Environ. Sci. Technol. 2001, 35, 1823-1829
Effect of Pedological Characteristics on Aqueous Soil Extraction Recovery and tert-Butyldimethylsilylation Yield for Gas Chromatography-Mass Spectrometry of Nerve Gas Hydrolysis Products from Soils MIEKO KATAOKA,† KOUICHIRO TSUGE,† HIROSHI TAKESAKO,‡ T A D A O H A M A Z A K I , § A N D Y A S U O S E T O * ,† National Research Institute of Police Science, 6-3-1 Kashiwanoha, Kashiwa, Chiba 277-0882, Japan, and Department of Agricultural Chemistry, Meiji University, 1-1-1 Higashi-mita, Tama-ku, Kawasaki, Kanagawa 214-8571, Japan, and National Institute of Agro-Environmental Sciences, 3-1-1 Kannondai, Tsukuba, Ibaraki 305-8604, Japan
Detection and identification of alkyl methylphosphonate (RMPA) and methylphosphonate (MPA) are performed to verify the existence of nerve gases by gas chromatographymass spectrometry (GC-MS) after tert-butyldimethylsilylation (TBDMS). However, it is sometimes difficult to detect RMPA and MPA in soils. This study examines the relationship between the pedological characteristics and the aqueous extraction recoveries and TBDMS derivatization yields of ethyl-, isopropyl- and pinacolyl methylphosphonate and MPA for 21 soil samples. The aqueous extraction recoveries were measured directly by capillary electrophoresis. Andosols showed low extraction recoveries, while Regosols and Fluvisols showed high recoveries. RMPA were extracted with higher recoveries than MPA from all soils. MPA could not be extracted from Andosols. Within the pedological characteristics, phosphate absorption coefficients showed a strong negative correlation with the extraction recoveries of all phosphonates. The levels of RMPA and MPA in aqueous soil extracts were also determined for eight soils by GC-MS after TBDMS. Compared to the aqueous extraction recoveries, the yields of TBDMS derivatives were low. Strong anion exchange led to a significant improvement in derivatization yields. The efficiencies of TBDMS derivatization were inversely correlated with the levels of alkaline earth metals extractable from soils when the three soils that possessed high total carbon were excluded.
Introduction The Chemical Weapons Convention, which went into force in April 1997, bans the production, stockpiling, and use of chemical weapons. In this context, it is important to verify the existence of chemical warfare agents (CWA) in suspected * Corresponding author telephone: (81)-471-35-8001; fax: (81)471-33-9159; e-mail:
[email protected]. † National Research Institute of Police Science. ‡ Meiji University. § National Institute of Agro-Environmental Sciences. 10.1021/es001529z CCC: $20.00 Published on Web 03/17/2001
2001 American Chemical Society
sites. Sarin, soman, tabun, and VX are representative organophosphorus neurotoxic compounds (1). In water, these nerve gases are readily hydrolyzed to the characteristic alkyl methylphosphonates (RMPA): isopropyl methylphosphonate (IMPA), pinacolyl methylphosphonate (PMPA), and ethyl methylphosphonate (EMPA) (2). RMPA are ultimately hydrolyzed to methylphosphonate (MPA). Tabun undergoes degradation by a slightly different pathway (Figure 1). Thus, it may be difficult to detect nerve gases directly in their active form from CWA verification sites. However, indirect evidence can be obtained via the identification of RMPA. The determination of MPA is also important because it is a stable hydrolysis product of RMPA and is also derived from synthetic intermediates or byproducts of nerve gas production such as dimethyl methylphosphonate, methylphosphonyl dichloride, and methylphosphonyl difluoride. The members of the Japanese Cult, AUM SHINRIKYO, caused the Sarin gas attacks in Matsumoto in 1994 and in the Tokyo Subway System in 1995, resulting in 19 deaths and numerous injuries (3). Our laboratory has been engaged in forensic investigations on such chemical terrorism cases and has performed analyses of various types of evidence samples, many of which consist of complex matrixes (4, 5). In terms of the analysis of RMPA and MPA in a complex matrix, various chromatographic methods have been developed and utilized for chemical verification, hazardous material monitoring, and forensic investigation (6-8). The method of gas chromatography-mass spectrometry (GC-MS) for the determination of volatile derivatives is well established, and this technique has used after tert-butyldimethylsilylation (TBDMS) (9) because of the ease of sample processing and high derivatization efficiency. In the course of our forensic investigations, we have noted a low detectability of RMPA and MPA. Because of its complex matrix, soil is one of the most difficult samples from which CWA and related compounds can be determined. It has been reported that GC-MS analysis of soils after derivatization suffered from serious problems, in terms of the detection of low levels of nerve gas hydrolysis products (10). Another paper also reported on the low recovery of RMPA and MPA from soils by GC analysis after trimethylsilylation (11). The low detectability of RMPA and MPA in soil samples in TBDMS GC-MS analysis can be attributed to the following two issues. One involves the interference in TBDMS derivatization by compounds which are co-extracted from soils. The other involves the relatively strong adsorption of RMPA and MPA to soils. In a previous study (12), we concluded that divalent cations, calcium (Ca2+) and magnesium (Mg2+), have a great effect on the detection of RMPA and MPA. We reported on the development of simple strong cation-exchange pretreatment methods to remove cations from soil extracts to improve detectability as is already used in Recommended Operating Procedures (13). Moreover, using a macroporous strong anion-exchanger resin, we established an efficient pretreatment method for the removal of interfering compounds including not only cations but also neutral substances (14). We also indicated that RMPA and MPA were extracted from three different types of soils with fairly reasonable recoveries except for MPA in alluvial soil and that the TBDMS derivatization yields were low and could be significantly improved by pretreatment with a strong cation or anion exchanger (12, 14). However, our previous studies were limited to only a few soil samples. In addition, the soils were involuntarily collected, and two of them were sampled from city environments. VOL. 35, NO. 9, 2001 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
1823
FIGURE 1. Hydrolysis of nerve gases. We now report on an investigation of the relationship between the pedological characteristics, aqueous extraction recoveries, and TBDMS derivatization yields of RMPA and MPA by using various types of soils sampled by National Institute of Agro-Environmental Sciences (NIAES) (15). The results show that the aqueous extraction recoveries for all phosphonates were inversely correlated with the phosphate absorption coefficient (PAC) and that the TBDMS derivatization yields in aqueous soil extracts were low for soils which contained high levels of extractable alkaline earth metals.
Experimental Section Reagents. Muromac MSA-1 (50-100 mesh) was obtained from Muromachi Chemical Industry (Tokyo, Japan). NMethyl-N-(tert-butyldimethylsilyl)trifluoroacetamide (MTBSTFA) was from Pierce (Rockford, IL). MPA, EMPA, and PMPA were from Aldrich Chemicals (Milwaukee, WI). IMPA was prepared as previously reported (12). All other chemicals were of analytical grade. All aqueous solutions were prepared with distilled, deionized water. MPA, EMPA, IMPA, and PMPA were dissolved in acetonitrile (400 µg/mL) and stored at -20 °C for use as stock solutions. A working solution was prepared by diluting the stock solution with additional acetonitrile. Soil Samples. The following three soil samples were used as previously reported (12). Soil no. 7 (volcanic soil) was sampled from the garden in our previous Institute (Chiyodaku, Tokyo). Soil no. 15 (sand) was sampled from a seashore in Okayama prefecture. Soil no. 21 (alluvial soil) was sampled from a shrine park in Kyoto City. An additional 18 soils collected from various areas of Japan were supplied from the Laboratory of Soil Genesis & Classification, NIAES. The pedological characteristics of the soil samples were defined by NIAES and us according to standard methods (16) and are depicted in Table 1. Soil samples were classified into five groups according to FAO-Unesco (17). Andosols (Kuroboku soil or volcanic soil) are nos. 1-8. Acrisols (red soils) are nos. 9 and 10. Luvisols (dark red soils) are nos. 11 and 12. Regosols (immature soil) are nos. 13-15. Fluvisols (alluvial soil) are nos. 16-21. Aqueous Extraction of Soil Samples. A total of 100 µL of an acetonitrile solution containing 70-100 µg of RMPA and 1824
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 35, NO. 9, 2001
MPA was spiked to 1.0 g of the soil samples and allowed to stand at room temperature for 4-5 h. Two milliliers of distilled water was added, and the resulting suspension was vortexed for 1 min and sonicated for 10 min at room temperature. Samples were centrifuged at 1500g for 5 min, and the resultant supernatant was filtered through a 0.45-µm cellulose acetate membrane. Aliquots (0.3 mL) were analyzed by capillary electrophoresis (CE), and also directly or after strong anion exchange (SAX), they were subjected to TBDMS derivatization and subsequent GC-MS analysis. Capillary Electrophoretic Determination of Nerve Gas Hydrolysis Products. RMPA and MPA in the aqueous soil extracts were determined as previously described (18) using a Quanta 4000E CE system (Waters, Milford, MA). The capillary column was a fused silica (75 µm i.d. × 60 cm), and the electrophoresis buffer was 100 mM boric acid, which contained 10 mM benzoate (pH 6.0). The voltage was set at 30 kV with a positive power supply. Detection was by indirect ultraviolet absorption at 254 nm, and the column was maintained at 25 °C. Samples were applied hydrostatically for 30 s. Capillary Electrophoretic Determination of Metal Cations. Concentrations of sodium, potassium, Ca2+, and Mg2+ ions in the aqueous soil extract were measured by CE as described previously (12). The capillary column used was a fused silica (75 µm i.d. × 60 cm), and the electrophoresis buffer was Ion Select Low Mobility Cation Electrolyte (Waters) containing 18-crown-8-ether and hydroxyisobutylic acid. The voltage was set at 20 kV with a positive power supply. Detection was indirect ultraviolet absorption at 214 nm. The column temperature was maintained at 25 °C. Samples were applied hydrostatically for 30 s. Strong Anion-Exchange Pretreatment. The aqueous soil extracts of the eight soil samples shown in Figure 3 were treated with SAX, using a macroporous Muromac MSA-1 resin as described previously (14). A total of 200 µL of the aqueous soil extract was applied to the resin column (2.3 cm × 0.5 cm i.d., 0.5 mL of resin) at a flow rate of 0.5 mL/min and washed with 10 mL of distilled water, followed by 3.5 mL of 0.1 M hydrochloric acid (HCl) solution. Analytes were eluted with an additional 3.0 mL of 0.1 M HCl solution. The washing and elution flow rate was manually maintained between 0.5 and
TABLE 1. Pedological Characteristics of the Soil Samples Used particle distribution (%) soil group
no.
location
depth (cm)
Andosols
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21
forest forest forest forest farm farm city garden paddy field forest forest farm farm farm forest seashore paddy field paddy field paddy field paddy field paddy field city park
0-10 69-89 12-22 48-82 0-16 118-151 0-5 0-10 0-12 25-57 0-19 55-75 0-26 0-9 0-5 0-12 0-10 0-11 37-90 0-20 0-5
Acrisols Luvisols Regosols
Fluvisols
a
Cation-exchange capacity.
b
pH
clay 20 µm
H2O
49.0 67.0 40.0 50.0 55.0 47.0 ndc 45.6 36.0 49.0 57.0 87.0 42.0 0.1 0.9 17.8 24.7 15.5 8.7 50.0 14.0
30.0 24.0 17.0 29.0 26.0 26.0 nd 35.5 12.0 11.0 33.0 11.0 43.0 0.6 0.7 20.3 39.3 25.6 15.5 45.0 13.3
21.7 9.5 44.0 21.7 19.4 27.2 nd 18.9 51.6 40.2 10.8 2.4 14.8 100.0 98.5 60.9 36.4 58.9 75.8 5.6 72.6
5.0 4.9 5.0 5.5 6.5 5.7 nd 5.8 4.2 4.6 7.5 7.7 7.9 4.9 7.3 5.7 6.7 5.6 6.5 6.4 6.4
exchangeable cation (mequiv kg-1)
KCl
total Ca (g kg-1)
total N (g kg-1)
Ca
Mg
K
4.1 4.1 4.5 5.2 5.6 5.5 nd 4.7 3.6 3.8 6.6 6.2 7.0 3.9 6.2 5.2 6.1 4.5 5.7 5.3 5.7
99.8 13.2 178.0 11.2 37.6 13.9 nd 35.0 45.0 1.4 17.8 7.3 nd 5.0 0.0 16.0 11.0 20.0 1.0 23.2 35.7
5.4 1.1 8.5 1.3 3.2 1.5 nd 2.8 3.3 0.3 1.8 1.3 nd 0.3 nd 1.3 0.1 1.4 0.1 2.1 nd
1.0 0.2 0.3 4.0 113.0 25.8 nd 265.0 5.0 5.1 205.0 152.1 601.0 3.7 32.4 117.6 155.6 43.0 13.5 141.0 159.4
2.0 0.9 0.9 12.8 43.1 31.9 nd 87.4 5.0 5.3 18.5 13.1 37.1 4.6 9.1 41.9 65.1 8.0 3.6 124.0 15.4
5.8 2.1 1.3 3.3 9.2 4.7 nd 0.0 4.1 2.3 10.6 2.9 6.4 0.7 1.9 2.3 3.7 4.3 0.4 18.4 8.3
Na
CECa (mequiv kg-1)
PACb
4.7 4.6 1.7 4.2 7.5 11.2 nd 0.8 1.6 0.8 7.0 3.4 2.3 0.4 22.5 3.6 7.1 1.6 0.0 30.1 1.2
444 401 401 160 198 225 nd 414 165 87 211 183 128 27 11 181 244 94 20 334 205
2230 1670 2570 2210 2240 2530 1758 1580 782 1088 1248 1427 1591 388 118 991 1060 495 98 1322 548
Phosphate absorption coefficient (P2O5 mg/100 g of soil). c Not determined.
FIGURE 2. Recoveries of alkyl methylphosphonates and methylphosphonates by aqueous extraction from soil samples. The recovery values were obtained after aqueous extraction and direct capillary electrophoretic analysis and represent averages (three determination) ( standard deviation. 1.0 mL/min. The eluted fraction was neutralized with sodium bicarbonate (pH approximately 7) and concentrated under reduced pressure at 50 °C on a rotary evaporator. tert-Butyldimethylsilylation and Gas ChromatographyMass Spectrometry. The aqueous soil extract or the SAX elution fraction was subjected to TBDMS derivatization followed by GC-MS as previously described (14). A total of 100 µL of the aqueous soil extract or the concentrated SAX eluate fraction was dried on a model VC-360 centrifugal concentrator (Taitec, Saitama, Japan) under reduced pressure at 50 °C in a 1-mL stoppered glass vial (Nichiden Rika Garasu, type MV-07, Tokyo, Japan). A total of 50 µL of MTBSTFA and 50 µL of acetonitrile, which also contained 12 µg/mL anthracene (internal standard (IS)), were added. The vial was then closed with a Teflon screw cap, homogenized by sonication for 5 min, and incubated at 60 °C for 1 h. A total of 1 µL of the mixture was applied to the GC-MS system, which consisted of an HP 6890 gas chromatograph combined with an HP 5973 quadrupole mass spectrometer (Yokowaga Analytical Systems, Tokyo, Japan). The stationary phase was a capillary column HP-5MS (30 m × 0.25 mm i.d., 0.25 µm thickness, J&W Scientific, Folson, CA). The carrier gas (helium)
flow rate and splitter ratio were adjusted to 0.8 mL min-1 and 50, respectively. The injection port, transfer line, and ion source were maintained at 250, 280, and 230 °C, respectively. Electron impact ionization (ionization energy 70 eV, ionization current 60 µA) was used as the ionization mode. The oven temperature was controlled by a program (initial temperature, 90 °C (1 min hold), then a ramp to 290 °C at 20 °C per min (5 min hold)). The acquisition mass range was 50-550, and sampling was 0.8 scan/s. Acquisition was started 4 min after sample injection. The extracted ion chromatograms were obtained at m/z 153 for the EMPA, IMPA, and PMPA derivatives; at m/z 267 for the MPA derivative; and at m/z 178 for IS. Statistical Analysis. Multiple regression analysis was performed using STATISTICA (StatSoft Inc., Tulsa, OK).
Results Aqueous Extraction of Nerve Gas Hydrolysis Products from Soils. In this experiment, we quantified the RMPA and MPA extracted from soils that had been allowed to stand for only 4-5 h after addition of the analyte acetonitrile solution. These VOL. 35, NO. 9, 2001 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
1825
TABLE 2. Metal Cation Concentrations in Aqueous Soil Extracts (mM) Andosols
Ca Mg K Na
1
2
3
4
5
0.06 0.11 0.13 0.42
0.02 0.02 0.06 0.25
0.04 0.15 0.29 0.16
0.02 0.05 0.01 0.19
0.51 0.42 0.20 0.16
6
7
0.08 12.8 0.29 4.90 0.01 1.64 0.24 6.66
Acrisols
Luvisols
8
9
10
11
12
13
Regosols
0.98 0.44 0.09 1.00
0.44 0.33 0.39 0.70
0.06 0.05 0.09 0.24
0.88 0.12 0.23 1.30
0.58 0.07 0.03 0.71
3.64 0.98 0.27 2.16
14
Fluvisols 15
0.05 1.24 0.06 2.03 0.08 0.22 0.15 10.6
16
17
18
19
20
21
0.86 0.38 0.12 0.98
0.99 0.54 0.10 1.25
0.74 0.26 0.50 0.51
0.16 0.05 0.09 0.20
2.65 2.82 0.44 5.00
0.34 0.11 0.65 0.15
TABLE 3. Influence of Pedological Characteristics on Aqueous Extraction Recoveries of Alkyl Methylphosphonates and Methylphosphonate from Soils coeff of detemination (R 2)
parameter/standardized partial regression coeff (β) PMPA IMPA EMPA MPA
PAC -0.65 PAC -0.76 PAC -0.75 particle distribution of clay -0.37
Ca2+
exchangeable 0.52 exchangeable Ca2+ 0.56 exchangeable Ca2+ 0.49 extracted Mg2+ 0.37
experiments were not intended to examine the exact adsorption isotherm. The spiked analyte levels onto the soils (70-100 µg/g) were below those that are generally used for the study of adsorption isotherms. The present CE method using benzoate-borate buffer provides accurate (within-day repeatability: 1.1% (PMPA) - 1.8% (IMPA), 30 µg/mL injection, n ) 4) and sensitive (detection limit, 2.5 µg/mL (S/N ) 3)) determination of RMPA and MPA in the aqueous soil extracts without interference of chloride and bicarbonate ions and with full recovery (14, 18). As shown in Figure 2, the aqueous extraction recoveries of PMPA, IMPA, and EMPA were similar in the respective soil samples, and the recoveries of MPA were considerably lower than those of RMPA. From all Regosols, Fluvisols, Luvisols, and two of the Andosols, RMPA were extracted with a nearly full recovery. The recoveries of RMPA were generally low in the case of the Andosols. From the Fluvisols and Regosols, MPA was extracted with considerably high recoveries ranging from 13 to 73%. In contrast, MPA could be extracted from the Andosols, Acrisols, and Luvisols in only limited recoveries below 13%. We statistically investigated the influence of pedological characteristics on the extraction recoveries of RMPA and MPA. The concentrations of alkaline metals and alkaline earth metals in the aqueous soil extracts (1.0 g of soil extracted with 2.0 mL of water) were examined and are shown in Table 2. Multiple linear regression analysis was performed by forward selection method on the correlation of the aqueous extraction recoveries with the values of the pedological characteristics (10 parameters as shown in Table 1 and 4 parameters as shown in Table 2) among the 19 soil samples. We eliminated three parameters (particle distribution of sand, pH (KCl), total nitrogen (TN)) from the predictor valuables because they were strongly correlated with particle distribution of clay, pH (H2O), and total carbon (TC), respectively (γ > 0.9). Within 21 soils, 2 soil samples lacked one or more examined parameter and were excluded from the statistical analysis. In Table 3, standardized partial regression coefficients of four parameters obtained after the third selection are presented. PAC gave strong negative correlation with the extraction recoveries of three RMPA. Exchangeable Ca2+, pH (H2O), and TC showed moderate positive correlation. The coefficients of determination were all high (R 2 > 0.88). In contrast, particle distribution of clay gave moderate negative correlation with the extraction recoveries of MPA. PAC showed negative correlation, and extracted Mg2+ and extracted K+ showed positive correlation. However, the coef1826
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 35, NO. 9, 2001
pH (H2O) 0.27 total C 0.16 pH (H2O) 0.23 PAC -0.36
total C 0.14 pH (H2O) 0.12 total C 0.18 extracted K+ 0.21
0.89 0.88 0.89 0.81
ficients of determination was not high (R 2 ) 0.81). Yields of tert-Butyldimethylsilyl Derivatives of Alkyl Methylphosphonates and Methylphosphonate from Aqueous Soil Extracts. The present GC-MS method provides accurate (within-day repeatability: 9.8% (PMPA) - 12.0% (IMPA), 0.8 µg in vial, n ) 7) and sensitive (detection limit: 50 ng in vial, S/N ) 3) determination of RMPA and MPA (12). From soil samples, the within-day repeatability was obtained between 8% (IMPA) and 24% (MPA) (3 µg/g of soil, n ) 5), and the detection limit was 0.1 (EMPA, IMPA) or 0.2 µg (MPA, PMPA)/g of soil (12). To investigate the relationship between pedological characteristics and the yields of TBDMS derivatives, eight soil samples of various soil types were selected and further examined. As shown in Figure 3A, the TBDMS derivatization yields of PMPA, IMPA, and EMPA were not so different in the respective soil samples, and the yields roughly increased with increasing hydrophobicity of the analytes. The yields of MPA were considerably lower than the others. The no. 5 Andosols sample showed high yields of RMPA (>60%), which were nearly the same as the aqueous extraction recoveries (Figure 2). SAX pretreatment did not therefore improve the yields of RMPA (Figure 3B). The yields of MPA were too low to evaluate quantitatively. In contrast, the no. 7 soil (Andosols) gave very low yields of RMPA (91%). SAX pretreatment therefore led to a significant improvement in the TBDMS derivatization yields of RMPA (>38%) and enabled the detection of MPA, even at low yields (2.3%). The low level of delectability in no. 7 soil can be explained from its extremely high Ca2+ levels (Table 2). Even after SAX pretreatment, the TBDMS derivatization yields were not equivalent to the aqueous extraction recoveries. The ratio was below 56%. It is likely that unknown compounds extracted from the soil were not removed by SAX pretreatment and suppressed the TBDMS derivatization. With respect to the no. 9 soil (Acrisols), despite the relatively high aqueous extraction yields (RMPA >67%, MPA 13%), the TBDMS derivatization recoveries were low (54%) and of MPA (>20%). This can be mainly attributed to the high levels of Ca2+ and Mg2+ in the aqueous extracts. In respect to the Fluvisols, nos. 17 and 20 soils showed high TBDMS derivatization yields for RMPA (59-88%), which were similar to the aqueous extraction recoveries (85-97%). MPA was detected in low yields (7%), which was probably due to high levels of alkaline earth metals in aqueous extracts. In contrast, no. 21 soil, which was sampled from a garden in Kyoto, showed rather high aqueous extraction recoveries for RMPA (>93%) and MPA (60%); the soil showed low TBDMS derivatization yields for RMPA (
