Analysis of explosives using electrospray ionization/ion mobility spectrometry (ESI/IMS)

Transcription

1 Talanta 50 (2000) Analysis of explosives using electrospray ionization/ion mobility spectrometry (ESI/IMS) G. Reid Asbury, Jörg Klasmeier, Herbert H. Hill Jr. * Department of Chemistry, Washington State Uni ersity, Pullman, WA , USA Received 6 May 1999; received in revised form 2 August 1999; accepted 3 August 1999 Abstract The analysis of explosives with ion mobility spectrometry (IMS) directly from aqueous solutions was shown for the first time using an electrospray ionization technique. The IMS was operated in the negative mode at 250 C and coupled with a quadrupole mass spectrometer to identify the observed IMS peaks. The IMS response characteristics of trinitrotoluene (TNT), 2,4-dinitrotoluene (2,4-DNT), 2-amino-4,6-dinitrotoluene (2-ADNT), 4-nitrotoluene (4-NT), trinitrobenzene (TNB), cyclo-1,3,5-trimethylene-2,4,6-trinitramine (RDX), cyclo-tetramethylene-tetranitramine (HMX), dinitro-ethyleneglycol (EGDN) and nitroglycerine (NG) were investigated. Several breakdown products, predominantly NO 2 and NO 3, were observed in the low-mass region. Nevertheless, all compounds with the exception of NG produced at least one ion related to the intact molecule and could therefore be selectively detected. For RDX and HMX the [M+Cl ] cluster ion was the main peak and the signal intensities could be greatly enhanced by the addition of small amounts of sodium chloride to the sprayed solutions. The reduced mobility constants (K 0 ) were in good agreement with literature data obtained from experiments where the explosives were introduced into the IMS from the vapor phase. The detection limits were in the range of g l 1 and all calibration curves showed good linearity. A mixture of TNT, RDX and HMX was used to demonstrate the high separation potential of the IMS system. Baseline separation of the three compounds was attained within a total analysis time of 6.4 s Elsevier Science B.V. All rights reserved. Keywords: Electrospray ionization; Ion mobility spectrometry; Reduced mobility constants; Baseline separation 1. Introduction Trace analysis of explosives is of major importance in two different fields. One is the threat of an illegal use of these compounds, which has led * Corresponding author. Tel.: ; fax: address: hhhill@wsu.edu (H.H. Hill Jr.) to major efforts in developing explosives detection systems [1 3]. The other is growing concern about the health risks associated with the release of explosives from military sites and former ammunition plants into the environment [4 6]. Ion mobility spectrometry (IMS) is one of the most promising analytical techniques currently investigated for the detection of explosives. The major advantages of IMS are its fast response times and /00/$ - see front matter 2000 Elsevier Science B.V. All rights reserved. PII: S (99)

2 1292 G. Reid Asbury et al. / Talanta 50 (2000) its sensitivity to many of the commonly used explosives due to their strong electron affinity [7]. Although a number of commercial IMS systems are already available, there are still some problems to overcome. The routine use of IMS as a detection method for explosives requires a reproducible response produced by clearly identified ions. The ion chemistry of the compounds, however, is different depending on the operation conditions (e.g. ionization method, drift tube temperature, drift gas) of the IMS system. Published IMS spectra often differ significantly due to temperature effects and the addition of reactants to the drift gas [8,9]. Hitherto, several different ions have been reported including molecular ions of the form M or [M-H], multi-component cluster-ions (e.g. [M+Cl ] or [M+NO 3 ] ) and fragment ions (e.g. NO 3 ). The exact nature of the species formed in the IMS can only be decisively determined using an IMS/MS combination. Thermally labile compounds like NG and EGDN tend to decompose at elevated temperatures and often only non-specific fragment ions (e.g. NO 2 and NO 3 ) are observed [1]. As hightemperature operation of the IMS provides several advantages such as, eliminating interference problems and enhancing the sensitivity, it would be desirable to establish such a system for the determination of explosives. The use of electrospray ionization (ESI) instead of the standard 63 Ni ionization source could also be advantageous. The combination of ESI with mass spectrometry (MS) has already been successfully applied to the analysis of explosives [10,11]. In our lab, recently a novel, cooled electrospray ionization system has been developed for the use with IMS. This enables the direct analysis of aqueous samples [12,13]. This methodology could provide a helpful tool for the screening of water samples suspected to be contaminated with explosives. Using our ESI-IMS-MS, we investigated the response characteristics and the detection limits for a number of different explosives. The mass spectral data were used to identify the IMS peaks to enable a comparison of the calculated K 0 -values with literature data. Also the ion patterns observed for RDX and HMX were compared to results of previously published ESI-MS studies. 2-ADNT, one of the initial metabolic intermediates of TNT formed after release of TNT into the environment [5], was included in the study to address the possible tracing of environmental contamination with an IMS method. 2. Experimental 2.1. Chemicals and sol ents The explosives used in this study included 2,4,6- trinitrotoluene (TNT), 2,4-dinitrotoluene (2,4- DNT), 1,3,5-trinitrobenzene (TNB), 2-amino-4,6- dinitro-toluene (2-ADNT), 4-nitrotoluene (4- NT), cyclo-1,3,5-trimethylene-2,4,6-trinitramine (RDX), cyclo-tetramethylene-tetranitramine (HMX), ethyleneglycol-dinitrate (EGDN), and nitroglycerin (NG). All compounds were purchased as standards in sealed ampoules from either Supelco (Bellefonte, PA) or Radian International (Austin, TX). They came dissolved in acetonitrile at a concentration of 1000 mg l 1 and were diluted with methanol water (9:1, v/v) to 10 mg l 1 or less. Solutions containing sodium chloride or ammonium acetate were prepared by diluting the stock standard solutions with an appropriate solution of the respective salt in methanol water (9:1, v/v). All solvents used were reagent grade (J.T. Baker, Phillipsburg, PA) Instrumentation The ESI/IMS/MS-system was constructed at Washington State University and has been described in detail elsewhere [12,13]. A total spray voltage of 2500 V was applied to the electrospray unit. The electrospray needle was kept cool by water cooled nitrogen flowing along the axis of the needle at a flow rate of about 60 ml min 1. This cooling process was necessary to eliminate solvent evaporation inside the needle prior to electrospray. A continuous flow of solvent (methanol water, 9:1, v/v) through the ESI source was provided by a dual piston syringe

3 G. Reid Asbury et al. / Talanta 50 (2000) pump (Brownlee Labs, Santa Clara, CA) with a flow rate of 5 l min 1. Samples were injected via a six-port injector (C6W, Valco Industries, Houston, TX) with an external injection loop. A detailed description of the IMS unit can be found in Wu et al. [13]. The instrument was operated in the negative mode, as explosives tend to produce negative ions. The electric field strength in the desolvation region and the drift region was 280 V cm 1 with a total drift voltage of 3650 V. The drift tube was operated at 250 C at atmospheric pressure. A counterflow of preheated dry nitrogen (800 ml min 1 ) was introduced as drift gas at the end of the drift region. This drift gas was further used to desolvate the electrosprayed ions in the desolvation region. The ion mobility spectrometer was interfaced to a C50-Q quadrupole mass spectrometer (ABB Extrel, Pittsburgh, PA 15238). The ions entered the MS via a 40-micron pin hole leak which served as the barrier between the atmospheric pressure of the IMS tube and the vacuum of the mass spectrometer. All mobility data was collected by replacing the stock preamplifier with a Keithley 427 amplifier (Keithley Instruments, Cleveland, OH) and sending the amplified signal to the data acquisition system, which was constructed at WSU. A detailed description of the IMS control and data acquisition system can be found in [12]. All of the IMS spectra shown in this paper were the average of 256 individual spectra. Mobility spectra were collected in a non-selective and a mass selective mode. In the non-selective mode all ions were allowed to pass the mass filter, whereas in the mass selective mode the mass spectrometer was set to allow only a single m/z to pass. All identifications of distinct ions were based on the mass spectral data. 3. Results and discussion 3.1. Low-mass background and breakdown ions Fig. 1. Background ions in the low-mass region of the IMS monitored for pure solvent (a), a solution of 10 mg l 1 RDX (b) and a solution of 2 mm NaCl in the solvent (c). Notation: 1, chloride (Cl ); 2, nitrate (NO 3 ); 3, formate (HCOO ); 4, nitrite (NO 2 ); 5, acetate (CH 3 COO ); 6, probably nitrite adduct (CH 3 OH+NO 2 ); 7, cluster of at least two unidentified ions (m/z 89 and 97). The low-mass region of the IMS spectra was studied in some detail in order to better characterize the high temperature atmospheric pressure ESI/IMS and its ion chemistry. The top spectrum of Fig. 1 shows the background signal of the IMS when spraying pure solvent. There are three distinct peaks (c1, c3, c5), which were identified by mass analysis to be Cl, HCOO, and CH 3 COO, respectively. These ions are most likely formed by ESI from trace impurities of the solvents (e.g. acetic acid, chloride salts). The middle spectrum of Fig. 1 shows the peaks observed in the low-mass region with a solution of RDX (10 mg l 1 ). Besides the three background peaks a number of other peaks appeared. Peaks labeled c2 and c4 were identified as nitrite and nitrate. They were probably formed by thermal breakdown of RDX in the hot desolvation region. All of the investigated explosives produced significant amounts of these two ions, which is not surprising since most of the compounds are thermally labile [1].

4 1294 G. Reid Asbury et al. / Talanta 50 (2000) reduction of the other background ions is probably a result of charge competition IMS spectra of standard solutions of explosi es Fig. 2. Ion mobility spectra of eight explosives with low mass filter set to 140. Concentration of standard solutions was 10 mg l 1 in methanol water (9/1, v/v), 2 mm NaCl was added for RDX and HMX. M/z values can be seen in Table 1. The peak labeled c6 has a mass of 78 amu and is suggested to be a (CH 3 OH+NO 2 ) adduct ion. The final peak (c7) appears to be a cluster of different ions. The mass spectral analysis revealed that it contains at least two ions with m/z 89 and 97. The identity of these peaks is speculated to be adducts of nitrite and nitrate with other breakdown products. The same mass peaks have been previously observed in an ESI/MS study, but were also unidentified [11]. The bottom spectrum of Fig. 1 shows the lowmass region for pure solvent containing 2 mm sodium chloride. As expected the chloride peak ( c1) dominates this spectrum. The simultaneous The different explosives were sprayed out of standard solutions at concentrations of 10 mg l 1 each. The respective spectra are shown in Fig. 2. They were taken with the low-mass filter set to 140, effectively eliminating any ions in the lowmass region from reaching the detector. Furthermore it should be noted that the spectra shown for RDX and HMX were taken with the addition of 2 mm NaCl to the standard solution, predominantly forming (M+Cl ) adduct ions (see discussion below). In the case of nitroglycerine only breakdown product ions (mainly nitrate) were produced and no ions could be detected in the high mass region. The thermal instability of NG in traditional IMS systems at temperatures above 150 C is known and has been reported before [1,9]. Obviously this is also valid for the ESI/IMS system at 250 C. TNB showed at least three different peaks with the M (m/z 213) being the dominant one. Mass spectral analysis revealed m/z values of 178 and 244 for the two other peaks seen in the IMS spectrum. The latter is speculated to be a (M+ CH 3 CO ) adduct formed by the molecule clustering with the methanol solvent. All the other compounds showed only one main peak being either (M-H) or M. The reduced mobility constants (K 0 ) of the dominant peaks for each compound were calculated and are listed in Table 1. It can be seen that they are in good agreement with previously reported literature values obtained from experiments where the compounds have been introduced into the IMS as vapors Sensiti ity enhancement by addition of Cl It is known that the sensitivity of IMS systems using a 63 Ni foil as ionization source to certain explosives can be enhanced by adding chlorinated reactants (e.g. dichloromethane) to the carrier gas [1,3,9]. This phenomenon has not yet been investigated with electrospray ionization. In our experi-

5 G. Reid Asbury et al. / Talanta 50 (2000) Table 1 K 0 -values of different product ions of the investigated compounds compared to literature data Compound m/z Species LoD ( g l 1 ) K 0 (cm 2 V 1 s 1 ) a Reference TNT 226 (M-H) (N 2 ) [8] 1.45 (air) [1 3] 2 ADNT 196 (M-H) ,4 DNT 181 (M-H) (N 2 ) [8] 1.57 (air) [1] 4-NT 136 (M-H) (air) [8] TNB 244 (M+CH 3 O ) n.c M ? 1.65 RDX 257 (M+Cl ) (air) [2,3] HMX 331 (M+Cl ) (air) [2] EGDN 152 M NG 226 (M-H) n.d (air) [2] 262 (M+Cl ) n.d (N 2 ) [9] a First column: measured in this study with N 2 as drift gas. Second column: reference data, drift gas given in parentheses. n.d., not detected; n.c., not calculated; LoD, limit of detection. ments the addition of small amounts of dichloromethane to samples of RDX and HMX showed no significant effects in the IMS, even with respect to the Cl peak. However, a huge increase in the peak intensities for the Cl peak and the (M+Cl ) adduct peak could be achieved by adding sodium chloride to the samples. This seems reasonable taking into account that when using a 63 Ni foil the ions are created in the gas phase from volatile species, whereas electrospray creates the ions in solution from dissolved species. Significant increases in signal intensity for the (M+Cl ) adduct were observed at concentrations as low as 0.02 mm NaCl. The optimal concentration was found to be about 2 mm with an overall increase in signal intensity compared to a pure standard solution of more than one order of magnitude (about 12 times). Figs. 3 and 4 show the different spectra for RDX and HMX with and without the addition of 2 mm NaCl. The increase in the intensity of the chloride peak after adding NaCl can be seen from Fig. 3. No M or (M-H) ions were detected in either case. However, the spectra clearly indicate a significant increase in the peak intensity with the addition of the chloride salt. Apparently, both compounds are able to form stable (M+Cl ) adduct ions with our operating conditions. None Fig. 3. Ion mobility spectra of RDX with (a) and without (b) addition of NaCl (2 mm). Concentration of standard solution was 10 mg l 1 in methanol water (9/1, v/v) each.

6 1296 G. Reid Asbury et al. / Talanta 50 (2000) refer to the same peak as an acetate adduct (M+ CH 3 COO ) due to the fact that they sprayed standard solutions containing 2 mm of ammonium acetate. With our experimental design we could not achieve any increase in the (M+59)- peak by adding 2 mm of ammonium acetate, but this might well be a temperature effect since the IMS was operated at 250 C Detection limits and linearity of response Fig. 4. Ion mobility spectra of HMX with (a) and without (b) addition of NaCl (2 mm). Concentration of standard solution was 10 mg l 1 in methanol water (9/1, v/v) each. Notation: 1, m/z 334, 2, m/z 331; 3, m/z 341; 4, m/z 355; and 5, m/z 357. of the other explosives produced measurable amounts of (M+Cl ) adduct ions. This is surprising for EGDN, as this compound has been shown to do so in vapor phase IMS systems [1,9]. The top spectra of Figs. 3 and 4 obtained with the pure standards also show the peak for the (M+Cl ) adduct, but the signals are much weaker. Whereas for pure RDX no other ions appeared in the IMS spectrum, HMX showed a couple of additional peaks. M/z of these peaks were identified to be 334, 341, 355 (M+59) and 357. Similar observations were made in previously published ESI/MS studies, but the identity of the respective ions remains in question. Yinon et al [11] labeled the (M+59) as a (M+NNO 2 -H) cluster ion, whereas Casetta and Garofolo [10] With the exception of NG, TNB and 4-NT all explosives produced signals significantly above the noise level (about 300 fa) at a concentration of 10 mg l 1. Detection limits for these compounds were determined as the concentrations giving peak heights three times the noise level. The calculated values for the different compounds are given in Table 1. They were found to vary between g l 1 which is similar to detection limits reported for HPLC systems [6,14]. With the parameters used for data acquisition this calculates to absolute IMS detection limits of pg, which compares to the lowest reported values for commercial vapor phase IMS systems [1,3]. The calibration curves of the compounds showed excellent linearity over a concentration range of at least two orders of magnitude having r 2 -values greater than These results are especially promising with respect to the need of establishing a quantitative IMS determination method for explosives Separation of three common explosi es by IMS A number of different methods for analyzing explosives from complex environmental matrices using liquid chromatographic and gas chromatographic separations have been reported so far [4 6,14,15]. The separation capabilities of ESI- IMS-MS demonstrating a resolving power similar to liquid chromatography has already been demonstrated [13]. As the IMS also shows high sensitivity for many of the explosives, the use of ESI/IMS for analyzing environmental samples was taken into consideration.

7 G. Reid Asbury et al. / Talanta 50 (2000) The experimental data showed measurable differences in the K 0 -values for most of the compounds (see Table 1). Thus, a mixture of three common explosives at concentrations of 1 mg l 1 (TNT) and 2 mg l 1 (RDX and HMX), respectively was analyzed to evaluate the separation capabilities of the instrument. In order to enhance the sensitivity of the system to RDX and HMX, 2 mm NaCl was added to the mixture. The resulting spectrum is shown in Fig. 5. Although the instrument was not optimized to give the highest possible resolution, the resolving power was sufficient to achieve baseline separation of the three compounds. The actual resolving power calculated from the data was 57 5 for the three peaks. In chromatographic terms, this corresponds to about theoretical plates. Additionally, it must be pointed out that the whole separation was achieved in less than 25 ms per single spectrum resulting in a total analysis time of only 6.4 s for averaging 256 spectra. This indicates that ESI-IMS could be a method to rapidly analyze environmental samples for mixtures of explosives at trace concentrations. Further investigations including improvement of resolution and analyzing real samples are currently in preparation. Acknowledgements This project was in part supported by research grants from the Federal Aviation Administration (Grant c97g009) and the U.S. Army Research Office (Grant DAAG ). Additionally, an ACS research fellowship was awarded to Reid Asbury sponsored by the Society of Analytical Chemistry of Pittsburgh. References Fig. 5. Ion mobility spectrum of a mixture of TNT (m/z 226), RDX (m/z 257) and HMX (m/z 331) showing baseline separation of these explosives. Concentrations were 1 mg l 1 for TNT and 2 mg l 1 for RDX and HMX in methanol water (9/1, v/v) additionally containing 2 mm NaCl. [1] L.L. Danylewych-May, C. Cummings, Explosive and taggant detection with Ionscan, in: J. Yinon (Ed.), Advances in Analysis and Detection of Explosives, Kluwer Academic Publishers, Dordrecht, 1993, pp [2] R.K. Ritchie, F.J. Kuja, R.A. Jackson, A.J. Loveless, L.L. Danylewych-May, Recent developments in IMS detection technology, in: G. Harding, R.C. Lanza, L.J. Myers, P.A. Young (Eds.), Substance Detection Systems, International Society for Optical Engineering, Bellingham, WA, 1994, pp [3] D.D. Fetterolf, Detection of trace explosive evidence by ion mobility spectrometry, in: J. Yinon (Ed.), Advances in Analysis and Detection of Explosives, Kluwer Academic Publishers, Dordrecht, 1993, pp [4] S. Barshick, W.H. Griest, Anal. Chem. 70 (1998) [5] S.D. Harvey, H. Galloway, A. Krupsha, J. Chromatogr. A 775 (1997) 117. [6] C.A. Weisberg, M.L. Ellickson, Am. Lab. 30 (1998) 32N. [7] R. Wilson, A. Brittain, Explosives detection by ion mobility spectrometry, in: J.E. Dolan, S.S. Langer (Eds.), Explosives in the Service of Man, Royal Society of Chemistry, Cambridge, 1997, pp [8] G.E. Spangler, P.A. Lawless, Anal. Chem. 50 (1978) 884. [9] A.H. Lawrence, P. Neudorfl, Anal. Chem. 60 (1998) 104.

8 1298 G. Reid Asbury et al. / Talanta 50 (2000) [10] B. Casetta, F. Garofolo, Org. Mass Spectr. 29 (1994) 517. [11] J. Yinon, J.E. McClellan, R.A. Yost, Rapid Commun. Mass Spectr. 11 (1997) [12] D. Wittmer, Y.H. Chen, B.K. Luckenbill, H.H. Hill, Anal. Chem. 66 (1994) [13] C. Wu, W.F. Siems, G.R. Asbury, H.H. Hill, Anal. Chem. 70 (1998) [14] U. Lewin, J. Efer, W. Engewald, J. Chromatogr. A 730 (1996) 161. [15] F. Garofolo, V. Migliozzi, B. Roio, Rapid Commun. Mass Spectrometr 8 (1994) 527..