Quantitative Analysis of Chemical Warfare Agent ... - ACS Publications

The recent implementation of the Chemical Warfare Convention (CWC) Treaty and ..... and extending the range of analytes to include biological warfare ...
0 downloads 0 Views 171KB Size
Anal. Chem. 1998, 70, 3598-3604

Quantitative Analysis of Chemical Warfare Agent Degradation Products in Reaction Masses Using Capillary Electrophoresis Alaa-Eldin F. Nassar,* Samuel V. Lucas, and Craig A. Myler

Battelle Memorial Institute, 2012 Tollgate Road, Bel Air, Maryland 21015 William R. Jones and Michael Campisano

Thermo BioAnalysis Corporation, 8 East Forge Parkway, Franklin, Massachusetts 02038 Lynn D. Hoffland

U.S. Army Material Command Treaty Laboratory, Aberdeen Proving Ground, Maryland 21010

ranged from 95 to 97% of the true incremental value (R2 ) 0.9999 for the linear regression).

Quantitative methods have been developed for the analysis of chemical warfare agent degradation products in reaction masses using capillary electrophoresis (CE). This is the first report of a systematic validation of a CE-based method for the analysis of chemical warfare agent degradation products in agent neutralization matrixes (reaction masses). After neutralization with monoethanolamine/ water, the nerve agent GB (isopropyl methylphosphonofluoridate, Sarin) gives isopropyl methylphosphonic acid (IMPA) and O-isopropyl O′-(2-amino)ethyl methylphosphonate (GB-MEA adduct). The nerve agent GD (pinacolyl methylphosphonofluoridate, Soman), [pinacolyl ) 2-(3,3-dimethyl)butyl] produces pinacolyl methylphosphonic acid (PMPA) and O-pinacolyl O′-(2-amino)ethyl methylphosphonate (GD-MEA adduct). The samples were prepared by dilution of the reaction masses with deionized water before analysis by CE/indirect UV detection or CE/conductivity detection. Migration time precision was less than 4.0% RSD for IMPA and 5.0% RSD for PMPA on a day-to-day basis. The detection limit for both IMPA and PMPA is 100 µg/L; the quantitation limit for both is 500 µg/L. For calibration standards, IMPA and PMPA gave a linear response (R2 ) 0.9999) over the range 0.5-100 µg/mL. The interday precision RSDs were 1.9, 1.0, and 0.7% for IMPA at 7.5, 37.5 and 75.0 µg/mL, respectively. Corresponding values for PMPA (again, RSD) were 2.9, 1.1, and 1.0% at 7.5, 37.5 and 87.5 µg/mL, respectively, as before. Analysis accuracy was assessed by spiking actual neutralization samples with IMPA or PMPA. For IMPA, the seven spike levels used ranged from 20 to 220% of the IMPA background level, and the incremental change in the found IMPA level ranged from 86 to 99% of the true spiking increment (R2 ) 0.9987 for the linear regression). For PMPA, the five spike levels ranged from 10 to 150% of the matrix background level, and similarly, the accuracy obtained

(1) Conference on Disarmament Special Committee on Chemical Weapons. CD/ CW/WP. 367 Geneva, October 7, 1991. (2) Davis, G. T.; Demek, M. M.; Sowa, J. R.; Epstein, J. J. Am. Chem. Soc. 1971, 93, 4093-4103. (3) Epstein, J.; Bauer, V. E.; Saxe, M.; Demek, M. M. J. Am. Chem. Soc. 1956, 78, 4068-4071. (4) Epstein, J.; Cannon, P. L.; Sowa, J. R. J. Am. Chem. Soc. 1970, 92, 73907393. (5) Courtney, R. C.; Gustafson, R. L.; Westerback, S. J.; Hyytiainen, H.; Chaberek, S. C.; Martell, A. E. J. Am. Chem. Soc. 1957, 79, 3030-3036. (6) Larsson, L. Acta Chem. Scand. 1957, 11, 1131-1142. (7) Hackley, B. E.; Plapinger, Jr. R.; Stolberg, M.; Wagner-Jauregg, T. J. Am. Chem. Soc. 1955, 77, 3651-3653. (8) Wagner-Jauregg, T.; Hackley, B. E.; Lies, T. A.; Owens, O. O.; Proper, R. J. Am. Chem. Soc. 1955, 77, 922-929. (9) Epstein, J.; Michel, H. O.; Rosenblatt, D. H.; Plapinger, R. E.; Stephani, R. A.; Cook, E. J. Am. Chem. Soc. 1964, 86, 4959-4963. (10) Epstein, J.; Plapinger, R. E.; Michel, H. O.; Cable, J. R.; Stephani, R. A.; Hester, R. J.; Billington, C.; List, G. R. J. Am. Chem. Soc. 1964, 86, 30753084. (11) Gustafson, R. L.; Martell, A. E. J. Am. Chem. Soc. 1962, 84, 2309-2316. (12) Swidler, R.; Plapinger, R. E.; Steinberg, G. M. J. Am. Chem. Soc. 1959, 81, 3271-3274. (13) Epstein, J.; Rosenblatt, D. H.; Demek, M. M. J. Am. Chem. Soc. 1956, 78, 341-343. (14) Hudson, R. F.; Keay, L. J. Chem. Soc. 1956, 477, 2463-2469. (15) Greenhalgh, R.; Heggie, R. M.; Weinberger, M. A. Can. J. Chem. 1970, 48, 1351-1357.

3598 Analytical Chemistry, Vol. 70, No. 17, September 1, 1998

S0003-2700(97)01387-5 CCC: $15.00

Since their introduction, and throughout their ongoing development, the study of chemical warfare agents and their degradation products has been an important area of chemical research and ongoing efforts to develop effective preventive measures, antidotes, and treatments for use against such agents. The recent implementation of the Chemical Warfare Convention (CWC) Treaty and the subsequent demand for verifiable compliance measurements has added to the need for reliable and rapid methods for the determination of chemical warfare agents and their degradation products. Previously reported studies examined the neutralization of chemical nerve agents.1-18 For example, Scheme 1 shows the

© 1998 American Chemical Society Published on Web 07/30/1998

Scheme 1. Hydrolysis Pathway for GB in H2O

Scheme 2. Solvolysis Pathway for GB in MEA

major hydrolysis pathway for the nerve agent GB (isopropyl methylphosphonofluoridate, Sarin) in aqueous monoethanolamine (MEA/H2O) as a displacement of the fluoro group on phosphorus by a hydroxide ion to form isopropyl methylphosphonic acid (IMPA).3,4 Scheme 2 shows the reaction of GB with MEA to give the GB-MEA adduct.15,16 The nerve agent GD (Soman) behaves analogously to give pinacolyl methylphosphonic acid (PMPA) and GD-MEA adduct. To study this MEA/water nerve agent neutralization system, a sensitive, rapid, accurate, and high-resolution method for these neutralization products is needed. Several papers have been published on the analysis of chemical warfare agent degradation products using high-performance liquid chromatography (HPLC) with fluorescence detection, and gas chromatography/mass spectrometry (GC/MS).19-29 The limitations of the methods used in these studies include the following: insufficient sensitivity, interference with naturally occurring components in the matrixes, required derivatization, or long analysis times. Microcolumn liquid chromatography and capillary (16) Vasil’ev, I. A.; Shvyryaev, B. V.; Liberman, B. M.; Sheluchenko, V. V.; Petrunin, V. A.; Gorskii, V. G. Mendeleev Chem. J. 1996, 39 (4), 3-10. (17) Dostrovsky, L.; Halmann, M. J. Chem. Soc. 1953, 101, 502-509. (18) Kingery, A. F.; Allen, H. E. Toxicol. Environ. Chem. 1995, 47, 155-184. (19) Purdon, J. G.; Pagotto, J. P.; Miller, R. K. J. Chromatogr. 1989, 475, 261272. (20) Harvey, D. J.; Horning, M. G. J. Chromatogr. 1973, 79, 65-74. (21) Smith, R.; Schager, J. J. J. High Resolut. Chromatogr. 1996, 19, 151-154. (22) Verweij, A.; Boter, H. L. Pestic. Sci. 1976, 7, 355-362. (23) Bossle, P. C.; Martin, J. J.; Sarver, E. W.; Sommer, H. Z. J. Chromatogr. 1983, 267, 209-212. (24) Shih, M. L.; Smith, J. R.; McMonagle, J. D.; Dolzine, T. W.; Gresham, V. C. Biol. Mass Spectrom. 1991, 20, 717-723. (25) Tornes, J. Aa.; Johnsen. B. A. J. Chromatogr. 1989, 467, 129-138. (26) Daughton, C. G.; Cook A. M.; Alexander, M. Anal. Chem. 1979, 51, 19491953. (27) Ruppel, M. L.; Suba, L. A.; Marvel. J. T. Biomed. Mass Spectrom. 1976, 3, 28-31. (28) Wills, E. R. J.; Hulst, A. G. J. Chromatogr. 1988, 454, 261-272. (29) Roach, M. C.; Ungar, L. W.; Zare, R. N.; Reimer, L. M.; Pompliano D. L.; Frost, J. W. Anal. Chem. 1987, 59, 1056-1059.

electrophoresis (CE) with flame photometric detection for the determination of a series of organophosphoric and organophosphonic acids in environmental samples have been reported.30,31 Recently, it has been reported that ion chromatography (IC) can be used for the analysis of the IMPA, PMPA, ethyl methylphosphonic acid (EMPA), and MPA.32-35 Through the use of cleanup steps (Ag-form cartridges) to reduce chloride and ethylenediaminetetraacetic acid (EDTA) to remove transition metals, and a preconcentration anion-exchange injection setup, IC detection limits in the lower or submicrogram per liter range and detectability in field samples (groundwater/soil leachate) in the singledigit to low double-digit microgram per liter range have been reported.32 However, this approach involves a rather demanding procedure and appears vulnerable to sample matrix issues when pushed to these limits. Additionally, the separation requires gradient elution, which adds substantially to the analysis time, makes retention times less reproducible, produces more waste solvent, and has a higher requirement in instrumentation sophistication. Others used CE as an alternative approach, as it offers short analysis times, little sample preparation other than dilution and filtration, and flexibility in formulating electrolytes to minimize matrix interference.36-38 Also, because the required daily solvent/ buffer usage is measured in milliliters as compared to liters with HPLC and IC, it becomes feasible to dramatically improve portability of instrumentation, as well as to allow analysis of much smaller samples. We recently reported the development of methods for alkylphosphonic acids and their monoesters by the reversal of electroosmotic flow in capillary electrophoresis.39,40 The cationic surfactant didodecyldimethylammonium hydroxide (DDAOH) was used for this procedure. Electrolytes were stable for at least three months. Effective separation was achieved for eight closely related alkylphosphonic acid analytes in less than 3 min. Freedom from interference due to common anions was obtained through the use of acidic pH electrolytes which eliminated interferences such as fluoride, carbonate anion, and humic acids found in soil sample (30) (a) Kientz, C. E.; Verweij, A.; de Jong, G. J.; Brinkman, U. A. T. J. Microcolumn Sep. 1992, 4, 465-475. (b) Kientz, C. E.; Verweij, A.; de Jong, G. J.; Brinkman, U. A. T. J. Microcolumn Sep. 1992, 4, 477-483. (31) Kientz, C. E.; Hooijschuur, E. W. J.; Brinkman, U. A. T. J. Microcolumn Sep. 1997, 9, 253-259. (32) Kingery, A. F.; Allen, H. E. Anal. Chem. 1994, 66, 155-159. (33) Oehrle, S. A.; Bossle, P. C. J. Chromatogr., A 1995, 692, 247-252. (34) Bossle, P. C.; Reutter, D. J.; Sarver, E. W. J. Chromatogr. 1987, 407, 400404. (35) Schiff, L. J., Pleva, S. G., Sarver, E. W., Mulick, J. D., Sawicki, E., Eds. Ion Chromatographic Analysis of Environmental Pollutants; Ann Arbor Science: Ann Arbor, MI, 1979; Vol. 2, pp 329-344. (36) Mercier, J.-P.; Morin, P.; Dreux, M.; Tambute, A. J. Chromatogr., A 1996, 741, 279-285. (37) (a) Robin, W. H.; Wright, B. W. J. Chromatogr., A 1994, 680, 667-673. (b) Cheicante, R. L.; Stuff, J. R.; Durst, D. H. J. Capillary Electrophor. 1995, 2, 157-163. (c) Cheicante, R. L.; Stuff, J. R.; Durst, D. H. J. Chromatogr., A 1995, 711, 347-352.(d) de Griend C. E. S.-v.; Kientz, C. E.; Brinkman, U. A. T. J. Chromatogr., A 1994, 673, 299-302. (38) Pianetti, G. A.; Taverna, M.; Baillet, Aa.; Mahuzier, G.; Baylocq-Ferrier, D. J. Chromatogr. 1993, 630, 371-377. (39) Nassar, A.-E. F.; Lucas, S. V.; Jones, W. R.; Hoffland, L. D. Anal. Chem. 1998, 70, 1085-1091. (40) (a) Nassar, A.-E. F.; Emery, A.; Hoffland, L. D. Proc. Am. Chem. Soc., Div. Environ. Chem. 1997, 37 (1), 39-42. (b) Hoffland, L. D.; Calloway, R.; Emery, A.; Nassar, A.-E. F. Proc. Am. Chem. Soc., Div. Environ. Chem. 1997, 37 (2), 41-43.

Analytical Chemistry, Vol. 70, No. 17, September 1, 1998

3599

leachates. The first interference, fluoride, is reactive with the fused-silica capillary at acidic pH. The second eliminated interference is carbonate/bicarbonate, both of which have pKa values well above the pH used. An additional benefit is that this acidic electrolyte also has enhanced shelf life because it does not absorb atmospheric CO2 (which drops the pH) as readily as alkaline pH electrolytes. This paper reports the validation of our CE/indirect UV method to quantitatively determine these alkylphosphonic acids (IMPA, PMPA), which are the major chemical neutralization products from the nerve agents GB and GD in the MEA/water destruction system. The criteria applied to the validation of our method consisted of demonstrating or establishing method specificity, limit of detection, limit of quantitation, linearity, precision, accuracy, robustness, and ruggedness. The samples were prepared by simple dilution of the reaction masses with deionized water before analysis with CE/indirect UV or conductivity detectors. The present work shows that the methods performed well in terms of sample solution stability, linear response over a wide range of concentrations, high precision and accuracy, and excellent recovery percentages. We also report herein the development of a new CE/ conductivity method for analysis of the secondary neutralization products, GB-MEA and GD-MEA adducts, which are cationic compounds in an acidic CE buffer system. EXPERIMENTAL SECTION Apparatus. All experiments were performed on either a Thermo CE, the conductivity system (now Thermo BioAnalysis Corp., Santa Fe, NM) Crystal 310 capillary electrophoresis system with a Crystal 1000 conductivity detector, using a 60 cm × 50 µm (i.d.) ConCap fused-silica capillary (Franklin, MA) or a HewlettPackard capillary electrophoresis system with UV detector (indirect UV system) using a 75 µm (i.d.) × 56 cm (effective length) fused-silica capillary. Both instruments have the capability to control the temperature of the fused-silica capillary. The electropherogram signals were recorded using 4880 software for the Thermo BioAnalysis system and HP Chem-Station software for HP System. Reagents. Alkylphosphonic acids and their monoesters were synthesized in our laboratory (U.S. Army Material Command Treaty Laboratory, Aberdeen Proving Ground, MD). Isopropyl methylphosphonic acid (IMPA) 98% pure, O-isopropyl O′-(2amino)ethyl methylphosphonate (GB-MEA adduct) 98% pure, and O-pinacolyl O′-(2-amino)ethyl methylphosphonate (GD-MEA adduct) 98% pure were custom synthesized by Radian International (Austin, TX). Commercial sources were as follows: pinacolyl methylphosphonic acid (PMPA) (pinacolyl ) 1,2,2-trimethylpropyl), Aldrich Chemical Co. (Milwaukee, WI); didodecyldimethylammonium bromide (DDAB) (>99%), Eastman Kodak (Rochester, NY); isooctylphenoxypolyethoxyethanol (Triton X-100), Union Carbide; phenylphosphonic acid and boric acid, Sigma Chemical Co. (St. Louis, MO); and acetic acid, J. T. Baker, Inc. (Phillipsburg, NJ). All reagent solutions and diluted samples were prepared using deionized water (Winokur Water System Corp., Barnstead, Dubuque, IA). Solutions were stored in Nalgene (polypropylene) plastic bottles. All buffers were degassed and filtered through a 0.45-µm cellulose nitrate membrane filter prior to use. All other chemicals were reagent grade. 3600 Analytical Chemistry, Vol. 70, No. 17, September 1, 1998

The didodecyldimethylammonium hydroxide (DDAOH) was prepared from the commercially available bromide salt by replacing the bromide with hydroxide using Dionex On-Guard A cartridges, (Dionex Corp., Sunnyvale, CA).41 The anion-exchange cartridges are supplied in the carbonate form and are converted to the hydroxide form by passing 10 mL of 1.0 M NaOH through the cartridge followed by 10 mL of deionized water. The cartridge is then used to convert a 10-mL aliquot of 25 mM DDAB to DDAOH. This conversion step eliminated a large interfering peak in the electropherogram due to bromide and was performed as needed. Procedures. New capillaries were pretreated with a 10-min rinse of deionized water, followed by 10 min of 0.5 M sodium hydroxide, then 10 min of deionized water, and finally a 10-min rinse of the analysis buffer. Rinses were performed at 2000 (conductivity system) or 900 mbar (indirect UV system). Samples and standards were introduced by keeping the (pressure × time) product at 300 mbar‚s for both detection methods. Constant voltage (negative polarity, detector side anodic) was used throughout to drive the separations, -25 and -30 kV for the conductivity system and indirect UV system, respectively. The wavelength used for indirect UV detection was 210 nm. The capillary temperature was ambient for the conductivity system and 40 °C (optimized) for indirect UV system. The electrolytes were 400 mM acetic acid (pH ∼2.5, nonadjusted) for conductivity detection (conductivity system) and 200 mM boric acid/10 mM phenylphosphonic acid/0.03 wt % Triton X-100/0.35 mM DDAOH for indirect UV detection (indirect UV system). Sodium hydroxide at 0.2 N was used to adjust the indirect UV electrolyte to pH 4.0. To obtain reproducible injections, the same injection sequence was followed for each sample: the capillary was rinsed with deionized water for 3 min and then with run buffer for 5 min prior to each injection. For the HP system, the capillary inlet and outlet levels must be equal to ensure that there is no siphoning in either direction. Standard Solutions. Stock solutions of IMPA and PMPA were prepared by dissolving 0.010 g of IMPA or PMPA in 10 mL of deionized water in a 10-mL volumetric flask (1.0 mg/mL). Calibration standards were prepared by serial dilution from these stock solutions. Sample Preparation. MEA/H2O reaction masses were diluted for CE analysis as follows: 0.2 g of the reaction mass into 10.0 mL of deionized water; further diluted with deionized water at 25 µL of this solution into 10.0 mL (overall dilution factor is 1.0 g of reaction mass to 20.0 L of water). Method Validation. Method validation in this work consists of demonstrating or establishing method specificity, limit of detection, limit of quantitation, linearity, precision, accuracy, robustness, and ruggedness.42 Most of the method validation experiments were carried out using reaction masses. The approach used to address each method validation parameter is briefly described as follows: (a) Specificity. The absence of interference from anions was expected in high levels with the G-agent/MEA/water system (fluoride and methylphosphonate). (41) Landers, J. P., Ed. In Handbook of Capillary Electrophoresis, 2nd ed.; CRC Press: Boca Raton, FL, 1996; Chapter 6. (42) For examples, see: (a) Dalal, P. S.; Reid, R.; Audibert, F.; Bhagat, H. R. Anal. Lett. 1993, 26 (4), 657-664. (b) Altria, K. D.; Walsh, A. R.; Smith, N. W. J. Chromatogr. 1993, 645, 193-196. (c) Pande, P. G.; Nellore, R. V.; Bhagat, H. Anal. Biochem. 1992, 204, 103-106.

(b) Limit of Detection (LOD) and Limit of Quantitation (LOQ). LOD is defined here as the analyte level giving a signalto-noise ratio of 3:1. LOQ is defined as 5-fold higher than the LOD. (c) Linearity. Calibration curves for both IMPA and PMPA at concentrations over the range 0.5-100 µg/mL were generated. (d) Precision. Three levels of calibration standards were repeated 5 times a day for 5 days to assess interassay precision. The five replicates on each day were further used as separate data sets for intraassay precision. (e) Accuracy. IMPA and PMPA were spiked into their respective (i.e., GB or GD) reaction masses: seven levels for IMPA ranging from 0.05 to 0.50 g of IMPA/g of reaction mass and five levels for PMPA ranging from 0.075 to 0.625 g of PMPA/g of reaction mass. Six spiking replicates per level were used in all cases. Comparison of the incremental IMPA/PMPA levels found (above the background level) with the true spiking level was used to assess method accuracy over the spiking range. Reagent blanks (MEA/water) and solvent blanks (buffer) were included on each day as control samples to ensure accuracy in qualitative and quantitative results. (f) Robustness. The effect of small variations on method performance of parameters, such as pH, temperature, and voltage was measured. CERTIFY III43 software was used to perform the necessary calculations on the analytical results obtained. This software uses weighted least-squares regression to calculate confidence bands. The function used for the weighting is the reciprocal of the variations, causing those values with lower variation to have more weight. Least squares is performed on the weighted data, and a regression equation is obtained. Confidence about the regression line is determined using the weighted parameters. The regression line is used to provide the reported (found) concentration from the raw data. RESULTS AND DISCUSSION Method Development for IMPA and PMPA. Recently,39,40 we reported that the alkylphosphonic acids and their monoesters can be separated within 3 min by reversing the electroosmotic flow. This approach uses acidic electrolytes and thereby eliminated the two most potentially troublesome interferences. The first potential interference, fluoride, is not observed in the separation because it reacts with the inner wall of the capillary at acidic pH.44 The second potential interference is carbonate/ bicarbonate which, because of its weak acidity (pKa1 ) 6.3), is also not observed at the acidic pH used. The shelf life of the buffers for both detection methods is three months, and this level of storage stability would be a great benefit in many other CE applications. Specificity (Selectivity). Selectivity is a measure of the degree of interference in the analysis of complex sample mixtures. Specifically, in this work, it is freedom from interferences present in the reaction masses which have high levels of fluoride and potentially high levels of MPA (as a degradation product of the monoesters). Carbonate/bicarbonate, usually present in aqueous (43) Program Manager for Chemical Demoralization Environmental Monitoring Office, Aberdeen Proving Ground, MD, July 1996. (44) Thermo BioAnalysis Corp., Franklin, MA 02038, unpublished results.

Figure 1. Analysis of (A) IMPA in the reaction mass sample from the neutralization of GB and (B) PMPA in the reaction mass sample from the neutralization of GD. Experimental conditions: buffer, 200 mM borate, 10 mM phenylphosphonic acid with cationic surfactant 0.35 mM DDAB and 0.03 wt % Triton X-100, pH 4.0; 75 µm (i.d.) × 56 cm (effective length) fused-silica capillary; peaks detected by indirect UV detection at 210 nm; injection, 10 s at 50 mbar; temperature, 40 °C; voltage, 30 kV (negative polarity).

systems, is also of concern. As noted above, fluoride and carbonate/bicarbonate are eliminated as potential interferences through the use of acidic CE buffers. Figure 1 shows example electropherograms for IMPA (A) and PMPA (B). In Figure 1, a blank reaction mass (no nerve agent neutralization), a reaction mass from neutralization of the respective nerve agent (GB or GD), a reference standard, and the reaction mass fortified with additional IMPA or PMPA are shown. Despite minimal sample preparation (simple dilution in water), this figure clearly indicates the advantage of the developed method for targeted analysis of IMPA and PMPA, as there was no interference in the electropherogram from other anions. Reaction mass samples were also spiked with methylphosphonic acid, 10 µg/mL, eluting in this system at ∼2.30 min. MPA is the terminal hydrolysis product of both IMPA and PMPA, and MPA caused no interference in the IMPA or PMPA analysis. Limit of Detection and Limit of Quantitation. The LOD is defined here as the analyte level that gives a signal-to-noise level ratio 3:1. The LOD for both IMPA and PMPA is 0.1 µg/mL. The LOQ is defined as 5-fold above the LOD to approximate the lowest concentration that can be determined with acceptable precision and accuracy. Thus, the LOQ is 0.5 µg/mL for IMPA and PMPA. These sub-part-per-million LOD/LOQ results suggest that the method might be useful for the environmental detection of species which can indicate the past use or manufacture of G- or V-type nerve agents. However, this type of application would possibly be complicated by the presence of a wider variety of potentially interfering anions which could be present in environmental samples. Linearity. Linearity was determined by analyzing standards at 11 different concentrations over the range, 0.5-100 µg/mL. Analytical Chemistry, Vol. 70, No. 17, September 1, 1998

3601

Table 2. Within- and between-Day Precision for PMPA at Three Levels % RSDa (n ) 5) for the given level, µg/mL day

7.5

37.5

87.5

1 2 3 4 5 interdayb

2.1 1.2 4.0 1.0 3.3 2.9

1.3 1.0 0.8 0.95 0.4 1.1

0.3 0.7 0.8 0.6 0.3 1.0

a RSD calculated on the basis of the peak area. b All 25 runs for each level.

Table 3. Analysis Results for Spiking IMPA in the Reaction Mass

Figure 2. Representative calibration curves for (A) IMPA and (B) PMPA. Table 1. Within- and between-Day Precision for IMPA at Three Levels % RSDa (n ) 5) for the given level, µg/mL day

7.5

37.5

75.0

1 2 3 4 5 interdayb

0.4 0.4 0.5 1.3 1.9 1.9

1.0 0.4 0.1 1.1 1.1 1.0

0.3 0.4 1.0 0.6 0.4 0.7

a RSD calculated on the basis of the peak area. b All 25 runs for each level.

Figure 2 shows calibration curves for IMPA and PMPA. The correlation coefficient for linear best fit was 0.9999 for both IMPA and PMPA. Precision. Method precision was assessed using data from five analysis replicates of three reference standards on each of five days. The three levels of these standards were low, middle, and upper values of the calibration range. The low and middle concentrations were 7.5 and 37.5 µg/mL for both IMPA and PMPA; the high range for IMPA was 75.0 µg/mL and for PMPA was 87.5 µg/mL. The intraassay (intraday) precision was determined at each level within each day by calculating the percent relative standard deviation (RSD) of the found concentrations. The interassay (interday) precision was determined at each level by calculating the RSD of the found concentrations from all five days combined (RSD from all 25 analyses). These precision results are given in Tables 1 and 2 for IMPA and PMPA, respectively. For IMPA, the intraassay precision ranged from 0.3 to 2.1% RSD over the three levels while the interassay precision ranged from 0.7 to 1.9% RSD (Table 1). The corresponding values for PMPA were 0.3-4.0% RSD intraassay precision and 1.0-2.9% RSD interassay precision (Table 2). Clearly, there is no significant difference between the intra-/interassay precision for both IMPA and PMPA at least over the limited, five-day period investigated 3602 Analytical Chemistry, Vol. 70, No. 17, September 1, 1998

targeta (g)

foundb (g) mean, n ) 6

RSD (%)

recovery (%)

0.010 0.020 0.025 0.040 0.050 0.070 0.100

0.0086 0.0172 0.0249 0.0379 0.0492 0.0643 0.0970

9.9 4.9 4.3 4.7 2.7 3.6 5.8

86.2 85.9 99.4 94.8 98.4 91.9 97.0

a Amount spiked into 0.20 g of reaction mass prior to sample workup and analysis. b After subtracting the background level of IMPA.

here. Migration time intraassay precision was less than 4.0% RSD for IMPA and 5.0% RSD for PMPA. Accuracy. Accuracy was assessed by spiking actual GB and GD reaction masses with known amounts of the respective destruction product (IMPA and PMPA). In each case, 0.2 g of reaction mass was first spiked with IMPA or PMPA and then prepared for analysis by serial dilution. For IMPA, the seven fortification levels ranged from 0.01 to 0.10 g per 0.20 g of reaction mass. These fortification levels correspond to 20-220% of the IMPA background level in the reaction mass used. For PMPA, the five fortification levels ranged from 0.015 to 0.125 g per 0.20 g of reaction mass, which corresponded to 10-150% of the PMPA background level. The data from these fortified samples were quantified against the full range calibration curve and the known background IMPA or PMPA level was subtracted to generate found values resulting from the fortification. Accuracies of these found values are tabulated as percent recoveries in Table 3 for IMPA and in Table 4 for PMPA. These data are graphically shown in Figure 3 where the net found amount is plotted against fortification amount. The Figure 3 plots for IMPA and PMPA are clearly linear and have an excellent 0,0 intercept. The linear correlation coefficients are 0.9987 for IMPA and 0.9999 for PMPA. The recovery percentages ranged from 86 to 99% for IMPA and 95 to 97% for PMPA. Robustness. Robustness in this work measured the degree to which the analysis results remain unaffected by small variations in method parameters such as pH, temperature, and voltage. At pH 4.0 ( 0.1, temperature 40.0 ( 0.5, and voltage -29.5 to -30.0, the migration times and peak response for IMPA and PMPA did not change, indicating that the method is robust within these limits.

Table 4. Analysis Results for Spiking PMPA in the Reaction Mass targeta (g)

foundb (g) mean, n ) 6

RSD (%)

recovery (%)

0.015 0.025 0.050 0.075 0.125

0.0148 0.0242 0.0480 0.0712 0.1193

6.8 3.7 3.6 3.7 2.7

95.5 96.7 96.0 94.9 95.4

a Amount spiked into 0.20 g of reaction mass prior to sample workup and analysis. b After subtracting the background level of PMPA.

Figure 4. Effect of temperature on the migration time of IMPA. Experimental conditions as Figure 1.

Figure 3. Linear regression of spiked vs found concentration for (A) IMPA and (B) PMPA when added to reaction masses.

Effect of Temperature and Voltage. The effect of temperature and voltage on the separation was assessed using a wider range of analytes: MPA, IMPA, PMPA, ethyl methylphosphonic acid, isobutyl methylphosphonic acid, cyclohexyl methylphosphonic acid, O-isopropyl ethylphosphonic acid, and O-2-ethylhexyl methylphosphonic acid. At 20, 25, 30, and 40 °C, migrations times decreased with increasing temperature. Elevation of the temperature reduces the viscosity of the electrolyte permitting faster separations; it also permits a larger sample plug for the same injection pressure and time. This explains why the areas did not decrease but rather remained constant as the migration times decreased (Figure 4). Over the range of 15-30 kV (negative polarity), peak area and migration time decreased with increasing voltage as the injectable mass remained constant. Thus, for this mixture of alkylphosphonic acids and monoesters, resolution improved with either a reduction in temperature or an increase in voltage. Ruggedness. Ruggedness is defined as a measure of reproducibility of test results from analyst to analyst. In this work, the samples have been prepared by several analysts, but the CE analysis was done by one analyst. Thus, our finding that analysis results were highly reproducible implies that the sample preparation step is rugged. We also checked the possibility of sample carry-over from a previous injection using the following analysis

sequence: reagent water blank, blank reaction mass spiked at 100 µg/mL IMPA or PMPA, reaction mass, and nonspiked blank reaction mass. No peak was observed in the final nonspiked blank reaction mass, indicating that there was no carry-over from the previous run. CE Buffer Stability. Previous work39 and this work show that the electrolytes used here are fully stable for at least three months, which is of great benefit in many other capillary electrophoresis applications. CE Analysis of GB-MEA and GD-MEA Adducts. A method using CE/conductivity detection for GB-MEA and GD-MEA adducts in these reaction masses was also developed in this work, but validation data similar to that for IMPA and PMPA are not yet available.45 These adducts are significant products in the reaction masses for GB and GD destruction in the MEA/water system. Their amounts range from about 5 to 90% of the starting GB/GD, depending on the exact chemical and physical conditions of the destruction trial. At pH ∼3, the GB-MEA and GD-MEA adducts are ionized due to protonation of the amino group. A 400 mM acetic acid electrolyte was found to produce sharper peaks for GB-MEA and GD-MEA adducts over lower ionic strength electrolytes as it improves sample stacking. With the high concentration acetic acid there was an initial concern with Joule heating so an Ohm’s law plot was obtained (voltage vs current) to find the optimum run voltage. It was found that the running current was linear from 5 to 25 kV for the 400 mM acetic acid electrolyte. The low running current of 7.5 uA for an applied voltage of 25 kV is due to the fact that only a small fraction of the acetic acid is dissociated at this pH. Standard addition was used to confirm the identity of the peak at ∼6.0 min in the GB reaction mass as the GB-MEA adduct. Figure 5 shows these confirmation data with electropherograms for the following: blank reaction mass (no nerve agent neutralization), GB reaction mass, reference standard, and GB reaction mass fortified with additional GB-MEA standard. Similar data were generated to confirm the identity of the GB-MEA adduct in GD reaction masses. Analysis of standards at low levels give a detection limit for GB-MEA adduct and GDMEA adduct of ∼0.075 µg/mL. No interfering cations were observed in the electropherograms of the reaction masses. Thus, this CE analysis system offers excellent sensitivity, freedom from interference, rapid analysis, and quantitative capability for these species. (45) Nassar, A.-E. F.; Lucas, S. V.; Jones, W. R.; Hoffland, L. D., in preparation.

Analytical Chemistry, Vol. 70, No. 17, September 1, 1998

3603

Figure 5. Analysis of GB-MEA adduct in the reaction mass sample from the neutralization of GB. Conductivity detection conditions: sample hydrostatically introduced into the capillary for 12 s at 25 mbar and a constant potential 25 kV (positive polarity). The electrolyte was 400 mM acetic acid (pH ∼2.5, nonadjusted), and the temperature was 25 °C.

CONCLUSIONS The current study shows that CE can be used as an alternative to IC in the analysis of anionic and cationic chemical warfare agent degradation products in reaction masses. The methods require little sample preparation other than dilution with deionized water. Analysis time is rapid with the analytes of interest obtained within 3 min (anions) or 6 min (cations) from time of injection. For anion

3604 Analytical Chemistry, Vol. 70, No. 17, September 1, 1998

analysis (IMPA and PMPA) the method has been thoroughly validated. Analyte detection limits are in the nanogram per milliliter range using pressure injections, and separations are robust with migration time variations less than 5.0% RSD on a day-to-day basis. The methods are versatile, permitting determination of G- or V-type nerve agent degradation products in any aqueous solution at nanogram per milliliter levels. In addition, it is shown that the interday and intraday reproducibility for the anions were in the range of about 1-3% RSD with accuracy reflected by spike recoveries in actual reaction masses of not less than 86%. Furthermore, the electrolytes used for this work are stable for at least three months. Future work will involve intra-/ interassays for the GB-MEA and GD-MEA adducts, lowering detection limits through investigation of electrokinetic means of sample preconcentration and extending the range of analytes to include biological warfare agents. ACKNOWLEDGMENT We thank the U.S. Army Material Command Treaty Laboratory, Aberdeen Proving Ground, MD, for their support.

Received for review December 30, 1997. Accepted June 25, 1998. AC9713870