VX-803

The identification of chemical attribution signatures of stored VX nerve agents using NMR, GC-MS, and LC-HRMS

Simon P.B. Ovenden, Renée L. Webster, Eva Micich, Lyndal J. McDowall, Nathan W. McGill, Jilliarne Williams, Shannon D. Zannatta

Sample 1

NMR, GC-MS and LC-MS

Source A

Sample 2

Source B

Sample 3 Source C

The identification of chemical attribution signatures of stored VX nerve agents using NMR, GC-MS, and LC-HRMS
Simon P. B. Ovenden*, Renée L. Webster, Eva Micich, Lyndal J. McDowall, Nathan W. McGill, Jilliarne Williams, Shannon D. Zannatta

Defence Science and Technology Group, 506 Lorimer St, Fishermans Bend, Victoria 3207, Australia

* Corresponding author email address: [email protected]

Abstract

The organophosphorous nerve agent VX is classified by the Chemical Warfare Convention (CWC) as a Schedule 1 chemical; namely a substance that is highly toxic with no use that is of benefit to society. Even with this classification, the nefarious use of the Schedule 1 chemical VX has been observed, as demonstrated in 2017 in Malaysia. Therefore, undertaking chemical analysis on samples of VX to identify chemical attribution signatures (CAS) for chemical forensics is required. To further understand the chemical profile of VX, and to aid in the identification of potential CAS, three in house synthesised stocks of VX were investigated. The three VX stocks analysed were synthesised in 2014, 2017 and 2018 using the same method, allowing for a comparison of data between each of the stocks at different stages of storage. As opposed to a majority of literature reports, these agent stocks were not stabilised, nor were they subjected to forced degradation. Using NMR, high resolution (HR) LC-HRMS, GC-(EI)MS and GC-(CI)MS to gain a full insight into the CAS profile, a total of 44 compounds were identified. Of these compounds, 30 were readily identified through accurate mass measurement and NIST library matches. A further seven were identified through extensive LC- HRMS/MS studies, with seven remaining unresolved. Several compounds, identified in minor amounts, were able to be traced back to impurities in the precursor compounds used in the synthesis of VX, and hence may be useful as CAS for source attribution.

Keywords

Chemical forensic signatures, chemical forensics, impurity profiling, VX, source attribution

1. Introduction

Over recent history, chemical warfare agents (CWAs) have been used in many situations against both opposing military forces and civilian populations [1-3]. Initially crude attempts were made, such as releasing chlorine gas from cylinders during trench warfare in World War I. With the development of more potent CWAs such as the vesicants (i.e. sulfur mustard and lewisite), and organophosphorous nerve agents (i.e. sarin, tabun and VX), in addition to advanced weaponry, the threat of CWAs increased significantly post World War II [1-4]. Throughout the 1980s Saddam Hussein’s regime in Iraq was notorious for using CWAs against military and civilian targets [1,2], most infamously against civilian Kurdish populations in Halabja in 1988 [1]. This attack was estimated to have killed up to 5000 people, and injured up to 10000 [1]. Eventually, the Chemical Weapons Convention (CWC) was ratified and came into effect in 1997. At the same time, the Organisation for the Prohibition of Chemical Weapons (OPCW) came into being to administer the implementation of the CWC. The ability to identify intact CWAs and degradation products from both environmental and laboratory/production facility samples is a desirable national capability for many countries. In fact, the ability to undertake these analyses and report findings has been the basis of OPCW proficiency tests conducted twice yearly since 1997.

Despite the CWC, regimes continue to use CWAs [3,4]. In recent years there have been numerous incidents involving the use of CWAs. This includes the use of VX for an assassination in Malaysia [4] and multiple incidents involving several different CWAs in Syria, with no person, group or regime admitting responsibility for their usage [3]. In response to the alleged use of CWAs in Syria, the OPCW instigated a Fact Finding Mission (FFM) from 2014 to gather evidence for the use of toxic chemicals against civilian populations. The evidence collected during the FFM led to the formation of the OPCW-UN Joint Investigative Mechanism (JIM). The purpose of the JIM was to attribute the CWA attacks to a person or persons, or a regime [3]. These investigations highlight the importance of having the tools to apply chemical forensics to environmental samples to identify chemical signatures and signals that may allow for attributing use.

Chemical forensics involves the application of analytical techniques, most commonly mass spectrometry based, to characterise materials and thereby retrospectively gain insights into their synthesis methods and possible source of origin [5]. In particular, chemical forensics relies on the identification of chemical attribution signatures (CAS), which can include residual synthetic precursors, impurities, reaction by-products and degradation products [6-8], metabolites [5] and other chemical markers such as stable isotopic ratios [9] and stereo isomeric ratios [10]. These can be used to distinguish samples of a material on the basis of their synthetic route, experimental conditions and source of precursor reagents used in production [7]. The most useful CAS are those which reliably occur in materials synthesised by a specific route, are resistant to change after coming into contact with other materials, and occur irrespective of the chemist, laboratory or facilities in which they are synthesised and stored [5].

Chemical forensics has traditionally been applied in combatting the illicit drug trade [6,11]. More recently, chemical forensics has been applied more broadly, incorporating a combination of illicit substances, pharmaceutical based agents, chemical and biological warfare agents and other toxic chemicals [6,7,12]. However, forensic investigation of highly toxic chemicals, such as chemical weapons, remains underdeveloped [8]. Despite this, there is a need for chemical forensics as applied

to the CWC Schedules of Chemicals. The nature of armed conflict is changing with the prospect of CWA use in asymmetrical conflicts and terrorism a continued possibility. This by its very nature makes it more important to identify and hold perpetrators to account [8,12].

As part of an ongoing program identifying CAS from stocks of CWAs of interest for chemical forensics, three existing stocks of VX of differing ages were analysed using nuclear magnetic resonance (NMR) spectroscopy, gas chromatography-mass spectrometry (GC-MS) with electron ionisation (EI) and chemical ionisation (CI), and liquid chromatography-high resolution mass spectrometry (LC-HRMS). There have been several previous literature reports on the chemical analysis of the degradation of VX samples by either GC-MS [13-18] and LC-HRMS [13,19]. However, none of these have used a suite of techniques to comprehensively analyse the chemical signature of VX. Understanding the trace chemical composition of these agent stocks allows for CAS to be identified. Discussed below are the results of the chemical analysis of these VX agent stocks, the different structural classes identified, and the structural elucidation of several compounds not previously reported.

2. Experimental

2.1 Safety and legal matters

Caution: Nerve agents are highly toxic and rigorous protective measures are required to work safely with these compounds. They need to be handled with extreme care. All work performed during these investigations were conducted in a designated laboratory within an appropriately equipped fumehood by highly trained staff. All glassware and laboratory consumables used in the preparation of these samples were appropriately decontaminated and disposed of after contact with the nerve agent. The synthesis of CWAs is restricted by the CWC. DST conducted the synthesis of VX in their protective purposes laboratory under the provisions of the CWC. Unauthorised preparation could be considered a criminal offence.

2.2 Materials

The VX analysed was synthesised three times via the same synthetic method during a five year period, in 2014 (VX2014), 2017 (VX2017) and 2018 (VX2018). The precursor chemicals 2-(N,N- diisopropylamino)ethylchloride hydrochloride, O,O’-diethyl methylphosphonothioate and phosphorous trichloride were obtained from Sigma-Aldrich. All chromatographic grade solvents (acetonitrile, dichloromethane, hexane, water) and derivatizing agents used in this analysis were obtained from Merck (Castle Hill, NSW, Australia). Deuterated chloroform (CDCl3) used for NMR analysis was supplied by Cambridge Isotope Laboratories Inc. (USA).

2.3 Sample Preparation

For NMR analysis, all VX agent stocks (approximately 10 mg) were dissolved in 1 mL of CDCl3. A 600 L aliquot was added to a 5 mm NMR tube immediately prior to analysis.
For all GC-(EI)MS on neat material, aliquots of all VX agent stocks were diluted to 100 ppm in hexane. For GC-(CI)MS, aliquots of all VX agent stocks were diluted to 500 ppm in hexane. Following these dilutions, 100 L aliquots of each diluted solution were added to a 2 mL glass vial with limited volume insert (LVI), capped and analysed. Similarly for LC-HRMS, aliquots of all VX agent stocks were diluted to 100 ppm in acetonitrile. Aliquots (50 L) of each diluted solution were added to a 2 mL glass vial with LVI, capped and analysed.

2.4 Trimethylsilyl derivatization of VX

To derivatize the samples, a 200 L aliquot of 100 ppm VX agent stocks was treated with 20 µL of N,O-bis(trimethylsilyl)trifluoroacetamide (BSTFA) with 1% trimethylchlorosilane (TMCS), and the solution heated at 40 °C for 30 min. On completion, the sample was split into two approximately 100
L aliquots, added to a 2 mL glass vial with LVI, and analysed by GC-(EI)MS.
2.5 Sample analysis

2.5.1 NMR Spectrometer

All NMR data were collected on a Bruker Avance NMR spectrometer (Bremen, Germany) operating at a 1H NMR frequency of 500.13 MHz running Bruker Topspin 3.2 NMR software. The spectrometer was equipped with a standard geometry 5 mm diameter BBO (Broad Band Observe)

probe head. All one- (1H, 13C, 31P) and two-dimensional NMR data (COSY, 1H-13C gHSQC, 1H-13C gHMBC) were collected using standard Bruker pulse sequences. The probe temperature was set to 298 K, and standard processing parameters were used. The 1H-31P Fast-gHMQC experiment was a non-standard Bruker pulse sequence run at 298 K, using recommended acquisition and processing parameters [20]. All NMR spectra were referenced to residual protons in CDCl3.

2.5.2 GC-FPD/(EI/CI)MS

An Agilent 7890A-5975C GC-MS (Santa Clara, USA) was used for all analyses. Dual analytical columns, both DB-5ms Ultra Inert 30 m × 0.25 mm × 0.25 µm were connected to a mass selective detector and a dual flame photometric detector. Ultra-high purity He was used as carrier gas and supplied at constant flow rate of 1.2 mL/min. For underivatised samples the GC oven was programmed from 40 °C (3 min hold) then ramped to 300 °C at 10 °C per min (3 min hold) with an MS solvent delay of 3.5 min. For the BSTFA-derivatised samples the GC oven was programmed from 40 °C (4 min hold) then ramped to 90 °C at 60 °C per min (1.2 min hold) then at 10 °C per min to 300
°C (5 min hold) with an MS solvent delay of 6.7 min. The FPD was heated to 200 °C with combustion gas flow rates at 75:100:50 mL/min (H2:air:N2). Injections of 1 µL were made with the injectors operated in pulsed splitless mode at 250 °C. Each set of samples were analysed with an accompanying quality control and retention index standard test mixture. EI data were collected in full scan mode from m/z 40-650 at 70 eV. CI data were collected in both positive (PCI) and negative (NCI) chemical ionisation modes with methane reagent gas. For PCI, the MS was operated at 210 eV and NCI analysis was at 225 eV. Data were acquired using Agilent MassHunter (version B.07) and analysed using MassHunter and Automatic Mass Spectral Deconvolution and Identification Software (AMDIS, Gaithersburg, MD, USA) version 2.72, with EI library matching performed against the OPCW Central Analytical Database (OCAD) and NIST libraries.

2.5.3 LC-HRMS

LC-HRMS data were collected on an Agilent 6540 QTOF mass spectrometer connected to an Agilent 1260 Infinity liquid chromatography system comprising of an in-line degasser, binary pump, auto-injector, column heater and diode array detector. For mass spectral data collection, the mass range scanned was from m/z 50-1000 at 1 spectra/s. Positive electrospray ionisation (ESI) was applied with a capillary voltage of 4000 V, a fragmentor voltage 180 V, gas temperature 300 °C , drying gas flow at 10 L/min, nebuliser pressure at 45 psi and sheath gas temperature 350 °C. All tandem MS experiments used fixed collision energy of 20 V. A 5 µL injection of each sample was made onto an Agilent 300SB-C18 column (2.1 mm x 100 mm, 1.8 µm, 300 Å) at 0.3 mL/min, with the column temperature set at 40 °C. Samples were subjected to linear gradient elution from water (+ 0.1% formic acid) to acetonitrile (+ 0.1% formic acid) over 26 min. Data were collected and analysed using Agilent MassHunter software.

2.6 Data Processing and Analysis

Peaks of interest were identified in the first case by analysis with AMDIS with matching performed against the NIST and OPCW Central Analytical Database (OCAD) vgwd_2018 mass spectral libraries using a nominal match threshold of 800. All compounds identified in this manner were designated a tentative match prior to further confirmatory analysis. Where authentic standards were available, these were used to assign compound identifications unambiguously. In those cases

where authentic standards were not available, evidence supporting identifications to a sufficient degree of certainty was sought through calculation of retention indices, derivatisation with BSTFA, determination of molecular weight using chemical ionisation mass spectrometry, determination of molecular formula using high resolution mass spectrometry and mass spectral interpretation.

3. Results and Discussion

Aliquots of the three VX agent stocks were characterised by GC-(EI)MS and GC-(CI)MS either neat or after BSTFA derivitisation, LC-HRMS and one- and two-dimensional NMR. Analysis of these data obtained from multiple platforms, in combination with NIST mass spectral library matching and LC-HRMS/MS data, allowed for the identification of 44 compounds, of which the structures of 37 were elucidated.

3.1 Analysis by nuclear magnetic resonance spectroscopy

Each VX agent stock was initially analysed by NMR to assess its purity, and gain an insight into the most abundant compounds present. Experiments run included 1H, COSY, 31P, 1H-31P gHSQC, 13C, 1H-13C gHSQC and 1H-13C gHMBC. The 1H and 31P NMR data for the three VX agent stocks are shown in Fig. 1. Readily identifiable in these data for VX2017 and VX2018 are resonances consistent with the presence of VX as denoted in Fig. 1a and b respectively. However these resonances were absent in the 1H NMR spectrum for VX2014 (Fig. 1a), as was the expected phosphorous resonance for VX in the 31P NMR spectrum (Fig. 1b). The corresponding two-dimensional data supporting the assignments is shown in the Supplementary material Fig A1.

Further analysis of the NMR data allowed for the identification of some impurities in all of the stocks analysed. For VX2014 and VX2017, analysis of the two-dimensional NMR data enabled the identification of ethyl methylphosphonic acid (9), in addition to bis-2-(N,N- diisopropylaminoethyl)disulfide (27). Key resonances aiding in the identification of these two compounds are shown in Fig. A1. The presence of these compounds is indicative of the degradation pathway when the pH is < 6 or > 10 [21,22].

3.2 Analysis by mass spectrometry

The 44 compounds identified by mass spectrometry in the three VX agent stocks are listed in Table 1. As no stabilisers were added to these agent stocks, all identified compounds were either a byproduct of the synthesis, a degradation product formed on storage, unreacted precursors, or contaminants from the precursors. Of the 44 compounds identified, the structures of 28 were readily identified through NIST library matches (> 80% match factor) of the GC-(EI)MS mass spectra, and LC-HRMS data (< 5 ppm mass accuracy). The structures for these compounds are shown in Supplementary material Fig. A2. An additional two compounds, O-ethyl N-isopropyl methylphosphonamidate (11) [16] and either 2-(N,N-diisopropylamino)ethyl ethyldisulfide or 2-[2- (N,N-diisopropylamino)ethylthio]ethanethiol (35) [16], remain tentative assignments as no meaningful mass spectral fragmentation data could be generated.

The structures of seven compounds remain unresolved. All seven compounds yielded fragment ions (m/z 114, m/z 72) in the GC-(EI)MS consistent with the presence of a 2-diisopropylamino-1- thioethyl moiety [14-16]. Three of the unknown compounds gave no phosphorous or sulfur signal in the FPD. Another three unknown compounds gave a FPD response only on the sulfur channel and one unknown compound only on the FPD phosphorous channel. A further seven compounds had their structures elucidated through the interpretation of LC-HRMS/MS data, and where possible compared with the literature. Previous gas phase fragmentation studies of both amiton (VG) [23] and VX [24] were beneficial in the interpretation of the generated LC-HRMS/MS data in this study.

The structures of VX, the highly toxic degradation product of VX, EA 2192, in addition to the seven elucidated compounds are shown in Fig. 1.
Some further observations could be made from the collected mass spectral data. The presence of 2-(N,N-diisopropylamino)ethylchloride (24) was identified in all three VX agent stocks. The observation of 2-(N,N-diisopropylamino)ethylchloride (24) could be a useful CAS for source attribution, as it is was one of the precursor compounds for the synthesis of VX. Several other compounds were identified in the VX agent stocks, including diisopropylamine (21) [13,15-17], 2- (N,N-diisopropylamino)ethanethiol (25) [13-17], 2-(N,N-diisopropylamino)ethanethial (22) [15,16] and 2-(N,N-diisopropylamino)ethyl vinyl sulfide (23) [15,16,25] (Supplementary material Fig. A2). The presence of these small reactive unsaturated sulfur containing compounds may explain the presence of some of the observed degradation compounds discussed below [25].

Previous literature reports on the GC-(EI)MS analysis of degraded VX samples highlighted the difficulty in identifying the structures of compounds containing the 2-diisopropylamino-1-thioethyl moiety [14-16]. The generated mass spectra from these compounds are generally information poor with the only EI fragment ions observed at m/z 114 and m/z 72 only [14-16]. Hence, in addition to the GC-(EI)MS data, either or both of GC-(CI)MS and LC-HRMS were needed to establish the molecular weight of identified compounds. Where possible GC-(EI)MS spectral matching using both NIST and OCAD databases, in addition to LC-HRMS and LC-HRMS/MS, were combined to confirm the structure of identified compounds. Again, the presence of fragment ions in LC-HRMS/MS data at m/z 128.1423, m/z 114.1280 and m/z 86.0961 allowed for the confirmation of the presence of the 2- diisopropylamino-1-thioethyl moiety in the compound of interest [24].

3.3 Phosphorous containing compounds

Using a combination of techniques, eighteen phosphorous containing compounds were identified (Table 1). VX was only detectable in VX2017 and VX2018, as was EA 2192. Conversely, ethyl 2-diisopropylaminoethyl methylphosphonate (12), ethyl methylphosphonic acid (9), diethyl methylphosphonate (10) and methyl phosphonic acid (8) (Supplementary material Fig. A2) were identified in all three agent stocks. The presence of both ethyl methylphosphonic acid (9) and EA 2192 suggested a change in pH on storage of the VX stocks [22]. As previously reported [22], EA 2192 is produced when the pH is in the range of 7 to 10. It is proposed that as EA2192 is produced, the pH drops (pH < 6) and drives the degradation reaction towards the formation of ethyl methylphosphonic acid (9) and 2-(N,N-diisopropylamino)ethanethiol (25) [22].

3.4 Phosphonothioates

In addition to VX and EA 2192, an additional eight phosphonothioates were identified (Table 1). Four of these were identified through either or both of LC-HRMS and NIST library matches on the EI fragmentation data, and where possible comparison with the literature. The compounds identified were O-ethyl methylphosphonothionate (13) in the LC-HRMS, and as the TMS derivative in the GC- (EI)MS), S-[5-diisopropylamino)-3-thiapentyl] methylphosphonothiolate (14) [19], O-ethyl[S-(5-
diisopropylamino)-3-thiapentyl] methylphosphonothiolate (15) [19], and S,S-bis[2- (diisopropylamino)ethyl] methyl phosphonotrithioate (16) [14,16,17]. The remaining four phosphonothioates required additional data to be identified.

O-ethyl S-(2-isopropylaminoethyl) methylphosphonothioate (1) ([M+H]+ obs m/z 226.1030, calc 226.1025,  2.2 ppm, C8H20NO2PS) was identified as a very minor component in the LC-HRMS data for both VX2017 and VX2018. Comparison of the observed molecular weight for VX and 1 showed the difference between these two compounds was a loss of m/z 42.0468, corresponding to C3H6 and therefore potentially an isopropylamino moiety. The weak intensity of the molecular ion of 1 precluded it from being further fragmented. However, it was hypothesised that the presence of the isopropylamino moiety on 1 as opposed to the diisopropylamino moiety as observed for VX could be as a consequence of minor amounts of 2-(N-isopropylamino)ethylchloride contaminating the precursor compound 2-(N,N-diisopropylamino)ethylchloride (24). To confirm this, a sample of precursor 24 used in the synthesis of VX was analysed by both LC-HRMS and NMR. Accurate mass LC-HRMS analysis showed that in addition to 2-(N,N-diisopropylamino)ethylchloride (24) at [M+H]+ m/z 164.1185 , a compound m/z 42.0467 amu less that also contained chlorine was observed at [M+H]+ m/z 122.0720 (Supplementary material Fig. A3a). Further NMR analysis (Supplementary material Fig. A3b) confirmed that 24 used in the VX synthesis was contaminated with minor amounts of 2-(N-isopropylamino)ethylchloride (approximately 2%). This observation may explain the minor amounts of O-ethyl S-(2-isopropylaminoethyl) methylphosphonothioate (1) detected. In this instance, the presence of CAS with an isopropyl moiety as opposed to a diisopropyl moiety may be applicable to source attribution of the precursor 2-(N,N-diisopropylamino)ethylchloride (24) used for the synthesis of VX.

From the LC-HRMS and GC-(EI)MS data collected for S-ethyl S-2-(diisopropylaminoethyl) methylphosphonodithioate (2) ([M+H]+ obs m/z 284.1271, calc 284.1266,  1.8 ppm, C11H26NOPS2) it was initially unclear if it was the S-ethyl or O-ethyl isomer [5,16,19], hence LC-HRMS/MS was undertaken to identify the correct isomer. The proposed fragmentation pathway for 2 is shown in Supplementary material Fig. A4. The key fragment ions at m/z 154.9752 (loss of N,N- diisopropylaminoethyl moiety, 2a), and m/z 94.9719 (loss of C2H5S, 2b) [24] readily identified 2 as the S-ethyl isomer.

Evident in the LC-HRMS/MS fragmentation of O-ethyl S-(2-diisopropylaminoethyl) phosphonothioic acid (3) ([M+H]+ obs m/z 270.1290, calc 270.1287,  1.1 ppm, C10H24NO3PS) were two parallel fragmentation pathways (Pathway A and B, see Fig. 3). The proposed fragmentation pathway for 3 is shown in Fig. 3. Pathway A yielded the ions 3a at m/z 128.1423, and 3b at m/z 86.0961; consistent with a 2-diisopropylamino-1-thioethyl moiety [24]. Pathway B yielded 3c at m/z 169.0057, corresponding to C4H10O3PS, consistent with the loss of the 2-diisopropylamino-1- thioethyl moiety and indicative of 3 being a phosphonothioate [21]. Further fragmentation to form 3d at m/z 140.9764 (loss of C2H4) and 3e m/z 96.9495 (loss of C2H4O) were further evidence of 3 being a phosphonothioate [23]. These data allow for the structure of O-ethyl S-(2- diisopropylaminoethyl) phosphonothioic acid (3) to be proposed as shown. While previously reported in the literature [26], this is the first report of mass spectral fragmentation data for 3.

The difference between O,O-diethyl S-(2-diisopropylaminoethyl) phosphonothioate (4) and O- ethyl S-(2-diisopropylaminoethyl) phosphonothioic acid (3) was the addition of C2H4 ([M+H]+ obs m/z 298.1600, calc 298.1600,  0.0 ppm, C12H28NO3PS). Again, two parallel fragmentation pathways were observed (Pathway A and B, see Supplementary material Fig. A5). Pathway A yielded the ion at m/z 128.1432 (loss of C4H9O3PS) to form the diagnostic 2-diisopropylamino-1-thioethyl fragment 4a at m/z 128.1432 [24], and indicated the additional C2H4 was associated with the phosphonothiolate

moiety. For the phosphonthioate moiety 4b, an analogous fragmentation pathway to amiton was observed [23] through the observation of fragments 4c at m/z 197.0396, 4d at m/z 169.0081, 4e at m/z 140.9766 and 4f at m/z 109.0050. The additional fragment ions 4g at m/z 96.9510 and 4h at m/z 90.9943 allowed for the structure of O,O-diethyl S-(2-diisopropylaminoethyl) phosphonothioate (4) to be proposed as shown. This compound was previously studied during SPME and GC-MS analysis of compounds related to the CWC [27].

The identification of O,O-diethyl S-(2-diisopropylaminoethyl) phosphonothioate (4) in the VX agent stocks is linked to the precursor compound P-methylphosphonous dichloride (MePCl2). P- methylphosphonous dichloride is prepared from phosphorous trichloride (PCl3). P- methylphosphonous dichloride (76 ˚C) [28] and PCl3 (77-79 ˚C) [29] have very similar boiling points making complete separation of residual PCl3 difficult. The remaining PCl3 would react similarly to MePCl2 leading to the formation of 4. O,O-diethyl S-(2-diisopropylaminoethyl) phosphonothioate (4) was identified in all VX agent stocks analysed. Given its inherent stability and ease of identification in the LC-HRMS data, it may well be exploited as a CAS for VX production by this method.

3.6 Pyrophosphonates

A total of four pyrophosphonates were unambiguously identified from the available GC-(EI)MS and LC-HRMS data (Table 1). Diethyl dimethylpyrophosphonate (17) was observed in all three VX stocks, and readily identified through OCAD library match. This was further supported by the observation of the [M+H]+ m/z 231.0544 in the LC-HRMS ([M+H]+ calc m/z 231.0551,  3.0 ppm), corresponding to the correct molecular formula C6H16O5P2. Although not observed in the BSTFA- derived material, both ethyl hydrogen dimethylpyrophosphonate (18) ([M+H]+ obs m/z 203.0230, calc m/z 203.0238,  3.9 ppm, C4H12O5P2) and dimethylpyrophosphonate (19) ([M+H]+ obs m/z 174.9916, calc m/z 174.9925,  5.1 ppm, C2H8O5P2) were also identified in the LC-HRMS data for VX2017 and VX2018. Additionally, minor amounts of O,O’-diethyl dimethylmonothiopyrophosphonothioate (20) ([M+H]+ obs m/z 247.0318, calc m/z 247.0323,  2.0 ppm, C6H16O4P2S) was also identified in the LC-HRMS data for VX2017 and VX2018.

3.7 Alkylsulfides, alkyldisulfides and isopropylaminothiaalkanes

A total of eighteen alkylsulfides, alkyldisulfides and isopropylaminothiaalkanes were identified from the three VX agent stocks (Table 1). As previously discussed, in both the GC-(EI)MS and LC- HRMS of all three agent stocks 2-(N,N-diisopropylamino)ethanethiol (25) [13-17] was identified. This was expected due to the presence of ethyl methylphosphonic acid (9). It is proposed that the formation of other sulfides and disulfides are a direct result of the reaction of this moiety either through dimerisation, or reaction with other reactive species present. The presence of a further four isopropylaminothiaalkanes diisopropylamine (21) [13,15-17], 2-(N,N-diisopropylamino)ethanethiol
(25) [13-17], 2-(N,N-diisopropylamino)ethanethial (22) [15,16] and 2-(N,N-diisopropylamino)ethyl vinyl sulfide (23) [15,16,25] has previously discussed. Further to this, the most abundant compound detected in VX2014 and VX2017 was bis-2-(N,N-diisopropylaminoethyl)disulfide (27) [13-17,19]. For VX2018, while VX was the most abundant, the presence of bis-2-(N,N- diisopropylaminoethyl)disulfide (27) [13-17,19] was readily established as the most abundant degradation product. The compounds bis-2-(N,N-diisopropylaminoethyl)sulfide (26) [14-17,19] and

bis-2-(N,N-diisopropylaminoethyl)trisulfide (28) [19] were also identified, but only in the older VX agent stocks VX2014 and VX2017.
Six isopropylaminothiaalkanes were identified in all VX agent stocks. The structures of these were identified using a combination of LC-HRMS, LC-HRMS/MS and comparison with the literature, and are shown in Supplementary material Fig. A2. The compounds identified were the tetrathiaalkane 1,12-bis(diisopropylamino)-3,6,7,10-tetrathiadodecane (29) [14,19], the
trithiaalkanes 1,9-bis(diisopropylamino)-3,4,7-trithianonane (30) [14,15,19,23], 1,8-
bis(diisopropylamino)-3,4,5-trithiaoctane (31) [19], 1,10-bis(diisopropylamino)-3,4,7,8-trithiadecane
(32) [19] and 1,11-bis(diisopropylamino)-3,6,7-trithiaundecane (33) [19], and the dithiaalkane 1,8- bis(diisopropylamino)-3,6-dithiaoctane (34) [19]. This left three compounds that needed further structural elucidation.

An additional disulfide, (2-diisopropylaminoethyl)(2-isopropylaminoethyl)disulfide (5) ([M+H]+ obs m/z 279.1921, calc m/z 279.1928,  2.5 ppm, C13H30N2S2) was elucidated through interpretation of LC-HRMS/MS data. The proposed fragmentation pathway for 5 is shown in Fig. 4. The observed fragmentation for 5 was very similar to that for bis-2-(N,N-diisopropylaminoethyl)disulfide (27) [14- 17,19,22], with the only differencing appearing to be the absence of an isopropyl moiety. On fragmentation of 5, the fragments 5a at m/z 220.1158 (C10H22NS2), and 5b at m/z 178.0726 (C7H16NS2), were observed. The difference between these two fragments was again the absence of an isopropyl moiety, an interpretation further confirmed through the observation of fragment 5c at m/z 118.0683 (C5H11NS). All remaining fragment ions observed in the LC-HRMS/MS data for 5 were analogues with those previously reported for bis-2-(N,N-diisopropylaminoethyl)disulfide (27) [24]. Given these data and the observation of 2-(N-isopropylamino)ethylchloride in the 2-(N,N- diisopropylamino)ethylchloride (24) precursor, it is proposed that the only difference between 5 and 27 was an absence of one isopropyl group. As 5 contains a moiety consistent with the minor impurity in the precursor 2-(N,N-diisopropylamino)ethylchloride (24), it could be a valuable CAS for source attribution of the precursor compounds for the synthesis of VX using this method.

Also identified was bis-2-(N,N-diisopropylaminoethyl)sulfoxide (6) ([M+H]+ obs m/z 305.2629, calc m/z 305.2626,  0.98 ppm) consistent with the molecular formula C16H36N2OS. The proposed fragmentation pathway for bis-2-(N,N-diisopropylaminoethyl)sulfoxide (6) is shown in Fig. 5. Again, two fragmentation pathways were observed. Following fragmentation pathway A (Fig. 5), the loss of m/z 101.1210 (C6H15N) to form 6a at m/z 204.1419 was consistent with the loss of a 2- diisopropylamino-1-thioethyl [24]. The formation of 6a was seen to be evidence for the possible presence of a sulfoxide. Fragmentation pathway B (Fig. 5) yielded the fragment 6b at m/z 174.1298, with further fragmentation consistent with the formation of 6c (m/z 132.0833), 6d (m/z 128.1438), 6e (m/z 114.1280), 6f (m/z 72.0808) and 6g (m/z 86.0962). The observed fragment ions on pathway B gave further support to the presence of a 2-diisopropylamino-1-thioethyl [24]. Considering the above data, it is proposed that 6 contains a sulfoxide. Hence the structure of 6 is proposed as shown. While the sulfonic acid of 2-(N,N-diisopropylamino)ethanethiol (25) has previously been reported [30], both the sulfone and the sulfoxide have not. To the best of our knowledge this is the first report in the literature for LC-HRMS data of bis-2-(N,N-diisopropylaminoethyl)sulfoxide (6).
The third compound was elucidated as 1-(N,N-diisopropylamino)-7-(N-isopropylamino)-3,- dithiaheptan-5-one (7) which yielded a molecular ion ([M+H]+ obs m/z 307.1874, calc 307.1872, 

0.65 ppm) consistent with the molecular formula C14H30N2OS2 and contained one double bond equivalent. The proposed fragmentation pathway for 7 is shown in Fig. 6, with the data again suggestive of two fragmentation pathways (Fig. 6). The presence of a weak ion at m/z 265.1362 suggested the loss of an isopropyl moiety from either end of 7 to form either 7a or 7b. Further fragmentation following pathway A yielded the diagnostic fragments 7c (m/z 128.1438) and 7d (m/z 114.1282) associated with a 2-diisopropylamino-1-thioethyl moiety [24]. Pathway B however seems unrelated to previously observed fragmentation profiles. The loss of an additional C5H13N from 7b to yield 7e (m/z 178.0344) corresponded to the loss of the 2-diisopropylamino-1-thioethyl moiety, leaving C6H12NOS2. Further loss of a sulfur formed 7f (m/z 146.0631), which then fragmented into two further ions. One corresponded to a further loss of sulfur to form 7g (m/z 114.0907), and was suggestive of the presence of a disulfide bond considering the loss of consecutive sulfurs. The other fragment formed was 7h at m/z 104.0166 (C3H6NOS) corresponding to the loss of an isopropyl moiety, followed by the loss of a sulfur to yield 7i at m/z 72.0444 (C3H6NO). The fragment 7j (m/z 86.0962) could be equally formed through loss of C2H4 from 7d, as well as decarboxylation of 7i. The proposed structures for fragments 7e and for 7k (m/z 160.1151), and the presence of a double bond equivalent, were interpreted to be supporting evidence for the presence of a carbonyl functionality located in the position shown. Considering these data, the structure of 1-(N,N-diisopropylamino)-7- (N-isopropylamino)-3-dithiaheptan-5-one (7) is proposed as shown. Alternative structures containing either a carbon-carbon double bond, or a carbonodithioate, were discounted as the proposed structures did not adequately explain the observed fragmentation data. Literature searches showed that this compound has not previously been reported.

4. Conclusions

The aim of this study was to chemically profile existing stocks of the CWA VX to further understand the degradation profile and gain an understanding of the CAS in the available agent stocks. A deeper understanding of the degradation profile would also allow for a better understanding of the chemical forensic information to be deduced from these materials. Using multiple chemical analytical techniques, including one- and two-dimensional NMR, both LC-HRMS and LC-HRMS/MS, and GC(EI)- and GC-(CI)MS, 44 compounds were identified in the three VX agent stocks analysed. Interpretation of LC-HRMS/MS data allowed for the identification of several compounds not previously reported in the literature.

Minor amounts of three compounds containing an isopropylamino moiety as opposed to a diisopropylamino moiety were identified. The identification of these minor compounds necessitated a chemical investigation of the precursor chemical 2-(N,N-diisopropylamino)ethylchloride (9). This investigation identified that 9 was contaminated with approximately 2% of the compound 2-(N- isopropylamino)ethylchloride. The identification of the N-isopropylamino containing compounds in the VX agent stocks, and the N-isopropylamino contaminant in 9, highlighted the potential usefulness of these compounds as CAS. The presence of N-isopropylamino side reaction products potentially links these minor compounds to the corresponding impurity in 9. Alternative techniques using stable isotope ratios are being explored as alternative confirmatory methods.

The identification of O,O-diethyl S-(2-isopropylaminoethyl) phosphonothioate (4) was another compound that has plausible applicability as a CAS. The formation of 4 occurs through reaction of residual PCl3 that is still present in the MePCl2 precursor, and then carried through the synthetic

pathway. Hence, O,O-diethyl S-(2-isopropylaminoethyl) phosphonothioate (4) is potentially a useful CAS linking the production of VX to a source of precursor MePCl2.
To the best of our knowledge this is the first investigation of VX that has fused the data from multiple analytical platforms to identify degradation products and possible CAS. It is also noteworthy that this analysis is the first conducted on neat VX agent stocks. No stabilisers were added to the agent, and the analysis was conducted on VX agent stocks that were not subjected to a forced degradation. Further studies are underway to confirm the robustness and applicability of the potential CAS identified, and to investigate the uniqueness of the CAS when compared to alternative synthetic pathways.

Acknowledgements

The authors would like to acknowledge Steven Torney, Dr Craig Brinkworth and Candace Greer for aiding in the collection of the GC(CI)-MS data, and Dr Harry Rose for his input and advice. Scott Blundell at the Monash University Analytical Platform is also acknowledged for contributions to the LC-HRMS data collection. The financial support of the Defence Science and Technology Group is gratefully acknowledged.

Appendix A. Supplementary material

Supplementary material to this article can be found online.

References

1. Halabja chemical weapons: A chance to find the men who armed Saddam. http://www.bbc.com/news/magazine-20553826, 2012 (accessed 8 August 2019).

2. S. Budiansky, Chemical weapons: United Nations accuses Iraq of military use, Nature 308 (1984) 483.

3. http://www.opcw.org/fact-finding-mission, 2019 (accessed 8 August 2019).

4. Kim Jong-nam: VX dose was 'high and lethal'. http://www.bbc.com/news/world-asia- 39096172, 2017 (accessed 3 October 2019).

5. B.P. Mayer, C.A. Valdez, A.J. DeHope, P.E. Spackman, A.M. Williams, Statistical analysis of the chemical attribution signatures of 3-methylfentanyl and its methods of production, Talanta. 186 (2018) 645-654.

6. B. Halford, Tracing A Threat. Chemical & Engineering News Archive, 90 (6) (2012) 10-15.

7. C.L. Mazzitelli, M.A. Re, M.A. Reaves, C.A. Acevedo, S.D. Straight, J.E. Chipuk, A systematic method for the targeted discovery of chemical attribution signatures: application to isopropyl bicyclophosphate production, Anal. Chem. 84 (15) (2012) 6661-6671.
8. K. H. Holmgren, R. Magnusson, S. Lindberg, C. Astot, R. Norlin, C.A. Valdez, A.K. Vu, A.M. Williams, A. Alcaraz, S. Hok, Part 1: Tracing Russian VX to its synthetic routes by multivariate statistics of chemical attribution signatures, Talanta (186) (2018) 586-596.
9. J.J. Moran, C.G. Fraga, M.K. Nims, Stable-carbon isotope ratios for sourcing the nerve-agent precursor methylphosphonic dichloride and its products, Talanta 186 (2018) 678-683.

10. C.G. Fraga, G.A. Pérez Acosta, M.D. Crenshaw, K. Wallace, G.M. Mong, H.A. Colburn, Impurity profiling to match a nerve agent to its precursor source for chemical forensics applications, Anal. Chem. 83 (24) (2011) 9564-9572.
11. B.P. Mayer, A.J. DeHope, D.A. Mew, P.E. Spackman, A.M. Williams, Chemical attribution of fentanyl using multivariate statistical analysis of orthogonal mass spectral data, Anal. Chem. 88 (8) (2016) 4303-4310.

12. D. Wiktelius, L. Ahlinder, A. Larsson, K.H. Holmgren, R. Norlin, P.O. Andersson, On the use of spectra from portable Raman and ATR-IR instruments in synthesis route attribution of a chemical warfare agent by multivariate modeling, Talanta 186 (2018) 622-627.

13. M.R. Gravett, F.B Hopkins, A.J. Self, A.J. Webb, C.M. Timperley, J.R. Riches, Fate of the chemical warfare agent O-ethyl S-2-diisopropylaminoethyl methylphosphonothiolate (VX) on soil following accelerant-based fire and liquid decontamination, Anal. Bioanal. Chem. 406 (5) (2014) 5121-5135.

14. P.A. D'Agostino, L. Provost, J. Visentini, Analysis of O-ethyl S-[2-(diisopropylamino) ethyl] methylphosphonothiolate (VX) by capillary column gas chromatography-mass spectrometry, J. Chromatogr. A. 402 (1987) 221-232.

15. D.K. Rohrbaugh, Characterization of equimolar VX–water reaction product by gas chromatography–mass spectrometry, J. Chromatogr. A. 809 (1998) 131-139.

16. D.K. Rohrbaugh, Methanol chemical ionisation quadropole ion trap mass spectrometry of O- ethyl S[2-(diisopropylamino)ethyl] methylphosphonothiolate (VX) and its degradation products, J. Chromatogr. A. 893 (2000) 393-400.

17. E.I. Savel'eva, I.G. Zenkevich, A.S. Radilov, Identification of O-Isobutyl S-(2- diethylaminoethyl) methylthiophosphonate chemical neutralization products in bitumen-salt matrices, J. Anal. Chem. 58 (2) (2003) 114-123.

18. C.A. Boulet, P.A. D'Agostino, Analysis of dimethylphyrophosphonate decomposition products of VX by GC-MS/MS and 31P NMR, Phosphorus Sulfur. 104 (1995) 93-101.

19. P.A. D'Agostino, J.R. Hancock, L.R. Provost, Analysis of O-ethyl S{2-(diisopropylamino)ethyl] methylphosphonothiolate (VX) and its degradation products by packed capillary liquid chromatography-electrospray mass spectrometry, J. Chromatogr. A. 837 (1999) 93-105.

20. H. Koskela, N. Grigoriu, P. Vanninen, Screening and identification of organophosphorus compounds related to the chemical weapons convention with 1D and 2D NMR spectroscopy, Anal. Chem. 78 (11) (2006) 3715-22.

21. Y.C. Yang, L.L. Szafraniec, W.T. Beaudry, D.K. Rohrbaugh, Oxidative detoxification of phosphonothiolates, J. Am. Chem. Soc. 112 (18) (1990) 6621-6627.

22. N.B. Munro, S.S. Talmanga, G.D. Griffin, L.C Waters, A.P. Watson, J.F. King, V. Hauschild, The sources, fate, and toxicity of chemical warfare agent degradation products, Environ. Health Persp. 107 (12) (1999) 933-974.

23. S. Ellis-Steinborner, A. Ramachandran, S.J. Blanksby, The fragmentation pathways of protonated Amiton in the gas phase: towards the structural characterisation of organophosphorus chemical warfare agents by electrospray ionisation tandem mass spectrometry, Rapid Commun. Mass Spectrom. 20 (2006) 1939-1948.

24. A.J. Bell, J. Murrell, C.M. Timperley, P. Watts, Fragmentation and reactions of two isomeric O-alkyl S-(2-dialkylamino)ethyl methylphosphonothiolates by electrospray ionization/ion trap mass ppectrometry, J. Am. Soc. Mass Spectrom. 12 (2001) 902-910.

25. D.K. Rohrbaugh, F.J. Berg, L. J. Szafraniec, D.I. Rossman, H.D. Durst, S. Munavalli, Synthesis and mass spectral charaterization of diisopropylamino-ethanethiol, -sulfides and -disulfides and vinyl sulfides, Phosphorous Sulfur. 149 (1999) 95-106.

26. E. Ghanem, Y.C. Li, C.F. Xu, F.M. Raushel, Charaterisation of a phosphodiesterase capable of hydrolyzing EA 2192, the most toxic degradation product of the nerve agent VX, Biochemistry 46 (31) (2007) 9032-9040.

27. M.A. Althoff, A. Bertscha, M. Metzulata, T.M. Klapötke, K.L. Karaghiosoff, Application of headspace and direct immersion solid-phase microextraction in the analysis of organothiophosphates related to the Chemical Weapons Convention from water and complex matrices, Talanta 174 (2017) 295-300.

28. W.J. Bailey, S.A. Buckler, Phosphorous compounds. I. Reduction of benzylphosphonium compounds with lithium aluminium hydride, J. Am. Chem. Soc. 79 (13) (1957) 3567-3569.
29. A.E. Arbuzov, F.G. Valitova, Preparation of pyrocatechol phosphorus monochloride, Trans. Kirov Inst. Chem. Tech. Kazan. 8 (1940) 12-15.

30. C.A. Valdez, R.N. Leif, S. Hok, B.R. Hart, Analysis of chemical warfare agents by gas chromatorgraphy-mass spectrometry: methods for their direct detection and derivitization approaches for the analysis of their degradation products, Rev. Anal. Chem. 37 (1) (2017) 1- 26.VX-803

Leave a Reply

Your email address will not be published. Required fields are marked *

*

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>