Analysis of chemical warfare agents by gas chromatography-mass spectrometry: methods for their direct detection and derivatization approaches for the analysis of their degradation products

Phosphonic acids

The main degradation products from the hydrolysis (and oxidative hydrolysis) of the OP-based nerve agents are the phosphonic acid half esters (Compounds 14, 25–28, Figure 4). Further hydrolysis of these esters either under basic or oxidative conditions leads to a common, ultimate product, namely MPA (15 in Figure 4) (Bizzigotti et al. 2012). Detection of these species in a mixture within the context of CWAs provides a key insight into the nature of the potential nerve agents present in it and provides testament for their past presence thus yielding important forensic information. Due to the lack of direct detection of these species by GC-MS means, their initial derivatization using silylation or methylation is a necessary requirement for analysis. In general, by employing the silylation technique, phosphonic acids have been modified to their tert-butyldimethylsilyl (TBDMS) ethers using N-(tert-butyldimethylsilyl)-N-methyltrifluoroacetamide (MTBSTFA) and to their trimethylsilyl (TMS) ethers employing BSTFA. Often times, the derivatization of the acids is rapid and efficient as highlighted by the silylation of the plant hormone regulator ethephon (2-chloroethylphosphonic acid). Ethephon was demonstrated to undergo a smooth conversion to its bis-tert-butyldimethylsilyl ether when heated at 45°C for 1 h in neat MTBSTFA. Using this approach, ethephon was detected in water samples, after copious co-evaporation with acetonitrile (ACN), in the range of 0.1–1 ng l⁻¹ with a reported limit of quantification of 0.1 ng l⁻¹ (Royer et al., 2006). Turning our attention to a more relevant report involving phosphonic acids related to CWAs, the same silylating agent was employed in the efficient derivatization of seven acids related to V- and G-series nerve agents. These acids included 14, 15, 25–28 in addition to the sodium salt of ethyl hydrogen dimethylaminophosphate, the main product from the first pass hydrolysis of the nerve agent tabun (GA). After determining the optimal conditions for their silylation (80°C for 45 min in ACN), the method was applied for their detection in water and soil samples with a reported limit of detection less than 5 pg using GC-ICPMS (Richardson & Caruso, 2007). Even though silylation is a simple and reliable derivatization technique, there exists room for improvement in its overall execution and practicality. For example during the silylation protocol, heating is necessary (typically between 65 and 80°C) for usually 1–2 h in order to ensure that a low concentration analyte is adequately derivatized. Lastly, the reagent itself is often employed as a reaction medium (i.e. neat) or in excess, thus yielding a large interfering signal in the GC chromatograph that may obscure other silylated analytes of interest. An alternative to silylation is methylation employing the universally known reagent, diazomethane. This method enjoys from two superior attributions relative to silylation, namely the mild conditions at which the derivatization takes place (normally ambient temperature) and the absence of interfering by-products in the final analysis. However, the drawback in this methodology actually lies in the nature of the reagent itself. Diazomethane is a highly reactive species that requires its fresh preparation before use if an efficient and high-yielding derivatization is desired. Despite careful storage under an inert atmosphere and refrigeration to extend its lifetime, it is recommended that after 4–5 days, a new batch of diazomethane should be prepared (Amphaisri, Palit & Mallard, 2011; Ghassabian et al., 2012; Harvey & Wahl, 2012). Furthermore, the preparation that at this time constitutes a recurring endeavor possesses explosive hazards that have pressured the scientific community into finding other alternatives to conduct methylations such as the use of trimethylsilyldiazomethane (Crenshaw & Cummings, 2004; Kemsley, 2011).

Figure 4:

Degradation pathways of nerve agents leading to the formation of highly diagnostic half phosphonic acid esters and methylphosphonic acid (15).

To this end, reports involving the methylation of phosphonic acids employing the stable salt TMO^.BF₄ (31, Figure 5) have started to slowly make a re-appearance in the literature. The use of oxonium salts as alkylating agents was introduced by the pioneering work of the Meerwein group who described their synthesis and subsequent applications in the alkylation of various functional groups including carboxylic acids (Jo Diem, Burrow, and Fry 1977 and Meerwein et al. 1937). However, it was not until 1998 that these salts experienced use in the realm of analytical chemistry when they were employed in the derivatization and qualitative detection of various carboxylic acids present in urine samples (Liebich, Gessele & Woll, 1998). Since then, the use of TMO^.BF₄ as a methylating agent in the field of analytical chemistry remained dormant until very recently, when work in our group demonstrated its use for the effective, efficient and clean methylation of a panel of phosphonic acids related to nerve agents followed by their unambiguous identification by GC-MS (Valdez, Leif & Alcaraz, 2016) (Figure 5). Interestingly, this method was applied for their methylation when spiked at a 10 μg g⁻¹ concentration in a complex matrix composed of various C₁₆–C₁₈ fatty acid methyl esters featured in organic samples during the 38th OPCW Test that our laboratory participated in. The mechanism for the methylation likely proceeds via nucleophilic attack of the phosphonic acid oxygen to one of the methyl groups of the trimethyloxonium salt generating the methylated product, the highly volatile dimethyl ether and tetrafluoroboric acid that is conveniently removed when neutralization of the mixture is carried out with aqueous sodium bicarbonate (Figure 5).

Figure 5:

Proposed mechanism for the methylation of pinacolyl methylphosphonic acid (14) by TMO^.BF₄ (31). The expedient derivatization yields a methyl ester that is now amenable for detection by GC-MS.

Sulfonic acids

The oxidative degradation of OP-based nerve agents caused by various oxidants such as bleach or hydrogen peroxide initially results in their hydrolysis by virtue of the water present in the overall process, but also in the subsequent oxidation of these products. For example, oxidative hydrolysis of the nerve agents VX and VR initially yields the 2-aminoethanethiol species that when present in an oxidative environment undergo conversion into their corresponding 2-aminoethylsulfonic acids. Sulfonic acid reactivity toward electrophilic reagents is expected to be comparable to that of their phosphorus-based counterparts as they both feature low pK_a values (usually in the vicinity of 2–3). Two of the most important sulfonic acids arising from nerve agent oxidative degradation are N,N-diisopropylaminoethyl sulfonic acid (29) and N,N-diethylaminoethyl sulfonic acid (30) that originate from the agents VX and VR, respectively (vide supra, Figure 4). Their presence in a given mixture strongly hints at the latent presence of these V-type nerve agents, and due to their crystalline nature and poor solubility in organic solvents such as methylene chloride or ACN, derivatization of the sulfonic acid group is mandated in order to carry out a successful GC-MS analysis.

Although methods for GC-MS analysis of sulfonic acids exist, there are a scarce number of reports on the analysis of 2-aminoethylsulfonic acids. Nevertheless, it is highly expected that the derivatization reactions employed for the more common, plain sulfonic acids can be similarly applied to 2-aminoethylsulfonic acids. One of the earliest reports involved sulfonic acids such as taurine and L-cysteic acid that were efficiently silylated with BSTFA at 110°C for 1 h and subsequently analyzed by GC-MS (Stokke & Helland, 1978). The authors found that prolonged heating of the reaction mixtures (up to 2 h) involving the taurine and cysteic acids led to the degradation of the silylated derivatives. Aside from this early report, there exists only one report where silylation has been used to derivatize sulfonic acid species that are close in structural relationship to 2-aminoethylsulfonic acids. This report describes the treatment of 2-amino-5-chlorotoluenesulfonic acid and once again L-cysteic acid but this time employing MSTFA to provide their TBDMS derivatives (Ng & Hupé, 1990). The conditions for the silylation involve a lower heating temperature (at 70°C) of the sulfonic acids (∼0.015 mmol) in ACN and in the presence of an excess mixture of tert-butyldimethylsilyl chloride (∼0.15 mmol) and MTBSTFA (∼0.5 mmol) for 24 h. However enough conversion to the silylated derivative had occurred after 1 h for detection by GC-MS. Therefore, one can expect that in a similar fashion, silylation of 2-aminoethylsulfonic acids associated with VX and VR can be accomplished employing MTBSTFA as well as BSTFA as the conditions enjoy an unparalleled similarity. Indeed it is believed that the analysis of 2-aminoethylsulfonic acids has likely been performed in a routine basis using silylation as the main derivatization tool, but those results have not made their appearance in the literature perhaps for their expected simplicity. However, as powerful as silylation may be as an initial screening reaction for the presence of these species, one must be careful not to rapidly dismiss their presence if the silylated product is not found. For example, in the 12th OPCW PT, N,N-diethylaminoethyl sulfonic acid (30) caused major problems to several participating laboratories. Although it was detected easily in the water sample that underwent the methylation reaction, it was not found in the silylated fractions when BSTFA and MSTFA were employed. In the end, it was found that 6 out of 19 labs failed to report this chemical in their report as a result of relying on silylation rather than methylation for this acid (Kuitunen, 2005).

Indeed, methylation has been an additional, reliable way of derivatizing sulfonic acids of this kind for their subsequent detection and identification by GC-MS. Methylations using diazomethane can be used to efficiently derivatize these acids as elegantly demonstrated by the Ley group en route to their total synthesis of taurospongin A (Hollowood, Ley & Yamanoi, 2002; Hollowood, Yamanoi & Ley, 2003). Due to the potentially explosive nature of its preparation and its short shelf life even when stored under an inert atmosphere and at 4°C, other methylating agents have been developed. As an alternative to diazomethane, our group employed TMO^.BF₄ (31) in the methylation of N,N-diisopropylaminoethyl sulfonic acid (29) and N,N-diethylaminoethyl sulfonic acid (30) in methylene chloride to furnish sulfonic acid methyl esters 32 and 33, respectively (Valdez, Leif & Alcaraz, 2016) (Figure 6). Although none of the components in the reaction, with the exception of the matrix itself composed of fatty acid methyl esters, were soluble in methylene chloride (TMO^.BF₄ is a salt and the sulfonic acids are virtually insoluble), the efficient methylation of 29 and 30 was readily accomplished. The methylated products (32 and 33) of these two important V-series agent markers were detected by GC-MS and verified by the instrument’s internal NIST database. A more in-depth discussion on this useful methylating salt can be found in the section discussing phosphonic acid derivatizations (vide infra). Alternatively, trimethylsilyldiazomethane can be used as an effective methylating agent, although no reports exist for its use solely in the derivatization of 29 and 30 for subsequent GC-MS analysis. Nevertheless, trimethylsilyldiazomethane has been used successfully in the mild and safer methylation of structurally similar 2-aminosulfonic acid-bearing analogs (Phelam, Patel & Ellman, 2014).

Figure 6:

Methylation of N,N-substituted aminoethyl sulfonic acids using TMO^.BF₄ in methylene chloride. The conversion to their methyl esters is rapid (<1 h) and at ambient temperature. An important feature of this methodology is the fact that even if the sulfonic acids are not soluble in DCM, they still undergo the methylation.

Pinacolyl alcohol (3,3-dimethyl-2-butanol)

PA (3,3-dimethyl-2-butanol, 16), in addition to MPA (14), is the other major hydrolysis product of the nerve agent GD (2). Its presence in a given sample signals the previous or latent presence of this nerve agent in addition to providing important chemical forensics information. As its importance in this field is well recognized, PA has been labeled by OPCW as a Schedule 2 compound. Due to volatility issues, PA is a difficult analyte to detect by GC-MS means specifically when it is present in low concentration (∼1–5 μl ml⁻¹) in mixtures containing several interfering species. Thus, analytical chemists have relied on its initial derivatization to accomplish the enhancement of its detection by GC-MS as well as configuring its structure to provide a more volatile product with different retention time and enhanced sensitivity by GC-MS. Common derivatization reactions on PA have been based on silylation using BSTFA or MTBSTFA to provide the TMS and the TBDMS derivatives, respectively. These derivatives possess higher boiling points and feature a more volatile profile relative to that of the native PA yielding analogs that elute at a later time in the GC column, a quality that becomes convenient if the PA signal is being obscured by a more abundant interference. As it is common with the use of these derivatization agents, heating of the mixtures up to 65°C for 2–3 h is necessary to effect the complete derivatization. This may create a problem during an OPCW PT when dealing with the derivatization of other analytes in the mixture that may be temperature sensitive. In addition, in the case of BSTFA, a TMS silyl ether of PA is produced that is prone to hydrolysis when facing acidic (pH ∼ 4) or basic conditions (pH > 11) (Green & Wuts, 2007). In an attempt of developing a practical and milder methodology that can yield a more stable silylated version of PA, our laboratory introduced the use of a combination approach using phenyldimethylsilyl chloride (PDMSCl, 34) and N-methylimidazole (NMI, 35) (Albo et al. 2014). The reaction of PA with PDMSCl in the presence of NMI generates a silyl ether that features a more extended time in the GC column and is inherently more stable than its TMS and TBDMS counterparts. The notion behind this approach is the use of the NMI as a nucleophilic activator of the PDMSCl to generate an intermediate silyl imidazolyl species (36) that is more reactive than the silyl chloride itself (i.e. in situ activation) (Figure 7). The concept was introduced by the Corey group back in 1972 by using imidazole as a reagent to enhance the tert-butyldimethylsilylation of several alcohols that show low reactivity for this modification under the most widely used condition (i.e. TBDMSCl and pyridine) (Corey & Venkateswarlu, 1972). In a more recent report, the Kiessling group introduced NMI as a suitable activator for uridine monophosphate during their synthesis of uridine diphosphate galactofuranose (Marlow & Kiessling, 2001). Thus, the reaction between PA (16) and this intermediate yields the phenyldimethylsilylated PA (37) and in the process regenerates the NMI for another round of silyl chloride activation. Even though it may appear that NMI solely plays the role of a catalytic reactivator in the process, excess of both reagents (PDMSCl and NMI) is employed to reduce the time for the derivatization and because portions of the NMI are needed to effectively remove the generated HCl (Figure 7). This approach was used successfully in the tagging of PA with the PDMS group in organic sample matrices from the 16th, 32nd and 34th OPCW PTs (Albo et al., 2014). The derivatization resulted in the production of a PA product that featured a larger retention time and was effective in increasing the analyte’s retention time away from the large interfering signals from the complex matrix.

Figure 7:

Proposed mechanism for the NMI-mediated phenyldimethylsilylation of PA (16) to furnish PDMS-PA (37). In this in situ activation of PDMSCl, NMI plays the dual role of (1) enhancing the reactivity of the silyl chloride through activation and (2) sequestering the generated HCl from the reaction. The efficient silylation of PA using this method occurs in 30 min at ambient temperature.

Sulfur mustards

The HD class of CWAs can be more accurately described as being composed of a number of different sulfur-based alkylating agents that share a common mechanism of action. The flagship compound within this family is HD (6), and the proposed mechanism for its alkylation properties is the intermediacy of a highly electrophilic episulfonium species (38) that arises from an intramolecular S_N2 displacement of one of the chlorine atoms by the sulfur atom (Figure 8). Close contact with these agents causes severe blistering on the skin, and although these are technically non-lethal agents, complications arising from their incapacitating power can lead to death if not treated properly. Decontamination technologies for these species involve the use of basic formulations that result in the hydrolysis of the agent and the production of TDG (17) that, in turn, is no longer a vesicant and non-toxic in nature (Bizzigotti et al., 2012). Oxidative decontamination has also found application with these types of agents; thus treatment of HD with oxidants (e.g. bleach) leads to the formation of its sulfoxide (TDGO, 18) and sulfone (thiodiglycol sulfone, 39) by-products. Although sulfoxides are viable intermediates during the oxidative process, their detection might be hampered by the fact that under the highly oxidative conditions (i.e. excess oxidant), they are further converted to their corresponding sulfone products (Figure 8). For this reason sulfoxides such as TDGO (18) are normally encountered during the analysis of urine and plasma samples obtained from mustard gas-affected individuals as they form a large part of the species arising from mustard gas metabolism. TDG and TDGO may be analyzed by LC-MS methods; however, it is the low levels of detection encountered with this approach (typically 10 ng ml⁻¹) that severely hamper it from wide usage when tackling these analytes. Interestingly, detection of intact TDG and TDGO can be accomplished using GC-MS; however, broadening of the peaks is typical for these kinds of compounds, thus making their derivatizations a necessary step in their analysis. The following section will focus on the most commonly employed HD (6) and its degradation products.

Figure 8:

Mode of alkylation by the sulfur mustards (shown for HD) and their main degradation products from their oxidative hydrolysis.

Thiodiglycol

The main product arising from the complete, basic hydrolysis of HD is TDG (17) (Figure 8). Detection of TDG by GC-MS requires its prior derivatization as demonstrated by its detection in water, serum and urine after its conversion with MTBSTFA (43) to bis-tert-butyldimethylsilylated TDG (44) (Figure 9A). Ohsawa et al. (2004) reported a 55% recovery of the derivatized material for the method that also featured an impressive limit of detection (LOD) of 5.4 ng ml⁻¹ for the water sample. The method involves the treatment of TDG with MTBSTFA (with 1% TBDMSCl) in the form of a co-solvent with ACN (1:1) followed by heating at 60°C for 1 h. The reported LOD for the two remaining matrices, serum and urine, were found to be 7 and 100 ng ml⁻¹, respectively, thus making it a viable method for the analysis of this HD product in aqueous samples. As briefly noted earlier, the use of fluorinated tags aids increases the sensitivity for detecting the analyte due to the electron-capturing ability of the fluorine atom (vide supra). Thus, it is not surprising that derivatization methods that introduce these powerful chemical reporters into TDG have been developed. Thus, in an early report derivatization of TDG with heptafluorobutyl anhydride (HFBA, 45) to provide the bis-heptafluorobutyl derivative (46) was described. The method was successfully applied to the derivatization of TDG present in rat urine samples after water removal and reaction with HFBA in dry ethyl acetate, in the presence of 5 Å molecular sieves to further scavenge any residual water, and heating the mixture to 60°C for 1 h (Jakubowski et al., 1990) (Figure 9B). After the derivatization was completed, the sample was dried, re-suspended in ethyl acetate and immediately analyzed by EI-GC-MS using selected ion monitoring (SIM). In another early report, extremely low LOD values (down to 1 ng ml⁻¹, 1 ppb) in urine samples were disclosed by reacting TDG with pentafluorobenzoyl chloride (PFBC, 47) in pyridine at ambient temperature for 5 min to furnish fluorinated TDG derivative 48 followed by analysis using negative ion chemical ionization GC-MS (Black & Read, 1988) (Figure 9C). In yet another example involving fluorine tag introduction into TDG, Popiel et al. (2014) employed trifluoroacetyl imidazole (TFAI, 49) to convert it to bis-trifluoroacetylated TDG (50). The reaction was carried out in methylene chloride under mild conditions (1 h at 30°C). Using this methodology, the authors reported the LOD value down to 0.01 ng ml⁻¹, which when compared to additional studies described in their paper is two orders of magnitude better than the one obtained when silylation (with BSTFA) is employed (Popiel et al., 2014) (Figure 9D). As an additional note, the chemistry group from Vertox laboratories performed a set of designed to compare the reactivity of three different trifluoroacetylating agents including TFAI to produce 50. The three agents were TFAI (49), trifluoroacetylbenzotriazole (TFABT, 51) and their then newly developed trifluoroacetylbenzimidazole (TFABI, 52) (Pardasani et al. 2004) (Figure 9E). During their work, TFAI and TFABI were found to be superior trifluoroacetylating agents than TFABT when running the derivatization of TDG and other homologous thio-containing alcohols closely related to the HD. Their explanation for the observed reactivity was the lower pK_a value exhibited by the imidazole and benzimidazole moieties in 49 and 52 (∼7.0) relative to the one exhibited by the triazole heterocycle in 51 (∼8.3). The reaction was found to proceed smoothly at ambient temperature and just 5 min, with little product yield improvement after extending the reaction time (Figure 9E).

Figure 9:

Derivatization of TDG employing BSTFA (A) and various fluorine-containing derivatization agents (B–E).

Thiodiglycol sulfoxide

Another by-product arising from the oxidative hydrolysis of HD, and also as a result of its metabolism in humans, is TDGO (18). Due to the similarity not only in structure but reactivity as well to TDG, it is not surprising to see that TDGO has been the subject of the same set of derivatizations for GC-MS analysis. However, the reactivity of the hydroxyl groups in TDGO is expected to be higher than that of the ones present in TDG based on inductive effects originating from the sulfoxide moiety. Due to the observed poor LOD values obtained when silylation is used as a derivatization means for TDGO, fluorinated tags have been more beneficial when their labeling and subsequent detection by GC-MS is needed. Thus, in showcasing one application of this kind, HFBA (45) used to convert TDGO into bis-heptafluorobutyrylated TDGO (53) was used and analyzed in urine samples using isotope dilution GC-MS-MS (Boyer et al. 2004). The derivatization involved the treatment of the sample, after complete water removal, with HFBA in ACN at 50°C for merely 30 min. Four years later, the Black group developed a method for TDGO detection in urine samples this time via its reaction with heptafluorobutyl imidazole (HFBI, 54). In this report, the protocol involves the heating of the dried sample containing the TDGO (as well as TDG) in ACN in the presence of HFBI at 50°C for only 30 min to furnish 53. Subsequent detection of the fluorinated TDGO was accomplished using isotope dilution GC-ion trap tandem mass spectrometry, reporting levels of TDG (obtained from the reduction of TDGO by titanium chloride in the urine sample) down to 104 ng ml⁻¹ (Riches, Read & Black, 2007) (Figure 10). During their analysis of TDG, the Popiel and Vanninen groups found that the reaction of TDGO employing TFAI (49) worked exceptionally well in converting it into its bis-trifluoroacetylated analog (55). The reaction was found to proceed in a number of organic solvents (including dichloromethane and ACN) and resulted in the rapid conversion (∼1 min at 30°C) of TDGO to 55 (Popiel et al., 2014). It is noteworthy to mention that for the specific GC-MS analysis of sulfoxide species such as TDGO, trifluoroacetylation using agents that produce acid by-product (such as trifluoroacetic anhydride or trifluoroacetyl chloride) result in the inefficient derivatization of the parent compound. The reason for the reduced derivatization efficiency is that sulfoxide species tend to undergo Pummerer-type rearrangements in the presence of the generated acid (i.e. trifluoroacetic acid), a pathway that is highly unlikely when the imidazole-based trifluoroacetylating agents such as TFAI are employed.

Figure 10:

Derivatization of TDGO (18) with fluorine-bearing tags for its analysis by GC-MS.

Nitrogen mustards

Just as the HD family, the nitrogen mustards are powerful vesicants. They are powerful alkylating agents that cause severe DNA damage by acting as non-specific crosslinking agents. Their mode of action is very similar to the one displayed by the HD, which involves the sequential displacement (S_N2-like) of their chlorine atoms via the intermediacy of an aziridinium ion. This intermediate displays highly electrophilic character and as such reacts with a large range of nucleophiles. Decontamination approaches aimed at nitrogen mustards have been similar to the ones undertaken for the HD and most CWAs, namely basic and oxidative hydrolysis. Degradation of the nitrogen mustards leads to their alcohol-containing products known as ethanolamines or 2-aminoethyl alcohols (Figure 11). GC-MS detection and analysis of these species in their intact form is possible; however, these have received less attention than the HD. Their ease of detection is a direct result of their volatility and GC equipped with specific nitrogen detectors that can provide accurate and sensitive readouts for the intact agents. Interestingly, as a result of their high reactivity towards various nucleophiles including water, it is their degradation products arising from hydrolysis (basic or oxidative) that have received the bulk of the attention as these are the most likely encountered species in a real-case scenario involving their use.

Figure 11:

2-Aminoethyl alcohol-containing products arising from the hydrolysis of the nitrogen mustards and the oxidative or basic hydrolysis of the V-series agents and the undertaken silylation strategies used for their analysis by GC-MS.

Ethanolamines (2-aminoethyl alcohols)

Of the various degradation products arising from the hydrolysis of the nitrogen mustards (HN1 (7), HN2 (8) and HN3 (56)), the ones bearing the 2-aminoethyl alcohol functionality stand out as the most prominent ones. The 2-aminoethyl alcohols arising from the hydrolysis of the nitrogen mustards HN1, HN2 and HN3 are EDEA (57), N-methyldiethanolamine (MDEA, 19) and triethanolamine (58), respectively (Figure 11). Also as importantly, the 2-aminoethyl alcohol functionality can be found to be present in the products originating from the complete hydrolysis of VX (4) and VR (5), namely N,N-diisopropylaminoethyl alcohol (59) and N,N-diethylaminoethyl alcohol (60) (Figure 11). Consequently, after realization of this link between these products to the aforementioned CWAs, it becomes clear why detecting these species in a given sample remains an important task in the overall study of these toxic species.

Detection of these hydrolysis products by GC-MS means is well reported and in these cases the initial derivatization of these analytes, for example via silylation, is needed. For example, MTBSTFA was used to silylate alcohols 19, 57 and 58 present in water, urine and blood samples with the reported LOD for the three alcohols in the water sample found to be at 2.5, 2.5 and 10 ng ml⁻¹, respectively (Ohsawa & Seto, 2006). The silylation reaction involved the complete drying of the aqueous mixture followed by the heating of the residue to 60°C for 1 h in the presence of excess MTBSTFA. The sample recovery for the water samples were found to be the highest at 88%, 88% and 79% for 57, 19 and 58, respectively, followed by recoveries in urine in the range of 72%–100% and recoveries from serum lying in the low 7%–31% for all three alcohols (Ohsawa & Seto, 2006). It is important to note that Ohsawa and Seto suggest that the addition of HCl to the aqueous mixture prior to evaporation to dryness increases the recovery of the alcohols during the derivatization step. This same acidic treatment was subsequently employed by the Alp group when derivatizing EDEA (19) and MDEA (57), using BSTFA, for their GC-MS analysis in rat urine samples with reported LOD values of 2.5 ng ml⁻¹ for EDEA and 1.6 ng ml⁻¹ for MDEA (Kenar & Alp, 2011). Similarly, heating of the alcohols in the presence of BSTFA in ACN was conducted at 60°C for 30 min to carry out their derivatization. As an alternate procedure to the well-established silylating methods mentioned above, our laboratory found that derivatization of several 2-aminoethyl alcohols with the phenyldimethylsilyl moiety can be accomplished under mild conditions with the in situ activation method employing phenyldimethylchlorosilane and NMI. The efficient silylation of a panel of 2-aminoethyl alcohols including the ones linked to the nitrogen mustards (19, 57, 58) was demonstrated. In addition, the same approach was utilized in the efficient derivatization of alcohols 59 and 60 that are the oxidative hydrolysis products of VX and VR, respectively, to furnish silylated products 61 and 62. The work demonstrated the advantage of using NMI over pyridine as a powerful nucleophilic base to activate the PDMSCl (Valdez, Leif & Hart, 2014a). The protocol involves the same conditions (ambient temperature and ∼30 min) as the one described in earlier work from our laboratory involving the silylation of PA (vide supra, 7).

Methylation as a measure of formulating suitable derivatives of 2-aminoethyl alcohols for GC-MS analysis has not been a widespread practice. As diazomethane is the most employed reagent for the introduction of the methyl moiety into analytes, it is its high degree of electrophilicity that inherently works against it when dealing with this class of substrates. Diazomethane is used in large excess leading to the derivatization of not only the hydroxyl groups but also the highly nucleophilic tertiary nitrogen center that ultimately converts the molecule into a quaternary salt thus severely handicapping its volatility and subsequent detection by GC-MS. Methylation of amino alcohols exists but these are conducted strongly under basic conditions (e.g. sodium hydride and methyl iodide), thus assuring that the hydroxyl groups are the most nucleophilic centers in the molecule even over other amine groups (Corda et al., 1994; Cacciapaglia et al., 2005).

Lewisites

Lewisites are arsenic-based CWAs that like nitrogen/HD also belong to the blistering agent class. The most common members of this class are Lewisite I (9), II (10) and III (11), all featuring an arsenic (III) center covalently bonded to at least one sp² carbon atom of the 2-chlorovinyl group (Figure 1). In the case of Lewisite III (11), all three substituents on the arsenic are 2-chlorovinyl units. Interestingly, the term Lewisite is commonly used to describe the organoarsenical mixture that results when arsenic trichloride reacts with acetylene gas in the presence of metal chloride catalysts such as aluminum chloride, mercury (II) chloride, antimony (III) chloride, or the combination of these in aqueous HCl media (Jarman, 1959; Jones, Rosser & Woodward, 1949a; Jones, Vallender & Woodward, 1949b). Thus, despite the use of several rounds of fractional distillations for purification, the composition of munition-grade Lewisites as well as Lewisite standards synthesized in a laboratory generally consists of ∼90% Lewisite I, ∼9% Lewisite II, and <1% Lewisite III. Direct detection of Lewisites in the intact form by GC-MS is extremely difficult due to the instability of these compounds under conditions where minimal moisture exists. Furthermore, the presence of alcohols, or for that matter thiols which possess an unmatched affinity for their arsenic center, may react to form unwanted adducts in the instrument aided by the heat of the GC port. Lastly, their detection in the intact form is further complicated by the deterioration of the GC stationary phases due to the acidity and corrosiveness of the samples themselves that is a direct consequence of their high sensitivity to moisture. This represents a major problem leading to the inability to achieve linear responses during their analysis resulting in the irreproducible quantitation of the intact species particularly when these are present in low concentrations. Interestingly, Lewisite III (11) is a non-vesicant, as its arsenic center is non-electrophilic and unreactive to O- and S-nucleophiles and stable to hydrolysis in aqueous media, which is the reason why it can be directly analyzed by GC-MS without derivatization. At higher concentrations, Lewisites I and II can also be analyzed by GC-MS without the need for derivatizations (Black & Muir, 2003; Epure, Grigoriu & Filipescu, 2010).

Fentanyls

Within the class of incapacitating agents, one of the most notorious members that have recently gained wide attention is the fentanyls. Some of these synthetic opioids display potencies that are far beyond the one exhibited by the most famous painkiller used in the clinical setting, namely morphine. For example, fentanyl (12), the flagship compound in this family exhibits potency of ∼100 times that of morphine (Figure 15). As expected other, more powerful analogs of this compound have been synthesized since Janssen’s landmark synthesis of 12, such as remifentanil (83), carfentanil (84) and sufentanil (85) with demonstrated potencies ranging from ∼200× to ∼1000× more than that exhibited by morphine.

Figure 15:

Selected members of the fentanyl class of synthetic opioids. The bracketed numbers represent their individual potencies relative to morphine.

These compounds act by binding transmembrane μ-opioid receptors on cell surfaces activating a cascade of intracellular signals that lead to their biological effect (Grass 1992a; 1992b; 2000). Commonly employed as potent analgesics during perioperative procedures in a clinical setting, they have emerged as the main choice for physicians when a patient is undergoing complex surgical procedure in addition to becoming key players in the management of pain for patients with cancer (Friedrichsdorf & Postier, 2014; Lee et al., 2014). Their rapid and effective biological action on the nervous system is the main reason for their use as incapacitating agents by the armed forces. A notable example of this use was the employment of two powerful members of this class in the aerosolized form, remifentanil (83) and carfentanil (84), to subdue Chechen terrorists during the Moscow Theater crisis in 2002 leading to 170 fatalities that included civilians (Coupland, 2003; Riches et al., 2012; Rieder et al., 2003; Wax, Becker & Curry, 2003). In addition, very recently their military use has transitioned into their illicit distribution in society resulting in numerous fatalities (Breindahl et al., 2017; Katselou et al., 2016; Rojkiewicz et al., 2016). Aside from their powerful neurological effects, a more concerning fact over these class of drugs by various agencies including the Drug Enforcement Agency is the efficient, straightforward and large-scale adaptability of their synthetic routes (Gupta et al., 2005; Valdez, Leif & Mayer, 2014b). Consequently, medical countermeasure approaches to alleviate their biological impact due to an overdose in an individual is one of the main branches aimed at starting a program to effect their capture and neutralization (Mayer et al., 2016). Due to their stability and resistance toward acid- or base-mediated hydrolysis (Garg et al., 2010), detection in their intact form is usually encountered. As a result of this, LC-MS has been the leading technique for their analysis and quantification greatly aided by fentanyl’s innate ultraviolet absorption and its aqueous solubility when converted to their hydrochloride or citrate salts (as commonly employed in the medical field) (Almoussa et al., 2011; Cooreman et al., 2010; Wang & Bernert, 2006).

In the realm of GC-MS, fentanyls themselves have not been a reason for method development; however, few reports exist where the free base can be directly detected and quantified using synthesized standards and deuterated analogs (Gardner et al., 2015; Kudo et al., 2013; Ohta, Suzuki & Ogasawara, 1999; Van Nimmen & Veulemans, 2007). Interestingly, GC-MS has played a key role in the analysis of products arising from their metabolism as elegantly demonstrated by the analysis of the major metabolite arising from the action of esterases on remifentanil (83). It is known that hydrolysis of the distal, primary ester occurs more readily than the more congested tertiary methyl ester in 83 to yield carboxylic acid-containing intermediate 86. Acid 86 was effectively silylated with BSTFA in conjunction with TMS chloride at 70°C for merely 30 min to provide silylated acid 87 that was detectable by GC-MS (Lehner et al., 2000) (Figure 16A). In a similar fashion, the N-dealkylated metabolic products referred to as nor-fentanyls have been analyzed after their adequate derivatization. In this specific scenario, metabolic dealkylation of sufentanil (85) and alfentanil (88) produces the piperidine-containing metabolite (89) that coincidentally is the same intermediate for both metabolic processes. Treatment of 89 with PFBC (47) furnishes the pentafluorobenzamide product (90) that showed enhanced detectability by GC-MS by virtue of the electron-capturing ability of the fluorinated tag (Valaer et al., 1997) (Figure 16B). This method allowed for the detection of the dealkylated metabolites of fentanyl, sufentanil and alfentanil as low as 0.3 ng ml⁻¹ in urine samples. As an alternate route to the analysis of nor-fentanyl (91), derivatization employing pentafluoropropionyl anhydride (92) yielded the pentafluoroamide product 93 that was easily detected by GC-MS (Figure 16C). The developed methodology allowed reported LODs of 0.08 ng ml⁻¹ for nor-fentanyl as well as nor-alfentanil 89 in urine samples (Strano-Rosi et al., 2011).

Figure 16:

Derivatization of products arising from the hydrolytic and/or metabolic breakdown of various fentanyl-related compounds. The use of fluorinated derivatization tags greatly enhances the detection of these species due to the unsurpassed electron-capture abilities of the fluorine atoms.

3-Quinuclidinyl benzilate

3-Quinuclidinyl benzilate, also known by its IUPAC name 1-azabicyclo[2.2.2]oct-3-yl 2-hydroxy-2,2-diphenylacetate (13), is an anticholinergic agent that affects both the peripheral and central nervous systems (Fusek et al. 2009). This chemical is commonly referred to by the NATO code BZ, although another common acronym is QNB, and is the only incapacitating agent classified as a Schedule 2 compound by the OPCW (Szinicz, 2005). The activity of BZ arises from its competitive inhibition of muscarinic receptors (Eglen, 2006), blocking ACh binding resulting in anticholinergic delirium, cognitive dysfunction, hallucinations and inability to function. BZ has been used as an incapacitating agent by the military for the reasons cited above and its dissemination in the aerosol form has been deemed effective as its primary route of absorption is through the respiratory system (Misik, 2013).

BZ can be analyzed directly by GC-MS without any prior treatment (Byrd et al., 1992; Byrd, Sniegoski & White, 1988; Kostianinen, 2000). Derivatization techniques such as silylation and methylation aimed selectively at the hydroxyl group of 13 are not known. There are two reasons that account for this behavior; the first one is the fact that the hydroxyl group is highly hindered (tertiary alcohol) and the second is the greater nucleophilicity and accessibility of the quinuclidine nitrogen. It is this nitrogen that often finds itself at the epicenter of reactions involving electrophilic reagents such as methyl iodide (94), for example, to generate the BZ-Me salts (95) that are more suitable for LC-MS analysis (Dolle et al., 2001) (Figure 17). Furthermore, the presence of the ester moiety joining the two halves of BZ additionally complicates matters as basic conditions usually employed to enhance the nucleophilicity of hydroxyl groups over nitrogen centers often result in the cleavage of this bond. For these reasons, and the fact that BZ is amenable for UV detection, LC-MS has been a better technique to carry out its analysis (Black & Read, 2005).

Figure 17:

Methylation of BZ resulting in the modification of the nitrogen center to generate BZ-Me salt (95) suitable for LC-MS applications.

Benzilic acid

BA is one of the two products arising from the breakdown of BZ (Byrd et al., 1992; Hull, Rosenblatt & Epstein, 1979; Sniegoski, Byrd & White, 1989). BA (21) exhibits poor gas chromatographic behavior, especially on common nonpolar general purpose column phases, such as the 5% phenyl-methylpolysiloxane column phase, and therefore derivatization is a necessity for its detection by GC-MS means. MTBSTFA has been a convenient derivatizing agent for BA and readily forms the bis-(tert-butyldimethylsilyl) BA (96) (Figure 18). The conditions used for this transformation involves heating of the mixture at 60°C for 30 min in a 1:1 chloroform:MTBSTFA mixture (Park et al., 2013). In this work, the authors report LOD values in the range of 0.08–0.01 ng ml⁻¹ for BA along with other CWA degradation products when spiked in water samples. Confirmation of the efficiency of the reaction, reached after several optimization studies on the method, was validated by using the protocol on the detection of BA in OPCW test samples. Methylation of BA has seldom found use as an alternative derivatization protocol but has employed diazomethane. The methylation involves the reaction of BA with diazomethane in acetone and yields the bis-methylated adduct (97) which possesses suitable volatility for GC-MS (Amphaisri, Palit & Mallard, 2011) (Figure 18). In their report, Amphaisri et al. also show their studies on the methylating power of two commercially available agents, namely, trimethylphenylammonium hydroxide (98) and trimethylsulfonium hydroxide (99). It was found that both of these reagents performed equally well but still below the more established diazomethane reagent despite their added benefit of their non-explosive nature. Thus, the thermally assisted methylations using these two reagents yielded significantly noisier chromatograms relative to their diazomethane counterpart; however, detection of BA and phosphonic acids related to CWAs was accomplished down to levels of 0.5 μg ml⁻¹. In addition to diazomethane, we found that in our hands the use of TMO^.BF₄ also works efficiently at providing the bis-methylated adduct 97 during the 40th OPCW PT that we participated in. The methylation under our conditions works at ambient temperature and for 2 h (Figure 18).

Figure 18:

Silylation and methylation strategies for benzilic acid for GC-MS analysis.

3-Quinuclidinol

3Q (20) is a bicyclic amino alcohol that is a key indicator for the presence of BZ due to its strong structural connection with this incapacitating agent. The alcohol in the underivatized form is not fit for direct GC-MS detection as its volatility profile is not appropriate in addition to the broadness of the peak that usually hinders detection even further. Derivatization certainly has provided great aid in the overall detection of 3Q by GC-MS means with silylation and carbamate generation presenting themselves as two most successful approaches. Unfortunately, and as alluded to earlier when BZ and BA were discussed, not many reports exist for its derivatization. One of the reasons for this apparent lack of reports is the fact that in this molecule, in addition to the alcohol functionality, there exists the nucleophilic tertiary nitrogen that possesses high affinity for electrophilic reagents such as BSTFA and diazomethane. Thus, it is often this high level of reactivity that is exploited when the sole modification of the hydroxyl group is required. This has been accomplished by temporarily blocking the amine using boron trihydride (100) resulting in the formation of amino-borane complex 101 (Figure 19). Complexes such as 101 are highly stable and labile only in the presence of acid (e.g. HCl), thus permitting the modification of the hydroxyl group, in this instance via methylation using sodium hydride and methyl iodide. Once the methylation of the hydroxyl moiety has been done, hydrolysis of the amino-borane complex using HCl furnishes the O-methylated 3Q product 102 in good yields (Stotter et al., 1987) (Figure 19). In the realm of silylations, MTBSTFA and BSTFA have found application for derivatizing 3Q involving a hollow fiber-protected liquid phase microextraction (HF-LPME) procedure that involved the in situ derivatization of the analyte when present in water. The hollow fiber served as a hydrophobic protective barrier from water to the derivatizing agent while the extraction/derivatization was underway. It was found that extraction of the aqueous matrix with a 1:1 chloroform:MTBSTFA followed by additional exposure of the fiber to MTBSTFA gave the highest uptake of 3Q in the protocol via its conversion to 3Q-TBDMS (103) (Lee et al., 2008) (Figure 19). Interestingly, it was found that the use of a 1:1 chloroform:BSTFA did not provide better uptake of 3Q relative to its MTBSTFA counterpart (i.e., conversion to 3Q-TMS 104), but it was a superior choice in the analysis of other aminoethyl alcohols. The authors report LOD of the various CWA degradation products for the HF-LPME procedure ranging from 0.04 to 0.36 μg l⁻¹.

Figure 19:

Derivatization strategies for 3-quinuclidinol.

Unfortunately for GC-MS analysis, methylation has not been an integral part of the available toolbox for the analysis of 3Q, and a reason for this again is the strong nucleophilicity of its nitrogen that on alkylation becomes a quaternary salt, thus annulling any chances of detection by GC-MS means. A clever way around this problem was devised by Vertox laboratories when they reported the use of p-tolylisocyanate (PTI, 105). The reaction between the hydroxyl moiety and the reagent produces a stable 3Q-carbamate (106) that is amenable for GC-MS analysis (Karthikraj et al., 2014) (Figure 19). The reaction between PTI and 3Q is rapid and efficient, which is the reason why heating is not needed. Furthermore, an added benefit of the protocol is the late eluting character of the carbamate in the analysis that becomes important when abundant, early eluting interferences are present in the mixture.