical Biological Center, Aberdeen Prov-ing Ground, Maryland 21010.

Received 7 May 2004; accepted 17 August 2004.

1084 APPLIED OPTICS � Vol. 44, No. 6 � 20 February 2005

Our recent study on RDX sensing shows that thesurface laser photofragmentation–fragment detec-tion (SPF–FD) technique can detect RDX residues insitu and in real time at ambient conditions.14 In thistechnique, a low-power laser operating in the UVphotofragments RDX on a surface, and a second low-power laser tuned to 226 nm ionizes the characteris-tic NO fragment by REMPI by use of its A–X �0, 0�transitions. An ion probe with miniature, square elec-trodes collects the resulting electrons and ions. Weobserve that exciting RDX at 248 nm yields a greaterion signal than exiting it at 266 or 308 nm. The de-tection limit is �14 ng�cm2 (a signal-to-noise ratio of3) at 248 nm and ambient conditions.

In this paper we report on a new ion probe thatincreases the SPF–FD technique’s sensitivity by afactor of 10, and we model the electric field of both oldand new probes in the presence of a substrate usingan ion optics simulation program. We extend theSPF–FD technique to the detection of HMX, CL20,and TNT residues; determine their limits of detection(LOD) at ambient conditions; and present a moredetailed study of the issues influencing sensitivity.We also report the UV absorption spectrum of CL20and measure its extinction coefficient at 248 nm.

2. Experimental Apparatus

Our previous publication contains the details of theSPF–FD technique.14 An abridged description of theexperimental apparatus follows. Briefly, a 248�nmlaser beam excites a thin film of energetic materialplaced on the surface of a quartz plate. We generatethis beam by doubling the output of a 10�Hz, Nd:YAG-pumped dye laser (Surelite III and LambdaPhysik, FL3002) and direct it perpendicular to the

substrate surface. We focus it with a 10�cm focal-length quartz lens to a 2.9 � 103 cm2 spot. A secondlaser beam, whose frequency is in the 224–226�nmrange, ionizes the resulting NO photofragment byREMPI. We generate this laser beam by doubling theoutput of a 10-Hz, Nd:YAG-pumped dye laser (Sure-lite II and Lumonics HyperEx-300) and direct it par-allel to the substrate surface. We also focus this beamwith a 10-cm quartz lens, but to a 5 � 105 cm2 spot.The pump and probe laser energies are in the10–20 �J range. A pulse generator (Stanford Re-search Systems, DG 535) set at 1 ms controls the timedelay between the lasers.

Figure 2 depicts two sets of miniature stainless-steel electrodes that collect the ions and electrons. Wedenote the set shown in Fig. 2(a) as VE for verticalelectrodes. They are perpendicular to the substratesurface and are approximately 15 mm � 15 mm insize with a 5�mm gap and �0.5 mm from the surface.A 700�V bias between them results in an averageelectric field of �140 V�mm. We denote the set in Fig.2(b) as HE for horizontal electrodes. They are parallelto the substrate surface and are similar to the verticalelectrodes, except that they have a 6�mm gap andcontain 2�mm holes for the passage of the excitationlaser beam. An 850�V bias between them yields anaverage field of �140 V�mm, the same as in the ver-tical electrodes. The lower plate is insulated and is�0.5 mm from the substrate surface. We mount eachprobe and substrate on an XYZ stage for precise,spatial alignment between the excitation and probelaser beams. A current amplifier (Keithley 427, gain106–107 V�A, time constant 0.01 ms) amplifies theelectrode signal, and a 125�MHz oscilloscope (Le-Croy, 9400) monitors it. A personal computer recordsthe signals from a boxcar averager (Stanford Re-search Systems, SR250).

We prepare thin films of RDX, HMX, CL20, andTNT by coating microscope slides with known vol-umes of dilute solutions of the target compound inacetone and then evaporating the solvent. For eachLOD measurement, we use four to ten slides with

Fig. 1. Structural formulas of selected energetic materials.

Fig. 2. Ionization probes. (a) Vertical electrodes (VEs) whose elec-tric field is parallel to the substrate surface. The excitation laserbeam is normal to the substrate, and the ionization laser is parallelto the substrate surface and 0.5 mm above it. (b) Horizontal elec-trodes (HEs) whose electric field is normal to the substrate surface.The excitation and probe beams are oriented as in (a).

20 February 2005 � Vol. 44, No. 6 � APPLIED OPTICS 1085

varying film thickness (surface concentration) of theselected explosive. We set the pump and probe lasersto 248 and 226.28 nm, respectively, and record theSPF–FD signal while moving the substrate so thateach laser shot samples a fresh film spot. A plot of ionsignal as a function of concentration yields the re-sponse curve for each compound. We record theSPF–FD spectra of the energetic materials by settingthe excitation laser to 248 nm while scanning theionization laser wavelength at a rate of 0.009 nm�sand simultaneously moving the substrate. Each spec-trum is an average of three spectra.

Our colleagues at the U.S. Army Research Labora-tory provided us with the energetic materials. High-pressure liquid chromatography analysis of theexplosives reveals an HMX purity of more than 99.9%and RDX, CL20, and TNT purities of more than 99%.A UV–visible spectrometer (Hewlett-Packard, Model8453) with Hewlett-Packard ChemStation softwarerecords the absorbance spectra of 2.25 � 105 M RDXand CL20 in acetonitrile (high-pressure liquid chro-matography grade) in the wavelength region of190–400 nm. The NO2 (485 parts per million in N2)and NO (0.097% in Ar) gases are from MathesonTri-Gas and Airgas, respectively.

3. Results and Discussion

Figure 3 shows the SPF–FD spectra of RDX, HMX,CL20, and TNT films, along with a REMPI spectrumof diluted NO gas, in the region of 225.8–226.85 nmat 298 K and 1 atm by use of the HE ionization probe[Fig. 2(b)]. We observe similar spectra with the VEprobe [Fig. 2(a)], but we do not observe any signalwith either probe if the 248�nm laser is off or if thereis no explosive on the substrate. The prominent spec-tral features of the explosives are similar to those ofNO and are attributed to NO rotational lines of theQ1 � P21, P1, and P2 � Q12 branches of the A–X �0, 0�band.14 The fact that the explosives yield a uniquemolecular fingerprint of NO is definitive proof that

SPF–FD successfully detects residues of low-vapor-pressure explosives at ambient conditions. The over-all SPF–FD mechanism involves the following steps:

R—NO2(s)→h(248 nm)

R � NO(X2�) � O, (1)

NO(X2�)→h(226 nm)

NO(A2��)→h(226 nm)

NO � (X2��)� e, (2)

in which step (1) represents the laser excitation of athin energetic film that yields ground-state NO andstep (2) represents the �1 � 1� REMPI of the NOfragment by means of its real intermediate A2�� state(� � 215 ns). We observe an enhancement in the NOionization because the intermediate-state’s energy isresonant with one 226�nm photon. Steps (1) and (2)depict efficient processes and require only a few mi-crojoules of laser radiation. The firing between thepump and probe lasers is set to 1 ms.

We determine the efficacy of both HE and VEprobes by testing them on RDX during identical op-erating conditions (same pump and probe laser en-ergy and focal point size). Our results show that theHE probe with horizontal electrodes is approximatelyan order of magnitude more sensitive than the VEprobe with vertical electrodes. An important factorthat contributes to the difference in sensitivity is theeffect of the substrate on the probe’s electric field. Weinvestigate this effect with SIMION 7.0, an ion opticsdesign program.15

SIMION calculates the electric field and scalar poten-tial between the electrodes in the presence of a sub-strate by solving the Poisson equation 2� � ���, inwhich � is the electric potential, ε is the absolutepermittivity, and � is the charge distribution. Amethod of images creates a valid solution of the elec-tric potential by use of the distribution of real chargeplus a simulated or image charge that is idealized tobe on the other side of the plane defined by the sub-strate surface and that fulfills the boundary condi-tions. The potential for the point charge close to thesubstrate is a combination of the charge and its mir-ror image charge multiplied by a constant that isdefined by �e 1��e � 1, in which �e is the dielectricconstant in the case of a linear isotropic homogeneousdielectric.

Figure 4 shows simulations of the VE and HEprobe’s electric field with a dielectric substrate closeto the probes. The top panels show some of the equi-potential surfaces for the probes, whereas the bottompanels show a two-dimensional slice of the three-dimensional picture depicting the equipotential sur-faces as equipotential lines. When the substrate isbrought close to the edge of the VE probe, a signifi-cant component of the electric field is oriented towardthe substrate surface and away from the probe, asshown by the equipotential lines (the force vectors arenormal to the equipotential lines and surfaces). Theelectrons or ions are forced away from the probe andare not collected. In contrast, the HE probe’s electric

Fig. 3. SLP–FD spectra of RDX, TNT, HMX, and CL20, and aREMPI spectrum of NO in the region of 225.8–226.8 nm. Thespectroscopic fingerprint of NO appears in the spectra of all theenergetic materials.

1086 APPLIED OPTICS � Vol. 44, No. 6 � 20 February 2005

field is mostly unperturbed as the substrate ap-proaches the electrodes, as shown in the bottom ofFig. 2(b), and all the charged species are collected.Thus the HE probe is more sensitive than the VEprobe. The HE probe also offers additional advan-tages over the VE probe. First, it can sample metallicsubstrates because the high-voltage electrode nearthe surface can be insulated, and second it can bemade to operate as a miniature ion mobility spec-trometer for increased selectivity.

Our experiments on NO gas corroborate our simu-lation results on the two ion probes. Figure 5 showsthe REMPI signal of 6 parts per million NO in N2from the HE probe [trace (a)] and VE probe [trace (b)]near a clean substrate. The flow rate of NO throughthe probes is �5 l�min; the pump and probe lasers areset to 248 and 226.282 nm, respectively; and theprobe beam is �0.5 mm above the surface, as in thethin-film experiments. Trace (a) has a 10��s rise timeand decay with a shoulder. We attribute the signal toboth photoelectrons (� � 0–20 �s) and positive ions(t � 20 �s) generated from the ionization of NO. Al-though the probe is biased to collect the negativecharges, the ions effect the opposite plate and gener-ate an image charge on the collector plate that man-ifests itself as a tail on the signal trace. The substratedoes not perturb the probe’s extraction field, and allthe charged species are collected. Trace (b) also has a10��s rise time, but an exponential decay. Its fullwidth at half-maximum is �20 �s. Trace (b), unlike

trace (a), arises only from photoelectrons generatedfrom the ionization of NO. The tail due to the positiveions is absent, and the overall signal is reduced.

Trace (c) in Fig. 5 shows the REMPI signal of NOwith the VE probe near a clean substrate and thelaser beam positioned between the electrodes at� 0.75 cm from the substrate (center of the elec-trodes). Positioning the probe beam at the electrode’scenter is ideal for detecting species in the gas phasebecause the field is uniform in this region and be-cause the species concentration is the same as in theregion close to the electrode’s edge. Trace (c) is similarto trace (b) (HE probe) and results from the collectionof both photoelectrons and positive ions of NO. Asexpected, the signals are approximately the same,suggesting that both probes are equally effective insampling species in the gas phase. However, the HEprobe is more sensitive than the VE probe when sam-pling surface species because the beam in the VEprobe must be steered from the center of the elec-trodes (a region of relatively low concentration of sur-face species but with a uniform electric field) to theelectrode’s edge near the surface, where the speciesconcentration is high but the electric field is per-turbed by the substrate. The overall signal is thusreduced, and the sensitivity of the VE electrodes forsurface sampling is lower.

Fig. 4. SIMEON simulations of the electric field in the VE and HEprobes with the probes near a dielectric substrate. (a) A three-dimensional perspective of the VE electrode’s electric field (toppanel) and a slice of the region between the plates showing thesubstrate and the equipotential lines within the plates (bottompanel). (b) The HE electrode’s electric field with a perspectivedrawing in the top panel and a two-dimensional slice in the bottompanel. In the SPF–FD sampling region, the substrate disturbs theelectric field in HE probe much less than in the VE probe.

Fig. 5. REMPI spectra at 226 nm of NO gas at ambient temper-ature and pressure. The HE and VE probes are near a cleansubstrate, and the ionization laser is fired 1 ms after the excitationlaser, as shown in the insert. The ionization laser beam passesthrough the center of the HE plates and near the edges of the VEplates, slightly above the substrate’s surface. Trace (b) from the VEprobe shows a weaker signal compared with trace (a) from the HEprobe because the substrate perturbs the electric field in the VEprobe much more than in the HE probe. Trace (c) is from the VEelectrodes with the probe beam passing between plates at�0.75 mm from the substrate surface (center of plates). The VEprobe’s electric field is nearly uniform because the substrate is notclose to the ionization region, and the resulting trace is similar totrace (a).

20 February 2005 � Vol. 44, No. 6 � APPLIED OPTICS 1087

The probe’s configuration and electric field orien-tation play a role in the extraction and collection ofcharged species from the 248�nm irradiation of theenergetic films. Figure 6 shows both the VE [trace (a)]and HE [trace (b)] probe signal from RDX at 248 nmwith the 226�nm laser off. The 248�nm laser beam isnormal to the VE probe’s electric field but parallel tothat of the HE probe. Both traces show the timeevolution of the charged species. The curves haveapproximately the same area but have differentshapes. Most of the signal in trace (a) occurs in thefirst 50 �s, whereas the signal in trace (b) is distrib-uted over 150 �s. We attribute the peaks to a combi-nation of photoelectrons (� � 15 �s) and molecularions (t � 15 �s). Although both probes collect the pho-toelectrons, the HE probe collects more of the ions.The origin and assignment of the positive ions are notwell known and are the subject of future inquiry.

Figure 7 shows response curves of the various ex-plosives at 248 and 226.28 nm for the pump andprobe laser wavelength, respectively. The straightlines are best fits to the data, which are representedby symbols. The responses are directly proportionalto the amount of material on the substrate for a fixedoptical setup and laser energy. In all cases, the signalis over a wide range of concentrations. The LOD isdefined by 3��R, where R is the response and � is theroot mean square of the noise. Table 1 shows the LODfor the various energetic materials. They are1.4 ng�cm2 for RDX, 2.0 ng�cm2 for HMX, 7.1 ng�cm2

for CL20, and 15.4 ng�cm2 for TNT. Ranking thecompounds by LOD yields a sensitivity ofRDX � HMX � CL20 � TNT. The RDX value of1.4 ng �cm2 corresponds to a sampling of�3.3 � 106 NO molecules (�0.4 fg of RDX) in the laserprobe volume14 and compares favorably with the200�pg value obtained by Cheng and co-workers.23

Table 1 shows that the LOD ratio of TNT to RDX is�11. This value is similar to the value of 10 obtainedby our group by one-laser PF–FD of gaseous RDX andTNT at 226 nm near ambient conditions.6 Probingthe NO fragment by REMPI yielded TNT and RDXLODs of 70 and 7 parts in 109, respectively.

The overall SPF–FD mechanism represented insteps (1) and (2) suggests that the LOD for each en-

Fig. 6. Total charge time plot resulting from the prompt, 248�nmexcitation of RDX with both the VE probe [trace (a)] and HE probe[trace (b)]. The 226�nm probe laser is off.

Fig. 7. RDX, HMX, CL20, and TNT response plots with the HEprobe. The excitation and probe lasers are set at 248 and226.28 nm, respectively.

Table 1. Energetic Materials with Their Limits of Detection andExtinction Coefficients at 248 nm and their R—NO2 Bond Dissociation

Energies

EnergeticMaterial

Limit ofDetection

at 248nm (ng�

cm2)

ExtinctionCoefficient

(ε)a at 248 nm(�103)

R—NO2 BondDissociation

Energy (kcal�mol)

RDX 1.4 7.6b, 7.2c, 6.8d 34.2e, 34.3f, 39.0g

HMX 2.0 9.4d 38.8h

CL20 7.1 15.0b 39.4i

TNT 15.4 13.6d, 14.2e �60j

a� � log �I0 �I� ��c, where I0 is the intensity of the incident light,I is the intensity of the transmitted light, � is the path length incentimeters, and c is the concentration in moles per liter.

bPresent study, acitonitrile solvent.cCabalo and Sausa, methanol solvent.14

dSchroeder and co-workers, ethanol solvent; an estimate frompublished spectra.16

eKamlet and co-workers, methanol solvent; an estimate frompublished spectra.17

fWu and Fried, calculation for gas-phase molecule.18

gKuklja and Kunz, calculation for molecule located near thesurface crystal; the authors report a slightly higher value of35.1 kcal�mole for a gas-phase molecule.19

hChakraborty and co-workers; calculation for gas-phase mole-cule.20

iRice, calculation for gas-phase molecule.21

jGonzalez and co-workers.22

1088 APPLIED OPTICS � Vol. 44, No. 6 � 20 February 2005

ergetic material depends on the amount of NO pro-duced at 248 nm [step (1)] and the amount of NOdetected at 226.28 nm [step (2)]. In our LOD mea-surements, the probe energy and optical setup are thesame for all the compounds; thus the amount of NOproduced at 248 nm depends on the absorption coef-ficient of the target compound at 248 nm and thegoverning mechanism that produces NO. Figure 8shows the UV absorption spectra of RDX and CL20 inthe region of 190–320 nm. To our knowledge, we arethe first to report the absorption spectra of CL20. Theextinction coefficients for RDX and CL20 at 248 nmare 7.6 � 103 and 1.5 � 104, respectively, and arelisted in Table 1. Table 1 also shows the 248�nmextinction coefficient of RDX of 7.2 � 103 from ourprevious study14; RDX, HMX, and TNT values of6.8 � 103, 9.4 � 103, and 13.6 � 103 at 248 nm, re-spectively, that we interpolated from the extinctioncoefficient curves of Schroeder and co-workers16; anda TNT value of 14.2�103 at 248 nm that we obtainedfrom the study of Kamlet et al.17 All the RDX values,as well as the TNT values, are in good agreementconsidering the different solvents used and theerror in interpolation at 248 nm. Ordering the com-pounds by absorbance at 248 nm yieldsCL20 � TNT � HMX � RDX. A priori, we expect thecompounds’ LOD order to parallel their absorbanceorder. Surprisingly, this is not the case: The CL20and TNT absorbtivities are higher than those of RDXand HMX, yet their sensitivities are lower. Also, theabsorbtivity of CL20 is approximately the same asthat of TNT, but its sensitivity is almost twice that ofTNT. Clearly, the molecule’s absorption coefficient at248 nm plays less of a role in its LOD than the mech-anism for generating NO.

The mechanisms involved in the 248�nm laser ir-radiation of RDX, CL20, HMX, and TNT on surfaces

are complex. They may include photothermal andphotochemical processes, as well as surface effects.Among the many suggested initial steps in the ther-mal decomposition of nitramines in the condensedphase, the most likely mechanism is the homolysis ofthe nitro functional group, which is weakly attachedto the remainder of the molecule. NO2 may then reactfurther to produce NO. Table 1 also lists the R—NO2bond dissociation energy for the four compounds. Ta-ble 1 shows that TNT has the highest bond dissocia-tion energy for R—NO2 scission by at least20 kcal�mol. In part, this is because the NO2 group inTNT is bonded to a carbon-atom-containing ring thatis more stable than the nitrogen-atom-containingring in the nitramines. Thus TNT releases its NO2less readily than RDX, HMX, and CL20, and its LODvalue is expected to be larger than that of the nitra-mines. Also, TNT has several alternative decomposi-tion pathways that compete with R—NO2 bondscission. They include oxidation of —CH3 to form an-thranil,20 nitro–nitrite isomerization,24 and catalysis.These pathways also decrease the initial productionof NO2 and contribute to TNT’s lower sensitivity rel-ative to the nitramines.

The R—NO2 bond dissociation energy of nitra-mines RDX, HMX, and CL20 is similar, around34–39 kcal�mol. However, CL20 has a LOD that is afactor of nearly 3 times greater than that of RDX andHMX. This suggests that the process of NO2 releasein these molecules is more complicated than the sim-ple cleavage of a single nitro functional group andmay involve the loss of more than one nitro groupfrom each molecule. In the case of RDX and HMX, theenergy for the ring’s C—N bond cleavage is loweredafter the removal of the nitro functional group, andfurther decomposition generating additional NO2 ispossible.25 In contrast, the C—N bond in CL20’s back-bone is stabilized following NO2 homolysis, and fur-ther decomposition is hindered.25,26 The backbone ofRDX and HMX is two dimensional, and it is stericallydifficult for the radical site to stabilize itself by inter-acting with its other parts. In the case of CL20, itscage structure promotes the stabilization of the rad-ical site by rearrangement or multiple bond forma-tion with other parts of the backbone and preventsadditional NO2 loss. Although our argument forCL20’s LOD being larger than that of RDX and HMXis plausible, other governing processes may be oper-able.

Figure 9 shows a SPF–FD spectrum of NO fromRDX with the HE probe, along with a spectral simu-lation of NO, in the region of 225.8–226.8 nm. Bothspectra reveal NO rotational lines of the Q1 � P21, P1,and P2 � Q21 branches of the A–X �0, 0� band. Amultiparameter computer program based on a Boltz-mann rotational distribution analysis generates thesimulation spectrum. Parameters include laser lineshape, rotational line strengths and energies, andtemperature.14 The best fit of the data, for which weused a Gaussian function for the laser line shape,yields a rotational temperature TR of 304 � 10 K.This value agrees well with the value of 325 � 25 K

Fig. 8. Absorbance curves of 2.25 � 105 M RDX and CL20 inacetonitrile in the 190–290�nm range.

20 February 2005 � Vol. 44, No. 6 � APPLIED OPTICS 1089

that we obtain with the VE probe14 and indicates thatthe NO fragment is thermally equilibrated by colli-sions with O2 and N2 in the time scale of the experi-ment, as expected from gas kinetic calculations.

We also probe the NO fragment for vibrationalexcitation and determine its vibrational temperature.Laser radiation near 226 nm excites theNO A–X �0, 0� transitions and probes theNO X2� �v� � 0� state, whereas laser radiation near224 and 237 nm excites the (1, 1) and (0, 1) transi-tions, respectively, and probes the �v� � 1� state. Weobserve significant signal from all the energetic ma-terials at 226 nm, but little, if any, at 224 nm or fromRDX at 237 nm. This indicates that NO is formedprimarily in its X2� �v� � 0� state with a vibrationaltemperature Tv of � 298 K. The millisecond timeof our experiment is sufficient to vibrationally relaxNO, which requires a few microseconds. Heflingerand co-workers observed vibrationally excitedNO X2��v� � 2� from the 248�nm photolysis of TNTvapor near ambient conditions.4,5 This is not surpris-ing because the time between the TNT photolysis andsubsequent laser-induced fluorescence detection ofNO (�10 ns) is less than the time required for NO tovibrationally relax. Also, TNT experiences fewer col-lisions when it decomposes in the gas phase com-pared with the condensed phase.

Unlike the 224�nm excitation of the energetic films,we observe a significant NO X2� �v� � 1� signal whenwe photolyze NO2 gas at 224 nm. In this case, thetime between the photolysis of NO2 and ionization ofNO ��6 ns� is insufficient to vibrationally thermalizeNO. We do not observe any v� � 1 signal from room-temperature NO, as expected, because its Boltzmann(v� � 1�v� � 0)) ratio is �105 at 298 K. We calculatean NO2 detection limit of �150 parts in109 at 224 nmfrom this and prior research.27 This suggests that lessthan 150 parts in109 of NO2 from the energetic filmsurvives �1 ms after the laser excitation pulse. Thusthe NO signal from the energetic films results prob-

ably from secondary reactions of NO2 rather than thegas-phase photolysis of NO2 emanating from the en-ergetic film. Patil and Brill show that NO2 is theprimary product from the rapid heating of CL20 (Ref.25) and RDX,28,29 and along with HCN, by HMX.28

They observe that NO, if present at all in the earlystages of the decomposition, increases rapidly as NO2reduces from secondary reactions. Geetha and co-workers corroborate these observations from their re-search on the thermal decomposition of CL20.26

4. Conclusion

We have demonstrated the analytical utility ofSPF–FD spectroscopy for detecting trace concentra-tions of RDX, HMX, CL20, and TNT on surfaces withLODs in the low nanogram per square centimeterrange at atmospheric pressure and room tempera-ture. A 248�nm laser excites the energetic film, and asecond laser operating near 226 nm facilitates thedetection of resulting NO fragment by REMPI. Bothprocesses are efficient and require only a few micro-joules of laser energy. The technique’s sensitivity de-pends in part on the probe geometry, with the HEprobe being a factor of �10 more sensitive than theVE probe. This is corroborated by our ion optics sim-ulations, which reveal that the substrate perturbs theelectric field of the HE probe less than that of the VEprobe.

The sensitivity of the SPF–FD technique also de-pends on the photochemical and photothermal pro-cesses yielding NO. NO is rotationally andvibrationally equilibrated in the time scale of ourexperiment and is formed probably from a secondaryreaction of NO2. TNT and CL20 have lower sensitiv-ities than RDX and HMX, whose sensitivities arecomparable. TNT’s stronger R—NO2 bond comparedwith that in the nitramines and decomposition path-ways that compete with R—NO2 homolysis mightcontribute to its low sensitivity. In the case of CL20,whose R—NO2 bond strength is comparable to that ofRDX and HMX, its cage structure likely inhibits theescape of additional NO2 groups, after primaryR—NO2 homolysis, and may contribute to its lowsensitivity.

In short, the SPF–FD approach with a HE probeexhibits great potential for the detection of trace en-ergetic materials on surfaces in real time and in situbecause of its high sensitivity and simplicity of in-strumentation. Research on the photochemical andphotothermal mechanisms involved in the UV irra-diation of explosive residues is continuing.

We thank B. Rice of the U.S. Army Research Lab-oratory (ARL) for calculating the N—NO2 bond en-ergy in CL20, R. Pesce-Rodriguez and P. Kaste ofARL for the energetic material samples, and A. Kot-lar and M. Schroeder of ARL for many helpful dis-cussions. We also thank the National ResearchCouncil Postdoctoral Research Associateship Pro-gram (Jerry Cabalo) and the ARL Director’s ResearchInitiative for support (Rosario Sausa).

Fig. 9. SPF–FD spectra of NO from RDX. The dashed curve is abest fit of the observed data, shown as a solid curve. The spectraare offset for clarity. A Boltzmann analysis yields a rotationaltemperature of 304 � 10 K.

1090 APPLIED OPTICS � Vol. 44, No. 6 � 20 February 2005

References and Notes1. A. D. Usachev, T. S. Miller, J. P. Singh, F. Y. Yueh, P. R. Jang,

and D. L. Monts, “Optical properties of gaseous 2,4,6-trinitrotoluene in the ultraviolet region,” Appl. Spectrosc. 55,125–129 (2001).

2. M. Todd, R. Provencal, T. Owano, B. Paldus, A. Kachanov, K.Vodopyanov, M. Hunter, S. Coy, J. Steinfeld, and J. Arnold,“Application of mid-infrared cavity-ringdown spectroscopy totrace explosives vapor detection using a broadly tunable(6–8 �m) optical parametric oscillator,” Appl. Phys. B 75, 367–376 (2002).

3. G. M. Boudreaux, T. S. Miller, A. J. Kunefke, J. P. Singh, F.Yueh, and D. Monts, “Development of a photofragmentationlaser-induced-fluorescence laser sensor for detection of 2,4,6-trinitrotoluene in soil and groundwater,” Appl. Opt. 38, 1411–1417 (1999).

4. D. Heflinger, T. Arusi-Parpar, Y. Ron, and R. Lavi, “Applica-tion of a unique scheme for remote detection of explosives,”Opt. Commun. 204, 327–331 (2002).

5. T. Arusi-Parpar, D. Heflinger, and R. Lavi, “Photodissociationfollowed by laser-induced fluorescence at atmospheric pressureand 24 °C: a unique scheme for remote detection of explosives,”Appl. Opt. 40, 6677–6681 (2001).

6. V. Swayambunathan, G. Singh, and R. Sausa, “Laserphotofragmentation–fragment detection and pyrolysis laser-induced fluorescence studies on energetic materials,” Appl.Opt. 38, 6447–6454 (1999).

7. V. Swayambunathan, R. Sausa, and G. Singh, “Investigationsinto trace detection of nitrocompounds by one- and two-colorlaser photofragmentation/fragment detection spectrometry,”Appl. Spectrosc. 54, 651–658 (2000).

8. B. C. Dionne, D. P. Rounbehler, E. K. Achter, J. R. Hobbs, andD. H. Fine, “Vapor pressure of explosives,” J. Energetic Mater.4, 447–472 (1986), and references therein.

9. R. B. Cundal, T. F. Frank, and C. Colin, “Vapor pressuremeasurements of some organic high explosives,” J. Chem. Soc.Faraday Trans. 1 74, 1339–1345 (1978).

10. J. M. Rosen and C. Dickinson, “Vapor pressures and heats ofsublimation of some high melting organic explosives,” J. Chem.Eng. Data 14, 120–124 (1969).

11. References 8–10 report the vapor pressure of HMX, RDX, andTNT. The vapor pressure of CL20 is not reported in the openliterature: It is probably much less than that of RDX at roomtemperature based on its molecular structure.

12. T. B. Tang, M. M. Chaudhri, C. S. Rees, and S. J. Mullock,“Decomposition of solid explosives by laser irradiation—amass-spectrometric study,” J. Mater. Sci. 22, 1037–1044(1987).

13. J. T. Dickinson, L. C. Jensen, D. L. Doering, and R. Lee,“Mass-spectroscopy study of products from exposure of cyclo-trimethylene trinitramine single-crystals to KrF excimerlaser-radiation,” J. Appl. Phys. 67, 3641–3651 (1990).

14. J. Cabalo and R. Sausa, “Detection of hexahydro-1,3,5-trinitro-1,3,5-triazine (RDX) by laser surface photofragmentation-fragment detection spectroscopy,” Appl. Spectrosc. 57, 1196–1199 (2003), and references therein.

15. SIMION is an electrostatic lens analysis and design programdeveloped originally by D. C. McGilvery at Latrobe University,Bundoora, Victoria, Australia (1977). SIMION 7.0 is a PC-based program developed by David Dahl of the Idaho National

Engineering and Environmental Laboratory. Additional infor-mation can be found at http://www.simion.com/.

16. W. A. Schroeder, P. E. Wilcox, K. N. Trueblood, and A. O.Dekker, “Ultraviolet and visible absorption spectra in ethylalcohol: data for certain nitric esters, nitramines, nitroalkyl-benzenes, and derivatives of phenol, aniline, urea, carbamicacid, diphenylamine, carbazole, and triphenylamine,” Anal.Chem. 23, 1740–1747 (1951).

17. M. J. Kamlet, H. G. Adolph, and J. C. Hoffsommer, “Stericenhancement of resonance. I. Absorption spectra of the alkyl-trinitrobenzenes,” J. Am. Chem. Soc. 84, 3925–3928 (1962).

18. C. J. Wu and L. E. Fried, “Ab initio study of RDX decomposi-tion mechanisms,” J. Phys. Chem. A 101, 8675–8679 (1997).

19. M. M. Kuklja and A. B. Kunz, “Electronic structure of molec-ular crystals containing edge dislocations,” J. Appl. Phys. 89,4962–4970 (2001).

20. D. Chakraborty, R. P. Muller, S. Dasgupta, and W. A. GoddardIII, “Mechanism for unimolecular decomposition of HMX(1,3,5,7-tetranitro-1,3,5,7-tetrazocine), an ab initio study,” J.Phys. Chem A 105, 1302–1314 (2001).

21. B. Rice, U.S. Army Research Laboratory, AMRSRD-ARL-WM-BD, Aberdeen Proving Ground, Md. (personal communication,2004). A preliminary density functional theory calculation atthe B3LYP/6-31G* level yields a CL20 N—NO2 bond strengthof 38.6 kcal/mol. The error limit is plus or minus a few kilo-calories per mole.

22. A. C. Gonzalez, C. W. Larson, D. F. McMillen, and D. M.Golden, “Mechanism of decomposition of nitroaromatics-laser-powered homogeneous pyrolysis of substituted nitrobenzenes,”J. Phys. Chem. 89, 4809–4814 (1985).

23. C. Cheng, T. E. Kirkbridge, D. N. Batchelder, R. J. Lacey, andT. G. Sheldon, “In-situ detection and identification of traceexplosives by Raman microscopy,” J. Forensic Sci. 40, 31–37(1995).

24. Y. Z. He, J. P. Cui, W. G. Mallard, and W. Tsang, “Homoge-neous gas-phase formation and destruction of anthranil fromo-nitrotoluene decomposition,” J. Am. Chem. Soc. 110, 3754–3759 (1988).

25. D. G. Patil and T. B. Brill, “Thermal decomposition of energeticmaterials 53. Kinetics and mechanisms of thermolysis of hexa-nitrohexazaisowurtzitane,” Combust. Flame 87, 145–151(1991).

26. M. Geetha, U. R. Nair, D. B. Sarwade, G. M. Gore, S. N.Asthana, and H. Singh, “Studies on CL20: the most powerfulhigh energy material,” J. Therm. Anal. Cal. 73, 913–922(2003).

27. R. L. Pastel and R. C. Sausa, “Spectral differentiation of traceconcentrations of NO2 from NO by laser photofragmentationwith fragment ionization at 226 and 452 nm: quantitativeanalysis of N—NO2 mixtures,” Appl. Opt. 39, 2487–2495(2000).

28. Y. Oyumi and T. B. Brill, “Thermal-decomposition of energeticmaterials 3. A high-rate, in situ, FTIR study of the thermolysisof RDX and HMX with pressure and heating rate as variables,”Combust. Flame 62, 213–224 (1985).

29. P. E. Gongwer and T. B. Brill, “Thermal decomposition ofenergetic materials 73. The identity and temperature depen-dence of ’minor� products from flash-heated RDX,” Combust.Flame 115, 417–423 (1998).

20 February 2005 � Vol. 44, No. 6 � APPLIED OPTICS 1091