2386
Langmuir 2005, 21, 2386-2390
Degradation Kinetics of VX on Concrete by Secondary Ion Mass Spectrometry John M. Williams* Battelle Salt Lake City Operations, West Valley City, Utah 84120
Brad Rowland and Mark T. Jeffery West Desert Test Center, Dugway Proving Ground, Dugway, Utah 84022
Gary S. Groenewold, Anthony D. Appelhans, Garold L. Gresham, and John E. Olson Idaho National Engineering and Environmental Laboratory, Idaho Falls, Idaho 83415-2208 Received August 18, 2004. In Final Form: December 24, 2004 At trace coverages on concrete surfaces, the nerve agent VX (O-ethyl S-2-diisopropylaminoethyl methyl phosphonothiolate) degrades by cleavage of the P-S and S-C bonds, as revealed by periodic secondary ion mass spectrometry (SIMS). The observed kinetics were (pseudo-) first-order, with a half-life of 2-3 h at room temperature. The rate increased with surface pH and temperature, with an apparent second-order constant of kOH ) 0.64 M-1 min-1 at 25 °C and an activation energy of 50-60 kJ mol-1. These values are consistent with a degradation mechanism of alkaline hydrolysis within the adventitious water film on the concrete surface. Degradation of bulk VX on concrete would proceed more slowly.
Introduction VX (O-ethyl S-2-diisopropylaminoethyl methyl phosphonothiolate, CAS registry number 50782-69-9) is a highly toxic nerve agent that is part of the United States chemical warfare agent munition stockpile that is presently undergoing destruction.1 At the present time, the U.S. is in the process of destroying VX-containing weapons in compliance with the Chemical Weapons Convention. However, the compound remains of interest because of possible use by rogue nations or terrorist groups. In fact, members of the Japanese terrorist group Aum Shinrikyo used VX to murder a disaffected cult member who was trying to escape.2 VX is particularly insidious because it is highly adsorptive: dermal contact with contaminated surfaces can result in toxic effects. For example, in an early study, VX attenuation was attributed to strong chemisorption to soil.3 Attempts to extract the compound using supercritical CO2 and to analyze it using thin-layer chromatography were confounded by its strong adsorptive tendency.4,5 In addition to strong surface adsorption, VX will undergo degradation from either environmental exposure6 or intentional decontamination efforts.7-9 The rate of VX degradation is of particular interest and has been determined for alkaline, (1) Carnes, S. A.; Watson, A. P. JAMA, J. Am. Med. Assoc. 1989, 262, 653. (2) Zurer, P. Chem. Eng. News 1998, 76, 7. (3) Verweij, A.; Boter, H. L. Pestic. Sci. 1976, 7, 355. (4) Kuitunen, M.-L.; Hartonen, K.; Riekkola, M.-L. J. Microcolumn Sep. 1991, 3, 505. (5) Witkiewicz, Z.; Mazurek, M.; Szulc, J. J. Chromatogr. 1990, 503, 293. (6) Kingery, A. F.; Allen, H. E. Toxicol. Environ. Chem. 1995, 47, 155. (7) Yang, Y.-C.; Szafraniec, L. L.; Beaudry, W. T.; Samuel, J. B.; Rohrbaugh, D. K. In Proc. of the 1994 ERDEC Sci. Conf. on Chem. and Bio. Defense Res.; Berg, D. A., Ed.; ERDEC-SP-036; 1996; p 375. (8) Yang, Y.-C.; Szafraniec, L. L.; Beaudry, W. T.; Rohrbaugh, D. K. J. Am. Chem. Soc. 1990, 112, 6621. (9) Yang, Y.-C.; Baker, J. A.; Ward, J. R. Chem. Rev. 1992, 92, 1729.
neutral, and acidic solutions in water. However, degradation pathways and rates occurring on surfaces have been less widely studied. It is known that surface degradation rates can be substantially different from solution degradation rates.3,6 In situ investigation of VX hydrolysis on concrete and mineral surfaces has been undertaken by Wagner and co-workers, who have used solid-state NMR to observe reactions on concrete chips, and MgO, CaO, and zeolite nanoparticles.10-14 However, there have been few direct investigations of hydrolysis kinetics on commonly encountered surfaces. The surfaces of common samples such as soil particles or concrete are highly heterogeneous, which makes characterization difficult. The occurrence of multiple degradation pathways also complicates investigation of VX behavior. It has been known for some time that it can undergo P-S, P-O, and S-C bond cleavages, leading to multiple hydrolysis products.6-9,15 Some of these are quite toxic, and others undergo secondary reactions that form yet other toxic compounds. Like VX, the resulting compounds may be highly adsorptive. Concrete is an especially important substrate for study, because of its wide use in construction of civilian and government installations. Determination of reaction pathways and degradation kinetics on this substrate will improve estimates of VX persistence after an exposure event. We have attacked this problem by applying static secondary ion mass spectrometry (SIMS)16 to the detection (10) Wagner, G. W.; Bartram, P. W. Langmuir 1999, 15, 8113. (11) Wagner, G. W.; Bartram, P. W.; Koper, O.; Klabunde, K. J. J. Phys. Chem. B 1999, 103, 3225. (12) Wagner, G. W.; Koper, O. B.; Lucas, E.; Decker, S.; Klabunde, K. J. J. Phys. Chem. B 2000, 104, 5118. (13) Wagner, G. W.; O’Connor, R. J.; Procell, L. R. Langmuir 2001, 17, 4336. (14) Wagner, G. W.; O’Connor, R. J.; Edwards, J. L.; Brevett, C. A. S. Langmuir 2004, 20, 7146. (15) Yang, Y.-C.; Berg, F. J.; Szafraniec, L. L.; Beaudry, W. T.; Bunton, C. A.; Kumar, A. J. Chem. Soc., Perkin Trans. 1997, 2, 607.
10.1021/la047933j CCC: $30.25 © 2005 American Chemical Society Published on Web 02/16/2005
Degradation Kinetics of VX on Concrete by SIMS
of VX on surfaces: the technique enjoys submonolayer sensitivity and yields compound-specific information on adsorbed VX and its degradation products. In previous research, identification of VX on soil17 and concrete particles18 was demonstrated using an ion trap (IT-) SIMS instrument. Further experiments showed that VX degrades to a variety of detectable compounds when in contact with concrete. The studies indicated that degradation pathways and kinetics could be discerned by temporally monitoring ion abundances for VX on concrete. Results on a single type of concrete have been reported recently;19 the present study expands this work to include concretes of different types and ages. These experiments have resulted in the direct measurement of VX degradation kinetics on concrete, and the calculation of a mean activation energy and kinetic model for the degradation process which are in reasonable agreement with kinetic parameters for hydrolysis in alkaline solutions. Experimental Methods Sample Preparation. Caution. VX is a highly lethal compound capable of killing or injuring at extremely minute doses. The compound must only be handled in approved chemical warfare surety laboratories by trained agent chemists. All sample preparation and analyses were performed at the chemistry laboratory of the U.S. Army West Desert Test Center (WDTC), Dugway Proving Ground, Dugway, UT; this laboratory is equipped with appropriate engineering and administrative controls for handling chemical warfare agents. Concrete samples were collected from the runway of the Salt Lake City International Airport (SLC) and from the Czech Republic. The Czech concrete was poured less than 2 years before for a related study, whereas the SLC concrete had been poured around 40 years before. The samples had not previously been exposed to VX. Before exposure, the samples were crushed with a hammer and then ground with a mortar and pestle. The surface areas of the concrete samples were measured to be 4.0 m2/g using N2 adsorption (BET method) for the SLC samples and 3.5 m2/g for the Czech samples.20 In addition, a SLC concrete sample was crushed and stored for 15 months to provide an aged crushed concrete sample. The VX used in the study was an analytical reference standard (U.S. Army, purity 94%), used as a 1 µg/µL standard solution in 2-propanol. Enough VX solution was applied onto crushed concrete samples to fully wet the sample (about 1 µL solution/mg concrete). This resulted in a deposition of about 1 µg VX/mg concrete ) 0.25 mg VX/m2 concrete, which corresponds to about 0.5 monolayer (assuming a molecular area of 95 Å2 for VX).21 Some samples were also spiked with 0.5 monolayer of tetrabutylammonium bromide (TBN, Aldrich), an inert organic salt, as an internal standard for VX quantitation.22 This method was superseded by the normalization method described below. Once the samples were spiked, the solvent was evaporated under a gentle stream of dry air or nitrogen at room temperature, (16) Groenewold, G. S.; Gianotto, A. K.; Ingram, J. C.; Appelhans, A. D. Curr. Top. Anal. Chem. 1998, 1, 73. (17) Groenewold, G. S.; Appelhans, A. D.; Gresham, G. L.; Olson, J. E.; Jeffery, M.; Wright, J. B. Anal. Chem. 1999, 71, 2318. (18) Groenewold, G. S.; Appelhans, A. D.; Gresham, G. L.; Olson, J. E.; Jeffery, M.; Weibel, M. J. Am. Soc. Mass Spectrom. 2000, 11, 69. (19) Groenewold, G. S.; Williams, J. M.; Appelhans, A. D.; Gresham, G. L.; Olson, J. E.; Jeffery, M. T.; Rowland, B. Environ. Sci. Technol. 2002, 36, 4790. (20) Adamson, A. W. Physical Chemistry of Surfaces; Wiley: New York, 1990. (21) The mass of VX corresponding to monolayer coverage was estimated by assuming that the effective molecular area of VX is 95 Å2, which is the area of a circle having a radius equal to half the length of the molecule on the surface (estimated at 5.0 Å). This radius estimate assumed that the molecule lies flat on the surface and was generated using molecular mechanics calculations (Cerius2, Molecular Simulations, Inc., San Diego, CA). If the molecule orients on the surface in an upright fashion or is coiled, then the coverages will be somewhat less than the values used here. (22) Gillen, G.; Hues, S. M. J. Am. Soc. Mass Spectrom. 1993, 4, 419.
Langmuir, Vol. 21, No. 6, 2005 2387 which required about 5 min. As soon as the samples were dry, approximately 1 mg of sample particles was affixed to a sample holder (the head of a #18 nail) with double-sided tape, and the remainder of the sample was contained and placed into a thermostated bath. The sample holder was then attached to the IT-SIMS probe for analysis as the initial data point (time ) 0). Samples from the same exposure batch were analyzed in a time series until the majority of the VX had degraded. Analyses were performed at ambient temperature (24-26 °C) and at various temperatures over the range of 30-55 °C When the analyses were complete, samples were immersed in bleach solution to neutralize any residual VX. This decontaminated waste was then treated and disposed of appropriately as hazardous waste. IT-SIMS. An IT-SIMS instrument located at the WDTC was used in the current study. The instrument has been described previously: briefly, it is based on a Teledyne Discovery 2 IT mass spectrometer (Mountain View, CA) that was modified for SIMS by the addition of a ReO4- primary ion gun, an offset dynode/ multichannel plate detector system, and an insertion lock for introducing the sample using the direct insertion probe.16-19,23 The vacuum housing containing the IT-SIMS, as well as the vacuum pump exhaust lines, was located in a chemical warfare agent surety hood where all agent operations were performed. In a typical MS1 experiment, the ion trap was operated at base radio frequency amplitude corresponding to a low mass cutoff of 40 amu. The ionization time was typically 250 ms, after which the mass spectrum was recorded. A single “scan” consisted of 20 summed spectra. A normal analysis would consist of about 15 scans, or 300 summed spectra. The primary ion dose for a typical analysis can be calculated at about 1.5 × 1012 ions/cm2, by using the ionization time and scan acquisition information above, the area irradiated (8 × 10-3 cm2), and the primary ion current (30 pA). At this dose, the surface of the sample is not considered to have been seriously perturbed;24 observed secondary ion yields remained relatively constant over the duration of the analysis.
Results and Discussion The most abundant ions derived from VX and its degradation products have been identified previously.17-19 Intact VX yields abundant ions at m/z 268 [VX + H]+, 139, 128, and 86. Its major degradation product diisopropylaminoethanethiol (DESH) yields abundant ions at m/z 160 [DESH - H]+, 128, 118, 114, and 86. Phosphonate degradation products were not monitored, such as ethyl methylphosphonic acid (EMPA) and S-2-diisopropylaminoethyl methylphosphonothioic acid (EA2192). EMPA is produced stoichiometrically along with DESH; EA2192 is a minor degradation product at neutral to alkaline pH typical of the concrete surface (yields of 10-22%).8,9 Although EMPA and EA2192 can be detected by SIMS on inert surfaces, they were difficult to detect on concrete, perhaps because of anion exchange and incorporation of these phosphonates into the ionic mineral matrix of the concrete. Future efforts may be made to monitor these degradation products by optimizing the IT for anionic secondary ion analysis. In a typical SIMS spectrum of 0.5 monolayer VX collected early in the degradation process, the intact molecule is observed in the protonated form [VX + H]+ at m/z 268 (Figure 1A). More abundant ions derived from VX were observed at m/z 128 and 86. The spectrum also contains lower abundance background ions and ions derived from the internal standard (TBN). The relative abundances of the VX-related ions were observed to change over time: the abundance of [VX + H]+ (m/z 268) decreased, as did m/z 139 and 128, while significant ions at m/z 160, 118, and 114 appeared and (23) Dahl, D. A.; Appelhans, A. D. Int. J. Mass Spectrom. 1998, 178, 187. (24) Briggs, D.; Hearn, M. J. Vacuum 1986, 36, 1005.
2388
Langmuir, Vol. 21, No. 6, 2005
Williams et al.
Figure 2. Typical semilogarithmic plot of Q over time for submonolayer VX on concrete, with identical linear trends for VX ions (m/z 268, 139) and degradation product ions (m/z 114, 118, 160). Note that m/z 86 deviates strongly from the linear trend for these data. Outliers (open symbols) were rejected before kinetic analysis.
Figure 1. Cation IT-SIMS spectra of VX. (A) 0.5 monolayer VX on concrete, 21 min after spiking. Ions derived from the internal standard are denoted by “is”. (B) VX on concrete after 163 min, showing a decrease in the abundance of [VX + H]+ and an increase in the abundance of ions derived from DESH. (C) Formation routes resulting in observation of m/z 128.
became more abundant (Figure 1B). The latter ions are derived from the P-S cleavage product DESH. However, the abundance of m/z 128 relative to the other DESH-derived ions (160, 118, and 114) in the spectra of concrete exposed to VX for long time periods was too great to be solely attributed to DESH. This is consistent with the existence of a second degradation process that operates by cleavage of the S-C bond (Figure 1C), forming a diisopropylaziridinium (DIAZ) product (as in alkaline solution).25 C-S cleavage has not been previously reported on metal oxide or zeolite substrates,10-12 although P-S cleavage was observed in those degradation studies. No standards for DIAZ were available, but by assuming the DIAZ and DESH sensitivities to be similar, we roughly estimate the degradation product distribution to be 30% DIAZ, 70% DESH. To monitor the ion abundances quantitatively, the ion intensities In(t) were normalized by the total intensity for all seven ions. This allows the degradation kinetics to be determined from changes in the fractional intensity Fn(t) at time t, for each ion.19
∑n In(t)
Fn(t) ) In(t)/
(1)
To account for growth of product ions as well as decay of VX ions, the absolute change in fractional intensities Qn(t) were calculated by subtracting Fn(t) values from the final values Fn(∞) (the limit as time f ∞) and then taking the absolute value:
The rate of VX degradation can be measured by monitoring ion abundances as a function of time. Plots of ln Q against time were linear for all ions monitored, indicating a first-order or pseudo-first-order reaction at the surface (m/z 86 sometimes showed nonlinear behavior). Figure 2 depicts a typical semilogarithmic plot of Q over time for all monitored ions. The slope of each plot is equal to -k, the rate constant for the first-order kinetic equation:
dQn(t)/dt ) -kQn(t)
(3)
The rate constants (k values) for the seven ions monitored were not significantly different within the error on the slope, for each degradation condition. Therefore, pooled estimates for k were obtained by taking the weighted mean of all seven k values (each weighted by its variance). Figure 3 shows the values of k for each condition tested (also listed in Table S1, Supporting Information). For all concrete types, an Arrhenius plot of ln k versus 1/RT resulted in a linear relationship with an activation energy of 50-60 kJ mol-1 (Figure 3). The activation energy agrees with values calculated by Yang and co-workers for hydrolysis of VX in alkaline solutions, where P-S cleavage producing DESH was calculated at 53-61 kJ mol-1, with an overall (composite) activation energy of 60-63 kJ mol-1.7 The alkaline nature of the concrete was verified by immersing the concrete in deionized H2O and measuring the pH, yielding pH values of 11.7 for the Czech concrete, 11.8 for fresh crushed SLC concrete, and 11.2-11.3 for old crushed SLC concrete (Table S2, Supporting Information). These values confirm a substantial population of basic surface sites and the presence of an alkaline film of adventitious water under normal humid air. To further explore this relationship, a model was fitted to the kinetic data of the form used by Yang for alkaline hydrolysis:7
(2)
k ) k0 + kOH[OH-] ) (A0 + AOH[OH-]) exp(-Ea/RT) (4)
Thus, Qn(t) is a quantity that decays toward 0 for all ions n.
(25) Yang, Y.-C.; Szafraniec, L. L.; Beaudry, W. T. Abstr. Pap. Am. Chem. Soc. 1995, 209, 7.
Qn(t) ) |Fn(t) - Fn(∞)|
Degradation Kinetics of VX on Concrete by SIMS
Langmuir, Vol. 21, No. 6, 2005 2389
Figure 3. Arrhenius plots for the degradation of submonolayer VX on each concrete type, with fits to the best fit model (solid) and the alkaline hydrolysis model (dashed). Error bars ((2σ) are shown where the error exceeds the size of the data marker. Table 1. Kinetic Model Parameters for VX Degradation on Concrete
parameter
best-fit results
alkaline hydrolysis (Yang)
A0 AOH Ea k0 ) A0 exp(-Ea/RT) at 25 °C kOH ) AOH exp(-Ea/RT) at 25 °C
9.6 × 106 min-1 2.6 × 109 M-1 min-1 55 ( 5 kJ mol-1 0.0023 min-1
62 kJ/mol 0.000 19 min-1
0.64 M-1 min-1
0.31 M-1 min-1
Table 1 shows the best-fit values for the parameters k0, kOH, A0, AOH, and Ea; the Ea values compare well to Yang’s values for alkaline hydrolysis, although the rate constants seem to differ by an offset in k0. This offset may correspond to a surface effect specific to concrete or to a bias introduced by our method of surface pH measurement. Figure 3 compares the data to the best fit model and to the offset Yang model; it is clear that both these models provide good fits to the data. These results indicate that the degradation of submonolayer VX on concrete is dominated by the mechanism of alkaline hydrolysis, probably within the adventitious film of water on the concrete surface and within its pores. These data and the alkaline hydrolysis model allow VX degradation rates to be estimated over the pH range typical of concretes. Concretes at several locations of military interest have a pH around 11-12 on a freshly exposed surface (Table S2, Supporting Information); the pH on an aged or weathered surface will be significantly lower. Equation 4 allows the rate constant to be estimated for two limiting cases: fresh concrete (pH 12) and extremely weathered concrete (neutral pH). These limits are calculated based on the best-fit parameters
Figure 4. Arrhenius plot for the degradation of submonolayer VX on concrete within extrapolated limits. For comparison, the observed rate constants on glass lie well below the rate determination limit, also shown. Concrete data symbols: fresh SLC (+), Czech (O), and old SLC ([).
above and are plotted in Figure 4 together with the data. The upper limit is expected to be a reliable estimate of the maximum rate of VX degradation on concrete. Although the data for our concrete samples fall within these limits, the lower limit is only hypothetical because it is a farther extrapolation and because there are effects that can lead to longer VX persistence on concrete: for example, passivation or contamination of the concrete with an inert layer and slower attenuation rates for higher coverages (bulk VX) on concrete, as reported by Wagner et al.13,14
2390
Langmuir, Vol. 21, No. 6, 2005
The effect of the basic concrete substrates in the degradation of VX was verified by comparing it to behavior upon neutral glass samples. Acid-washed 250-µm diameter glass beads (Supelco) were spiked with VX solution, allowed to dry, and then analyzed in the same manner. On glass the Q values were practically constant over time; the rate constants k were not significantly different from 0 (the k values were less than their associated uncertainties). These results are included in Table S1 (Supporting Information) and Figure 4. In addition, there was a complete absence of DESH in the spectra on glass even at long times, and the m/z 128 ion abundance did not increase relative to m/z 268 [VX + H]+. Thus, on the glass beads, VX was either not degrading or degrading at a rate that was orders of magnitude slower than that which occurred on concrete. This observation agrees with the known stability of VX stored in glass. These results agree closely with observations by Wagner et al. of the degradation of VX on crushed concrete by 31P MAS NMR spectroscopy, where they observed the apparent first-order degradation of VX with a half-life of 2.2 h at room temperature, up to a reactive capacity of about one monolayer.13 Subsequent VX attenuation proceeded with a half-life of 28 days to 3 months. The degradation of VX on concrete can also be compared to its behavior on MgO and CaO. Wagner and co-workers spiked neat VX onto high-surface-area MgO and CaO and measured a bimodal attenuation behavior using 31P NMR.11,12 They interpreted their results in terms of very rapid P-S cleavage on the metal oxide surface, which was over within 2 h, followed by a much slower disappearance process. This latter process was also reported to be due to surface P-S cleavage but was limited by the rate of VX evaporation from passivated sites to active basic sites. In both NMR experiments, the bimodal kinetics were interpreted in terms of the heterogeneous reaction system (surficial vs bulk) of liquid VX with the solid substrate. The SIMS results reported here have isolated the faster surficial process on concrete and provided strong evidence that this reaction is dominated by the mechanisms of aqueous alkaline hydrolysis. Conclusions Submonolayer VX rapidly degrades on the surface of concrete in the absence of any visible solvent. The reaction products, the reaction rates, and the temperature dependence bear a strong similarity to degradation reactions occurring in aqueous alkaline solution. In fact, the degradation of submonolayer VX on concrete seems to be
Williams et al.
dominated by alkaline hydrolysis within the adventitious water film on the concrete surface and in its pores. At an ambient temperature of 25 °C, VX could be expected to degrade to 1% of its initial concentration within 9-33 h and to 1 part in 106 within 26-100 h, depending on the alkalinity of the concrete. The rate constants are greater at higher temperatures, with the rate doubling about every 10 °C. It should be emphasized that the samples used here were crushed concrete, which may be substantially more alkaline than intact weathered concrete as encountered in the environment. Further assessment of the reactivity of weathered concrete surfaces represents an important direction for additional studies. This paper emphasizes that VX on concrete poses a potential long-term threat.26-28 The extent, severity, and duration of this threat and the appropriate actions for decontamination and remediation are questions which require accurate information on the persistence and fate of VX on intact surfaces. By reducing the degradation kinetics to a simple model dependent upon two parameters, the temperature and surface pH, the results presented here may assist in answering such questions. These results indicate that trace amounts of VX have only limited persistence on common concrete surfaces, up to about 5 days at 25 °C, owing to alkaline hydrolytic action. These results confirm that trace-level VX contamination (below about 0.5 mg/m2) will not pose a persistent hazard on such surfaces. Acknowledgment. We gratefully acknowledge support and guidance from Mr. Bill Davis, West Desert Test Center (WDTC), Dugway, UT, and the support of the Joint Science and Technology Panel for Chemical and Biological Defense Supporting Science and Technology Business Area, Edgewood Chemical Biological Center (ECBC), Edgewood, MD. Supporting Information Available: Degradation rate constants k for submonolayer VX on crushed concrete (Table S1) and comparison of pH for several concrete samples (Table S2). This material is available free of charge via the Internet at http://www.pubs.acs.org. LA047933J (26) Karalliedde, L.; Wheeler, H.; Maclehose, R.; Murray, V. Public Health 2000, 114, 238. (27) Reutter, S. A.; Olajos, E. J.; Mioduszewski, R. J.; Watson, A. Edgewood Research, Development & Engineering Center Report; ERDEC-SP-017; Aberdeen Proving Ground, MD, 1994. (28) Jenkins, R. A.; Buchanan, M. V.; Merriweather, R.; Ilgner, R. H.; Gayle, T. M.; Watson, A. P. J. Hazard. Mater. 1994, 37, 303.
