Simultaneous Raman Spectroscopy− Laser-Induced Breakdown

Anal. Chem. 2010, 82, 1389–1400

Simultaneous Raman Spectroscopy-Laser-Induced Breakdown Spectroscopy for Instant Standoff Analysis of Explosives Using a Mobile Integrated Sensor Platform Javier Moros, Juan Antonio Lorenzo, Patricia Lucena, Luciano Miguel Tobaria, and Jose´ Javier Laserna* Department of Analytical Chemistry, Faculty of Sciences, University of Ma´laga, E29071 Ma´laga, Spain A novel experimental design combining Raman spectroscopy and laser-induced breakdown spectroscopy (LIBS) in a unique integrated sensor is described. The sensor presented herein aims to demonstrate the applicability of a hybrid dual Raman-LIBS system as an analytical tool for the standoff analysis of energetic materials. Frequencydoubled 532 nm Nd:YAG nanosecond laser pulses, first expanded and then focused using a 10× beam expander on targets located at 20 m, allowed simultaneous acquisition of Raman-LIBS spectra for 4-mononitrotoluene (MNT), 2,6-dinitrotoluene (DNT), 2,4,6-trinitrotoluene (TNT), cyclotrimethylenetrinitramine (RDX), C4 and H15 (plastic explosives containing 90% and 75% of RDX by weight, respectively), and Goma2-ECO (Spanish denominated dynamite class high explosive mainly composed of ammonium nitrate, nitroglycol, and dinitrotoluene among other compounds), sodium chlorate, and ammonium nitrate. With the use of a Cassegrain telescope, both Raman and LIBS signals from the same laser pulses were collected and conducted through a bifurcated optical fiber into two identical grating spectrographs coupled to intensified charge-coupled device (iCCD) detectors. With the use of the appropriate timing for each detection mode, adjustment of the laser power on the beam focal conditions is not required. The ability of the present single hybrid sensor to simultaneously acquire, in real time, both molecular and multielemental information from the same laser pulses on the same cross section of the sample at standoff distances greatly enhances the information power of this approach. In its broadest sense, sensor fusion refers to the integrated use and design of multiple sensors, such integration being generally for purposes of achieving either better performance or reduced penalties than can be obtained with the collection of sensors, each separately used.1 An important branch of sensor * To whom correspondence should be addressed. Phone: +34 952 13 1881. Fax: +34 952 13 2000. E-mail: [email protected]. (1) Robison, S. R. The infrared & Electro-optical Systems Handbook. Emerging Systems and Technologies; SPIE Optical Engineering Press: Bellingham, WA, 1993; Vol. 8. 10.1021/ac902470v  2010 American Chemical Society Published on Web 01/19/2010

fusion is sensor data fusion,2 i.e., combining the data from two or more sensors to obtain an integrated, and possibly more accurate, assessment of the sensed target. One of the most demanding applications of spectroscopic techniques is the detection and identification of explosives located at significant distances from the instrument. The issue of explosive residue analysis requires the utilization of sound techniques with the purpose of providing a precise identification of chemical identity, free from the uncertainties caused by a variable measurement background, a changing transmission atmosphere, and a relatively large list of candidate compounds to be used in detonating devices. For this end, fused sensors may provide an improved solution in terms of analysis speed and necessary identification confidence when compared with individual analytical tools. In recent years, different methods based on standoff Raman and standoff laser-induced breakdown spectroscopy (LIBS) have been developed for analyzing explosives from several to tens of meters distance. Sharma et al.3 reported their initial efforts to use a small portable Raman system for standoff detection and identification of various types of organic chemicals. Its use for identifying unknown compounds by measuring standoff spectra of two explosives (cyclotetramethylene tetranitramine and triaminotrinitrobenzene) at a distance of 10 m was demonstrated. Gaft and Nagli4 developed and tested a Raman system for the field remote detection and identification of minimal amounts of different explosives on relevant surfaces at a distance of up to 30 m, whereas Carter et al.5 designed and demonstrated a standoff Raman system for detecting high-explosive materials at distances up to 50 m also in ambient light conditions. In reference to LIBS, Lopez-Moreno et al.6 used for the first time an open-path system working under a coaxial configuration to demonstrate the feasibility of detecting energetic materials in the field. The authors showed promising results for the discrimi(2) Shah, P. V.; Singh, A.; Agarwal, S.; Sedigh, S.; Ford, A.; Waterbury, R. Proc. SPIE 2009, 7303, 730329-1–730329-12. (3) Sharma, S. K.; Misra, A. K.; Sharma, B. Spectrochim. Acta, Part A 2005, 61, 2404–2412. (4) Gaft, M.; Nagli, L. Opt. Mater. 2008, 30, 1739–1746. (5) Carter, J. C.; Angel, S. M.; Lawrence-Snyder, M.; Scaffidi, J.; Whipple, R. E.; Reynolds, J. G. Appl. Spectrosc. 2005, 59, 769–775. (6) Lo´pez-Moreno, C.; Palanco, S.; Laserna, J. J.; De Lucia, F. C., Jr.; Miziolek, A. W.; Rose, J.; Walters, R. A.; Whitehouse, A. J. Anal. At. Spectrom. 2006, 21, 55–60.

Analytical Chemistry, Vol. 82, No. 4, February 15, 2010

1389

nation of explosive residues placed on a vehicle surface located at 45 m from the instrument. Gottfried et al. identified explosives at 20 m using the ability of double-pulse LIBS to reduce the amount of air entrained in the plasma achieving signal enhancement.7 Recently, the possibility of detecting explosives behind an obstacle has been demonstrated.8 The detection statistics associated to explosive residues placed behind glass and plastic barriers located at 30 m was reported. The idea of extracting information on both Raman scattering and atomic emission from a sample using a single instrument has been around for a while.2 From the stand point of sensor fusion, this idea is appealing since both sensing modes offer complementary information on the identity of sample constituents, namely, molecular identity to be extracted from scattering data and atomic abundance in the sample to be inferred from atomic emission spectra. The implementation of this concept is quite feasible since both spectroscopic techniques interrogate samples using pulsed laser beams of identical characteristics and employ dispersive spectrographs on overlapping spectral ranges requiring approximately the same spectral resolution. Although the timing requirements in both instances differ broadly, the question can be solved using the capabilities offered by gated multichannel detectors. A number of papers have been published demonstrating this concept, mainly centered on geological studies and on the cultural heritage field.9-15 Initial attempts comprised close contact laboratory analysis and involved first the acquisition of the Raman spectrum and then the LIBS spectrum after refreshing the sample position and modifying the laser pulse energy or the beam focal conditions.9-12,15 Unfortunately, under the conditions of this sequential data acquisition mode, Raman and LIBS information may not evidence any relation as the spectral data do not belong to the same interrogated target area. Combined Raman and LIBS data from minerals using a single 532 nm pulsed laser source at a standoff distance of around 9 m have been also reported.13,14 Spectra thus obtained show features corresponding to both Raman scattering and atomic emission. Although increased information can be gathered from a given sample position, significant spectral overlap between Raman bands and atomic lines may occur in the Stokes region of the Raman spectrum. Unfortunately, with the aforementioned approaches, the complementary benefits of both techniques are not exploited in full. The need to capture multispot data from different regions of the (7) Gottfried, J. L.; De Lucia, F. C., Jr.; Munson, C. A.; Miziolek, A. W. Spectrochim. Acta, Part B 2007, 62, 1405–1411. (8) Gonza´lez, R.; Lucena, P.; Tobaria, L. M.; Laserna, J. J. J. Anal. At. Spectrom. 2009, 24, 1123–1126. (9) Giakoumaki, A.; Osticioli, I.; Anglos, D. Appl. Phys. A: Mater. Sci. Process. 2006, 83, 537–541. (10) Courre`ges-Lacoste, G. B.; Ahlers, B.; Rull Pe´rez, F. Spectrochim. Acta, Part A 2007, 68, 1023–1028. (11) Escudero-Sanz, I.; Ahlers, B.; Courre`ges-Lacoste, G. B. Opt. Eng. 2008, 47, 033001. (12) Wiens, R. C.; Sharma, S. K.; Thompson, J.; Misra, A. K.; Lucey, P. G. Spectrochim. Acta, Part A 2005, 61, 2324–2334. (13) Sharma, S. K.; Misra, A. K.; Lucey, P. G.; Wiens, R. C.; Clegg, S. M. Spectrochim. Acta, Part A 2007, 68, 1036–1045. (14) Sharma, S. K.; Misra, A. K.; Lucey, P. G.; Lentz, R. C. F. Spectrochim. Acta, Part A 2009, 73, 468–476. (15) Osticioli, I.; Mendes, N. F. C.; Porcinai, S.; Cagnini, A.; Castellucci, E. Anal. Bioanal. Chem. 2009, 394, 1033–1041.

1390


specimen often causes difficulties in sensor fusion due to the spatial heterogeneity of the target. In this paper we present the first experimental sensor permitting to separately obtain standoff, instant, simultaneous Raman spectroscopy and LIBS data from single spots on the sample. Instrumental design of the mobile integrated sensor platform, the critical issues involved on spectral data acquisition, as well as possible conflicting parameters affecting the performance of both measurement modes are detailed. All studies described were performed indoor using a partially open corridor providing a firing range up to 50 m long. EXPERIMENTAL SECTION Experimental Setup for Dual Raman-LIBS Sensing. Figure 1 shows the experimental setup used for instant and simultaneous Raman and LIBS spectroscopy based standoff analysis. The instrument used a Quantel Brilliant Twins Q-switched Nd:YAG laser (10 Hz, 400 mJ/pulse, at 532 nm, 5.5 ns pulse width) as a radiation source. The beam was first expanded and then focused on a well-defined spot on the distant sample using a 10× large output beam expander (Special Optics, Wharton, NJ). After interaction with the target, inelastic scattering and plasma emission were collected using a homemade Cassegrain telescope (167 cm in length and 24 cm in diameter), which permits us to focus the return light onto the tip of a bifurcated optical fiber (600 µm in diameter mounted on a precision linear stage). Fibers were then coupled to a pair of identical gated Czerny-Turner spectrometers, model Shamrock sr-303i (303 mm focal length, f/4, 100 µm slit), each fitted with an Andor iStar intensified CCD detector (1024 × 1024 pixel, 26 mm2 pixel, intensifier diameter 25 mm). The Raman spectrometer used a 300 grooves per mm grating, whereas the LIBS spectrometer used a 150 grooves per mm grating. The spectral coverage was from 534 to 825 nm and from 235 to 828 nm, respectively. As shown in the inset of Figure 1, a holographic SuperNotch filter (Kaiser Optical systems Inc.) was placed in front of the fiber to remove the Rayleighscattered light at 532 nm. Two pulse and delay generators (Berkeley Nucleonics model 565-4C) were also included to aid in the synchronization of the experiment. The entire system is mounted on a wheeled cart, which is easily transportable, also fitted with leveling feet for guaranteeing device stability once it is located in the desired location. Additionally, laser and telescope are also equipped with a pair of pneumatic leveling isolation mounts to avoid oscillations on the focusing point on the target and on the light collection point at the optical fiber. Data acquisition was carried out using offered software for imaging and spectroscopy from Andor installed in a pair of personal computers. Data obtained were exported in appropriate format and processed using Matlab (The Mathworks Inc., South Natick, MA). Samples. Explosive materials were provided by Laboratorio Quı´mico Central de Armamento (Ministry of Defense, Madrid, Spain) and included 2,4,6-trinitrotoluene (TNT), cyclotrimethylenetrinitramine (RDX), C4 and H15 (plastic explosives containing 90% and 75% of RDX by weight, respectively), and Goma2-ECO (Spanish denominated dynamite class high explosive mainly composed of ammonium nitrate, nitroglycol, and dinitrotoluene among other compounds). Additionally, the explosive components sodium chlorate (NaClO3) and ammonium nitrate (NH4NO3),

Figure 1. Experimental setup of the standoff dual Raman-LIBS sensor: (A) Nd:YAG laser (532 nm); (B) beam expander; (C) telescope; (D) laser power sources; (E) pulse and delay generators; (F) spectrographs; (G) bifurcated optical fiber coupled into a collimating lens; (H) holographic SuperNotch filter; (I) personal computer. The inset shows the telescope optical layout. For additional details see text.

purchased from Fluka (Madrid, Spain), were used. The study also includes 4-mononitrotoluene (MNT) and 2,6-dinitrotoluene (DNT). Bulk samples were prepared for the analysis in their most appropriate form according to their explosive nature. Inorganic substances, MNT, and DNT were used as pellets. After their grinding in an agate mortar, 1.5 g amount of the chemicals was compacted using a hydraulic press under a base pressure of 10 Tm. With this procedure, cylindrical pellets of ca. 200 mm2 in area, 6 mm thick, were obtained. RDX, C4, H15, and Goma2-ECO were prepared from 1 g of material stocked together on the surface of glass microscope slides (76 mm × 26 mm) keeping into consideration that density and area reached were similar to those established for the previously cited pellets. Finally, TNT was carefully melted in a crucible using a burner and avoiding prolonged exposure to the flame. Once melted, it was directly deposited as a bulk on a glass slide. It should be highlighted that the different compacting grades of the bulk materials assayed may significantly affect the laser-matter interaction.

RESULTS AND DISCUSSION Timing Considerations. In Raman scattering interaction of photons with the molecule of interest and re-emission of scattered photons occur almost simultaneously. Thus, when a nanosecond laser pulse is used, the Raman lifetime remains in the nanosecond time regime and photons are only scattered or emitted during the laser pulse width. On the contrary, atomic emission that follows laser ablation is a much slower process (heating, vaporization, atomization, ionization, and radiative emission must take place), the rate of plume formation being associated to power density at the target surface, physicochemical properties of the sample, as well as ambient gas characteristics.16 A time-resolved study of the acquired signal is consequently the first critical issue when simultaneous molecular and multielemental analytical information needs to be separately collected, also aiming at suppressing long-lived fluorescence photons. In the first place, timing parameters for discriminating between Raman scattering and molecular fluorescence were (16) Miziolek, A. W.; Palleschi, V.; Schechter, I. Laser-Induced Breakdown Spectroscopy (LIBS) Fundamentals and Applications; Cambridge University Press: New York, 2006.


1391

Figure 2. Standoff temporal behavior for DNT and NH4NO3 out-focused (A) and in-focused (B) Raman scattering signals as well as LIBS emission signal (C) after excitation using a Nd:YAG (532 nm) laser pulse (width ∼5.5 ns). Each Raman spectrum is the result of 25 accumulated laser pulses using 0.25 and 4 GW cm-2 irradiance values for out-focused and in-focused studies, respectively. Each LIBS signal belongs to the ensemble-average of spectra resulting from 10 laser pulses using 4 GW cm-2. Exposure time values of 5.5 and 500 ns, for Raman and LIBS, were, respectively, used. Note that for better visualization time grows from left to right in the Raman plots and from right to left in the LIBS plots. More details are in the body of the text.

evaluated. If the detection system is properly timed so as to detect only those photons scattered or emitted during the laser pulse, Raman photons will be collected while rejecting the majority of fluorescence. For this study Raman spectra were acquired using a 5.5 ns gate width at different delays from the external trigger input (considered 0 time) to the opening of the camera intensifier tube. The external trigger input was supplied by the Q-switch output signal of the laser. Each spectrum consisted of the accumulation of 25 laser shots. The study involved DNT and NH4NO3 as model compounds using 500 mJ laser pulses 1392


focused to a 40 mm2 spot (0.25 GW cm-2). Results are depicted in Figure 2A. Raman scattering instantaneously appears at 0 ns delay. Here, it should be noted that the delay time of approximately 132 ns introduced by the working distance of 20 m (6.6 ns delay per every forward-backward meter traveled by light between the laser and the target) is practically compensated by the propagation time spent through some of the system components (delay generators, cables, etc.). Raman bands can be well-appreciated up to about 10 ns delay, Raman and possible fluorescence effects being clearly lost for a gate delay of 20 ns for both substances.

Figure 3. Contour dual Raman-LIBS plots for the standoff response of some of the most significant substances from each family of energetic materials and related compounds under study as a function of different irradiance conditions. Intensity behavior for Raman response was evaluated by checking normalized net intensity (signal minus background) for bands related to NO2 symmetric stretching (DNT), ν(C-N-C) symmetric ring-breathing vibration mode (H15), NO3- symmetric stretching (Goma2-ECO), and ClO3- symmetric stretching (NaClO3) vibrations. LIBS response surfaces were built using net signals for C2 system (DNT), HR (656.4 nm) (H15 and Goma2-ECO), and Na (589.8 nm) (NaClO3) lines. The respective experimental conditions for Raman acquisition and LIBS acquisition were as follows: delay time values, 100 ps and 800 ns; exposure time values, 5.5 ns and 5 µs. Raman spectra were accumulated for 25 laser shots. LIBS spectra represent the ensemble-average of the same laser pulses.

Additionally, it is possible to check that, for DNT, laser photons have enough energy and fluorescence is excited, its slight presence at gate delays longer than 2 ns being appreciated. The main effect of this fluorescence is that the Raman signature of the material “rides” on top of a broad nonspecific baseline, and thus, significant growth in the background signal was observed. On the contrary, NH4NO3 shows no evidence of the effect. Thus, from these results gate delays shorter than 2 ns for a gate width of 5.5 ns should be selected as optimal timing parameters for Raman photons acquisition.

However, as our aim was centered on the simultaneous collection of Raman and LIBS information from the same laser event and, in consequence, on the same target spot, timing parameters for Raman signals acquisition were evaluated at irradiance values causing laser-induced breakdown to occur. For this purpose, the beam was now tightly focused on the target. The irradiance reached was then 4 GW cm-2. Results on temporal evolution of Raman signal under these conditions are illustrated in Figure 2B. Under high-irradiance conditions, a plasma is formed, and in consequence, bremsstrahlAnalytical Chemistry, Vol. 82, No. 4, February 15, 2010

1393

Figure 4. Evolution of standoff Raman net signal-to-noise ratio (S/N) vs different number of accumulated laser pulses for some of the substances under study. Mean and standard deviation (error bars) were obtained from five independent series of each one accumulation value. These plots were built using the same Raman bands employed for the Raman contour plots of Figure 3. Experimental conditions: delay time, 100 ps; exposure time, 5.5 ns; irradiance value, 4 GW cm-2.

ung and unspecific recombination emission take place as a function of the ablation threshold of each substance. The Raman signature of the material “rides” on top of a broad nonspecific tilted baseline. This effect takes more and more relevance for large gate delays, and it contributes to totally mask the weakest Raman bands at delay times >10 ns. Thus, for simultaneous monitoring Raman scattering and atomic emission, the delay time for the Raman channel should be set large enough to allow Raman signal to develop and, at the same time, short enough to avoid masking of the molecular fingerprint by the LIBS background. As a tradeoff, the gate delay was set to 100 ps, whereas the gate width was 5.5 ns, i. e. the nominal laser pulse width. Timing parameters for proper acquisition of LIBS signals are inferred from Figure 2C. A gate delay of about 1 µs is required for avoiding the unspecific background. Moreover, while for organic substances such as DNT as shown here, the LIBS signal lasts longer, for inorganic analytes the signal should be acquired during just a few microseconds after the fate of unspecific background. As a trade-off, a delay time of 0.8 µs and a gate width of 5 µs result adequate for acquisition of LIBS data from both organic and inorganic analytes. Irradiance Studies. The success in acquiring simultaneous information on Raman scattering and atomic emission from a single laser event is linked to the energy gradient created along the target by the laser beam. When the laser pulse impacts the target surface, a combination of processes occurs, among them ablation and laser-induced breakdown in the central, most intense, part of the beam and molecular scattering in the outer region of the beam, where the irradiance is below the ablation threshold 1394


of the material.17 Thus, the optimal combination of pulse energy and beam focal conditions resulting in effective excitation of both phenomena must be found. For this purpose, behavior of the most characteristic LIBS emission lines as well as the most intense Raman bands for each material was evaluated by focusing pulses of variable energy on spots of different size, in order to create a wide range of irradiance conditions. As an example, Figure 3 depicts irradiance effect plots on dual Raman-LIBS simultaneous signals for some representative substances. As shown, Raman features were observed along the whole range of area values tested. However, best Raman signals for all substances are acquired when spot values on target range from 5 to 10 mm2 and pulse energies are comprised between 500 and 600 mJ. It should be noted that no Raman scattering is observed for pulse energies below 380 mJ, independently of the cross section interrogated. Regarding LIBS, analytically useful emission signal takes place when high laser pulse energy is focused on a reduced spot area. For area values larger than 5 mm2, the breakdown irradiance threshold (located at about 2 GW cm-2 for most substances tested) is not reached, and consequently, no LIBS data can be acquired. Thus, with the aim of simultaneously collecting Raman and LIBS data, a 560 mJ laser pulse focused to a 2.5 mm2 area spot ensured an adequate trade-off for further studies. Under these experimental (17) Van Vaeck, L. In Encyclopedia of Analytical Science; Townshend, A., Poole, C. F., Worsfold, P. J., Eds.; Academic Press: New York, 2005; pp 237249.

conditions, irradiance is well above the threshold level required for inducing standoff breakdown in all solids under study, whereas the irradiance is low enough for avoiding the broad nonspecific tilted baseline (associated to the continuum background) that totally masks the Raman signature of the material. Efficiency of Raman Signal Acquisition. As reported in a previous paper, single-shot LIBS spectra of several energetic

materials located at a 45 m from the instrument can be readily obtained.3 Differently, the signal from several laser shots must be accumulated before a useful Raman signal-to-noise ratio (S/N) builds up. This is particularly relevant in standoff Raman spectra acquisition. It should be kept in mind that the number of Raman spectra accumulated should be set to a minimum in order to cope with the single-shot capability of LIBS.

Figure 5. Continued. Analytical Chemistry, Vol. 82, No. 4, February 15, 2010

1395

Figure 5. Raw instant, simultaneous and dual standoff Raman (accumulated) and LIBS (averaged) spectra of (A) MNT, (B) DNT, (C) TNT, (D) RDX, (E) C4, (F) H15, (G) Goma2-ECO, (H) NH4NO3, and (I) NaClO3, at 20 m distance obtained from 10 laser pulses (560 mJ pulse-1 on a 2.5 mm2 spot). More acquisition details across the body of the text.

Figure 4 plots the Raman S/N as a function of the number of accumulated laser pulses for some substances under study. Net signal (signal minus background) was calculated as the mean signal for five independent analyses, whereas noise magnitude was calculated as the root-mean-squared (rms) value of the signal for a defined spectral region where only background was present. As expected, for all substances the S/N grows with the number of Raman spectra accumulated. A somewhat different behavior is observed for the four compounds shown in the figure. Differences are due to the distinct degree of compactness of each chemical, their different heterogeneities, and the laser ablation that occurs while the Raman spectra are accumulated. Thus, DNT and H15 show a well-defined increasing 1396


slope. However, low initial S/N values indicate that several laser pulses accumulation is required. On the contrary, S/N values for Goma2-ECO and NaClO3 exhibit no significant change. In these cases, initial S/N values are clearly higher than those for DNT and H15 and they seem to remain practically constant (consider error bars) throughout the range of accumulated laser pulses assayed. Thus, for an adequate trade-off between clear and useful molecular fingerprint coping with multielemental LIBS data, it was considered that 10 laser pulses on the target are appropriate. It should be noticed that, although the use of accumulated laser pulses is required for Raman signal acquisition, the traditional ensemble-average of spectra resulting from the same 10 laser pulses was used for LIBS spectral data processing.18,19

Table 1. Assignment of Standoff Raman Spectral Features for Explosives and Related Compounds under Study observed Raman shift (cm-1) Raman vibrational modes

MNT

DNT

TNT

RDX

C4

H15

Goma2-ECO NH4NO3 NaClO3

ClO3- deformation NO2 wagging 632 NO3- asymmetric bending NO2 scissoring 858 792 C-N-C symmetric ring-breathing 876 880 876 ClO3 symmetric stretching NO3 symmetric stretching H-C-C in plane bending 1100 1200 N-N stretching 1214/1272 1214/1262/1308 1212/1268 NO3- asymmetric stretching NO2 symmetric stretching 1336 1358 1360 NO3- asymmetric stretching C-C stretching 1510 NO2 asymmetric stretching 1594 1525 1530/1614 C-H stretching 2920/3070 2940/3098 2940/2992 2960 2937 + NH4 symmetric stretching

618 717 935 1040

1040 1284 1414

3185

Table 2. Standoff LIBS Spectral Features for Explosives and Related Compounds under Studya LIBS emission line CN C2

H(I) N(I) O(I) Na(I)

Ca(II) a

wavelength, nm

MNT

DNT

TNT

358.5 388.3 416.6 436.7 468.4 512.8 515.4 558.4 486.1 (β) 656.4 (R) 742.5 744.4 747.0 777.2 777.4 777.5 498.4 569.0 589.8 616.2 393.4 396.8

x x x x x x x x

x x x x x x x x

x x x x x x x x

x x x x x x x x x x x x x



RDX

C4

H15

Goma2-ECO

x

x x x

x x x

x

x x

x

x x

x x x x x x x x x x x x x x

x x x x x x x x x x x x x x x

x x x x x x x x x x x x x x x


NH4NO3

NaClO3



Note: LIBS bold assignments indicate possible interferences which can appear in the ideal spectra.

Dual Raman-LIBS Spectra. Figure 5 shows raw standoff (20 m) simultaneous Raman-LIBS spectra of all substances under study acquired after bulk target excitation. Raman spectra correspond to the accumulation of 10 laser pulses, whereas LIBS (18) Radziemski, L. J.; Loree, T. R.; Cremers, D. A.; Hoffman, N. M. Anal. Chem. 1983, 55, 1246–1252. (19) Rusak, D. A.; Castle, B. C.; Smith, B. W.; Winefordner, J. D. Crit. Rev. Anal. Chem. 1997, 27, 257–290. (20) Lin-Vien, D.; Colthup, N. B.; Fateley, W. G.; Grasselli, J. G. Infrared and Raman Characteristic Frequencies of Organic Molecules; Academic Press: London, 1991. (21) Socrates, G. Infrared and Raman Characteristic Group Frequencies: Tables and Charts, 3rd ed.; John Wiley & Sons: New York, 2001. (22) Lewis, I. R.; Daniel, N. W., Jr.; Griffiths, P. R. Appl. Spectrosc. 1997, 51, 1854–1867. (23) Torres, P.; Mercado, L.; Cotte, I.; Herna´ndez, S. P.; Mina, N.; Santana, A.; Chamberlain, R. T.; Lareau, R.; Castro, M. E. J. Phys. Chem. B 2004, 108, 8799–8805. (24) Rajkumara, B. J. M.; Ramakrishnan, V.; Rajaram, R. K. Spectrochim. Acta, Part A 1998, 54, 1527–1532. (25) Alves, W. A.; Faria, R. B. Vib. Spectros. 2003, 31, 25–29.

spectra result from averaging the signal obtained for the same 10 laser pulses. Observed Raman scattering frequencies identified for each material are summarized in Table 1.20-25 As seen in Figure 5, the Raman spectrum of MNT is saturated under the aforementioned experimental conditions. A spectrum is shown obtained at lower irradiance magnified 20-fold. H15 exhibits the poorest Raman fingerprint, extremely affected by the presence of the plasticizer additive. Regarding LIBS, Table 2 lists the most prominent LIBS features observed. As shown in Figure 5, nitroaromatic substances (MNT, DNT, TNT) exhibit sequences of the CN violet system (originated not only from the decomposition of nitrogen-containing products but also from the interaction of the C2-containing plume with N2 from the surrounding air) and the C2 Swan system (originate from C2Hx fragments directly released from any structure present in the target compound or from carbon-carbon recombination in the hot plasma). Additionally, medium- and Analytical Chemistry, Vol. 82, No. 4, February 15, 2010

1397

Table 3. Standoff Dual Raman-LIBS Sensor Ablation Featuresa

a Mean values and their corresponding standard deviations for crater volume and ablation rate were estimated from five independent craters generated by indicated number of laser pulses on the same bulk material. RSD means relative standard deviation. The field of view of each photomicrography is 4.3 × 3.4 mm2.

low-intensity emission atomic lines of H, N, and O can be also appreciated.26,27 These results are also interesting to remark the high similarity exhibited by the LIBS spectra of RDX and NH4NO3 in spite of their different nature. Goma2-ECO is mainly composed by ammonium nitrate. However, its spectrum shows remarkable differences with that of its main constituent NH4NO3. Particularly noticeable are a pair of very specific, broad, and reasonably strong emission lines centered at 554.7 and 617.3 nm. Since the detailed composition of this material is unknown for us, these emission features could be related to either the plasticizer or products generated from reactions during laser events on mixtures of organic substances and ammonium nitrate. Emission lines from impurities present in the samples and in the surrounding air, such as Ca, Na, and/or K, are also noticed in the standoff LIBS spectra. Standoff Raman-LIBS Sensor Features. Once the possibility of simultaneous standoff Raman and LIBS data acquisition from (26) Portnov, A.; Rosenwaks, S.; Bar, I. J. Lumin. 2003, 102, 408–413. (27) Gottfried, J. L.; De Lucia, F. C., Jr.; Munson, C. A.; Miziolek, A. W. J. Anal. At. Spectrom. 2008, 23, 205–216.

1398


the same laser events was demonstrated, some features, including target mass ablation rate and spectral variability derived from the use of our designed sensor, were tested. Table 3 lists data on ablation rates and depicts pictures of the ablation craters formed. Depending on the material considered crater volumes range from 1.4 to 5.0 mm3, thus evidencing the broad differences in ablation behavior, as a function of the nature and characteristics of the sample. Compounds analyzed as pellets (MNT, DNT, NH4NO3, and NaClO3) exhibit similar ablation rates in the low hundred micrograms per shot with the lowest values in mass relative standard deviation (RSD). Thus, the compacting grade attained with pellets seems to be important in ensuring a homogeneous ablation rate in spite of the difference in physicochemical properties of the compounds considered. The shot-to-shot variability in ablation rate for these materials remains close to 10%. Contrarily, the ablation rate of plastic materials (C4, H15, and Goma2-ECO) is extremely variable and much larger than that of pelletized compounds. The ablation rate of TNT is similar to that of pelletized materials, but its RSD is much larger. Contrarily, RDX shows the

Table 4. Standoff Spectral Variability for the Most Significant Standoff Raman-LIBS Signals of the Bulk Substances under Studya RS

LIBS

-1

compound

Raman band (Raman shift, cm ) Raman net intensity RSD (%) emission lines, (wavelength, nm) LIBS net intensity RSD (%)

MNT

NO2 symmetric stretching (1336)

108471 ± 51360

47b

DNT


28117 ± 3654

13

TNT


24899 ± 4973

20

RDX

C-N-C symmetric ring-breathing (876)

17145 ± 5772

23

C4


11980 ± 3597

30

H15


11822 ± 2724

23

Goma2-ECO NO3- symmetric stretching (1040)

45462 ± 11434

25

NH4NO3

NO3- symmetric stretching (1040)

230440 ± 29561

13

NaClO3

ClO3- symmetric stretching (935)

171758 ± 56923

33

CN (388.3) C2 (515.4) H(I)R (656.4) N(I) (744.4) O(I) (777.4) CN (388.3) C2 (515.4) H(I)R (656.4) N(I) (744.4) O(I) (777.4) CN (388.3) C2 (515.4) H(I)R (656.4) N(I) (744.4) O(I) (777.4) CN (388.3) H(I)β (486.1) H(I)R (656.4) N(I) (744.4) O(I) (777.4) CN (388.3) H(I)β (486.1) H(I)R (656.4) N(I) (744.4) O(I) (777.4) CN (388.3) H(I)β (486.1) H(I)R (656.4) N(I) (744.4) O(I) (777.4) H(I)β (486.1) (554.7) (617.3) H(I)R (656.4) N(I) (744.4) O(I) (777.4) H(I)β (486.1) H(I)R (656.4) N(I) (744.4) O(I) (777.4) Na(I) (589.8) N(I) (744.4) O(I) (777.4)

2411 ± 948 3015 ± 954 1572 ± 1289 162 ± 99 240 ± 181 2037 ± 764 2003 ± 698 958 ± 742 123 ± 73 212 ± 150 1857 ± 662 598 ± 234 470 ± 365 68 ± 34 143 ± 104 590 ± 297 1112 ± 696 3079 ± 1822 177 ± 118 516 ± 323 910 ± 575 741 ± 815 2050 ± 2335 96 ± 51 314 ± 374 1207 ± 413 641 ± 399 1737 ± 1093 140 ± 192 266 ± 183 521 ± 586 897 ± 274 1178 ± 326 1622 ± 1829 67 ± 74 236 ± 274 1355 ± 1353 3963 ± 3690 168 ± 188 591 ± 596 21522 ± 3291 3407 ± 1034 8803 ± 2981

39 32 82 61 75 38 35 77 59 71 36 39 78 50 73 50 63 59 67 63 63 110 114 53 119 34 62 63 137 69 112 30 28 113 110 116 100 93 112 101 15 30 34

a Mean net signal intensity ± its standard deviation as well as its corresponding relative standard deviation. Values have been calculated from an overall set of 50 standoff lasers pulses using definitive experimental conditions detailed on the body of the text. Raman values were calculated using five independent spectra obtained after 10 accumulated laser pulses; each one as well as LIBS results were obtained from five independent spectra obtained after 10 averaged laser pulses each one. b Raman MNT features were calculated from spectra obtained using lower laser pulse energy in order to avoid saturation effects. Atmospheric conditions measured during data collection: temperature 32.7 °C and relative humidity 30%.

largest ablation rate, whereas the variability remains close to that of the pelletized samples. It should be noted that in none of the compounds tested was a detonation initiated under the intense laser field. With the aim of inspecting whether the aforementioned ablation rates directly correlate with the spectral signals, the intensities of the main Raman and LIBS spectral features were measured. The results are listed in Table 4. Due to the broad differences in the properties and formulations of the compounds tested, it is difficult to extract definitive conclusions. However, some general trends can be noticed. For explosives tested as pure compounds, the ablation rate is reasonably reproducible, with values close to 10% RSD (except for TNT). The variability in LIBS signals is generally lower in this case. Contrarily, explosives formulated as mixtures of compounds (C4, H15, and Goma2-ECO) exhibit large variations in both the ablation rate and LIBS intensities. It

should be noted that a large ablated mass per pulse does not necessarily result in increased LIBS intensity. This fact is clearly seen for RDX and Goma2-ECO, with ablation rates of milligrams per pulse and intensities similar to those of the other explosives. This observation leads to the important conclusion that most of the material ablated is ejected from the sample surface in the form of nonemitting particles. This is also true for the remaining explosives, but there is no clear evidence to extend the arguments to all compounds and additional experimentation is needed. Raman intensities do not apparently correlate with ablation rates. This is to be expected since the Raman signals arise from the intact, nonablated section of the sample that interacts with the laser beam. But recalling the results in Figure 3, it is clear that, for identical pulse energy, larger spots, still resulting in plasma emission, cause increased Raman signals. In other words, Analytical Chemistry, Vol. 82, No. 4, February 15, 2010

1399

conditions that favor Raman emission tend to reduce LIBS intensities and vice versa. CONCLUSIONS A standoff fused Raman-LIBS sensor for energetic materials analysis has been presented. Under well-defined timing conditions simultaneous acquisition of molecular and multielemental spectral data of a substance from the same laser event is demonstrated. This approach exploits the energy gradient created along the target surface by the laser beam to extract vibrational data from the outer part of the interacting surface and atomic information from the ablated mass. Consequently, the information gathered belongs to the same sample area inspected by the laser beam. The system shows promise in situations requiring simultaneous spectral and compositional information from the same spot of a sample surface, for instance, in the standoff analysis of human fingerprints. Further studies on the effectiveness of this fused sensor system for explosive residue detection, its robustness under changing weather conditions (atmospheric

1400


temperature, relative humidity, wind, ...), as well as strategies for reducing the spectral variability are under development. Research is also in progress to develop chemometric tools for exploiting the complementary information of Raman spectroscopy and LIBS for explosive identification purposes, using data fusion strategies. ACKNOWLEDGMENT This research was supported by project CTQ2007-60348 of the Spanish Ministerio de Ciencia e Innovacio´n. Part of the equipment used here has been purchased with funds of the contract DELIBES between Indra Sistemas, S. A. and the University of Malaga. Funding of project OPTIX (reference 218037) of the European Commission is also gratefully acknowledged.

Received for review December 18, 2009. AC902470V

October

30,

2009.

Accepted

Simultaneous Raman Spectroscopy− Laser-Induced Breakdown

Recommend Documents