Extraction of Nerve Agent VX from Soils - Analytical Chemistry (ACS


Extraction of Nerve Agent VX from Soils - Analytical Chemistry (ACS...

4 downloads 265 Views 94KB Size

Anal. Chem. 2004, 76, 2791-2797

Extraction of Nerve Agent VX from Soils Ce´cile Montauban, Arlette Be´gos, and Bruno Bellier*

Section Analyse Chimique, centre d’e´ tudes du Bouchet 5, rue Lavoisier, BP3 F-91710 Vert-le-Petit, France

O-Ethyl S-(2-diisopropylaminoethyl) methylphosphonothiolate (VX; Figure 1), one of the most toxic chemicals ever produced by man,1 and related V-type compounds, have been prepared and adopted in the 1950s as potential chemical warfare agents. These agents are more toxic than other agents such as sarin or tabun, especially by percutaneous exposure, and are reported to be highly persistent.2 Stockpiling or use of V-agents is now strictly limited to “non prohibited purposes”, as defined by the Chemical Warfare Convention,3 whose verification regime relies on inspections, onsite analysis, or sampling and off-site analysis in a network of expert laboratories. Besides, V-agents have also been used for terrorist purposes in Japan.4 Therefore, methods allowing us to extract, detect, and identify VX and related compounds are crucial, both for civilian and for military protection, and for obtaining juridical proofs of illegal use, preparation, or stockpiling of such agents.

In general, retrospective identification of chemical warfare agents relies on detection and identification of characteristic degradation products, which can be, for organophosphorus agents, methylphosphonic acid (MPA), or O-alkyl (e.g., isopropyl (degradation of sarin), pinacolyl (soman) ...) methylphosphonates.5 However, these compounds may not allow us to discriminate between the potential parent agents: indeed, MPA or O-ethyl methylphosphonic acid may emerge from the degradation of both ethyl methylphosphonofluoridate and VX.6 Determining whether a residual intact agent is present in the sample is most important with respect to the high persistence of VX and its associated potential lethality. Analysis of VX or its Russian analogue O-isobutyl S-(2diethylaminoethyl) methylphosphonothiolate (R-VX; Figure 1) is not a problem by itself, and several methods are now documented in the scientific literature for its selective and specific detection in liquid organic7 or aqueous matrixes (with liquid chromatographic separation8 or organic solvent extraction prior to gas chromatographic analysis9). These methods involve gas chromatographic separation with phosphorus-selective detection or mass spectrometry, or liquid chromatography hyphenated to mass spectrometry. What apparently remained as a problem was extraction of VX from complex matrixes, such as soils, which is a reference matrix for environmental sampling and would be doubtlessly submitted to analysis in case of an alleged use on the field. In contrast to G-agents, VX can interact with a matrix through its sulfur and especially nitrogen atoms, the latter being eventually involved in strong ionic interactions. It is also much less volatile than G-agents, thus rendering thermal desorption much less effective for extraction from solid samples. A few authors describe attempts to extract VX from soils, with limited, if any, efficiency. The reference work in this area used 32P-radiolabeled VX at a spiking level of 200 mg‚kg-1.10 Water extraction hardly allowed recovery of more than 50% of total radioactivity from humic sand and clayey peat, and even less from humic loam, which suggests a strong adsorption of the compound,

* To whom correspondence should be addressed. E-mail: bruno.bellier@ dga.defense.gouv.fr. Tel. +33.1.69.90.84.21. Fax +33.1.69.90.84.69. (1) Updated estimated toxicity levels on man are given by: Hartmann, H. Regul. Toxicol. Pharmacol. 2002, 35, 347-356. (2) Franke, S. Manual of military chemistry, 2nd ed.; Deutscher Milita¨r. Verlag: East Berlin, 1977. (3) Convention on the Prohibition of the Development, Production, Stockpiling and Use of Chemical Weapons and on their Destruction; The Technical Secretariat of the Organization for the Prohibition of Chemical Weapons, The Hague, Netherlands, 1993. (4) Katagi, M.; Nishikawa, M.; Tatsuno, M.; Tsuchihashi, H. J. Chromatogr., B 1997, 689, 327-333.

(5) Mesilaakso, M.; Rautio, M. In Encyclopedia of Analytical Chemistry; Myers, R. A., Ed.; J. Wiley & Sons: Chichester, U.K., 2000; p 899. D’Agostino, P. A.; Provost, L. J. Chromatogr. 1992, 589, 287-294. (6) Munro, N. B.; Talmage, S. S.; Griffin, G. D.; Waters, L. C.; Watson, A. P.; King, J. F.; Hauschild, V. Environ. Health Perspect. 1999, 107, 933-974. (7) D’Agostino, P. A.; Provost, L. R.; Visintini, J. J. Chromatogr. 1987, 402, 221-232. (8) D’Agostino, P. A.; Hancock, J. R.; Provost, L. R. J. Chromatogr., A 1997, 837, 93-105. (9) Stan’kov, I. N.; Sergeeva, A. A.; Derevyagina, I. D. J. Anal. Chem. 2000, 55, 988-990; Bonierbale, E.; Debordes, L.; Coppet, L. J. Chromatogr., B 1997, 688, 255-264. (10) Verweij, A.; Boter, H. L. Pestic. Sci. 1976, 7, 355-362.

The development and optimization of a method allowing the extraction of intact organophosphorus chemical warfare agent O-ethyl S-(2-diisopropylaminoethyl) methylphosphonothiolate (VX) from several types of soils are presented here. This involved the selection of an appropriate buffer to bring the sample to a pH close to the pKa of VX but sufficiently low to avoid its basic hydrolysis. Buffering with Tris (pH 9) and subsequent extraction of the aqueous layer by a 85:15 (v/v) hexane/dichloromethane mixture allows rapid and sensitive flame photometric detection of VX at spiking levels lower than 10 µg‚g-1, even after 3 months of aging. Extraction yields were close to 60% in complex matrixes. This method also allows recovery and identification of a characteristic degradation product of VX, bis(2-diisopropylaminoethyl) disulfide, which appears to be formed during the aging process. The performance of this method is far better than that of OPCW reference operating procedure, which does not allow extraction of detectable amounts of VX (spiked at 10 µg‚g-1) in one of the soils used for this study.

10.1021/ac035441q CCC: $27.50 Published on Web 04/20/2004

© 2004 American Chemical Society

Analytical Chemistry, Vol. 76, No. 10, May 15, 2004 2791

Figure 1. Structure of VX, Russian-VX, and internal standard.

an inappropriate extraction method, or both. Moreover, only a small fraction of this radioactivity was labeled VX, especially in the case of clayey peat and humic loam, which accounts for a significant degradation of VX in this context. In a further study, poor recovery of radioactivity from clayey peat and humic loam spiked with both 32P- and 35S-labeled VX led to the conclusion that VX was tightly bound to these types of soils.11 A more recent report12 indicates that very high spiking levels (0.1-10 mg‚g-1) might lead to an easier recovery and concludes that VX would rapidly degrade on the soil chosen for this study. However, the analytical method used (cholinesterase inhibition) limits the information on the specific extraction of intact VX. Further attempts eventually used technological advances such as automated extraction methods: Kutuinen and co-workers reported that supercritical carbon dioxide extraction of soil spiked with VX was completely ineffective, whereas ultrasound-assisted extraction with dichloromethane led to a recovery of 0.4% in one out of three experiments, VX remaining undetected in the two other cases.13 A Norwegian team reported that VX can only be detected in soil samples spiked at very high levels (0.5 mg‚g-1), with poor recovery yields (8 ( 6%).14 A more recent and extensive study by the same team,15 involving eight types of soils spiked with 20 µg‚g-1 agent and dichloromethane as extraction solvent, indicates that VX can be correctly extracted from sandy loam (16.2 ( 1.3%) and turf (34.2 ( 3.7%), the latter being a very particular case. All other soils lead to poor yields ranging from 0.7 to 3.2%. The reference document for laboratories working on chemical warfare agent analysis suggests the use of 1% (v/v) triethylamine in methanol for extraction of polar compounds amenable to GC analysis;16 that this method should be convenient for VX extraction is specified by the most comprehensive compendium on this topic.17 Proposed solvents also include acetone and ether, in addition to hexane or dichloromethane.18 A decade ago, an interlaboratory comparison test brought out some other information about VX extraction from a soil whose composition was not given. In this test, soil spiked with 100 µg‚g-1 VX and other contaminants was analyzed by 10 laboratories using (11) Kaaijk, J.; Frijlink, C. Pestic. Sci. 1977, 8, 510-514. (12) Kingery, A. F.; Allen, H. E. Toxicol. Environ. Chem. 1995, 47, 155-184. (13) Kutuinen, M.-L.; Hartonen, K.; Riekkola, M.-L. J. Microcolumn Sep. 1991, 3, 505-512. (14) Tørnes, J. A.; Opstad, A. M.; Johnsen, B. A. Int. J. Environ. Anal. Chem. 1991, 44, 227-231. (15) Hussain, F.; Tørnes, J. A.; Strømseng, A. Proc. 7th Int. Symp. Prot. Chem. Biol. Warfare Agents, 2001. (16) Rauttio, M., Ed. Recommended operating procedures for sampling and analysis in the verification of chemical disarmament, The Ministry of Foreign Affairs of Finland, Helsinki, 1994. (17) Kutuinen, M.-L. In Encyclopedia of Analytical Chemistry; Myers, R. A., Ed.; J. Wiley & Sons: Chichester, U.K., 2000; p 1060. (18) MacNaughton, M. G.; Brewer, J. H. Environmental chemistry and fate of chemical warfare agents; Southwest Research Institute: San Antonio, TX, 1994.

2792

Analytical Chemistry, Vol. 76, No. 10, May 15, 2004

numerous solvent systems.19 Semiquantitative measurements suggest that extraction with dichloromethane, after mixing with sodium sulfate, allows better recoveries (40-70%) than all other solvent systems (0-30%); however, these results must be interpreted with great care, due to the discrepancy between the analytical conditions used in the different laboratories. Complementary investigations by the Finnish laboratory VERIFIN, published in the same report,19 lead to the conclusion that VX (spiking level not reported) is preferentially extracted with chloroform or dichloromethane, satisfactorily from sand (41-61%), but with poor recoveries when spiked in humus or clay (5-18%). Acetone and diethyl ether give even poorer, if measurable, yields. Very recently, Hook et al. reported encouraging results showing that solid-phase microextraction20 was suitable for extracting VX in the headspace of clothing material samples and noted that additionnal work would be required to adapt their method to trace-level analysis and to complex media such as that presented by soil. On the basis of these observations, it seemed appropriate to develop and optimize a method allowing extraction of VX and other V-type nerve agents from various types of soils with satisfactory yields, at low pollution levels. The main aims of this study were the selection of an appropriate buffer to obtain a soil buffer solution at a pH sufficiently close to the pKa of VX to allow leaching, but sufficiently low to avoid basic hydrolysis. The organic liquidliquid extraction step, necessary to allow GC and GC/MS analysis, was subsequently optimized, bearing in mind that these techniques are the most widely available in analytical laboratories dealing with CWC verification. The resulting optimized method was then compared to former reference procedures and applied to the Russian analogue of VX. MATERIALS AND METHODS CAUTION! VX and R-VX as well as some of their impurities and degradation products are exceedingly toxic cholinesterase inhibitors and should be handled only in approved chemical warfare surety laboratories by properly trained personnel. Reagents. VX, R-VX, ( and O-methyl O-n-octyl methylphosphonate (Figure 1) were synthesized at the chemical synthesis facility of the centre d’e´tudes du Bouchet, in strict compliance with the regulations of the Chemical Weapons Convention. VX was >98% pure (R-VX >94%), as determined by GC and 31P NMR. Solvents used (Prolabo, Nogent-sur-Marne, France) were certified for pesticide residue analysis, if possible, or else of the highest quality available. (19) International Interlaboratory Comparison (Round-Robin) Test for the Verification of Chemical Disarmament. F.1. Testing of Existing Procedures, Ministry of Foreign Affairs of Finland, Helsinki, 1990. Detailed data are available as Supporting Information. (20) Hook, G. L.; Kimm, G.; Betsinger, G.; Savage, P. B.; Swift, A.; Logan, T.; Smith, P. A. J. Sep. Sci. 2003, 26, 1091-1096.

Table 1. pH of Soil-Buffer Mixtures pH

soil none local soil (5 g) tropical soil (5 g)

vol of buffer (mL)

Tris buffer

5 5 10 15 5 10 15

9.0 8.3 8.6 8.8 7.2 8.0 8.5

Borax buffer

hydrogenocarbonate buffer

hydrogenophosphate buffer

9.0 7.3 8.2 8.5 5.0 6.8 7.5

9.7 7.2 7.4 7.5 6.4 6.7 6.8

10.8 6.8 7.3 7.5 5.9 6.5 6.9

Mineral salts used were from Sigma-Aldrich (St. Quentin Fallavier, France). Buffer solutions were prepared according to the literature.21 The pH of soil-buffer mixtures (Table 1) was measured by means of a electronic pH-meter (Mettler Delta 345) after 20 min of ultrasound-assisted agitation. Soils (Nature, Spiking, and Aging). Two soils were used for this study, whose properties (measured by the French Institut National de la Recherche Agronomique, Olivet, France) are given in Table 2. Local soil (collected in the vicinity of our laboratory) is a neutral silt loam, and tropical soil (collected in French Guyana) is an acidic sandy silt loam with high organic content. Before spiking, these soils were allowed to dry at 40 °C for 1 week and were sieved at 2 mm. A third, commercially available, type of sample (sea sand purified by acid and calcined/Merck, ref 1.07712) was selected to allow further comparison by other laboratories. For spiking, 4 × 250 µL of a dichloromethane solution of VX at the desired concentration (100 or 1000 µg‚mL-1) was added to a batch of 100 g of soil. The resulting mixture was homogenized, the solvent gently evaporated at ambient pressure and temperature, and then the sample kept at room temperature under ambient light conditions (night/day cycle) in tightly closed vials until experimentation. Contamination level was chosen to be in conformity with the NATO standards for field contamination, as well as the scenarios usually encountered during OPCW Interlaboratory Proficiency Tests. Extraction Procedures. Manual ultrasound-assisted extraction was performed as follows: 5 g of soil was mixed with the chosen buffer (5-15 mL), homogenized (1-min vortex), placed in an ultrasound bath for 20 min, then again homogenized (1 min, vortex). An organic solvent (10 mL) was then added to the resulting mixture (1 min, vortex), and the whole was centrifuged (4000 rpm, 10 min). The organic phase was collected, and the remainder was extracted two more times with organic solvent (10 mL) following the same protocol. The combined organic fractions were dried on sodium sulfate, filtered (PTFE 0.45 µm), and concentrated under a nitrogen stream (30 °C, 1 bar) to a final volume of 1 mL (or eventually less in the case of extracts from aged samples). Unless otherwise specified, all extractions were performed on three separate samples from the same batch of soil. Two other extracting methods were tested as previously described.16,19 Attempts at pressurized solvent extraction were done by means of an ASE 200 extractor (Dionex, Jouy-en-Josas, France) at a (21) CRC Handbook of Chemistry and Physics, 82d ed.; CRC Press: Boca Raton, FL, 2001, p 8-43,

pressure of 150 bar and extraction temperature of 60 °C (Tris buffer) or 40 and 100 °C (mixture of tris buffer and solvent). Gas Chromatographic Analysis. Quantitative analysis was performed by gas chromatography using O-methyl O-n-octyl methylphosphonate (Figure 1) as internal standard. The apparatus was a Varian 3400 GC equipped with a Rtx-5MS capillary column (5% diphenyl, 95% dimethylpolysiloxane of 30-m length, 0.32-mm i.d., 0.25-µm film thickness, Restek, Evry, France) and a flame photometric detector operating in phosphorus mode [splitless injection, injector temperature 250 °C, detector temperature 280 °C, gas flow 1 mL‚min-1, temperature program 50 °C (1 min) and then 10 °C‚min-1 to 260 °C (8 min)]. Quantification of VX with the chosen internal standard (25 µg‚mL-1 of extract), was performed for VX concentrations ranging from 5 to 100 µg‚mL-1 (r2 ) 0.9996, limit of quantification 3 µg‚mL-1). Each extract was analyzed in triplicate Mass Spectrometric Analysis. Confirmation of the structure of VX and eventual degradation products was achieved by GC/ MS with a system (Agilent 6890/5973, with electron impact ionization or ammonia chemical ionization) equipped with a VF5MS column (95% dimethyl, 5% diphenylpolysiloxane, 30-m length, 0.25-mm i.d., 0.25-µm film thickness, Varian, les Ulis, France) using the following GC conditions: splitless injection, injector temperature 270 °C, gas pressure 9 psi, temperature program 50 °C (1 min) and then 10 °C‚min-1 to 275 °C (11.5 min). RESULTS AND DISCUSSION Selection of Buffer. The presence of a basic nitrogen (pKa ) 8.8) in the leaving group of VX was anticipated to be a crucial point for chemisorption onto soil particles and formation of strong hydrogen bonds. Thus, direct extraction by an organic solvent was not a fully satisfying method, with soils having pH much lower than the pKa of VX, where VX should be present as a strongly sequestered quaternary salt. However, acidic or basic extraction media had to be avoided due to instability of VX at high and low pHs. A preconditioning of the soils with a slightly basic buffer had several advantages, first shifting the equilibrium from VX-H+ to VX, thus authorizing organic extraction, but also having a saltingout effect, by forming ion pairs with adsorption sites on the soil particles. Indeed, addition of salts such as buffer solutions could help by dissolving humate salts that entrap analytes of interest.22 Such a use of salts as extraction assistants has been documented for various applications; commonly used salts include sodium sulfate or ethylenediaminetetraacetate, tetrasodium salt.23 The results obtained with the various buffers are reported in Table 1. The high organic content of Guyana soil could play a buffering role; in an intrisincally acidic matrix, this could counteract the effect of the added basic buffer, limit its buffer effect, and explain the lower pHs obtained than with local soil.24 These results prompted us to use Tris (pH 9) in a first series of experiments, this buffer leading to the most important increase of pH. (22) Crescenzi, C.; Di Corcia, A.; Nazzari, M.; Samperi, R. Anal. Chem. 2000, 72, 3050-3055. (23) David, M. D.; Campbell, S.; Li, Q. X. Anal. Chem. 2000, 72, 3665-3670. Guo, F.; Li, Q. X.; Alcantara-Licudine, J. P. Anal. Chem. 1999, 71, 13091315. (24) Noami, M.; Kataoka, M.; Seto, Y. Anal. Chem. 2002, 74, 4709-4715.

Analytical Chemistry, Vol. 76, No. 10, May 15, 2004

2793

Table 2. Composition of the Soils Used for This Study

soil

sand (%)

silt (%)

clay (%)

organic matter content (%)

pH (water suspension)

K+ (mequiv/ 100 g)

Mg2+ (mequiv/ 100 g)

local soil tropical soil

28.9 52.9

51.1 29.2

20 17.9

2.5 7.7

6.9 4.8

0.28 0.21

0.36 0.18

Table 3. Recovery of VX from Fortified Soil Extracts (Tris Buffer), with Various Organic Solvents

soil local soil (5 g)

tropical soil (5 g)

a

vol (Tris buffer) (mL)

pH (mixture buffer soil)

solvent (liq-liq extraction)

VX recovery (%)

5 5 5 5 15 5 5 5 5 15

8.1 8.1 8.1 8.1 8.8 7.2 7.2 7.2 7.2 8.5

hexane ethyl acetate hexane/ethyl acetate (1:1 v/v) hexane/dichloromethane (85:15 v/v) hexane/dichloromethane (85:15 v/v) hexane ethyl acetate hexane/ethyl acetate (1:1 v/v) hexane/dichloromethane (85:15 v/v) hexane/dichloromethane (85:15 v/v)

13 36 30 47 61 nda 10 ,10b 10 54

nd, not detected. b Detected below limit of quantification.

Table 4. Recovery of VX after Liquid-Liquid Extraction of Tris Buffer Extracts of Soils Spiked with 10 µg‚g-1 VX and Comparison with Other Reference Methods soil local soil (5 g)

tropical soil (5 g) calcined sea sand (5 g)

preconditioning (ultrasound)

organic extraction solvent

5 mL of Tris 15 mL of Tris 5 g of sodium sulfateb 5 g of sodium sulfateb nonec 15 mL of Tris 15 mL of Tris

hexane/dichloromethane (85:15 v/v) hexane/dichloromethane (85:15 v/v) hexane/dichloromethane (85:15 v/v) dichloromethane methanol/triethylamine (99:1 v/v) hexane/dichloromethane (85:15 v/v) hexane/dichloromethane (85:15 v/v)

VX recovery (%) 47.2 ( 1.5 59.5 ( 1.5 nda nda nda 42.5 ( 4.0 16.0 ( 1.0

RSD (%) 3.2 2.5

9.4 9.4

a nd, not detected. b Mixture of soil (5 g) and Na SO (5 g) extracted with organic solvent (20 mL) according to ref 19. c Soil (5 g) extracted with 2 4 methanol/triethylamine 99/1 v/v (3 × 5 mL) according to ref 16.

Selection of Solvent for Liquid-Liquid Extraction. Unspiked soil (5 g) was extracted with 5 mL of Tris (pH 9), the subsequent extract fortified with VX (50 µg), and the resulting mixture extracted with various organic solvents (3 × 10 mL). Working on soil extracts, rather than directly on soils, allowed us to specifically study the liquid-liquid extraction step. Solvents were selected based on the available literature for organophosphorus nerve agent extraction. Pure hexane was clearly insufficient, although it had been reported to allow excellent VX recovery from water solutions mixed with high concentrations of “salting-out” reagents.25 However, salt concentrations used by Stan’kov et al.25 are much higher than the concentration of Tris we used and may account for the difference observed. Therefore, dichloromethane, reported as allowing VX extraction from complex mixtures,26 was used in combination with hexane, which allowed us to recover VX satisfactorily (Table 3). Hexane/dichloromethane extracts displayed an intense, symmetrical VX chromatographic peak, while an important broadening (25) Stan’kov, I. N.; Sergeeva, A. A.; Derevyagina, I. D.; Konovalov, K. V. J. Anal. Chem. 2003, 58, 160-164. (26) Savel’eva, E. I.; Zenkevich I. G.; Radilov, A. S. J. Anal. Chem. 2003, 58, 114-123.

2794 Analytical Chemistry, Vol. 76, No. 10, May 15, 2004

of the latter was noticed with ethyl acetate extracts, which can generate a higher uncertainty for quantification. Thus, hexane/ dichloromethane was preferred for further experiments. Parameters Influencing VX Recovery. Three main properties of the buffer could have a major influence on VX recovery: pH, ionic strength, and chemical nature. The influence of pH was studied by an increase of the volume of Tris buffer from 5 to 15 mL (Tables 3 and 4). These results evidence that a significant elevation of pH (all other parameters remaining unchanged) leads to a concomitant elevation of VX extraction yield. To further ascertain the respective roles of pH and chemical nature of the buffer, a local soil sample spiked with VX (10 µg‚g-1) was mixed with 10 mL of Borax buffer (final pH 8.22) and compared to extracts with 5 mL of Tris buffer + 5 mL of deionized water (to yield the same volume of extract and give comparable processing steps, final pH 8.27). VX recovery was also comparable in both cases (Borax, 27.2 ( 1.6%, Tris + water, 32.5 ( 2.5%), clearly indicating that the chemical nature of the buffer is not at stake, while pH of the buffer is confirmed to govern the extraction process.

Last, the significant decrease in recovery observed when diluting Tris (which has a little effect on pH) suggests that the ionic strength of the buffer also somewhat influences the extraction. However, buffers are necessary to produce a smooth elevation of pH, avoiding basic hydrolysis of VX, and are generally associated with high ion concentrations: therefore, this parameter was not further studied. Since VX cannot be conveniently transferred from aqueous solutions at pHs below 8 to organic solvents, other buffers (hydrogenocarbonate, hydrogenophosphate) were not further studied; Tris was retained for the following experiments, with respect to its superior capacity to bring both soils at the desired pH. Loss of VX during the Extraction Process. Additional experiments were performed to determine why high proportions of VX (40-60% when spiked at 10 µg‚g-1, and even more at lower spiking level) were not recovered. Apart from the intrinsic efficiency of the extraction solvents, two possible causes for reduced recovery were identified: losses during extract processing steps and VX stability in buffer solutions. Therefore, a hexane/ dichloromethane mixture (85:15 v/v, 30 mL) fortified with 50 µg of VX was treated (filtration, concentration) as organic extracts obtained in the extraction process. Quantification of VX indicated that roughly 20% of the initial quantity was lost during the filtration and concentration step (30 to 1 mL or less), most probably by adsorption on the PTFE filter and glassware, which was already reported by others;20 the very low volatility of VX suggests that evaporation has to be marginal. Thus, the maximal expectable recovery may be 80%, which indicates that the extraction step by itself allows extraction of 55-75% of VX from the soil. The same protocol was used to assay losses due to evaporation and filtration for samples spiked with 1 µg‚g-1 VX. In this case, these steps seemed to be responsible for nearly all unrecovered VX and allowed a maximal theoretical recovery of ∼25%. Then, since the extraction step lasts 22 min, the stability of VX in Tris buffer medium was also assayed; no significant difference was observed whether a Tris solution of VX was extracted immediately (79.8 ( 1.9%) or after 22 min (78.1 ( 2.3%). Thus, degradation during the process is negligible. Ultrasound-Assisted Extraction of VX from Contaminated Soils. Soils spiked with 10 µg‚g-1 VX were extracted 24 h after spiking with 15 mL of Tris buffer following the manual ultrasoundassisted protocol described previously. Results (Table 4) indicate that the extraction yield is very good with both soils, as well as repeatability, compared to previous literature data discussed in the introduction section. Apparently, deep interactions of VX as well as degradation on local soil occur quite slowly, since recovery is not significantly changed when extracting a 24-h-old spiked soil or a fortified soil extract (59-61%). A second series of samples were spiked with 1 µg‚g-1 VX and extracted 24 h after spiking with 15 mL of Tris. VX was recovered in a 25.7 ( 3.7% yield from local soil and a ∼10% yield from tropical soil (VX was clearly detected and identified by GC/MS analysis, albeit below quantification limit). The important losses during extract processing described previously give a satisfactory justification for this important decrease. Finally, and to allow further comparison by other laboratories, our method was applied to a standardized calcined sea sand

Figure 2. Recovery of VX in aged local and tropical soils.

sample spiked with 10 µg‚g-1 VX. Recovery was lower than in both complex soils used formerly, yet VX was clearly detected by GC/FPD. Application to Russian VX. The extraction method optimized for VX (with 15-mL Tris addition) was applied to extract Russian VX spiked (10 µg‚g-1) in the two reference soils, and in the calcined sand, used in this study. Recovery was slightly lower than that obtained with VX but was in all cases satisfactory (local soil, 46.0 ( 0.1%, tropical soil 25.3 ( 0.6%, sea sand 21.3 ( 1.8%). Losses due to concentration and filtration were evaluated, and the theoretical maximal recovery was determined to be 77%. Unsuccessful Extraction Attempts. To compare our method with previously recommended extraction processes, we performed an extraction attempt of VX from tropical soil (spiking level 10 µg‚g-1) following the OPCW recommended procedure (i.e., using 1% (v/v) triethylamine in methanol) or after mixing this sample with an equal mass of sodium sulfate.19 The OPCW procedure was tedious, with a particularly difficult filtration step, and led to severe damage of GC columns, due to the presence of concentrated triethylamine in the injected solution. In both cases, the amount of VX extracted was below the detection limit of the GC/ FPD method. Pressurized solvent extraction was reported by our laboratory27 and others28 to be suitable for extraction of chemical warfare agents and their degradation products. Unfortunately, all attempts to extract VX by this technique [with Tris buffer alone or mixed (1/2, v/v) with hexane/dichloromethane (85/15, v/v) or ethyl acetate] proved unsuccessful, although the conditions used were reported to be suitable for extraction of organophosphorus pesticides,29 as well as other persistent organic pollutants, from soils.30 The simultaneous use of nonmiscible solvents, and the high salt concentration, seem to be unsuited to this apparatus and were probably responsible for the multiple troubles encountered (blockade of the system, long and tedious filtration steps); moreover, drastic temperature and pressure conditions may lead to increased degradation or interactions of VX with the matrixes used in this study. Aging of VX in Soils. The optimized protocol (buffering with 15 mL of Tris, pH 9) was applied to dry soils spiked with 10 µg‚g-1 VX after different durations of aging, at ambient temperature. The results, given in Figure 2, show that the most important decrease (27) Chaudot, X.; Tambute´, A.; Caude, M. J. Chromatogr., A 2000, 866, 231240. (28) Beck, N. V.; Carrick, W. A.; Cooper, D. B.; Muir, B. J. Chromatogr., A 2001, 907, 221-227. (29) Lopez-Avila, V. Crit. Rev. Anal. Chem. 1999, 29, 195-230.

Analytical Chemistry, Vol. 76, No. 10, May 15, 2004

2795

Figure 3. Electron impact ion chromatograms (top, total; middle top, extracted at m/z ) 114) of an extract from a 96-day old local soil sample and corresponding mass spectra (middle bottom, electron impact; bottom, chemical ionization) of VX in this extract (IS, internal standard).

(50%) of VX recovery takes place during the first two weeks. This decrease can be due either to reinforced solute-matrix interactions or to degradation of the agent. Although degradation products have not been thoroughly investigated, VX appears to (30) Bjo ¨rklund, E.; Nilsson, T.; Bøwadt, S. Trends Anal. Chem. 2000, 19, 434445.

2796 Analytical Chemistry, Vol. 76, No. 10, May 15, 2004

be degraded at a moderate pace, if degraded at all, by both complex soils tested here, taking into account that the method allows us to extract a theoretical maximum of 80% of the initial spiked quantity. These results are not consistent with previous reports, indicating that the degradation rate of VX in soils could be as high as 50% per day12 or 90% after 2 days;10 however, as we

already stated, all previous reports dealing with the environmental fate of VX conclude a rapid degradation since no intact product is extracted, yet do not clearly prove that the method used would have allowed us to extract VX if some of it remained undegraded. Interestingly, Groenewold and co-workers, using a completely different approach for the determination of VX (direct ionization of VX on the surface of soil particles using ion trap secondary ion mass spectrometry), observed that VX remains virtually undegraded after a 6-month storage at 4 °C of spiked samples,31 which would support the results we present in this paper. Likewise, recent work performed on sandy soil samples (about 92% sand, 8% clay) spiked with amiton [O,O-diethyl S-(2-diethylaminoethyl) phosphorothiolate], a phosphate analogue of V-agents, shows that 40% of the initial spiked quantity may still be recovered after 28 days.32 Last, mass spectrometric experiments targeted on 96-day samples proved that the quantity of VX extracted (10%) is sufficient for an unambiguous characterization by electron impact and chemical ionization (Figure 3), in conformity with the requirements for analyses conducted under the aegis of the Organization for the Prohibition of Chemical Weapons. It is noteworthy that bis-2-diisopropylaminoethyl disulfide, a significant product emerging from VX degradation, is also identified by these techniques; no other relevant degradation compound was detected in the organic extracts analyzed by GC or GC/MS. Identification of this product is consistent with previous statements suggesting that it would be a major signature of VX33 and supports the interest of recent developments aiming at the determination of VX contamination through detection of this compound in soils.34 CONCLUSIONS The procedure presented here allows the extraction of VX from two different types of complex soils (with medium to high silt and clay contents, which are reputed to be highly penalizing for (31) Groenewold, G. S.; Appelhans, A. D.; Gresham, G. L.; Olson, J. E.; Jeffery, M.; Wright, J. B. Anal. Chem. 1999, 71, 2318-2323. (32) Borrett, V. T.; Gan, T.; Lakeland, B. R.; Leslie, D. R.; Mathews, R. J.; Mattsson, E. R.; Riddell, S.; Tantaro, V. J. Chromatogr., A 2003, 1003, 143155. (33) Small, M. J. Technical Report 8208, USAMBRDL, Fort Detrick, 1983. (34) Hook, G. L.; Kimm, G.; Koch, D.; Savage, P. B.; Ding, B.; Smith, P. A. J. Chromatogr., A 2003, 992, 1-13. (35) Lemarie´, L.; Sokolowski, M. S.; Wickramage, C. Proc. 3d Singapore Int. Symp. Prot. Toxic Subst. 2002. (36) Groenewold, G. S.; Williams, J. M.; Appelmans, A. D.; Gresham, G. L.; Olson, J. F.; Jeffery, M. T.; Rowland, B. Environ. Sci. Technol. 2002, 36, 47904794.

extraction, in particular in the field of chemical warfare agents35) at low spiking levels, even in samples aged up to 3 months, with excellent sensitivity, selectivity, and repeatability, and in a limited time of manipulation. The concept underpinning the procedure, namely, buffering the matrix to a pH close to the pKa of VX, theoretically allows an extension to all acidic or neutral matrixes. This procedure is also valid for the Russian analogue of VX and can presumably be adapted to all S-dialkylaminoalkyl O-alkyl alkylphosphonothiolates (V-agents). Our results clearly demonstrate the inadequacy of the OPCW recommended operating procedure, as well as other standard protocols, and suggest that the previously admitted behavior of VX on soils (i.e., rapid degradation) could rather reveal the poor efficiency of extraction methods used in these studies. The capability of our method to allow unambiguous mass spectrometric identification of traces of VX 3 months after spiking of a dried soil is a determinant progress for OPCW purposes, since it is most probable that samples collected under its responsibility will generally have been polluted for several weeks. A limitation to our results is that dry soils were used for spiking, which will not be always the case in real sampling; further work is now in progress to evaluate the impact of soil humidity on VX recovery. Complementary experiments are also necessary to fully characterize polar degradation products that may emerge during the aging process and thus discriminate VX degradation from adsorption in depth to soil particles. Recent advances using solid-phase microextraction20 suggest the use of this technique to recover VX from the buffer extract, instead of using liquid-liquid extraction. Last, application to basic matrixes such as concrete will be considered, which will allow to complete stimulating results obtained using direct surface interrogation.36 ACKNOWLEDGMENT We warmly thank Laurent Botti, Sylvie Laiguillon, and Jacky Dissard for GC/MS identification of VX byproducts, and Lucien Coppet for numerous helpful discussions. SUPPORTING INFORMATION AVAILABLE Physicochemical properties of VX and detailed results from an interlaboratory comparison test. This material is available free of charge via the Internet at http://pubs.acs.org. Received for review December 5, 2003. Accepted March 6, 2004. AC035441Q

Analytical Chemistry, Vol. 76, No. 10, May 15, 2004

2797