1. Introduction

At the U.S. Army Research Laboratory, we are interested in detecting trace amounts of explosive residues at standoff distances in the field. The ability to detect these residues could indicate the presence of hidden explosive devices. We have been investigating Laser Induced Breakdown Spectroscopy (LIBS) for this purpose[1–3]. LIBS has several attractive features – real-time data collection is possible, no sample preparation is needed, and it can be configured for standoff detection. However, trace explosive residue detection on surfaces presents several challenges for LIBS. The laser pulse interrogates the substrate and the surrounding atmosphere as well as the residue. For explosives, this is especially undesirable because most explosive materials predominantly contain carbon, hydrogen, oxygen, and nitrogen. Atmospheric oxygen and nitrogen entrained in the plasma will contribute to the atomic emission from the oxygen and nitrogen from the explosive sample. In addition, if the substrate is an organic material the carbon and hydrogen will interfere with the atomic emission from the explosive. A potential solution to reduce interference from the substrate and surrounding atmosphere and to minimize imprints is the substitution of femtosecond pulses for the typically used nanosecond pulses.

A femtosecond laser pulse deposits all of its energy into the material before any is transferred to the surrounding lattice [4, 5]. Thus, the material is rapidly converted to ionized gas. Because the energy is deposited on a short time scale, the laser does not interact with the plasma. By contrast, a nanosecond pulse interacts with several transient states of matter. As a result, the use of femtosecond pulses leads to lower ablation thresholds, higher efficiencies of ablation, and ablation craters with less surrounding thermal and mechanical damage (i.e. less heat affected zone) [6, 7]. Many studies have characterized the use of a femtosecond pulse for laser ablation. Because less mechanical and thermal damage occurs, more precise machining can be obtained [8–13]. The advantages of femtosecond pulses for ablation may also be desirable for certain LIBS experiments [14]. Several studies have been performed in order to characterize the properties and performance of femtosecond LIBS. The majority of these experiments have been performed on metallic surfaces: copper, brass, aluminum, etc. [6, 7, 15–18]. Other studies have used femtosecond pulses for LIBS analysis of bacterial samples and organic coatings [19–23]. Two key advantages associated with femtosecond LIBS have been identified: negligible air entrainment and reduced background continuum emission [16, 19, 24]. Both were attributed to the lower temperature of the femtosecond-pulse-induced plasma. For explosives detection, minimal air entrainment and reduced continuum would be beneficial. Initial femtosecond LIBS spectra of explosives have been obtained by Dikmelik et al. from TNT residue using a 1 mJ femtosecond pulse [25]. Two molecular species were observed, CN and C₂, and no atomic emission was observed. In this work, we have expanded on the previously published femtosecond LIBS explosives detection studies by sampling additional types of explosives and investigating the possible advantages of using femtosecond ablation at different laser energies.

2. Experimental

The femtosecond LIBS system is shown schematically in Fig. 1. Briefly, a Ti:Sapphire (Ti:S) oscillator (Coherent, Vitesse) generates a femtosecond pulse that is used to seed the Ti:S amplifier (Coherent, Hidra-25). A Nd:YLF pump laser (Coherent, Evolution-15) is used to amplify the output energy of the femtosecond laser pulse. The output of the Ti:S amplifier is a train of 800 nm, 1 mJ, <120 femtosecond pulses at a 1 kHz rep rate. An additional Nd:YAG laser (Continuum, Powerlite Precision II 8000) can be used to further amplify the femtosecond pulse energy, generating a pulse train of 20mJ femtosecond pulses at a 10 Hz rep rate. The femtosecond laser pulses pass through a convex focusing lens (100 mm f.l.) and are reflected onto the sample using a dichroic mirror.

Nanosecond pulses were generated by a Nd:YAG (Big Sky, CFR400) laser at 1064nm. The pulses pass through a convex focusing lens (100 mm f.l.) and the same dichroic mirror that reflects the 800 nm femtosecond pulses. The two pulse trains are overlapped at the dichroic mirror so they are collinear. The LIBS plasmas are formed at the same target spatial location for each laser. A pierced parabolic mirror collects the plasma emission and focuses the light onto a 600 μm fiber optic. The light is delivered into an echelle spectrometer (Catalina Scientific, SE200) or a Czerny Turner spectrometer (Andor, SR-163 Series) through a 25 micron pinhole or a 50 micron slit, respectively. An ICCD detector (Apogee Alta or Andor iStar) was used to record the LIBS spectra. The spectra from the Czerny Turner spectrometer were collected with a 300 ns gate delay and a 5 μs gate width and cover the wavelength range from 230–410 nm. The spectra collected from the echelle spectrometer cover the wavelength range from 200–850 nm. For the echelle ICCD, the gate delay and gate duration were 50 ns and 10 μs, respectively. Bulk explosive samples of RDX (Cyclotrimethylenetrinitramine, C₃H₆N₆O₆), C-4 (91% RDX, 9% plasticizer and binder), and Composition-B (36% TNT, 63% RDX, and 1% wax) were provided by colleagues at the U.S. Army Research Laboratory. We prepared explosive residue samples by applying a small amount (<1 mg) to the aluminum foil substrate and crushing/spreading it with a Teflon® block over ~25 cm². Excess RDX was knocked off the substrate so only residue that adhered to the metal substrate was left behind. No quantification was attempted for this work.

 

Fig. 1. Femtosecond laser induced breakdown spectroscopy experimental setup

Download Full Size | PDF

3. Results

We measured LIBS spectra from several bulk explosives (RDX, C-4, and Composition-B) and blank aluminum foil using 10mJ femtosecond pulses with a laser fluence of ~30 J/cm². Twenty spectra were recorded for each sample using the echelle spectrometer. The average spectrum of each explosive is shown in Fig. 2. In each spectrum from the explosives, we observed the constituent elements – carbon, hydrogen, nitrogen and oxygen. In addition, the molecular CN is observed, presumably formed via recombination of carbon and nitrogen from the sample and surrounding atmosphere. We also observe continuum emission in each spectrum, despite the 50 ns gate delay. The periodic pattern of the continuum is due to the sensitivity of the ICCD for each segment of the spectrum dispersed by the echelle optical path. Oxygen and nitrogen atomic emission are also observed in the aluminum spectrum due to the surrounding atmosphere.

We collected LIBS spectra from a residue of RDX on an aluminum substrate using the Czerny Turner spectrometer. For each individual spectrum, a fresh surface of the residue was presented to the laser pulse, since the first shot would consume the thin layer of explosive. The laser energy and fluence for the femtosecond and nanosecond pulses were both 10 mJ and ~30 J/cm². For the nanosecond pulse and the femtosecond pulse, we collected and averaged twenty LIBS spectra each. A comparison of the spectra is shown in Fig. 3. A small imprint (<0.5 mm) is observed in the substrate for each of the two pulse durations. The presence of the explosive residue is indicated by the carbon atomic line at ~248 nm and the CN molecular fragment at ~388 nm. The plasma from the femtosecond pulse produces larger carbon atomic emission intensity. There is minimal (if any) carbon emission from the nanosecond pulse. By contrast, the neutral aluminum emission intensity is greater for the plasma from the nanosecond pulse, even saturating the detector at ~308 and ~396 nm. The emission intensities from the CN molecular are similar for both pulse durations.

 

Fig. 2. Femtosecond LIBS spectra of bulk explosives collected using an echelle spectrometer Note the intensity axis is different for aluminum. The laser line at 800 nm is observed in the explosive spectra.

Download Full Size | PDF

 

Fig. 3. LIBS spectra of RDX residue on aluminum substrate collected from a Czerny turner spectrometer using a nanosecond pulse (black) and femtosecond pulse (red) both with a 30 J/cm² fluence. Inset shows expansion of the carbon atomic emission peak portion of the spectra.

Download Full Size | PDF

Next, we reduced the energy of the femtosecond laser. A plasma could be generated with a 3.2 J/cm² fluence using a 1 mJ femtosecond pulse with our current experimental setup. By contrast, a nanosecond pulse less than 4 mJ (a fluence of 12 J/cm²) will not generate a plasma on an explosive residue with our current experimental setup. We collected low fluence femtosecond LIBS spectra of RDX residue and a blank piece of aluminum, shown in Fig. 4. The imprint in the aluminum created by the plasma is not observable by eye. The only evidence of RDX residue in the spectrum is the CN molecular fragment, which is not present in the blank aluminum spectrum. The continuum emission is less than that shown in Fig. 3, however, the carbon peak at ~248 nm is too weak to observe. The aluminum atomic emission peaks are smaller compared to the blank aluminum foil due to the presence of the explosive residue on the surface of the aluminum.

 

Fig. 4. Femtosecond LIBS spectra collected from a Czerny Tuner spectrometer of RDX residue on aluminum (red) and blank aluminum (black) with a 3.2 J/cm² laser fluence.

Download Full Size | PDF

4. Discussion

By using higher energy femtosecond pulses, we were able to observe the constituent atomic elements present in several explosive samples, in both bulk and residue forms. We also compared the LIBS spectra of RDX residue obtained from plasmas generated by femtosecond and nanosecond pulses with comparable laser fluences. The carbon atomic emission intensity relative to the aluminum atomic emission intensity is greater for the femtosecond spectrum. It is possible this is indicative of more residue being sampled relative to the substrate, but it could also be a result of the differences in temperature between the nanosecond (5800±700 K) and femtosecond (8500±660 K) plasmas. We calculated the temperatures using the Boltzmann equation and Al I atomic emission lines present in Fig. 3.

As mentioned earlier, advantages attributed to the femtosecond induced plasma that could improve the detection of explosive residue include minimized background continuum and negligible atmospheric entrainment. However, at the higher laser fluence used in Fig. 2, we observe continuum background emission (even with a 50 ns gate delay) and oxygen and nitrogen atomic emission originating from atmospheric entrainment in the blank aluminum spectrum. The laser energy and fluence we used in these experiments is larger than the experiments described in Ref [19], our gate delay is shorter (50 ns vs 100 ns) and we are using a metallic substrate instead of an organic substrate. These factors could all lead to the observation in Ref [19] of less continuum emission and air entrainment. It has been demonstrated that the continuum intensity (duration) and temperature are more dependent on laser energy and optimal gate timing than pulse duration [17, 26]. For the experimental parameters used in Fig. 2, gateless LIBS would be hard to achieve. As laser fluence is increased for a femtosecond laser pulse, the advantages of femtosecond pulses are diminished for explosive residue sampling. In addition, for ablation applications where spatial resolution is important, higher energy femtosecond pulses could lead to structural and heat damage similar to that observed when longer pulse durations are used for ablation [8, 12, 27, 28].

The use of femtosecond pulses does allow the generation of plasmas at lower laser fluences. Thus, advantages for explosive residue detection could be realized due to the lower laser fluence – less continuum background, less air entrainment, and less substrate interrogation (as observed by comparing the imprint left by the 10 mJ femtosecond pulse with the 1 mJ femtosecond pulse imprint). It is also beneficial from a cost analysis point of view since compact fiber laser systems can be used to generate low energy femtosecond pulses instead of more costly bench top laser systems [18]. But, by using low energy pulses, spectral emission intensity is sacrificed, as evidenced by the spectra generated by 1 mJ femtosecond pulses in Fig. 4. As observed in previous work using 1 mJ femtosecond pulses, only molecular species were evident [25].

Future work will involve several steps to determine if femtosecond pulses provide any significant advantage over nanosecond pulses for explosive residue discrimination. We have demonstrated here, however, that any advantages over conventional nanosecond LIBS are only likely to be realized at the very low fluences possible with femtosecond pulses. The efficiency of light collection, the light throughput to the detector, and the detector efficiency must be optimized in order to collect the low signal generated by small laser fluences. The optimal laser fluence must be found that maximizes the advantages realized by low energy femtosecond pulses but minimizes the sacrifice of signal. Finally, the discrimination between explosive materials and background material at the optimal femtosecond laser fluence must be compared to nanosecond LIBS.