Sampling and Comparison of Extraction Techniques Coupled with Gas Chromatography–Mass Spectrometry (GC-MS) for the Analysis of Substrates Exposed to Explosives

Abstract

Explosive-detecting canines (EDCs) show high sensitivity in detecting explosives that they are trained to detect. The ability of canines to detect explosive residues to the parts per trillion level can sometimes result in nuisance alerts. These nuisance alerts can occur when various materials (i.e., substrates) are exposed to volatile organic compounds (VOCs) present in explosive mixtures, leading to contamination—the unintended absorption or adsorption of VOCs by the substrate. Chemical constituents such as taggant, plasticizer, and residual solvent in explosives are often composed of VOCs that canines are trained on to detect explosives. Composition C-4 (C4) is a common explosive that EDCs are trained to detect and hence is this study’s focus. Common VOCs of interest emitted from C4 include 2,3-dimethyl-2,3-dinitrobutane (DMNB), 2-ethyl-1 hexanol (2E1H), and cyclohexanone. In this study, we developed a protocol for comparing different substrates such as cotton, cardboard, wood, sheet metal, and glass that were exposed to volatiles from C4. 1-bromooctane (1-BO) was used as a single-odor compound to compare the complex odor originating from C4. Triplicates of substrates such as cotton, wood, cardboard, sheet metal, and glass were exposed to 1 g of C4 in a paint can for one week and the substrates were then extracted using various extraction methods such as liquid injection, direct SPME, and headspace analysis coupled with gas chromatography–mass spectrometry. An extraction time study was performed to determine the optimal extraction time for SPME analysis, and it was found to be 20 min. Comparison of extraction methods revealed that SPME surpassed other techniques as DMNB was found on all substrates using SPME. It was observed that porous substrates such as wood and cardboard have a higher retention capacity for volatiles in comparison to non-porous substrates such as sheet metal and glass. Finally, swabbing was evaluated as a sampling technique for the substrates of interest and the extracts were analyzed using the total vaporization–solid phase microextraction (TV-SPME) technique. No volatiles associated with C4 were identified on conducting a GC-MS analysis, suggesting that swabbing is not an ideal technique for analysis of substrates exposed to C4.

1. Introduction

Humans have relied on canines for over 12,000 years for their ability to detect odors, using them primarily as hunting companions [1]. Canines have since demonstrated high sensitivity and selectivity for detection of contraband such as drugs and explosives [2,3,4,5]. Explosive-detecting canines (EDCs) are routinely used for real-time detection of explosives as they can reliably identify chemical odors emitting from explosive mixtures. Their ability to be trained to identify new target odors, portability, and ease of handling make them an ideal detector of explosives [6,7]. EDCs alert based on the odor of volatile organic compounds (VOCs) present in an explosive mixture [8].

VOCs such as taggants are added to the explosives at the time of manufacture to facilitate their detection and identification. Taggants are energetic volatile organic compounds such as 2,3-dimethyl-2,3-dinitrobutane (DMNB), 2-nitrotoluene, 4-nitrotoluene, and ethylene glycol dinitrate (EGDN) that serve as means of identification of explosives for both canines and analytical instruments [9]. Due to the high volatility of taggants and other VOCs in explosives and high sensitivity shown by EDCs, invisible residues of explosives can also be detected which can lead to “nuisance alerts”. Additionally, training aids and other substrates that may be exposed to VOCs emitting from explosives during various stages such as manufacturing, transport, and storage, run the risk of becoming cross-contaminated [6]. This cross-contamination can lead to an increase in the instances of nuisance alerts which can be time and resource consuming. Therefore, it is important to study the extent of contamination that various substrates may exhibit on being exposed to explosives.

Most explosives have a complex odor due to the presence of numerous VOCs present in an explosive mixture. Composition C-4 (C4) is a plastic explosive commonly used for training canines and contains VOCs such as DMNB, 2-ethyl-1-hexanol (2E1H), and cyclohexanone [10,11]. The volatiles emitted by C4 can adsorb on various surfaces and hence there is a need to devise optimal sampling techniques and extraction procedures for determination of contamination on these surfaces. For that reason, this study will focus on a range of substrates exposed to C4. A substrate can be described as a surface where chemical processes such as adsorption and absorption occur. In parallel, 1-bromooctane (1-BO) will also be studied as a single-odor compound due to its growing application as a universal detector calibrant (UDC) for EDCs in recent times [12,13].

Previous studies have focused on development and optimization of various highly sensitive analytical techniques for the analysis of explosive mixtures [11,14,15,16,17,18]. Among these analytical techniques, gas chromatography coupled with mass spectroscopy (GC-MS) is a recognized gold standard for analysis of VOCs with high sensitivity and selectivity [19]. Ref. [20] compared different modes of ionization, namely, electron ionization (EI), negative ion chemical ionization (NICI), and positive ion chemical ionization (PICI) detection methods for detection of organic explosives such as dinitrotoluene (DNT), trinitrotoluene (TNT), and 1,3,5-trinitro-1,3,5-triazine (RDX) and concluded that detection limits are lower by NICI for all explosives tested, with the exception of RDX [10]. Lai et al. in 2010 employed SPME-GC/MS and detection by ion mobility spectrometry (IMS) for detection of VOCs in plastic explosives. The limit of detection (LOD) for target compounds such as n-butyl acetate and DMNB ranged from 1.5–2.5 ng [20]. Rodriguez and Almirall in 2021 used portable GC-MS coupled with capillary microextraction of volatiles (CMV) for continuous vapor sampling of VOCs such as 3-nitrotoluene (3-NT) and 2,4-dinitrotoluene (2,4-DNT) associated with explosives. They were able to achieve rapid sampling and preconcentration of sub-ng levels of volatiles in field scenarios using a 10 min GC-MS method comparable with benchtop instruments [21]. Calabrese et al. (2024) utilized a quartz crystal microbalance and HS-SPME-GC-MS/ECD for analysis of substrates such as polystyrene, cardboard, steel, and glass exposed to trinitrotoluene (TNT), triacetone tri peroxide (TATP), cocaine, and hexamethylene triperoxide diamine (HMTD). They observed that volatiles were retained the most by cardboard followed by polystyrene and glass [22]. Additionally, a multitude of previous studies have analyzed the elemental composition and structure of substrates using scanning electron microscopy with energy-dispersive X-ray spectroscopy (SEM-EDX) which can play a role in the sorption process for different analytes [23,24,25,26].

Despite these advancements, the effectiveness of GC-MS for explosive residue detection also relies on the storage conditions, sampling techniques, and extraction parameters employed prior to the instrumental analysis [27]. Therefore, there is an imperative need for evaluating sampling techniques and establishing the most optimized GC-MS methods for detection of VOCs associated with explosives.

One of the most frequently used sampling techniques involves swabbing of the substrate to collect explosive residues, particularly if the substrate is large and cannot be moved [19,27]. The key parameters involved in the swabbing method are the type of substrate, extraction solvent, and extraction time needed to obtain maximum recovery of VOCs. Previous studies have investigated a range of these parameters, and the results have differed depending upon the nature of the substrate and the analyte of interest [28,29,30,31,32,33,34].

Various extraction methods such as liquid injection, headspace analysis, and different solid phase microextraction (SPME) techniques such as total vaporization SPME (TV-SPME) and headspace SPME (HS-SPME) are used for VOCs associated with explosives. While each extraction method has its merits, its suitability for use depends upon numerous factors such as the sample matrix, volatility of the analytes, sample concentration, need for extraction solvent and possibility of field sampling to name a few.

Currently, there are no established GC-MS methods for analysis of substrates exposed to volatiles emitting from C4. Substrates like cotton, wood, cardboard, metal, and glass have unique surface properties that enable them to bind or retain other substances. The research described in this paper focuses on the sampling and comparison of extraction techniques coupled with GC-MS for the analysis of substrates exposed to this explosive. Given the widespread presence of substrates such as cotton, glass, sheet metal, cardboard, and wood, these substrates are of interest for this study.

2. Materials and Method

2.1. Chemicals and Materials

1-Bromooctane (1-BO) was purchased from Thermo Fisher Scientific (Waltham, MA, USA). Composition C-4 was obtained from Omni Explosives (Marion, AR, USA) using a Bureau of Alcohol, Tobacco, Firearms, and Explosives (ATF) A user of explosives license was issued to the corresponding author. Chloroform and 1,4-dinitrobenzene were purchased from Fisher Scientific. All chemicals had a purity of 99%. Unlined paint cans of 32 oz capacity were purchased from Grainger (Lake Forest, IL, USA).

A range of substrates were used in this study. Glass substrates having dimensions of 3.5 cm × 0.5 cm were cut out from glass slides obtained from Fisher Scientific. Cardboard substrates were cut into pieces of dimensions of 3.9 cm × 1.9 cm for this study. Cotton balls (100% cotton) purchased from Meijer (Indianapolis, IN, USA) were used as cotton substrates. Wood substrates having dimensions of 4.2 cm × 1.3 cm were cut out from craft sticks manufactured by Horizon USA (New Providence, NJ, USA). Sheet metal substrates having dimensions of 3.3 cm × 1.9 cm were used in this study. All inlet liners were purchased from Restek (Centre County, PA, USA). Cotton tip applicators were purchased from Puritan (Guilford, ME, USA).

2.2. Sample Preparation

Triplicate samples of each substrate were hung inside separate paint cans using a thread and secured by Scotch tape on the outside as shown in Figure 1. Triplicates of each substrate were exposed to approximately 1 g of C4 in one paint can and 320 µL of 1-BO in another paint can for liquid injection experiments. The amount of 1-BOw was reduced to 250 µL for all other extraction methods due to carry over. These paint cans were set aside for one week. This preparation was used for all analysis methods detailed in Section 2.2.1, Section 2.2.2, Section 2.2.3, Section 2.2.4 and Section 2.2.5.

2.2.1. Solvent Extraction of the Substrates

The substrates were extracted using 1 mL of 100 ppm 1,4-dinitrobenzene in chloroform, which was used as an internal standard. The substrates in the internal standard solution were vortexed for five minutes and 700 μL of the extract was placed into 2 mL autosampler vials with inserts.

2.2.2. Swabbing of the Substrates

To evaluate the role of swabbing as a technique for VOC collection for C4, cotton tip applicators were used as demonstrated by Sauzier et al. in 2016 [28]. The cotton tip applicators were briefly immersed in isopropanol, which was allowed to evaporate for 20 s. The applicators were then rubbed back and forth on both sides of the substrate for 10 s. Wood, cardboard, sheet metal, and glass were chosen as substrates for this part of the study. The tips were then placed in 12 mL screw-top vials and extraction was performed using 1 mL of acetone. All vials were placed on a VWR Standard Analog Shaker Table (Illinois, United States) at speed 3 for 20 min. Finally, 60 µL of the extract was transferred to a 20 mL headspace vial for TV-SPME analysis. The TV-SPME equation shown below was used to determine the volume of extract placed in the SPME vial:

$$Vs= \left({\displaystyle \frac{\left(10A-\frac{B}{T+C}\right)}{RT}}V\right)\left({\displaystyle \frac{M}{\rho }}\right)$$

In this equation, Vs is the volume of the liquid extract, A, B, and C are the Antoine constants for the solvent, R is the ideal gas constant (L bar/K mol), T is the temperature (K), V is the volume of the vial (mL), M is the molar mass of the solvent (g/mol), and ρ is the density of the solvent (g/mL).

2.2.3. Direct SPME

The substrates were removed from the paint cans using forceps and placed in a 20 mL SPME vial for SPME analysis.

2.2.4. Headspace

The substrates were removed from the paint cans using forceps and placed in a 20 mL headspace vial for headspace analysis.

2.2.5. Change in Weight of the Substrate

Adsorption of volatiles on the substrates can result in deposition of invisible residue on the substrates. To test whether there is any change in the weight of substrates before and after exposure to C4, carboard was selected as it showed high retention for all volatiles of interest. Triplicates of cardboard substrates were weighed and then set up as mentioned in Section 2.2 for one week. They were weighed again after exposure using a Mettler Toledo analytical balance and a paired t-test was performed to determine any notable change in their weight.

2.3. GC-MS Parameters

An Agilent 6890N GC coupled to an Agilent 5975 Inert Mass Selective Detector (Santa Clara, CA, USA) was used for all experiments. Hydrogen was used as a carrier gas and a Gerstel MultiPurpose Sampler (MPS) was used as an autosampler. An Agilent HP-MS Ultra Inert 30m Column (Santa Clara, CA, USA) having an inner diameter of 0.250 mm, and a 0.25 μm film thickness was employed for all experiments. All experiments were run in electron ionization (EI) mode with an ionization energy of 70 electron volts (eV).

2.3.1. Liquid Injection

Liquid injection was performed using a straight inlet liner of 4.00 mm inner diameter. A chloroform blank was run in between each sample. First, 1 µL of the sample was injected into the gas chromatograph, and the flow rate was set at 2.5 mL/min. The initial oven temperature was set at 30 °C, held for 1 min, followed by 15 °C/min until the final temperature of 250 °C was achieved. All substrates were run in splitless mode and electron ionization (EI) was used as the mode of ionization. A scan range of m/z 40–m/z 400 was used, with a solvent delay of 1.50 min.

2.3.2. Total Vaporization–Solid Phase Microextraction (TV-SPME) of the Swabs

The vials were incubated at 60 °C for 2 min and the extraction time was 20 min. The pre-fiber-bakeout was performed for 7 min and post-fiber-bakeout was for 5 min. The inlet temperature was set at 220 °C in splitless mode. The initial oven temperature was set at 60 °C and held for one minute, followed by a ramp of 20 °C/min to a final temperature of 300 °C where it was held for one minute. The transfer line was set at 250 °C and a scan range of m/z 40–m/z 400 was used, with no solvent delay.

2.3.3. Solid Phase Microextraction

SPME analysis was performed using a tapered inlet liner of 2.00 mm inner diameter. A 65 μm polydimethylsiloxane/divinylbenzene (PDMS/DVB) StableFlex/ss SPME fiber was used for extraction.

The vials were incubated for 2 min at 60 °C while agitated for 15 s on, 5 s off. The sample extraction times were 1, 5, 7, 10, 15, and 20 min and the desorption time was 60 s. The inlet temperature was set to 220 °C and was operated with a 20:1 split ratio. The initial oven temperature was 60 °C and was held for 1 min, then the temperature was ramped at 20° C/min to 300 °C where it was held for 1 min. The transfer line was set to 250 °C. The quadrupoles were kept at 150 °C. A scan range of m/z 40–m/z 400 was used, with no solvent delay. One air blank and a 20 min SPME fiber bakeout were completed between each sample.

2.3.4. Headspace Analysis

Headspace analysis was performed using a tapered inlet liner of 4.00 mm inner diameter and a PAL NWH-02-01F 2500 uL headspace syringe (Santa Clara, CA, USA) was used for analysis.

The syringe was held at 65 °C and the injection volume was set at 1000 µL. The vials were incubated for 2 min at 60 °C while agitated for 15 s on, 5 s off. The inlet temperature was set to 250 °C and was operated in splitless mode. The initial oven temperature was 60 °C and was held for 1 min, then the temperature was ramped at 10 °C/min to 230 °C where it was held for 1 min. The transfer line was set to 250 °C. The quadrupoles were kept at 150 °C. A scan range of m/z 40–m/z 400 was used, with no solvent delay. One air blank was run, and the headspace syringe was purged with nitrogen for 120 s between each sample.

3. Results and Discussion

Triplicates of each substrate were analyzed using the methods mentioned above. Mean and standard deviation of the absolute peak areas obtained for all analytes of interest present in the exposed substrate were calculated and plotted in the charts shown in this section. A comparison between all substrates was made based on the peak areas obtained after analysis.

3.1. Liquid Injection

Results obtained from this extraction method for all substrates studied are shown in Figure 2. The peak areas obtained of the standard volatile 1-BO and volatiles emitted by C4 such as DMNB, cyclohexanone, and 2E1H suggest that volatiles are retained distinctly by each substrate.

Cardboard showed the largest peak areas for DMNB and 1-BO which suggests that it has an excellent retention capacity for these volatiles. All substrates retained DMNB from C4 except glass. The largest peak area for cyclohexanone was obtained for wood which also showed good retention of DMNB and 1-BO. No peak for cyclohexanone was found for cotton and glass. 2E1H was not found to be extracted from any of the substrates studied using liquid injection. All substrates exhibited large peak areas for the standard odor compound, 1-BO. These results suggest that porous substrates such as cardboard, cotton and wood have higher retaining capacity for volatiles than non-porous substrates such as sheet metal and glass. In contrast to non-porous substrates that have a single-layer structure, porous substrates composed of cellulose such as cotton, cardboard, and wood have a layered structure that increases the available surface area and thereby bring about an increase in sorption of volatiles.

3.2. TV-SPME of Swabs

Swabbing was not found to be effective as a sampling technique for all the substrates exposed to C4 in this study. No peaks for any volatiles emitted from C4 such as DMNB, cyclohexanone and 2E1H were found on swabbing the substrates and extracting them using acetone for TV-SPME analysis. However, 1-BO was present on all substrates as shown in Figure 3. The recovery of 1-BO was highest for wood followed by cardboard, sheet metal, and glass. The large error bar present for wood can be attributed to carry over during the run due to the highly volatile nature of 1-BO.

3.3. SPME

The extraction time study for the SPME fiber, using C4 samples as the primary explosive material, was conducted across various substrates, including cotton, wood, metal, glass, and cardboard. Figure 4 highlights the extraction results for cotton, with data for other substrates available in the Supplementary Materials. A 20 min extraction time was found to be optimal across all substrates. The SPME fiber exhibited logarithmic trendlines for the extraction of DMNB, 2E1H, and cyclohexanone, which are the expected volatiles of C4. The logarithmic trendline for DMNB suggests a rapid initial uptake by the SPME fiber, slowing as it approaches saturation, a behavior consistent with SPME fibers. Although 2E1H and cyclohexanone also followed logarithmic trendlines, their concentrations were much lower due to their lower volatility compared to DMNB. These differences underscore the importance of optimizing extraction times to ensure reliable results, leading to the adoption of a 20 min extraction time for all subsequent SPME testing to ensure consistent analyte recovery across different sample matrices.

The extraction efficiency of volatiles from C4 was evaluated across substrates—cardboard, wood, cotton, glass, and metal—using 20 min SPME extraction times. Figure 5 compares peak areas of 1-BO with the C4 volatiles DMNB, cyclohexanone, and 2E1H. Cardboard showed the highest retention and release of volatiles, followed by wood, with cotton retaining significant DMNB and 2E1H but less cyclohexanone. The metal had large peak areas for 1-BO and DMNB but showed no 2E1H or cyclohexanone, while glass had the lowest retention overall. 1-BO consistently produced large peak areas, making it a reliable standard. DMNB was detected across all substrates, confirming its role as a key C4 volatile, whereas 2E1H and cyclohexanone were less consistent, particularly on non-porous metal and glass. The findings suggest that porous substrates like cardboard, wood, and cotton are more effective in retaining and releasing volatile compounds compared to non-porous substrates like glass and metal.

3.4. Change in Weight of the Substrate

On measuring the weight of cardboard substrate before and after exposure to C4, it was observed that the weight of the substrate increased. Cardboard was chosen for this study as it demonstrated the largest peak areas for the volatiles of interest. However, on performing paired t-tests for three cardboard substrates, p-values of 0.071, 0.055, and 0.023 were obtained. Thus, only one of the substrates can be confirmed as having increased in mass at the 95% confidence interval. Our results align closely with the findings of Calabrese et al. (2024), who observed a higher sorption of analytes on porous substrates like cardboard and polystyrene, as measured using a quartz crystal microbalance [22].

4. Conclusions

This study sheds light on the commonly found substrates that can act as nuisance alerts when exposed to C4. A comparison of various extraction methods was performed to achieve the best extraction efficiency for all substrates and a reference odor compound, 1-BO, was used in all experiments. An extraction time study was conducted to determine the optimal extraction time for SPME methods, and it was found that 20 min is ideal for extraction of volatiles emitted from C4.

In general, SPME outperforms liquid injection for all substrates (see Figure 6). SPME shows larger peak areas for DMNB on cardboard, wood, cotton, glass, and sheet metal, indicating that SPME is more effective at detecting DMNB on these substrates. The difference in its sensitivity is especially notable for glass as liquid injection failed to detect DMNB for it but SPME was able to register a clear peak. The only exception is sheet metal, where there is no significant difference between the peak areas for SPME and LI. No peaks for 2E1H were observed using liquid injection for any substrate. The larger peak areas obtained for 1-BO across all substrates for liquid injection in comparison to SPME can be attributed to the greater volatility of 1-BO. Hence, it can be concluded that SPME is a superior extraction method for detecting volatiles emitting from C4. Our results are consistent with findings by Calabrese et al. (2024) that the amount of sorption varies based on the substrate and cardboard retained the highest amount of volatiles out of all substrates tested in this study [22].

In addition, swabbing was tested as a sampling technique for the exposed substrates and no peaks for volatiles from C4 were found. Peaks for 1-BO were found for all substrates, indicating that it is not an ideal sampling method for these substrates. Future studies can be conducted on optimization of a sampling technique for substrates of interest since no method currently exists for collection of residues from large sample sizes.