1. Introduction

The interaction of femtosecond laser pulses with materials is substantially different from that of nanosecond laser pulses, and offers a number of advantages that have been extensively documented for micromachining applications [1]. Improvements in material removal such as increased reproducibility, reduced material deposition around the ablation area, and a smaller heated volume around the ablation region have motivated the initial studies [2,3] that used femtosecond lasers for laser-induced breakdown spectroscopy (LIBS), an analytical technique based on the spectral analysis of optical emission from a laser-induced plasma [4]. For molecular species, ultrafast excitation has the potential to improve the analytical capability of LIBS, since mass spectrometry studies have shown that fragmentation is reduced in ultrafast laser ionization of gas-phase molecules [5] and high mass fragments and clusters are formed by femtosecond laser irradiation of solid-phase species [6]. If optical emission can be obtained from the molecules formed as a result of ultrafast excitation, specificity of detection could be improved.

LIBS has a number of properties that makes it attractive for the detection of explosives. In terms of real-time detection, it takes approximately one second to acquire a broadband spectrum for LIBS with the primary limitation being the read-out time of the spectrometer CCD array [7]. LIBS requires a very small amount of material for analysis and is therefore considered to be nondestructive. Since the plasma formation process and spectral analysis in LIBS are both optical, only optical access to the material is required and LIBS can be performed from large standoff distances. Femtosecond LIBS of metallic samples from a 25 m distance has been demonstrated using a mobile laser system [8].

LIBS has been applied to the analysis of a number of explosives including trinitrotoluene (TNT), cyclotrimethylene-trinitramine (RDX), cyclotetramethylene-tetranitramine (HMX), and pentaerythritol tetranitrate (PETN) [9,10]. The laser sources used to study these materials were Q-switched lasers with pulse widths on the order of 10 ns and a pulse energy of 30 mJ for the first study [9] and 350 mJ for the second one [10]. In these studies, optical emission from the constituent elements of the explosive (C, H, N, and O) provided the basis for the detection of the explosives. By employing algorithms based on the analysis of line intensity ratios, the explosives could be discriminated from other samples. In this paper, we report femtosecond LIBS results for TNT deposited on aluminum substrates. Nanosecond LIBS results for the same samples are also presented for comparison.

2. Experiment

The LIBS apparatus we use in our experiments is shown in Fig. 1 and contains two laser systems that we employ for LIBS studies. In this work, we have primarily used a Ti:Sapphire chirped-pulse amplifier. The amplifier output consists of 150 fs duration pulses with 1 mJ of energy at a repetition rate of 1 kHz. The amplifier is seeded by a mode-locked Ti:Sapphire laser with a pulse width of 100 fs and an average power of 600 mW at a center wavelength of 800 nm. The nanosecond laser is a Q-switched Nd:YAG that has a pulse width of 7 ns, a pulse energy of up to 650 mJ and a wavelength of 1064 nm. The pulse energy was reduced to 90 mJ by adjusting the Q-switch delay of the laser, to bring the fluence level down to typical nanosecond LIBS experimental conditions. The laser outputs were focused with lenses of focal lengths 21.5 cm and 15 cm for the nanosecond and femtosecond beams, respectively, and directed by mirrors towards the sample. The corresponding pulse fluence levels on the sample surface are estimated to be 200 J/cm² for femtosecond excitation and 4300 J/cm2 for nanosecond excitation.

The optical emission from the plasma is collected by a fused-silica lens, coupled into a multimode optical fiber, and delivered to an Echelle-type spectrometer. The spectrometer is operated with a UV dispersion module that covered the wavelength range of 200–785 nm with a resolution of 0.02 nm. The spectra are recorded using a computer interfaced intensified charge-coupled device (ICCD) camera. The image intensifier in the camera is controlled by gating electronics, and a digital delay generator internal to the camera head provides gating of the recorded spectrum with an adjustable delay and width. The laser-induced breakdown event and the recording of the plasma emission spectrum were synchronized using the synchronization and delay generator (SDG) electronics of the femtosecond laser system. For the case of femtosecond excitation, each observation consisted of acquiring 15 single-shot spectra and accumulating those spectra during analysis in order to increase the signal-to-noise ratio. Since the number of photons detected in a single-shot spectrum is low for femtosecond excitation, accumulation of single-shot spectra allows a more complete spectrum to form.

 

Fig. 1. Schematic diagram of the experimental apparatus used to collect LIBS signals. ICCD: intensified charge-coupled device.

Download Full Size | PDF

Samples of explosives were prepared for LIBS by depositing solutions of explosives onto aluminum substrates and allowing the solvent to evaporate. Solutions of TNT in acetone (50 mg/ml) were obtained from a vendor (AccuStandard). The explosive that remained on the substrate after the evaporation of the solvent was melted and recrystallized to form a uniform layer on the substrate. The resulting TNT layer was thin enough to be transparent. Because a single laser shot removed the TNT completely at the central portion of the target spot, the sample was translated in the transverse direction after each laser shot to obtain a fresh spot for analysis.

3. Results and discussion

Figure 2 shows the nanosecond LIBS signal for TNT with single-shot excitation, for detection gate delay and width values of 1 µs and 2 µs, respectively. The four strongest peaks in the spectrum are aluminum emission lines from the substrate with wavelengths of 308.22, 309.28, 394.42, and 396.16 nm. Emission from the elemental constituents of the explosive are observed as follows: carbon peak at 247.86 nm, hydrogen peak at 656.56 nm, nitrogen peak at 747.02 nm, and a double-peaked spectral feature due to oxygen with peak locations at 777.32 and 777.52 nm. The oxygen-related spectral feature is a product of three closely spaced transitions of neutral oxygen at wavelengths of 777.19, 777.42, and 777.54 nm [11].

 

Fig. 2. Nanosecond LIBS signal obtained from TNT on an aluminum substrate. The detection gate delay and width values were 1 µs and 2 µs, respectively.

Download Full Size | PDF

The femtosecond LIBS signal for TNT is shown in Fig. 3 (a) for detection gate delay and width values of 100 ns and 1 µs, respectively. The detection gate delay and width were set to be shorter than the corresponding values for the nanosecond case because continuum emission from femtosecond laser-induced plasmas was observed to decay faster in earlier studies [2,3], and elemental emission lifetimes were also observed to be shorter [3]. Similar to the nanosecond LIBS result for TNT, the two strongest peaks in the spectrum are aluminum emission lines from the substrate at wavelengths of 394.42 nm and 396.16 nm. Two other aluminum emission lines at 308.22 and 309.28 nm are also present, but compared to the nanosecond LIBS result, these lines have smaller intensities relative to the strongest aluminum lines.

Emission from the elemental constituents of the explosive are noticeably absent from the femtosecond LIBS signal. Instead, molecular emission from CN and C₂ is observed as shown in the expanded view of the 370–400 nm wavelength region in Fig. 3 (b) for CN emission. This spectral feature is the product of a sequence of electronic transitions from the second excited electronic level to the ground electronic level (B²Σ→X²Σ) with no change in the vibrational quantum number (Δν=0) [12]. Transitions originating from different vibrational levels in the excited electronic level lead to distinct emission bands in the spectrum. The strongest emission is due to the (0, 0) band originating from the lowest vibrational level in the excited electronic level and is observed at 388.32 nm. The vibrational temperature of the CN species determines the relative peak amplitudes of the various emission bands, and the rotational temperature determines the shape of a given vibrational band, most cleanly observed for the (0, 0) band of the CN spectrum. LIBS studies that have reported CN emission from other samples include nanosecond LIBS measurements on DNT using a pulse energy of 10 mJ at a wavelength of 532 nm, focused by a 7.5 cm focal length lens [13], and femtosecond and nanosecond LIBS measurements on bacteria with pulse energies of 4.5 mJ focused by a 3.0 cm focal length lens [14,15]. The C₂ emission spectrum is also a sequence of electronic transitions (d³Π_g→a³Π_u) with Δν=0, with the strongest emission observed at 516.54 nm. This emission has also been observed in the studies mentioned above [13–15], as well as in one of the two studies on nanosecond LIBS of explosives [10].

 

Fig. 3. Femtosecond LIBS signal obtained from TNT on an aluminum substrate: (a) over the complete detection wavelength range; (b) expanded view of emission from CN. The detection gate delay and width values were 100 ns and 1 µs, respectively.

Download Full Size | PDF

The absence of atomic emission from the elemental constituents of the explosive in the case of femtosecond excitation is consistent with the reduced fragmentation observed in femtosecond laser mass spectrometry experiments on nitroaromatic molecules [5,6,16]. In comparison, the longer timescale of nanosecond excitation is known to result in severe fragmentation of nitroaromatic molecules [16]. As a consequence, we observe emission only from the elemental constituents of the explosive in the case of nanosecond LIBS of TNT.

The role played by the substrate in the interaction is also important to note for both femtosecond and nanosecond excitation. The thin layer of explosive on the substrate has weak linear absorption for both laser wavelengths. A significant portion of the laser energy is absorbed by the aluminum substrate and then transferred to the TNT layer either directly by excited electrons or indirectly by phonons, as described for prototype adsorbate-substrate systems in the literature [17]. We observe that the strongest peaks in both nanosecond and femtosecond LIBS signals are aluminum emission lines.

4. Conclusion

We have presented results for both femtosecond and nanosecond LIBS of TNT on aluminum substrates. The emission for nanosecond LIBS of TNT is characterized by the elemental constituents of the explosive. For the fluences used here, femtosecond LIBS did not yield constituent elemental emission for the explosive, rather emission from CN and C₂ molecules appeared as the marker for the explosive TNT. Aluminum emission lines from the substrate are also observed for both nanosecond and femtosecond excitation indicating the role played by the substrate in the interaction. The results shown demonstrate that femtosecond LIBS is a promising technique for remote detection of explosive residues. Future work with other organic compounds could assess the specificity of detection that can be obtained with femtosecond LIBS.