2.1.

Optical Setup

Figure 1 shows a sketch of the Raman laser and optical setup. A Continuum Sunlite EX-OPO FX-1 laser tunable source provided the 264 nm (10 Hz repetition rate, 3-ns pulse length, $<0.3{\mathrm{cm}}^{-1}$ linewidth) excitation wavelength. The laser beam was relayed by several high UV reflective mirrors to the sample. The backscattered Raman signal was collected by a 12-cm parabolic UV mirror at 45 deg standing $\sim 60\mathrm{cm}$ far from the sample and then sent through an optical fiber ($600\mu \mathrm{m}$ resolution) connected to a Horiba iR550 spectrometer ($2400\text{line}/\mathrm{mm}$ holographic grating, blaze 250 mm, 0.06 nm resolution, CCD $1024\times 256\text{-}\text{pixel}$, maximum resolution 0.01 nm). A 266 nm long-pass filter was placed in front of the fiber to reject the incoming UV laser radiation.

Fig. 1

Raman lab   optical setup (M = mirror, L = lens, PM = parabolic mirror, F = fiber, OF = optical filter).

2.2.

Samples Preparation

Explosive precursors samples were prepared over different colored textile materials: urea (Fig. 2), sodium perchlorate (Fig. 3), ammonium nitrate (Fig. 4), and sodium nitrate (Fig. 5). The preparation of the samples was a challenging task to provide realistic scenarios. A standard drop-cast method [REF] was not suitable because the solution tended to penetrate into textile without leaving all the traces on the surface. Another difficulty was due to the lab configuration: the sample must stand vertically, hence the precursors’ material must sustain gravity during the measurements. For this reason, a new method was employed: standardized amount of 2, 1, 0.5, and 0.25 mg and traces ($\mu \mathrm{g}$ range) for each respective precursor were first weighted and then pressed at 5 tons over colored cotton textiles. The applied method was found successful in both having the precursors’ crystal on the textile surface and in having the sample standing. The sample areas were prepared to be below the upcoming laser beam size (elliptic shape $6\mathrm{mm}\times 3\mathrm{mm}$, 2 mJ per pulse, $14\mathrm{mJ}/{\mathrm{cm}}^2$). All the samples were first analyzed to verify their distribution using an Olympus LEXT OLS 4000 3D microscope with a $50\times$ magnification. In addition, these images were used to quantify the overall amount of the sample over the background surface. Traces were photographed after all tests to observe a photo deterioration of the chemical compound or any possible background photobleaching.

Fig. 2

Brown textile sample of urea at different nominal concentrations (a) traces (microscope image ×50); (b) 0.25 mg; (c) 0.5 mg; (d) 1 mg; and (e) 2 mg.

Fig. 3

Blue textile sample of sodium perchlorate at different nominal concentrations (a) traces (microscope image ×50); (b) 0.25 mg; (c) 0.5 mg; (d) 1 mg; and (e) 2 mg.

Fig. 4

Yellow textile sample of ammonium nitrate at different nominal concentrations (a) traces (microscope image ×50); (b) 0.25 mg; (c) 0.5 mg; (d) 1 mg; and (e) 2 mg.

Fig. 5

Red textile sample of sodium nitrate at different nominal concentrations (a) traces (microscope image ×50); (b) 0.25 mg; (c) 0.5 mg; (d) 1 mg; and (e) 2 mg.

2.3.

Sample Image Analysis

For each precursor measurement, it was possible to quantify the amount probed. An arbitrary threshold was applied using ImageJ2 software¹⁰ to each microscope image to force each pixel to be either black or white as shown in Fig. 6. White indicates the presence of precursor crystal while black indicates its absence. Knowing the size of the picture ($2583\times 2574\mu \mathrm{m}$) and the percentage of white pixels, it was possible to calculate the area covered by each precursor.

Fig. 6

Applied black and white threshold of microscope image 50× magnification of selected images of (a) urea; (b) sodium perchlorate; (c) ammonium nitrate; and (d) sodium nitrate.

Then knowing from literature¹¹ the penetration depth and the respective given density, the effective amount measured was finally computed and summarized in Table 1. The penetration depth at 266 nm was found in literature for urea and ammonium nitrate ($11\mu \mathrm{m}$) while for sodium perchlorate and sodium nitrate the maximum penetration depth was assumed to be the depth crystal measured with the microscope (average $60\mu \mathrm{m}$). From Table 1 It is possible to calculate the sensitivity of this experiment. The minimum effective amount measured on traces for all the tested explosive precursors was found to be: $2.1\mu \mathrm{g}$ for urea, $14.5\mu \mathrm{g}$ for sodium perchlorate, $3.4\mu \mathrm{g}$ for ammonium perchlorate, and $100.7\mu \mathrm{g}$ for sodium nitrate.

Table 1

Effective measured amount in μg for all samples.

3.1.

Raman Spectra

Deep UV Raman spectra of the selected explosive precursors [urea (Fig. 8), sodium perchlorate (Fig. 9), ammonium nitrate (Fig. 10), and sodium nitrate (Fig. 11)] were detected over a remote distance of 60 centimeters. The overall testing time was 3 s for each experimental point (10 integrated nanosecond pulses acquired at 10 Hz, three accumulations). All spectra were background subtracted. The textile backgrounds signals were recorded at the same operating conditions and shown in Fig. 7. It can be noticed that Raman signal was not detected from the colored dyes used to change the fabric color. Therefore, it is not expected any interference due to peak overlap. A little bit of fluorescence is observed for the yellow color textile used (yellow curve). In this case, the observed effect was about 10 counts shift upward on the right between 1200 and $1600{\mathrm{cm}}^{-1}$. However, this effect is neglectable comparing the Raman signal strength so no significant fluorescence interference is expected during the detection of the explosive precursors as well. In addition to the five different amounts of chemicals, a measurement of the bulk compound (labeled as f) is also provided for comparison. The bulk consisted in a $1\mathrm{cm}\times 1\mathrm{cm}$ standard chemical QS-Suprasil cuvette filled with each chemical respectively. Figure 8 shows the Raman spectra of urea. A distinct Raman peak simultaneously identifies several chemicals in a sample even if the spectral signatures seem closer. The strongest Raman vibrational ${\nu }_1$ mode (${\mathrm{NH}}_2\text{-}\mathrm{CO}\text{-}{\mathrm{NH}}_2$ planar) for urea was detected at $1007{\mathrm{cm}}^{-1}$.¹² This spectral line is chosen to be the specific signature for urea since it appears for all the concentrations. Above 0.5 mg, a secondary peak appears on the right-hand side of to the strongest one and it can be distinguished from the background ($1172{\mathrm{cm}}^{-1}$, ${\mathrm{NH}}_2\text{-}\mathrm{CO}\text{-}{\mathrm{NH}}_2$ planar vibrational mode). For concentrations above 1 mg [Figs. 6(d) and 6(e)], three main peaks can be distinguished: $1465{\mathrm{cm}}^{-1}$ (asymmetric NCN stretch), $1536{\mathrm{cm}}^{-1}$ (${\mathrm{NH}}_2$ bend) and the close 1578, and $1646{\mathrm{cm}}^{-1}$ (CO stretch).¹² Figure 9 shows the Raman spectra of sodium perchlorate. A main Raman vibrational ${\nu }_1$ mode (Cl-O symmetric stretch) at $948{\mathrm{cm}}^{-1}$¹³ can be noticed for all the concentrations tested. A secondary double peak (${\nu }_3$ antisymmetric stretch) at 1087 and $1148{\mathrm{cm}}^{-1}$ is distinguishable above 0.5 mg. A weak ${\nu }_4$ (antisymmetric band) peak placed on the left-hand side of the strong ${\nu }_1$ at $630{\mathrm{cm}}^{-1}$ is barely noticeable. Figure 10 shows the Raman spectra of ammonium nitrate. We notice a characteristic ${\nu }_1$ band (symmetric stretch) at $1041{\mathrm{cm}}^{-1}$ for all the samples.⁶ Above 0.5 mg, some secondary vibrational modes appear respectively from left to right at $1285{\mathrm{cm}}^{-1}$ (${\nu }_3$), a double peak at 1408 and $1452{\mathrm{cm}}^{-1}$. In the region between 1400 and $1500{\mathrm{cm}}^{-1}$ those vibrational modes are related to the ${\mathrm{NH}}_4$ deformation and the ${\mathrm{NO}}_3$ stretching.¹² Figure 11 shows the Raman spectra of sodium nitrate. For all the different amount probed, the two vibrational modes at $1064{\mathrm{cm}}^{-1}$ (${\nu }_1$)¹⁰ and $1383{\mathrm{cm}}^{-1}$ (${\nu }_3$) are always visible. The vibrational mode at $1668{\mathrm{cm}}^{-1}$ vanishes at the background noise level; however, both ${\nu }_1$ and ${\nu }_3$ bands can be used to clearly identify the presence of this precursor over the others. The stronger Raman signal recorded for traces was urea, followed by ${\mathrm{NaClO}}_4$, ${\mathrm{NH}}_4{\mathrm{NO}}_3$, and ${\mathrm{NaNO}}_3$. As expected the bulk materials show the strongest signal-to-noise ratio (S/N) and the signal strength keeps decreasing lowering the amount of the chemicals. In all the recorded spectra, a significant interfering fluorescence was not observed and photo degradation was minimized keeping the average laser energy below 2 mJ per pulse. However, despite the laser energy was relatively low, it was not kept under the maximum permissible skin and eye exposure limits of $3\mathrm{mJ}/{\mathrm{cm}}^2$ (MPEs) for a nanosecond laser source.¹⁴ The eye-safe requirement would be an important target if the system will be employed in public spaces and it will be addressed in the near future.

Fig. 7

Background of colored textiles (brow, blue, yellow, and red) recorded for 3 s.

Fig. 8

UV Raman spectra probed from the urea sample of Fig. 2. at different concentration: (a) traces; (b) 0.25 mg; (c) 0.5 mg; (d) 1 mg; (e) 2 mg; and (f) pure.

Fig. 9

UV Raman spectra probed from sodium perchlorate sample of Fig. 2. at different concentration: (a) traces; (b) 0.25 mg; (c) 0.5 mg; (d) 1 mg; (e) 2 mg; and (f) pure.

Fig. 10

UV Raman spectra probed from ammonium nitrate sample of Fig. 3. at different concentration: (a) traces; (b) 0.25 mg; (c) 0.5 mg; (d) 1 mg; (e) 2 mg; and (f) pure.

Fig. 11

UV Raman spectra probed from the sodium nitrate sample of Fig. 4. at different concentration: (a) traces; (b) 0.25 mg; (c) 0.5 mg; (d) 1 mg; (e) 2 mg; and (f) pure.

3.2.

Detection Limits

Figure 12 represents the S/N versus the effective mass amounts previously calculated from each explosive precursor in Table 1. For every sample tested, the S/N was calculated as $\mathrm{S}/\mathrm{N}=2\times H$ signal/h background.¹⁵ The value $h$ background was calculated as the difference of maximum and minimum value in a region without the presence of the Raman signal. Since every explosive precursor has a different Raman signature, it was not possible to choose a unique region to determine h background. For this reason, the range was chosen from 1220 to $1420{\mathrm{cm}}^{-1}$ for both urea and ${\mathrm{NaClO}}_4$, from 80 to $1000{\mathrm{cm}}^{-1}$ for ${\mathrm{NH}}_4{\mathrm{NO}}_3$, and from 1120 to $1320{\mathrm{cm}}^{-1}$ for ${\mathrm{NaNO}}_3$. Figure 12 also shows the limit of detection (LOD) and the limit of quantification (LOQ) respectively defined as three times and 10 times the S/N.¹⁵ It can be noticed that for all the cases S/N is $>3$, so the detection is always possible for all precursors even in the smallest amounts. The LOQ shows that it is possible to quantify the acquired spectra for urea for the lowest masses measured. For the other precursors there are limitation: for sodium perchlorate, quantification is possible above $15\mu \mathrm{g}$; for ammonium nitrate, more than $30\mu \mathrm{g}$ are necessary; and for sodium nitrate, $200\mu \mathrm{g}$ are required due to its low signal intensity.

Fig. 12

Raman S/N over species concentration (a) urea; (b) ammonium nitrate; (c) sodium perchlorate; and (d) sodium nitrate. 3 s acquisition time.