Liquid chromatography/mass spectrometric analysis of explosives: RDX adduct ions

Transcription

1 RAPID COMMUNICATIONS IN MASS SPECTROMETRY Rapid Commun. Mass Spectrom. 2003; 17: Published online in Wiley InterScience ( DOI: /rcm.1006 Liquid chromatography/mass spectrometric analysis of explosives: RDX adduct ions Alexei Gapeev, Michael Sigman and Jehuda Yinon* National Center for Forensic Science, University of Central Florida, P.O. Box , Orlando, FL , USA Received 28 January 2003; Revised 27 February 2003; Accepted 27 February 2003 In liquid chromatography/mass spectrometry (LC/MS) of 1,3,5-trinitro-1,3,5-triazacyclohexane (RDX), attachment of an anion to the analyte molecule is the major way of producing characteristic ions under electrospray ionization (ESI) and atmospheric pressure chemical ionization (APCI) conditions. The formation of RDX cluster ions in LC/MS and the origin of the clustering agents have been studied. In order to determine whether the clustering anions originate from selfdecomposition of RDX in the source or from impurities in the mobile phase, isotopically labeled RDX ( 13 C 3 -RDX and 15 N 6 -RDX) and isotopically labeled glycolic acid, acetic acid, ammonium formate and formaldehyde have been used in order to establish the composition and formation route of RDX adduct ions produced in ESI and APCI sources. The results showed that, in ESI, self-decomposition of RDX plays no role in adduct ion formation; rather, RDX clusters with formate, acetate, hydroxyacetate, and chloride anions present in the mobile phase as impurities at ppm levels. In APCI, part of the RDX molecules decompose yielding NO 2 species which in turn cluster with a second RDX molecule producing abundant [MþNO 2 ] cluster ions. Copyright # 2003 John Wiley & Sons, Ltd. Liquid chromatography/mass spectrometry (LC/MS) is already a well-established technique for the analysis of trace levels of explosives in complex matrices, such as postblast residues 1,2 and soil extracts. 3,4 LC/MS mass spectra of RDX (1,3,5-trinitro-1,3,5-triazacyclohexane), a widely used military heterocyclic nitramine explosive, have been studied by several groups. 1,3 6 However, both electrospray ionization (ESI) and atmospheric pressure chemical ionization (APCI) seem to produce rather inconsistent mass spectra of RDX, even under similar experimental conditions, resulting in the production of an array of ions. Furthermore, relative abundances of the ions fluctuate to a significant extent, depending on analyte concentration, the presence of impurities in the mobile phase, and contamination of the LC/MS system. 7 *Correspondence to: J. Yinon, National Center for Forensic Science, University of Central Florida, P.O. Box , Orlando, FL , USA. jyinon@mail.ucf.edu In order to be a useful analytical technique for the analysis of RDX, mass spectrometry must produce reproducible spectra while tolerating minor variations in experimental conditions. Mass spectral consistency is of paramount importance for accurate quantitation. In gas-phase ionization methods, such as electron and chemical ionization, RDX forms well-characterized and reproducible mass spectra Post-column additives which bind with nitramines, such as chloride, nitrite, trifluoroacetate (TFA), have been used to control the ionization process and produce easily identifiable [nmþx] adduct ions, 6 where X is an additive anion and n ¼ 1 or 2. However, in many cases when the concentration of the additive was low, the adduct ion was accompanied by an additional adduct ion formed from impurity anions present in the mobile phase, resulting in the formation of two types of adducts, namely, RDX bound to anions already present in the mobile phase and RDX bound to the additive. Increase in the additive concentration often resulted in mass spectra that were difficult to interpret because of the dominance of the additive ions. A negative-ion ESI mass spectrum of RDX was acquired by flow injection into an eluent consisting of 5 mm ammonium acetate in methanol/water (75:25), 12 resulting in ions with m/z values of 267, 268, 281, and 297. The most abundant ion at m/z 281 ([Mþ59] ) was assigned as an RDX-acetate adduct. In a recent paper describing optimization of APCI conditions for the detection of high explosives, 13 the negative ion APCI mass spectrum of RDX in methanolic solution contained low-abundance ions at m/z 221 ([M H] ) and 222 ([M]. ), and abundant ions at m/z 267 and 268 interpreted as Copyright # 2003 John Wiley & Sons, Ltd.

2 944 A. Gapeev, M. Sigman and J. Yinon [MþNO 2 H]. and [MþNO 2 ], respectively. When NH 4 Cl was used as an additive, the instrument response was significantly improved and there was only one major ionic species in the mass spectrum, [MþCl] at m/z 257/259. In a study on the environmental fate of RDX, 4 the ESI mass spectrum of a water/methanol solution of RDX produced a very intense ion forming the base peak at m/z 297 ([Mþ75] ), and an ion at m/z 267 ([Mþ45] ). These ions were interpreted as [MþNO 2 þch 3 O 2H] and [MþNO 2 H], respectively. In a study on analysis of munitions residues in water, the ESI mass spectrum of RDX in methanol/water containing 10 mm ammonium formate 3 yielded a major ion at m/z 267 ([Mþ45] ), interpreted as an [RDXþCHOO] adduct ion. High field Fourier transform ion cyclotron resonance (FTICR) mass spectrometry makes possible unambiguous assignment of elemental composition of the trapped ions. ESI-FTICRMS was used to establish compositions of the ions produced by ESI-MS of RDX solution in acetonitrile, containing two different additives, ammonium acetate (NH 4 CH 3 COO) and ammonium hydroxide (NH 4 OH). 14 The ions at m/z 267 ([Mþ45] ) and 281 ([Mþ59] ), produced in the mass spectrum, were assigned as [RDXþHCOO] and [RDXþC 2 H 3 O 2 ], respectively. The ion at m/z 297 ([Mþ75] ) was assigned as [RDXþHCOOþH 2 CO]. Table 1 summarizes the adduct ions observed in LC/MS spectra of RDX with different additives. It can be seen that the use of a post-column additive does not necessarily result in a spectrum containing a single (or dominant) ionized molecular species. The purpose of the present study was to understand the role of additives and impurities in the formation of adduct ions in the ESI and APCI mass spectra of RDX, the factors that affect adduct formation, and to explain the discrepancies in the results obtained in various previous studies. EXPERIMENTAL Acetic acid, 13 C 1 -acetic acid, glycolic acid, formic acid, 13 C 1 - formic acid, 13 C 1 -bromoacetic acid, 13 C 1 -formaldehyde, ammonium hydroxide, ammonium chloride and methanol ( Purge and Trap grade) were obtained from Sigma-Aldrich (Milwaukee, WI, USA) and were used as received, except as noted. Methanol and water (HPLC grade) were obtained from Burdick & Jackson (Muskegon, MI, USA). Isotopically labeled glycolic acid was synthesized 15 by hydrolysis of isotopically labeled bromoacetic acid in a sealed ampoule at 1208C. Ammonium formate and 13 C 1 -ammonium formate were synthesized from formic acid and 13 C 1 -formic acid, respectively. 16 RDX and HMX standards were purchased as 1 mg/1 ml solutions in acetonitrile from Supelco (Bellefonte, PA, USA). RDX in solid form was obtained from FBI Laboratories (Washington, DC, USA) and was further purified by crystallization from glacial acetic acid followed by drying in vacuum. 13 C 3 -RDX was purchased from Cambridge Isotope Laboratories (Andover, MA, USA) as 1 mg/1 ml acetonitrile solution. Acetonitrile was evaporated under vacuum before making stock solutions. 15 N 6 -RDX was purchased from Applied Biosystems (Boston, MA, USA) as solid material. MS analyses were performed using a LCQ DUO ion trap mass spectrometer (Thermo-Finnigan, San Jose, CA, USA). Helium was used as damping and collision gas, while nitrogen was used as sheath and auxiliary gas. RDX was dissolved in methanol/water (1:1) and was injected into a Table 1. Adduct ions in LC/MS mass spectra of RDX Mobile phase Additive Instrument ESI or APCI Ions Ion assignment Reference Acetonitrile/water (50:50) Methanol/water (75:25) 2 mm ammonium acetate [(NH 4 )COOCH 3 ] 5 mm ammonium acetate [(NH 4 )COOCH 3 ] Quadrupole ESI [Mþ59] [RDXþCH 3 COO] 1 Quadrupole ESI [Mþ44] 12 [Mþ45] [Mþ59] [RDXþCH 3 COO] [Mþ75] Ion Trap APCI [Mþ35] [RDXþ 35 Cl] 13 [Mþ37] [RDXþ 37 Cl] Methanol/water (50:50) 0.5 mm ammonium chloride (NH 4 Cl) Methanol-water (50:50) No additive Ion Trap APCI [M 1] [RDX H] 13 M [RDX] [Mþ45] [RDX NO 2 H] [Mþ46] [RDX NO 2 ] Methanol/water (50:50) No additive Triple quadrupole ESI [Mþ45] [RDXþNO 2 H] 4 [Mþ75] [RDXþNO 2 þch 3 Oþ2H] Methanol/water (50:50) 10 mm ammonium Ion trap ESI [Mþ45] [RDXþCHOO] 4 formate (NH 4 HCOO) Methanol-water (50:50) No additive Ion trap ESI [M 1] [M H] 5 [Mþ45] [MþNO 2 H] Acetonitrile Acetonitrile 0.5% ammonium acetate (NH 4 CH 3 COO) 0.5% ammonium hydroxide (NH 4 OH) [Mþ46] [MþNO 2 ] [Mþ59] [MþNNO 2 H] FTICR ESI [Mþ45] [RDXþCHOO] 14 [Mþ59] [RDXþCH 3 CO 2 ] FTICR ESI [Mþ62] [RDXþNO 3 ] 14 [Mþ75] [RDXþH 2 COþHCO 2 ] [Mþ89] [RDXþH 2 COþCH 3 CO 2 ]

3 RDX adduct ions in ESI and APCI 945 Figure 1. ESI mass spectrum of RDX mL/min flow of the same solvent through a 5-mL sample loop, unless noted otherwise. The capillary potential and the heated capillary temperature in the ESI mode were set at 4 kv and 1758C, respectively. RESULTS AND DISCUSSION RDX adduct ion composition in ESI Figure 1 shows the ESI mass spectrum obtained by injection of 20 ml of a 1-ppm solution of RDX into a 0.15-mL/min water/methanol (1:1) flow. [2Mþ75] and [Mþ75] are the most abundant ions in the RDX concentration range of 25 to 10 ppm. Lower RDX concentrations lead to lower [2Mþ75] / [Mþ75] abundance ratios. At higher RDX concentrations formation of the [2Mþ75] dimer becomes more favorable, so that the [2Mþ75] /[Mþ75] ratio increases. Other ions present in the mass spectrum are [Mþ45],[Mþ59] and [2Mþ35] and [2Mþ37]. The origin of the chloride anion is probably from the water; the ESI mass spectrum (not shown) of RDX in pure methanol, without water, did not show any chloride adduct ions. On another hand, the use of pure water as mobile phase resulted solely in [MþCl] and [2MþCl] ions. In order to establish whether the adduct ions contain RDX decomposition fragments, we acquired ESI mass spectra with 13 C 3 -RDX and 15 N 6 -RDX, using identical experimental conditions. The mass spectra are shown in Figs. 2 and 3, respectively. When 13 C 3 -RDX was used (Fig. 2), there were shifts of 3 Da for each of the [Mþ45],[Mþ59], and[mþ75] ions, and shifts of 6 Da for each of the [2Mþ35], [2Mþ37], [2Mþ59] and [2Mþ75] ions. When 15 N 6 -RDX was used (Fig. 3), shifts of 6 Da were observed for [Mþ45],[Mþ59],and[Mþ75], and shifts of 12 Da were observed for [2Mþ35], [2Mþ37], [2Mþ59] and [2Mþ75] ions. These results confirmed that [2Mþ35], [2Mþ37], [2Mþ59] and [2Mþ75] indeed contain two RDX molecules, and that RDX decomposition plays no role in ion formation in the ESI source. [Mþ75] ion composition The elemental composition of the ion at m/z 297 was determined by FTICR 14 to be [RDXþC 2 H 3 O 3 ]. However, it is impossible to deduce any structural information based solely on this finding. As was pointed out in the original work, 14 RDX adducted with one formate anion (HCO 2 ) plus one formaldehyde molecule (H 2 CO) would have this elemental composition. It is unlikely that a formaldehyde molecule participates in formation of the adduct ion at m/z 297 because this molecule does not readily form an anion and therefore will not attach to the RDX molecule. Addition of 13 C-labeled formaldehyde to the mobile phase up to 250 ppm did not result in any change in the mass spectra, which is another indication that formaldehyde is not a constituent of the [Mþ75] ion. Glycolic (hydroxyacetic) acid, CH 2 (OH)COOH, forms an anion [M H] that fits the C 2 H 3 O 3 elemental composition. ESI-MS of RDX with 0.5 mm 13 C 1 -glycolic acid as an additive resulted in a mass spectrum containing two ions at m/z 298 and 520. These ions correspond to one and two RDX molecules clustered with one 13 C 1 -glycolic acid anion. A competition between 13 C 1 -glycolic acid and glycolic acid, present in methanol as impurity, was used to confirm the composition of the ion at m/z 297. Figure 4 shows a plot of abundances of the ions at m/z 297 and 298 versus 13 C 1 -glycolic acid concentration in the mobile phase. The abundances become equal at 4 ppm 13 C 1 -glycolic acid, the concentration at which RDX adduct ions are equally distributed between labeled and unlabeled (impurity) glycolic acid. Assuming that the probabilities of formation of RDX adduct ions with labeled and unlabeled glycolic acid are equal, i.e., excluding any possible isotope effects, this concentration can be considered as the upper limit of glycolic acid impurity in

4 946 A. Gapeev, M. Sigman and J. Yinon Figure 2. ESI mass spectrum of 13 C 3 -RDX. the mobile phase. Taking into account the water content in the mobile phase, the glycolic acid impurity in methanol is estimated as 8 ppm (upper limit). [Mþ45] ion composition ESI-FTICRMS accurate mass measurement established that the [Mþ45] ions are composed of one RDX molecule, one hydrogen atom, two oxygen atoms and one carbon atom. These species were proposed to be formate anions clustered with one RDX molecule. 14 The formate anions were thought to be a product of the reaction between water and HCN from RDX self-decomposition. However, as previously shown, our 13 C 3 -RDX and 15 N 6 -RDX ESI data do not support the RDX self-decomposition origin of formic acid in the mobile phase. Furthermore, in a recent ab initio study, 17 it was indicated that the HCN route of RDX self-decomposition has a threshold 12 kcal/mol higher than that for N N homolytic bond cleavage with NO 2 elimination. The only other possible formate source is the mobile phase itself. 13 C-Labeled ammonium formate was used to Figure 3. ESI mass spectrum of 15 N 6 -RDX.

5 RDX adduct ions in ESI and APCI 947 Another species with CH 3 COO elemental composition, the acetate anion, can be assumed to be present in the mobile phase along with formic acid. Competition for RDX between acetate, present as an impurity in methanol, and added 13 C 1 - labeled acetic acid, was used to estimate the acetate content in the mobile phase. The water/methanol mixture contained 30 ppm acetate, which corresponds to 60 ppm acetate in pure methanol (upper limit). When using Purge and Trap grade glass-distilled methanol instead of HPLC-grade methanol in the mobile phase, the abundance of the [Mþ45] and [Mþ59] ions was lower, which is an additional indication of the source of these adduct ions. Figure 4. Plot of ion abundances for m/z 297 and 298 as a function of 13 C 1 -glycolic acid concentration in the mobile phase. estimate the formate impurity concentration in methanol. [MþCHOO] and [Mþ 13 CHOO] ion abundances become equal at 50 ppm 13 C-formate concentration in the mobile phase. Assuming that 12 C- and 13 C-formate anions have equal affinity for RDX, and excluding any possible isotope effects, we can conclude that the mobile phase has an upper limit of 50 ppm formate as an impurity, corresponding to 100 ppm formate impurity (upper limit) in HPLC-grade methanol. [Mþ59] ion composition The [Mþ59] ions were assigned previously as RDX clustered with deprotonated oxiran-2-ol, a product of the reaction between acetonitrile and water. 14 Substitution of RDX from the stock acetonitrile solution with solid RDX resulted in an identical mass spectrum. Thus, by excluding acetonitrile, we have eliminated deprotonated hydroxyoxirane as a possible clustering agent. APCI: RDX decomposition APCI is a soft gas-phase ionization technique, which produces ions at atmospheric pressure by action of a corona discharge in a spray of vaporized analyte from solution. Due to a large ( ) solvent-to-analyte ratio, even a small amount of impurity (several ppm) with affinity for RDX may result in the production of adduct ions. On the other hand, RDX is known to decompose at 2308C, 18 with the loss of NO 2 as the lowest energy route, 17 so that formation of [RDXþNO 2 ] in the gas phase is feasible even in the heated capillary during evaporation. Negative chemical ionization (NCI) of RDX in the gas phase results in abundant [MþNO 2 ] adducts from RDX self-decomposition. 19 As both APCI and NCI rely on ion production from gas-phase analyte molecules, one may expect similarities in the mass spectra. The APCI mass spectrum resulting from an injection of 20 ml of 50-ppm RDX solution in water/methanol (1:1) into a 0.3-mL/min flow is shown in Fig. 5. The use of 15 N 6 -RDX resulted in the mass spectrum shown in Fig. 6. The ion at m/z 267 becomes m/z 273 (shift of 6 Da), whereas there is a shift of 7 Da for the ion at m/z 268. These differences in mass shifts Figure 5. APCI mass spectrum of RDX.

6 948 A. Gapeev, M. Sigman and J. Yinon Figure 6. APCI mass spectrum of 15 N 6 -RDX. prove that in the APCI source the RDX molecule decomposes, generating NO 2 species which in turn cluster with a second RDX molecule to form [MþNO 2 H]. Addition of 0.1 mm of formate or chloride in the mobile phase suppresses NO 2 adduct formation, thus making possible the formation of a desired adduct ion. CONCLUSIONS LC/MS is a powerful technique for the analysis of trace levels of RDX, which relies on formation of cluster ions of analyte molecules with an anion formed from an additive or background molecule. RDX self-decomposition and mobile phase impurities were proposed to be the sources of anions to account for the RDX adducts in ESI and APCI mass spectrometry. Results obtained with isotopically labeled RDX showed that RDX self-decomposition is not involved in the formation of adduct ions in the ESI source. In ESI, RDX was found to cluster with formate, acetate, hydroxyacetate, and chloride anions, which are present in the mobile phase as impurities at ppm levels. In APCI some of the RDX molecules decompose, yielding NO 2 species which in turn cluster with other RDX molecules, thus producing abundant [MþNO 2 ] cluster ions. In ESI, addition of an additive such as ammonium nitrate, in an amount of about 0.1 mm, will suppress adduct ions due to background contaminants and form a major [MþNO 3 ] ion. REFERENCES 1. Casetta B, Garofolo F. Org. Mass Spectrom. 1994; 29: Yinon J. Forensic Sci. Rev. 2001; 13: Cassada DA, Monson SJ, Snow DD, Spalding RF. J. Chromatogr. A 1999; 844: Beller HR, Tiemeier T. Environ. Sci. Technol. 2002; 36: Yinon J, McClellan JE, Yost RA. Rapid Commun. Mass Spectrom. 1997; 11: McClellan JE. Fundamentals and Applications of Electrospray Ionization-Quadrupole Ion Mass Spectrometry for the Analysis of Explosives, PhD thesis, University of Florida, Gainesville, FL, McClellan JE, Mulholland JJ, Murphy JP, III, Yost RA. Proc. 46th ASMS Conf. Mass Spectrometry and Allied Topics, Dallas, TX, Yinon J. Biomed. Mass Spectrom. 1974; 1: Zitrin S, Yinon J. In Advances in Mass Spectrometry in Biochemistry and Medicine, vol. 1. Spectrum Publications: New York, 1976; Yinon J. J. Chromatogr. A 1996; 742: Yinon J, Zitrin S. Modern Applications in Analysis of Explosives, John Wiley: Chichester, Schreiber A, Efer J, Engewald W. J. Chromatogr. A 2000; 869: Evans CS, Sleeman R, Luke J, Keely BJ. Rapid Commun. Mass Spectrom. 2002; 16: Wu Z, Hendricson CL, Rodgers RP, Marshall AG. Anal. Chem. 2002; 74: Eichloff RJ. Liebigs Ann. Chem. 1905; 342: Ingersoll AW. Org. Synth. 1943; 2: Chakraborty D, Muller RP, Dasgupta S, Goddard WA. J. Phys. Chem. A 2000; 104: Lee J-S, Hsu C-K, Chang C-L. Thermochim. Acta 2002; 392/ 393: Yinon J. J. Forensic Sci. 1980; 25: 401.