Anal Bioanal Chem DOI 10.1007/s00216-013-7410-3

RESEARCH PAPER

High-efficiency headspace sampling of volatile organic compounds in explosives using capillary microextraction of volatiles (CMV) coupled to gas chromatography–mass spectrometry (GC-MS) Wen Fan & José Almirall

Received: 1 August 2013 / Revised: 27 September 2013 / Accepted: 30 September 2013 # Springer-Verlag Berlin Heidelberg 2013

Abstract A novel geometry configuration based on sorbentcoated glass microfibers packed within a glass capillary is used to sample volatile organic compounds, dynamically, in the headspace of an open system or in a partially open system to achieve quantitative extraction of the available volatiles of explosives with negligible breakthrough. Air is sampled through the newly developed sorbent-packed 2 cm long, 2 mm diameter capillary microextraction of volatiles (CMV) and subsequently introduced into a commercially available thermal desorption probe fitted directly into a GC injection port. A sorbent coating surface area of ∼5×10−2 m2 or 5,000 times greater than that of a single solid-phase microextraction (SPME) fiber allows for fast (30 s), flow-through sampling of relatively large volumes using sampling flow rates of ∼1.5 L/min. A direct comparison of the new CMV extraction to a static (equilibrium) SPME extraction of the same headspace sample yields a 30 times improvement in sensitivity for the CMV when sampling nitroglycerine (NG), 2,4-dinitrotoluene (2,4-DNT), and diphenylamine (DPA) in a mixture containing a total mass of 500 ng of each analyte, when spiked into a litervolume container. Calibration curves were established for all compounds studied, and the recovery was determined to be ∼1 % or better after only 1 min of sampling time. Quantitative analysis is also possible using this extraction technique when the sampling temperature, flow rate, and time are kept constant between calibration curves and the sample.

Published in the topical collection Microextraction Techniques with guest editors Miguel Valcárcel Cases, Soledad Cárdenas Aranzana and Rafael Lucena Rodríguez. W. Fan : J. Almirall (*) Department of Chemistry and Biochemistry and International Forensic Research Institute, Florida International University, 11200 SW 8th St. OE 116A, Miami, FL 33199, USA e-mail: [email protected]

Keywords Capillary microextraction of volatiles (CMV) . Gas chromatography–mass spectrometry (GC-MS) . Dynamic headspace sampling . Explosives

Introduction Headspace analysis of volatiles has enjoyed wide applicability as an analytical technique for its simplicity and general use for sampling gas and liquid and/or solid matrices while requiring no solvent during the sample preparation [1, 2]. While direct headspace sampling can lead to low sensitivity, depending on the polarity or volatility of the analytes [3], preconcentration techniques such as purge and trap (PT) [4], solid-phase microextraction (SPME) [1], and, more recently, planar solid-phase microextraction (PSPME) [5, 6] have been found to provide adequate sensitivity for many applications. The equilibrium-based SPME method has been applied widely after its invention in the early 1990s [7–10] because of the improvements in sensitivity and the potential for quantitative analysis. The SPME sampling fiber, however, has a relatively small surface area and phase volume [10] and is best used for static sampling of gas-phase volatiles in a closed system or from aqueous solutions under equilibrium conditions. Although modifications can be made to achieve dynamic headspace sampling with the use of an inside needle capillary adsorption trap (INCAT) [11] or solid-phase dynamic extraction (SPDE) [12], the sampling volume is limited to the shape and internal surface area of the syringe. A purge and trap dynamic headspace sampling method uses Tenax or other solid sorbent to trap headspace compounds when a flow of purge gas collects the volatiles from liquids or solids [4]. The purge and trap setup is relatively complicated, normally requiring an external thermal desorption unit and the adsorption of water on the trap may degrade the column and damage the MS

W. Fan, J. Almirall

detectors [13]. In addition, the high desorption flow rates require cryogenic focusing to transfer a tight plug to the head of a chromatographic column [14]. A new design for dynamic sampling using a planar geometry, coined a PSPME device, permits for dynamic sampling and preconcentration that has been applied for the analysis of both explosives and drugs with excellent sensitivity and extraction efficiency [6] due to the increased surface area [7]. The PSPME design and geometry has been limited to coupling to the thermal desorption inlets of ion mobility spectrometers (IMS), which limit the chromatographic resolution, compound identification and selectivity, quantitative analysis, as well as other limitations normally associated with IMS detection such as low dynamic range and potential for detector saturation. In this study, we report, for the first time, a novel capillary microextraction of volatiles (CMV) configuration that can be used for dynamic headspace analysis and coupled to gas chromatography mass spectrometry (GC-MS). The microextraction device has a significantly increased surface area and phase volume in comparison to the widely used SPME extraction and benefits from a small enough size that it can fit into a thermal desorption probe that can be placed in a GC injector. The CMV preconcentration device has its both ends open, allowing the use of a vacuum pump to allow air flow through the device for dynamic sampling, resulting in excellent recovery with only a 1-min extraction time in comparison to a 30min static extraction using the traditional SPME technique. Moreover, a calibration curve can be constructed to achieve quantitative analysis by keeping the extraction parameters constant between the calibration curve conditions and the unknown sample. We report the preparation of the CMV devices for the first time, and we also report the sampling/preconcentration performance of the CMV when sampling spiked standards that contain the volatile compounds nitroglycerine (NG), 2,4dinitrotoluene (2,4-DNT), and diphenylamine (DPA), which are associated with the presence of the low explosives (smokeless powders). The dynamic extraction performance of CMV for the analytical sampling and preconcentration of these gasphase volatiles is also compared to that of the SPME fibers under static sampling conditions. Finally, the CMV devices were used to sample for the presence of smokeless powders (SPs) and military explosives in open containers to determine the utility of CMV to detect explosives when coupling CMVextraction to GC-MS detection and analysis.

Experimental section Preparation of CMV Sorbent-coated glass filter disks were prepared using the method described elsewhere [15] and precut into rectangular

pieces measuring ∼2 mm by 2 cm. A glass capillary with an inner diameter of 2 mm was cut into 2-cm-long glass tubes and filled with seven (7) of the ∼2-mm-wide and 2-cm-long strips of the sorbent-coated strips. A magnified photo of the packed capillary glass tubes or the CMV device is shown in Fig. 1a, b. Materials Calibrations for the GC-MS instrument were performed using standard solutions of NG (Cerilliant Corporation, Round Rock, TX, USA), 2,4-DNT (Alfa Aesar, Heysham, Lancs), and DPA (Acros Organics, NJ, USA) diluted using Optima grade methanol (Fisher Scientific, Fair Lawn, NJ, USA) to concentrations ranging from 2 to 50 μg mL−1. Standard solution concentrations ranging from 40 to 150 μg mL−1 were spiked and used for headspace extraction and consisted of compounds that are present over the headspace of smokeless powders. Small quantities of smokeless powders which included Alliant Unique (Alliant Powder, Radford, VA, USA) and IMR 4198 (IMR Powder Co., Shawnee Mission, KS, USA) were placed in quart-sized metal cans (All-American Containers, Miami, FL, USA) and packing cardboard boxes (38 L) (Lowe’s, Miami, FL, USA) for headspace analysis. Military explosives (C4 and TNT wrappers) were sampled at a government facility by placing small amounts of explosives and wrappers in plastic bags and sampling the air above the plastic bags in open systems. Polydimethylsiloxane (PDMS) SPME fibers (SUPELCO, Bellefonte, PA, USA) were used in parallel studies in comparison to CMV devices. Sampling and introduction to GC-MS The CMV was conditioned in an oven at 250 °C for 1 h, allowed to cool to room temperature, and connected to a handheld air monitoring vacuum pump (Escort Elf Air Sampling Pump, Zefon International Inc., Ocala, FL, USA) that provides a flow through the CMV device of 0.1–1.5 L per min (LPM). After typical sampling times of between 30 s and 1 min, the CMV device was disconnected from the tubing and inserted into a probe for thermal desorption in an Agilent Technologies 7890A GC injector. The thermal separation probe (Agilent Technologies Inc., Santa Clara, CA, USA) can be coupled with the GC injector using a commercially available adapter. Standard solutions containing the analytes were spiked into quart-sized cans, and smokeless powders were placed in either the quart cans or cardboard boxes and sampled after a 10-min equilibration between the spike and the headspace for the solutions and after a 24-h equilibration for the solid smokeless powders. The can lid was opened, and the CMV was held at one end of the can opening

High-efficiency headspace sampling using CMV-GC-MS Fig. 1 a, b These photographs depict the dimensions of a CMV device. The inner capillary diameter is ∼2 mm, and the glass tube length is ∼2 cm long. b An enlarged image of one end of the CMV device better illustrates the packed sol–gel-coated glass filters within the glass tube. c Schlieren flow visualization of the CMV device connected to a vacuum pump and sampling the headspace of an open container using a circular motion and d visualization of the headspace sampling of the same sized container with a hole in the lid of the can

(Fig. 1c). A second sampling configuration was used in which the headspace was sampled through a ∼5-mm-diameter hole on the lid of the can (Fig. 1d). The handle flap on the cardboard boxes was opened in order to have access for headspace sampling using the CMV device. In order to evaluate their retention performance, the CMV devices were analyzed at different time intervals after sampling for up to 67 h. GC-MS/micro-electron capture detector (μ-ECD) An Agilent Technologies 7890A GC system was coupled to both a 5975C inert XL mass spectrometer detector (MSD) and a micro-electron capture detector (μ-ECD) by using a twoway splitter with makeup gas connections controlled by a pneumatics control module (PCM). The split ratio of the MSD to the μ-ECD was set at 5:1. The GC front injector used a split/splitless port and was used for direct injection and SPME analysis in split mode. A second injector was fitted with a probe adapter which was used to thermally desorb the CMV devices at 180 °C. The GC column was a DB-5ms Ultra Inert (8 m×0.25 mm×0.25 μm). The GC oven temperature ramp started at 40 °C, was held for 1 min and ramped at 15 °C/ min to 200 °C and held for 1 min, ramped again at 15 °C/min to 240 °C, held for 6.5 min and ramped at 25 °C/min to 270 °C, ramped at 5 °C/min to a final temperature of 280 °C, and held for 4 min. The total analysis time was 29.33 min.

Results and discussion CMV devices The CMV device is constructed of an open-ended capillary glass tube packed with precut 2-mm wide strips of PDMScoated glass filters that have been previously described [5]. The CMV device has an inner diameter of 2 mm, and the length of the CMV is adjusted to fit into any probe that can be inserted into the injector port of a GC. For this study, CMV devices were fashioned to fit into the Agilent thermal desorption unit with a length of 2 cm, which contains about 0.230 g of coated glass filters. The ∼0.05-m2 PDMS-coated surface is calculated to contain a phase volume of 50 mm3, which is significantly greater than a single SPME fiber (surface area, 9.4×10−6 m2 and phase volume, 0.612 mm3) [10]. Direct injection A mixture containing a total of 30 ng of NG, 30 ng of 2,4DNT, 20 ng of DPA, and 20 ng of EC was spiked onto one end of the CMV device and then placed directly into the thermal desorption unit. Figure 2a shows the targeted compound peaks in the gas chromatogram using the MSD. Calibration curves were generated for NG, DPA, and DNT and shown in Fig. 2b– d. These results are compared with the calibration curves obtained using a direct injection of the same mixture with

W. Fan, J. Almirall

Fig. 2 Gas chromatogram (a) of targeted compounds which was direct spiked on the CMV device. Calibration curves generated for NG (b), 2, 4-DNT (c), and DPA (d) by using both CMV (squares) and GC autosampler (diamonds)

the use of an autosampler employing the same split ratio. The integrated peak areas are comparable for both extraction

methods for NG and DPA, while CMV results in better overall sensitivity for DNT.

Fig. 3 Headspace extractions with 1-min dynamic sampling times using a CMV device (squares) and 30-min static sampling time using SPME fiber (diamonds) show comparable quantitative analysis results for a NG, b 2,4-DNT, and c DPA with a much reduced sampling time

High-efficiency headspace sampling using CMV-GC-MS Table 1 Recovered mass (as percent) of NG, DNT, and DPA after a 1-min dynamic extraction using the CMV device and compared with 30- and 10-min static extractions using the SPME fiber Recovery (%)

CMV (1 min)

SPME (30 min)

SPME (10 min)

NG DNT DPA

1.5 1.4 1.3

0.5 1.7 1.3

0.5 0.1 0.5

Headspace extraction of standard solutions Headspace extraction of mixtures containing the target volatiles in concentrations ranging from 40 to 150 μg mL−1 was conducted after a 10-min equilibrium period at room temperature (23 °C). The CMV device was placed at the opening of the can, collecting the headspace in a circular motion for 1 min during dynamic extraction of the headspace in the can at a flow rate of 1.5 L/min. A parallel study was conducted using static SPME extractions of the same mixture solutions for 10- and 30min extractions at 23 °C. Figure 3 shows the comparison of a 1min dynamic CMV extraction to a 30-min static SPME extraction resulting in similar extraction/detection performance. The recovery of NG, 2,4-DNT, and DPA extracted on the CMV device can be calculated by using the calibration curves generated from the direct injection (Fig. 2b–d). EC has a relative low vapor pressure (6.5×10-6 torr at 25 °C) [16] compared to NG (4.4×10-4 torr at 25 °C) [17], 2,4-DNT (2.1 × 10-4 torr at 25 °C) and DPA (6.4 × 10-4 torr at 25 °C) [18], and any preconcentration

method (including the CMV method presented), at room temperature, results in insufficient amounts of mass collected for detection in GC-MS given the short equilibrium and dynamic sampling times described in this scenario (fast sampling of open systems and large volume containers). The recovery of a particular analyte was obtained by dividing the amount detected with the amount available and is summarized in Table 1. For a 1-min extraction with the CMV, the NG recovery is 1.4 % compared to the 0.5 % with a 30-min SPME extraction, while for DNT and DPA, the recovery percentages are 1.4 and 1.3 % for CMV and 1.7 and 1.3 % for SPME. Less extraction times with SPME fibers (10 min) resulted in significantly lower recovery of 0.5, 0.1, and 0.5 % for NG, 2,4-DNT, and DPA, respectively. This comparison experiment shows the improved extraction and recovery due to a higher surface area and phase volume even with significantly reduced sampling times. Extractions of high mass loadings (1,500 ng for NG and DNT and 1,000 ng for DPA) resulted in saturation even after short (1 min) extractions. When low mass loadings were sampled, extraction times were linearly correlated to extracted amounts with varying linear ranges, depending on the partition coefficients of the analytes of interest as shown in Fig. 4, suggesting that quantitative analysis is possible when the extraction parameters are held constant between a calibration series and the unknown when conducting dynamic extractions. Panels a and b of Fig. 4 show that even after a 60-min extraction, the CMV devices have not been saturated for NG and 2,4DNT and the recovery rate can reach as high as 20 and 15 %, respectively, with no breakthrough observed. For DPA, the

Fig. 4 Extended dynamic headspace extraction suggesting quantitative results for a NG, b 2, 4-DNT, and c DPA

W. Fan, J. Almirall

Fig. 5 Dynamic headspace sampling using CMV with a an open system (a) and a semi-closed system (b)

extracted amount plateaued (reached an equilibrium) after a 20min extraction resulting in the familiar single-fiber SPME extraction curve shown in Fig. 4c, which represents a relatively low partition coefficient in PDMS/air. Compared to SPME extraction in a closed system, the CMV devices are used in an open system where the lid was removed and the container was completely open. The experiments described above and shown in Table 1 were conducted using open systems and resulted in some sample loss during the sampling process. When the dynamic extraction was conducted with a hole on the lid (Fig. 5b) instead of completely open (Fig. 5a), the recovery was found to increase by a factor of 2.6 compared to the results shown above and led to better precision. Smokeless powders and military explosives analysis (retaining power of CMV) In order to evaluate the analytical performance of the CMV, a small amount (10 and 100 mg) of Alliant Unique and IMR 4198 smokeless powders was placed in quart cans and cardboard boxes. After a 24-h equilibrium time, the lids were removed and the CMV devices were used to sample the headspace for only 1 min. For the Alliant Unique brand smokeless powder, NG and DPA (Fig. 6a) were identified in the headspace and DNT (Fig. 6b) was identified in the IMR 4198 which corresponded to the previously reported and expected results. Due to the small amounts of DPA present in the IMR 4198,

DPA can be only detected when larger amounts of smokeless powders are present. Long-term retention of the analytes after extractions was tested by sealing the CMV devices in aluminum foil and stored at 23 °C for 67 h prior to analysis. At least 70 % of the DNT and DPA mass extracted was retained in the CMV devices after 67 h of storage. These studies suggest that the CMV devices can be readily sealed and transported to laboratory facilities for subsequent analysis when on-site GC-MS analysis is not available. The CMV devices were also taken to a local airforce base for sampling of military explosives. The CMV devices were used to sample the wrappers of C4 and TNT explosives and sealed in aluminum foil and transported to the laboratory for analysis in the GC-MS. The sampling scheme included a 24-h equilibration time after the C4 and TNT wrappers were placed in plastic bags followed by sampling with CMV devices at the opening for 1 min. After sampling, the CMV devices were kept in the aluminum foil for 72 h before any analysis was conducted. The results showed that TNT wrappers contained 2,4-DNT in the headspace and C4 wrappers contained the taggant 2,3-dimethyl-2,3-dinitrobutane (DMNB) over the headspace.

Conclusions The novel sorbent-filled mini capillary tubes (CMV) described here were found to be an inexpensive alternative to

Fig. 6 Gas chromatograms of All Unique (a) and IMR 4198 (b) smokeless powder headspace extraction

High-efficiency headspace sampling using CMV-GC-MS

SPME static sampling while providing some significant advantages over static SPME sampling. Fast sampling times of 1 min (versus 30-min SPME extraction) provided good sensitivity, and the CMV was designed to allow for dynamic sampling of open systems or partially open systems. The significantly increased surface area and phase volume of the CMV over SPME resulted in quantitative sampling over a limited mass loading range and can be correlated to the initial mass or concentration in the headspace, when the sampling conditions are held constant for a calibration series and the unknown sample. The CMV devices were also shown to retain at least 70 % of the mass of the analytes after storage at 23 °C for more than 60 h. Finally, the utility of the CMV was shown for the detection of the target analytes to indicate the presence of the low explosives when as low as 10 mg of smokeless powder samples was placed in a quart-sized can and as low as 1 g of smokeless powder was placed in a 38-L container (the size of a small suitcase). The CMV devices also showed detection of 2,4-DNT and DMNB in military explosives (TNT and C4) and wrappers, sealed in plastic bags. The novel geometry CMV device is a quantitative, dynamic headspace analysis technique that can be applied for the analysis of a headspace over different explosives and may be developed for other applications such as clinical, food, and environmental analysis and biomedical testing. Acknowledgments The University Graduate School (UGS) at Florida International University (FIU) is acknowledged for partially funding Wen Fan on this study with a Dissertation Evidence Acquisition Fellowship. Matthew Staymates at National Institute of Standards and Technology (NIST) is also acknowledged for providing the Schlieren flow visualization of headspace sampling using a CMV device shown in Fig. 1c, d. The local Air Force base is also acknowledged for providing access to a laboratory containing the military explosives for headspace analysis.

References 1. Pawliszyn J (1997) Solid phase microextraction: theory and practice. Wiley-VCH, New York 2. Arthur CL, Pawliszyn J (1990) Solid phase microextraction with thermal desorption using fused silica optical fibers. Anal Chem 62(19):2145–2148. doi:10.1021/ac00218a019 3. Snow NH, Slack GC (2002) Head-space analysis in modern gas chromatography. TrAC Trends Anal Chem 21(9–10):608–617. doi: 10.1016/S0165-9936(02)00802-6

4. Zlatkis A, Lichtenstein HA, Tishbee A (1973) Concentration and analysis of trace volatile organics in gases and biological fluids with a new solid adsorbent. Chromatographia 6(2):67–70. doi:10.1007/ bf02270540 5. Gura S, Guerra-Diaz P, Lai H, Almirall JR (2009) Enhancement in sample collection for the detection of MDMA using a novel planar SPME (PSPME) device coupled to ion mobility spectrometry (IMS). Drug Test Anal 1(7):355–362. doi:10.1002/dta.81 6. Guerra-Diaz P, Gura S, Almirall JR (2010) Dynamic planar solid phase microextraction-ion mobility spectrometry for rapid field air sampling and analysis of illicit drugs and explosives. Anal Chem 82(7):2826–2835. doi:10.1021/ac902785y 7. Fan W, Young M, Canino J, Smith J, Oxley J, Almirall J (2012) Fast detection of triacetone triperoxide (TATP) from headspace using planar solid-phase microextraction (PSPME) coupled to an IMS detector. Anal Bioanal Chem 403(2):401–408. doi:10.1007/s00216012-5878-x 8. Pragst F (2007) Application of solid-phase microextraction in analytical toxicology. Anal Bioanal Chem 388(7):1393–1414. doi:10. 1007/s00216-007-1289-9 9. Ulrich S (2000) Solid-phase microextraction in biomedical analysis. J Chromatogr A 902(1):167–194. doi:10.1016/S0021-9673(00)00934-1 10. Pawliszyn J (1999) Applications of solid phase microextraction. RSC chromatography monographs. Royal Society of Chemistry, Cambridge 11. McComb ME, Oleschuk RD, Giller E, Gesser HD (1997) Microextraction of volatile organic compounds using the inside needle capillary adsorption trap (INCAT) device. Talanta 44(11): 2137–2143. doi:10.1016/S0039-9140(97)00093-3 12. Musshoff F, Lachenmeier DW, Kroener L, Madea B (2002) Automated headspace solid-phase dynamic extraction for the determination of amphetamines and synthetic designer drugs in hair samples. J Chromatogr A 958(1–2):231–238. doi:10.1016/S00219673(02)00317-5 13. Urbanowicz M, Zabiegała B, Namieśnik J (2011) Solventless sample preparation techniques based on solid- and vapour-phase extraction. Anal Bioanal Chem 399(1):277–300. doi:10.1007/s00216-0104296-1 14. EPA (1984) Method for the determination of volatile organic compounds in ambient air using Tenax adsorption and gas chromatography/mass spectrometry (GC/MS). http://www.epa.gov/ ttn/amtic/files/ambient/airtox/to-1.pdf. Accessed 27 Sept 2013 15. Guerra P, Lai H, Almirall JR (2008) Analysis of the volatile chemical markers of explosives using novel solid phase microextraction coupled to ion mobility spectrometry. J Sep Sci 31(15):2891–2898. doi:10.1002/jssc.200800171 16. ChemSpider Centralite. http://www.chemspider.com/ChemicalStructure.6567.html. Accessed 10 Aug 2013 17. Furton KG, Myers LJ (2001) The scientific foundation and efficacy of the use of canines as chemical detectors for explosives. Talanta 54 (3):487–500. doi:10.1016/S0039-9140(00)00546-4 18. Joshi M, Delgado Y, Guerra P, Lai H, Almirall JR (2009) Detection of odor signatures of smokeless powders using solid phase microextraction coupled to an ion mobility spectrometer. Forensic Sci Int 192(1–3):135–135