Anal. Chem. 2003, 75, 4639-4645
Determination of Nitroaromatic Compounds in Air Samples at Femtogram Level Using C18 Membrane Sampling and On-Line Extraction with LC-MS C. Sa´nchez,†,§ H. Carlsson,†,‡ A. Colmsjo 1 ,† C. Crescenzi,† and R. Batlle*,†,§
Department of Analytical Chemistry, Stockholm University, 106 91 Stockholm, Sweden, FOI, Swedish Defense Research Agency, Department of Energetic Materials, 147 25 Tumba, Sweden
This paper explores the use of C18 solid-phase extraction membranes for sampling very low concentrations of nitroaromatic compounds in the atmosphere. After sampling, analytes trapped in the membrane are desorbed online directly by a chromatographic mobile phase. The analytes are then separated onto a porous graphitic carbon (PGC) HPLC column. Finally, they are analyzed by an LC-MS/MS detector equipped with an atmospheric pressure chemical ionization (APCI) interface. The method was validated by controlled exposure of the membranes to standard gaseous mixtures of 2,4,6-trinitrotoluene (TNT) and 2,4-dinitrotoluene (2,4-DNT). The developed method was fully characterized, and no breakthrough was observed when sampling volumes up to 9.2 m3. Analyses of membranes following medium- and long-term storage demonstrated that samples could be stored on the C18 membranes without degradation or losses. In addition, the results obtained with this technique were compared with those obtained by a gas chromatographic method in which analytes were collected on Tenax TA and thermally desorbed. The developed method allows sampling at flow rates of 15 L/min and has method detection limits in the femtogram/liter range, with a relative standard deviation lower than 10%. An additional advantage of this method is that it separates most of the TNT and DNT isomers, as demonstrated by applying the method to the analysis of headspace over military-grade TNT explosives. Between 60 and 80 million antipersonnel (AP) mines are currently in place in over 70 countries. Designed to kill or maim humans, they injure an estimated 1200 persons and kill another 800 every week. An estimated 300 000 persons around the world have been disabled by them. In addition to lives that are lost due to explosions, the mere suspicion that landmines may be present prevents the use of large areas that could otherwise be used for agricultural or social infrastructure. Removing landmines from such areas is known as humanitarian demining, and the system * Corresponding author. Fax: + 46 (0) 8 156391. E-mail: ramon.batlle@ anchem.su.se. † Stockholm University. ‡ Swedish Defense Research Agency. § Current address: Department of Analytical Chemistry, Centro Polite´cnico Superior, Maria de Luna 3, 50018 Zaragoza, Spain. 10.1021/ac034278w CCC: $25.00 Published on Web 07/16/2003
© 2003 American Chemical Society
used must have a detection and removal rate better than 99.6% to meet United Nations specifications.1-3 AP landmines are typically shaped in the form of a membrane or cylinder, with diameters from 20 to 125 mm, lengths from 50 to 100 mm, and 20 g of explosive upward. They can be encased in metal, but their metal content can be as low as 0.1 g in modern mines, which are often encased in materials such as plastic or wood.4 Thus, detection using metal detectors is not feasible in practice, because if they were tuned to be sufficiently sensitive, there would be high rates of false alarms. Since AP mines are usually shallowly buried, at a maximum depth of 50 mm, a potentially useful detection approach currently being investigated is the chemical detection of vapors that evolve from the explosives and are transported to the surrounding air or soil in the immediate vicinity of the mines.5 Being a chemical multianalyte method, this approach is less prone to false alarms than physical nonspecific methods, such as metal detection, but high sensitivity is required, since the concentration of molecules expected to reach the gas phase in real field conditions is extremely low.6 Today there is no reliable on-site equipment for landmine detection in the field, beside well-trained dogs,7 but they can only work for limited periods of time, and they tire quickly in hot weather. Therefore, active research is being pursued into mine detection using sensors,8-10 solid-phase microextraction,11,12 or solid sorbents.13-15 (1) Bruschini, C.; Gros, B.; Guerne, F.; Piece, P.; Carmona, O. J. Appl. Geophys. 1998, 40, 59-71. (2) Newman, R. D.; Mercer, M. A. Int. J. Occup. Environ. Health 2000, 6, 243248. (3) Combrinck, M. Afr. Earth Sci. 2001, 33, 693-698. (4) Hussein, E. M. A.; Waller, E. J. Appl. Radiat. Isot. 2000, 53, 557-563. (5) Yinon, J. TrAC, Trends Anal. Chem. 2002, 21, 292-301. (6) Chambers, W. B.; Rodacy, P. J.; Jones, E. E.; Gomez, B. J.; Woodfin, R. L. Proceedings of the SPIE Conference on Detection and Remediation Technologies for Mines and Minelike Targets III, Orlando, FL, 1998; pp 453-461. (7) Furton, K. G.; Myers, L. J. Talanta 2001, 54, 487-500. (8) Albert, K. J.; Myrick, M. L.; Brown, S. B.; James, D. L.; Milanovich, F. P.; Walt, D. R. Environ. Sci. Technol. 2001, 35, 3193-3200. (9) Bakken, G. A.; Kauffman, G. W.; Jurs, P. C.; Albert, K. J.; Stitzel, S. S. Sens. Actuators B 2001, 79, 1-10. (10) Cremer, F.; Schutte, K.; Schavemaker, J. G. M.; Den Breejen, E. Inform. Fusion 2001, 2, 187-208. (11) Barshick, S. A.; Griest, W. H. Anal. Chem. 1998, 70, 3015-3020. (12) Jenkins, T. F.; Leggett, D. C.; Miyares, P. H.; Walsh, M. E.; Ranney, T. A.; Cragin, J. H.; George, V. Talanta 2001, 54, 501-513. (13) Sigman, M. E.; Ma, C.; Ilgner, R. H. Anal. Chem. 2001, 73, 792-798.
Analytical Chemistry, Vol. 75, No. 17, September 1, 2003 4639
Common analytical techniques for the analysis of explosives include gas chromatography with either electron capture, nitrogen phosphorus detection, or mass spectrometry12,13,16 and highperformance liquid chromatography with ultraviolet or MS detection.14,17-21 Other approaches, such as amperometric gasphase sensing22 or optical techniques,23-25 have also been employed. The use of unmodified solid-phase extraction (SPE) cartridges for active air sampling, including its application to the analysis of energetic materials, has already been described.16,26-28 A major drawback of the method is that it is not possible to accommodate high air flows due to restrictions imposed by cartridge geometry and consequent backpressure. Thus, prolonged operating times are required to sample large volumes, which is often essential in order to detect traces of the analytes. To overcome this problem, solid-phase extraction membranes can be used instead of columns. In SPE membranes (or disks), sorbent material is embedded in poly(tetrafluoroethylene). Advantages and drawbacks of using SPE membrane to extract liquid samples have been summarized in a recent paper.29 These membranes have also been used in overpressured liquid chromatography,30 passive sampling,31 and assessing the chemical permeation of protective clothing.32 In air sampling, the major advantages of SPE membranes over conventional SPE cartridges arise from their higher permeability and surface area-to-volume ratios, which allow much higher sampling rates. In the literature, few reports have been presented about the use of SPE membranes for air sampling.17,33-37 The study presented here had several aims. The main goal was to evaluate commercially available SPE membranes in order (14) Batlle, R.; Carlsson, H.; Holmgren, E.; Colmsjo ¨, A.; Crescenzi, C. J. Chromatogr., A 2002, 963, 73-82. (15) Hable, M. A.; Sutphin, J. B.; Oliver, C. G.; McKenzie, R. M.; Gordon, E. F.; Bishop, R. W. J. Chromatogr. Sci. 2002, 40, 77-82. (16) Batlle, R.; Carlsson, H.; Tollba¨ck, P.; Colmsjo ¨, A.; Crescenzi, C. Anal. Chem. Accepted. (17) Sasano, R.; Furusyo, Y.; Matsumura, T. Kuki Seijo 1999, 37, 239-244. (18) Harvey, S. D.; Clauss, T. R. W. J. Chromatogr., A 1996, 753, 81-89. (19) Gates, P. M.; Furlong, E. T.; Dorsey, T. F.; Burkhardt, M. R. TrAC, Trends Anal. Chem. 1996, 15, 319-325. (20) Cassada, D. A.; Monson, S. J.; Snow, D. D.; Spalding, R. F. J. Chromatogr., A 1999, 844, 87-95. (21) Schreiber, A.; Efer, J.; Engewald, W. J. Chromatogr., A 2000, 869, 411425. (22) Zhao, X.; Yinon, J. J. Chromatogr., A 2002, 946, 125-132. (23) Buttner, W. J.; Findlay, M.; Vickers, W.; Davis, W. M.; Cespedes, E. R.; Cooper, S.; Adams, J. W. Anal. Chim. Acta 1997, 341, 63-71. (24) Janni, J.; Gilbert, B. D.; Field, R. W.; Steinfeld, J. I. Spectrochim. Acta A 1997, 53, 1375-1381. (25) Yang, J. S.; Swager, T. M. J. Am. Chem. Soc. 1998, 120, 11864-11873. (26) Koostra, P. R.; Herbold, H. A. J. Chromatogr., A 1995, 697, 203-211. (27) Magnani, F.; Lattanzi, L.; Maione, M. Chromatographia 1998, 47, 57-62. (28) Ramil, M.; Rodrı´guez, I.; Cela, R. J. Chromatogr., A 2002, 963, 65-71. (29) Thurman, E. M.; Snavely, K. TrAC, Trends Anal. Chem. 2000, 19, 18-26. (30) Regnault, C.; Delvordre, P.; Postaire, E. J. Chromatogr. 1991, 547, 403409. (31) Kingston, J. K.; Greenwood, R.; Mills, G. A.; Morrison, G. M.; Bjo¨rklund, L. J. Environ. Monit. 2000, 2, 487-495. (32) Vo, E.; Berardinelli, S. P.; Boeniger, M. Appl. Occup. Environ. Hyg. 2001, 16, 729-735. (33) Fujino, T.; Goto, I.; Kii, K. Oita-ken Eisei Kankyo Kenkyu Senta Nenpo 2001, 28, 24-47. (34) Saito, I.; Onuki, A.; Seto, H. Earozoru Kenkyu 2001, 16, 209-216. (35) Akanashi, M.; Kitabayashi, Y.; Ohashi, Y.; Sekiguchi, T. Tochigi-ken Hoken Kankyo Senta Nenpo 2000, 5, 91-97. (36) Furusho, Y.; Sasano, R.; Kuriyama, K.; Matsumura, T. Kurin Tekunoroji 1999, 9, 57-62. (37) Stuff, J. R.; Cheicante, R. L.; Dupont, H.; Ruth, J. L. J. Chromatogr., A 1999, 849, 529-540.
4640 Analytical Chemistry, Vol. 75, No. 17, September 1, 2003
Figure 1. Air sampling setup. (1) Teflon ring (o.d. 47 mm); (2) C18SPE membrane (o.d. 47 mm); (3) stainless steel net (o.d. 42 mm). For clarity, the drawing is not to scale.
to detect energetic materials and related compounds via active air sampling. Further goals were to develop and evaluate a simple on-line desorption method avoiding any pretreatment step, and finally, to characterize the explosives, separate the isomers of TNT and DNT, and accurately identify them by LC-MS/MS. EXPERIMENTAL SECTION Chemicals and Materials. The nitroaromatic reference substances were obtained from several sources. 1,2-, 1,3-, and 1,4dinitrobenzene (DNBs) were purchased from Dr. Ehrenstorfer (Augsburg, Germany). 2,3-, 2,4-, 2,5-, 2,6-, 3,4-, and 3,5-dinitrotoluene (DNTs) and 2,4,6-trinitrotoluene (TNT) were obtained from the Swedish Defense Research Agency (Tumba, Sweden). Trinitrobenzene (TNB) was purchased as a 100 µg/mL acetone solution from Promochem (Wesel, Germany). Stock solution (75125 ng/µL) was prepared in acetonitrile. Reference standard solutions were prepared as needed by dilution of this solution and added, at known concentrations, to the blank solid-phase extraction membranes. Spiked filters were wrapped in aluminum foil and allowed to equilibrate for at least 4 days at - 4 °C before use. 2,4-Dinitrofluorobenzol (DNFB, injection standard) was obtained from the Swedish Defense Research Agency. A stock solution (5400 ng/µL) as well as working dilutions (10.6 ng/µL) were prepared in acetonitrile. Methanol and toluene (Suprasolv quality for liquid chromatography) were obtained from Merck, (Darmstadt, Germany). Acetonitrile (Chromasolv quality for liquid chromatography) was purchased from Riedel de Ha¨en (Seelze, Germany). Air Sampling. Air sampling was performed with the setup illustrated in Figure 1. The sampler consisted on an anodized aluminum holder containing a 47-mm Empore octadecyl solidphase extraction (SPE) membrane obtained from Varian (Walnut Creek, CA). In this system, the SPE sampling membrane (used as received) is kept in place by two Teflon rings, and the whole system is supported by a stainless steel net. Sampling was carried out in an active mode using a VDE 0530 pump (KNF Neuberger, Freiburg, Germany). The flow rate was set to 20 L/min, yielding a sampling capacity of 15 L/min, as a result of the backpressure of the membrane. This value compares favorably to the maximum flow achievable when using SPE cartridges, which was found to be limited to 3.0 L/min.
Table 1. Multiple Reaction Monitoring (MRM) Conditions analyte
Figure 2. On-line extraction setup. Continuous lines, on-line desorption and HPLC-MS analysis; dashed lines, internal standard loading and filling of the extraction cell with water.
To validate the SPE membrane sampling method, a continuous vapor generator (constructed in-house) that can generate atmospheres with low concentrations of TNT and 2,4-DNT was used. The system is temperature-controlled, which permits a wide range of equilibrium conditions to be generated, and the humidity can also be varied at will, allowing a wide range of environments to be simulated. The vapor concentrations can be adjusted by gasblending to obtain concentrations of explosives as low as 20 pg/ L. Further, the system is equipped with a sampling manifold for sampling air with different samplers. On-Line Desorption Setup. The setup used for on-line desorption is depicted in Figure 2. After the sampling was completed, SPE membranes were removed from the holder, cut into four pieces, and introduced into the polyether ether ketone (PEEK) extraction cell (30 × 4.6 mm, internal volume 1 mL; Jour Research, Onsala, Sweden). The cell was then connected to the system and filled with distilled water using the auxiliary pump (Shimadzu LC10AD, Shimadzu Corp., Kyoto, Japan). In this way, the air remaining in the system was displaced to prevent it from disturbing the chromatography or causing fluctuations in pressure. Finally, the cell was inserted into the chromatographic system by switching valve 1. At the same time, 10 µL (10.6 ng/µL) of the internal standard (DNFB) was injected using valve 2. Valve 2 was also used to inject the external reference mixture. No traces of the analytes were detected in water added in excess of the volume required to fill the cell, and no carryover was observed in second chromatographic runs when using the same membrane. Liquid Chromatography-Mass Spectrometry. The HPLC system consisted of a Varian 9012 HPLC gradient pump (Varian, Walnut Creek, CA) and a Hypercarb analytical column 100 × 4.6 mm (5 µm) purchased from Thermo Quest (Cheshire, UK). The chromatographic conditions for the analysis were as follows: solvent A was a mixture of water/acetonitrile/methanol (50:40: 10, v/v), and B was methanol/acetonitrile/toluene (73:25:2, v/v). The initial percentage of solvent B was 40%, flow 1.0 mL/min, held isocratically for 3 min. The concentration of B was then raised linearly to 100% at a flow rate of 2.0 mL/min over 22 min and held for a further 10 min. The column was then washed with 100% B at 2.5 mL/min for 2 min, after which the mobile was returned to its initial composition over 1 min, and the column was equilibrated with this mixture for 10 min.
1,2-DNB 1,3-DNB 1,4-DNB 2,3-DNT 2,4-DNT 2,5-DNT 2,6-DNT 3,4-DNT 3,5-DNT 1,3,5-TNB 2,4,6-TNT DNFB (IS)
molecular ion
collision energy (%)
167.9
15
181.9
15
213.0 226.9 186.1
18 15 15
fragments selected (% abundance) 138 (100); 108 (37) 138 (100); 108 (32) 138 (100); 108 (13) 152 (100); 122 (41) 152 (100); 165 (34) 152 (100); 122 (11) 152 (100); 122 (34) 182 (100); 152 (31) 152 (100); 122 (13) 183 (100); 125 (14) 210 (100); 197 (34) 140 (100); 156 (38)
The mass spectrometer used was a triple quadrupole Quattro Micro (Micromass, Manchester, U.K.), equipped with an atmospheric pressure chemical ionization (APCI) interface, operating in negative ionization mode. Interface conditions were set to the following values: source temperature, 130 °C; APCI probe temperature, 300 °C; desolvation gas flow, 112 L/h; cone gas flow, 0 L/h; corona, 30 µA; cone voltage, 20 V; extractor, 2 V; RF lens, 0.3 V. To determine the best interface conditions, single MS in full scan monitoring mode (100-450 m/z) was employed. The accepted values were those providing the most intense signals of the molecular ions (m/z 168, 181, 186, 213, 226, and 227, corresponding to DNBs, DNTs, DNFB (IS), TNB, and TNT, respectively). Then, tandem mass spectrometer parameters were optimized (for each analyte) in multiple reaction monitoring (MRM) acquisition mode by varying the collision energies in the range between 8 and 30. The value selected for each fragment was that giving the highest signal-to-noise ratio. The optimized value of the collision energy, expressed in terms of the manufacturer’s nominal relative collision energy (%), together with the fragment selected for each analyte and the internal standard, are shown in Table 1. Figure 3 shows a typical chromatogram obtained for a direct liquid injection of 20 µL (75 ng injected) of the standard solution. Gas Chromatography-Thermal Desorption. Thermal desorption samples were collected on prepacked Perkin-Elmer stainless steel tubes (length, 8.9 cm; 0.64 cm o.d.; Perkin-Elmer, Norwalk, CT) filled with Tenax TA (60-80 mesh) and were analyzed with a gas chromatograph coupled to a Perkin-Elmer ATD 400 automatic thermal desorber. Desorption parameters were as follows: desorption temperature, 320 °C; desorption time, 4 min; trap low temperature, -30 °C; trap high temperature, 225 °C; and trap hold time, 8 min. Gas chromatography was performed on an Agilent 6890A (Agilent Technologies, Wilmington, DE) chromatograph equipped with a thermionic detector held at 275 °C (TS-1 type bead, Detector Engineering & Technology, Walnut Creek, CA) and a 30-m × 0.25-mm DB-1701 analytical column coated with a 0.15-µm film (J&W Scientific). The GC oven was programmed as follows: 2 min at 50 °C, ramped at 40 °C/min to 175 °C, then at 20 °C/min to 290 °C. RESULTS AND DISCUSSION Air Sampling. As explained in the Introduction and the Experimental Section, the main attraction of using C18 SPE Analytical Chemistry, Vol. 75, No. 17, September 1, 2003
4641
Figure 3. Standard chromatogram (amount injected, 75 ng).
membranes for air sampling was the opportunity they provide to exploit a much higher sampling flow rate, as compared to conventional cartridges. However, the possibility that breakthrough may occur in the sampler when large volumes of air are sampled at high flow rates had to be tested. For this purpose, two C18 SPE membrane holders were connected in series. A sampling volume of 9.2 m3 of the standard gas containing 2,4DNT and TNT with a concentration of 0.4 ng/L (total amount of analytes collected, 3.68 µg) was then collected at a flow rate of 11 L/min from the continuous vapor generator described in section 2.2. The two membranes were then analyzed separately. No breakthrough from the first membrane was detected for 2,4-DNT and
