Laboratory Setup for Long-Term Monitoring of the Volatilization of

D. Waysbort*, E. Manisterski, H. Leader, B. Manisterski, and Y. Ashani. Israel Institute for Biological Research, P.O. Box 19, Ness Ziona, Israel. Env...
0 downloads 0 Views 137KB Size
Environ. Sci. Technol. 2004, 38, 2217-2223

Laboratory Setup for Long-Term Monitoring of the Volatilization of Hazardous Materials: Preliminary Tests of O-Ethyl S-2-(N,N-Diisopropylamino)ethyl Methylphosphonothiolate on Asphalt D. WAYSBORT,* E. MANISTERSKI, H. LEADER, B. MANISTERSKI, AND Y. ASHANI Israel Institute for Biological Research, P.O. Box 19, Ness Ziona, Israel

Contrary to commonly used pesticides, the rate of volatilization of extremely toxic chemicals such as the nerve agent O-ethyl S-2-(N,N-diisopropylamino)ethyl methylphosphonothiolate (VX) cannot be readily obtained under environmental conditions due to its high mammalian toxicity that would require extraordinary precautions. An alternative is a laboratory setup that would be used to obtain environmentally relevant data required for risk assessment studies. In this paper we describe a newly designed climatic hood that enables control of temperature, humidity, and air velocity within less than (0.5% fluctuations during continuous operation. The performance of the evaporation system together with the sampling and analytical procedures produced a meaningful concentration profile of vapors obtained from a 15 mg sample of VX dispersed as small droplets over a 10 × 16 cm piece of asphalt road. The released vapors amounted to approximately 30% of the applied mass, and its time course was best fitted to a triexponential curve with rate constants changing over time from 2.2 to 0.03 h-1. The asphalt enhanced a specific degradation pathway of VX that is relatively minor in aqueous solutions. Results provide the first data on the volatilization of VX from samples of asphalt road, and offer an insight into VX behavior in the environment.

Introduction Organophosphorus (OP) cholinesterase (ChE) inhibitors are commonly used as pesticides and are manufactured worldwide and stockpiled in large quantities. Since they constitute considerable environmental risk, numerous reports have been published on their terrestrial fate and rate of volatilization from a variety of soil matrixes (1). In contrast, information on the rate of volatilization from urban surfaces of some of the most toxic OPs that have been produced and stockpiled for nonpeaceful purposes is quite scarce. Events of recent years (e.g., Iraq-Iran war in the 1980s and chemical terrorist attacks in Japan in 1994-5) have increased awareness of chemical terrorism and threats of possible use of nerve agents against a civilian population. Thus, it is important to quantify * Corresponding author phone: 972-8-9381554; fax: 972-89381490; e-mail: [email protected]. 10.1021/es030382j CCC: $27.50 Published on Web 02/24/2004

 2004 American Chemical Society

and evaluate parameters that will establish databases for use in risk assessment of exposure of the general population to highly toxic OPs. For the very toxic (2) and relatively persistent (3) nerve agent O-ethyl S-2-(N,N-diisopropylamino)ethyl methylphosphonothiolate (VX), one can speculate that low-level longterm exposure will produce variable degrees of inhalation hazards in unprotected population (4). The degradation of VX on environmental surfaces is reasonably documented (3, 5, 6); however, data on its ability to vaporize as a function of droplet size, temperature, humidity, air velocity, and type of surface are scarce. Sound and scientific knowledge of the time course in air of OPs such as VX is necessary for the evaluation and prediction of long-term consequences following contamination of urban surfaces. Yet, the high mammalian toxicity of certain chemicals such as VX requires extraordinary safety precautions that practically preclude experiments in the free environment. Thus, an alternative approach would be a laboratory experiment; in this paper we describe an experimental system that was designed and built for the determination of the volatilization of a few milligrams of tested material from small surfaces (10 × 16 cm). The stability of the system in terms of temperature, humidity, and air velocity together with the sampling and analysis procedures enabled measurement of the time dependence of vapor over at least 14 days. Results of the measurements with VX suggest that the evaporation chamber can provide reliable estimates of vapor concentrations of compounds that volatilize from various surfaces, under controlled weather conditions. Data on the time course of VX volatilization from asphalt road samples are reported here for the first time.

Materials and Methods Principles and Structure of the Experimental Setup. In view of the relatively low vapor pressure of many OPs, including VX, and their strong adsorption to commonly used laboratory tubing, it was important to select suitable components that would minimize interactions between the vapors and the line path that leads from the tested surface to the sampling device. Further, refined collection techniques are required to guarantee high efficiency of vapor accumulation over 10 h of sampling and reasonable stability of the collected solute. To minimize further the loss of vapors at the low concentration range, the tubings and valves were heated by external jackets of circulating warm water. The evaporation unit that holds the tested surface (Figure 1, chamber B) was placed in a climatic glovebox (chamber A) that was designed to serve as a hood that would comply with rigorous safety regulations. The two chambers were made of stainless steel with the following dimensions: chamber A, 75 × 74 × 52 cm; chamber B, 16 × 10 × 6 cm. The 6 cm height of chamber B was designed to accommodate platforms to support tested surfaces with variable heights. The pathway of the thermostated humidified air of chamber A into chamber B (units 2-6) was built in such a manner that air flows from chamber A to chamber B through a flexible PVC line (i.d. 1.6 cm) via a rectangular stainless steel inlet (1 × 10 cm) into a stainless steel tunnel of 16 cm length. Thus, the air is forced to flow evenly above the contaminated surface. Chamber B is kept under reduced atmospheric pressure relative to chamber A (∆P ) 8 mbar). Heat exchangers were embedded in the walls of chamber A. Chamber B was mounted on an aluminum block that was VOL. 38, NO. 7, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

2217

FIGURE 1. A schematic diagram of airflow circuits in climatic chamber A, evaporation unit B, and samplers C: 1, inlet of external wet/dry air; 2, air entrance to filters; 3, two commercial NBC filter canisters containing about 100 g of impregnated activated charcoal; 4, flow meter; 5, thermocouples (5a, 5b, 5c); 6, hygrometer; 7, contaminated surface; 8, heated lines; 9, pneumatic valves; 10, glass manifold; 11, glass samplers and thermostatic holder of the samplers; 12, critical orifice; 13, easy access external samplers. Each dotted line is comprised of a chemical filter (3), a pneumatic valve (9), and a critical orifice (12). placed on the floor of chamber A so that a heat exchanger underneath could maintain the desired temperature of the tested surface. The required air temperature and humidity of chamber B are provided by chamber A. The latter is equipped with a wall fan to maintain homogeneous climatic conditions in the entire space of chamber A. The desired humidity was obtained by mixing wet and dry air (Figure 1, unit 1), and the temperature was controlled by thermostatic baths (Lauda RK8). Airflow from chamber A through chamber B was driven by use of a filter-protected central vacuum pump. Air was guided into chamber B through two commercial nuclearbiological-chemical (NBC) filter canisters containing about 100 g of impregnated activated charcoal (unit 3) that eliminated any possible contamination from vapors or aerosol. The line that leads the filtered air into chamber B is equipped with a mass flow meter (unit 4, Bronkhorst HiTec, Ruurlo, Holland), type K thermocouples (unit 5, ElconMamab, Israel), and a hygrometer (unit 6, Hy-Cal, engineering humidity transmitter, California). Air velocity is controlled by use of a critical orifice. The air that flows above the contaminated surface (unit 7) passes through heated pneumatic valves (units 8 and 9) and is sampled through short heated stainless steel and glass lines (units 8 and 10) into glass samplers (units 11). These samplers are organized in chamber A in two sets of parallel manifold lines. Each line contains six samplers that can hold 15-100 mL of collecting fluid, which is described below. The exact volume was determined in accordance with the air temperature and the rate of airflow and in most cases was 30 mL/sampler. Sampler lines are equipped with a 5 L/min critical orifice (units 12) for two purposes: (a) to collect a 2218

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 7, 2004

sufficient amount of vapors that will allow a reasonable detection limit and (b) to minimize foaming. The samplers were dipped in a specially fitted thermostatic bath cooled to approximately 1 °C, which permitted maintenance of the internal solution at 5-7 °C. This minimized loss of solute by evaporation, and reduced hydrolysis of the accumulated OP during extended collection time. To prevent adsorption of vapors on the line walls, the air pathway from the evaporation unit B to the samplers C is heated by circulating hot water (approximately 80 °C) in external jackets. In addition, two samplers were situated in thermostatic baths outside chamber A, in a space that is part of the regular hood (unit 13). To be able to work at temperatures below ambient, an additional hood door was built to meet the specific requirements. Thus, the door’s wall was made of triple-layer Perspex that created two flat compartments with two holes equipped with hand gloves. In the inner compartment (facing the hood), cooling solvent is circulated at the temperature of the air desired inside chamber A. Water at sufficiently low temperature to prevent ice accumulation on the surface of the door was circulated in the external compartment. Thus, a clear view of chamber A was possible even when the inside temperature was as low as -5 °C. Experimental parameters were monitored throughout the run and recorded continuously by a computerized system. The signal outputs from the sensors (temperature, humidity, and flow rate) were processed by a Fluke Hydra Series A2620 model (John Fluke Mfg. Co.) and stored in the computer for later retrieval. Stability of the Climatic Hood. Changes of less than (0.2 °C in temperature and 95% pure, a conclusion that was confirmed by titration of a predetermined concentration of human butyrylcholinesterase solution (see below). Analysis of VX in Samplers. The concentration of VX in the samplers was determined by enzymatic assay, following a series of intermediate dilutions in Millipore-quality water. VX in a final concentration range from 0.05 to 0.3 ng/mL was incubated in 1 mL of 1.5 nM purified human butyrylcholinesterase (15) in 10 mM Hepes buffer, pH 7.8, containing 0.05% bovine serum albumin, 1 mM EDTA, and 0.01% sodium azide. Residual enzyme activity was determined after 20 h at 25 °C by the spectrophotometric method of Ellman et al. (16) using butyrylthiocholine iodide as substrate and monitoring the increase of OD at 412 nm. A standard linear titration curve was constructed from the known concentration of fresh solutions of VX by plotting residual activity vs VX concentration. The error was estimated at less than 25%. Extraction and Analysis of VX from Asphalt. The recovery of VX residues from asphalt was carried out by immersing the whole tested block (from the dynamic experiments; see below), or smaller pieces (obtained from static experiments; see below), in methanol and shaking for 1 h. The methanol was removed by decantation, and the same procedure was repeated with fresh methanol and shaking for 2 h. After the second decantation, a third portion of methanol was added, and the asphalt was shaken overnight. The three methanol extracts were combined and evaporated under reduced pressure at 45 °C. The thick oily content in the evaporation flask was dissolved in 20 mL of CHCl3, dried over anhydrous Na2SO4, and evaporated. The remaining residue was treated 3× with 25 mL of methanol, and the combined fractions were evaporated under reduced pressure at 45 °C. Finally, the flask content was dissolved in 0.5 mL of CDCl3. The

TABLE 1. Differential and Total Amount of VX Collected in the Air from VX on an Inert Stainless Steel Platea collectn cycle

time,b min

amt of VX,c mg

cumulative amt of VX,d mg

1 2 3 4 5 6 7 8 9 10 11

60 120 180 243 131e 288 431 954 1421 1880 2405

2.857 2.986 2.270 2.064 0.0206 0.0046 0.00097 0.00046 0.00034 0.00023 0.00011

2.857f 5.843 8.113 10.177 10.198 10.202 10.203 10.204 10.204 10.204 10.204

a VX ) 15 mg; 40 °C and 0.1-0.2% relative humidity; air velocity 0.7 0 m/s. b The numbers indicate time elapsed from t ) 0 to the beginning of the collection cycle. c Collected from the air per specified cycle. d Cumulative amount of VX, including the specified cycle. e The plate was removed. Cycle 5 and on represent collection of desorbed VX vapors from the system’s walls. The numbers are time elapsed from removal of the plate and the beginning of the sampling cycle. f Numbers rounded to third decimal point.

concentration of VX was determined by 31P NMR spectroscopy following addition of a known amount of O,O,O-triethyl phosphate (TEP). Compared to extraction in benzene, methanol provided extracts free of bituminous components that interfere with 31P NMR spectroscopy at 121.65 MHz (General Electric, GN 300 WB). The NMR protocol was validated by showing that more than 95% of standard VX solutions containing 0.45-8.1 mg/sample could be accounted for by use of a known amount of TEP and comparing areas under the 31P NMR signals. To further validate the extraction protocol, methanol was replaced by benzene that caused complete disintegration of the asphalt, dissolving the main organic components. Following removal of the insoluble particles (e.g., gravel, sand) by filtration, the concentration of VX in the benzene extracts was assayed by 31P NMR as described above.

Results Volatilization of VX from an Inert Surface. Initial control tests in a VX-free evaporation chamber B did not detect any anti-cholinesterase compound in the samplers that could interfere in the interpretation of results from subsequent experiments. To validate and ascertain the various quantitative aspects of the newly designed system, the first volatilization experiments with VX were carried out from a chemically inert polished stainless steel plate. The observation that more than 95% of the applied VX was recovered (see below) is consistent with the assumption that the system will enable an almost complete mass balance of the vapor phase in future experiments. Data from a representative experiment are summarized in Table 1. A 15 mg sample of VX was deposited in droplets on a round-shaped stainless steel plate (9.6 cm diameter) and evaporated at 40 °C (0.1-0.2% relative humidity) using an air velocity of 0.7 m/s. After about 250 min, the plate was removed and washed several times with ethanol until no VX was detected. The amount of residual VX (2.83 mg) was determined by enzymatic assay as described. The efficiency of the samplers was 98% for sampling periods not exceeding 1 h, as judged from the VX vapors that escaped into a second sampler connected in tandem. When the sampling duration lasted 8-10 h, the efficiency decreased to 90%. The stability of VX in the sampler solution was such that after 8 h the concentration found by the enzymatic method required correction due to a 20-30% loss in VX content. This was based on the decrease of known concentrations of VX in samplers that were bubbled with VX-free VOL. 38, NO. 7, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

2219

FIGURE 2. Time course of volatilization of 15 mg of VX from an inert surface at 40 °C, 0.1-0.2% relative humidity, and 0.7 m/s air velocity. The ordinate depicts concentration in units of micrograms or nanograms of VX per liter of air (note the differences on the ordinate scale of the graphs). (A) Point-to-point curve depicting the evaporation in the first 4 h. The arrow indicates removal of the contaminated plate. (B) Initial rate of VX desorption from the system’s walls. The abscissa denotes time elapsed after removal of the contaminated plate. Data points were fitted to a monoexponential decay curve with t1/2 ) 62 min. (C) Terminal phase of VX desorption. The abscissa denotes time elapsed after removal of the plate. Data points were fitted to a monoexponential decay curve with t1/2 ) 614 min.

TABLE 2. Cleansing of the Evaporation Lines from Adsorbed VXa sourceb

rel humidity, %

time to VX cumulative source duration,d concn,e amt of VX removalc h pg/L desorbed,f µg

inertg

0.1-0.2 245 min

inertg

258 min 11 days 11 days

2 asphalt 2 asphalt 70

50 130 75 55 50

5 0.6 2 3 10

27.3 34.6 0.195 0.270

a 40 °C at 0.7 m/s air velocity. b 15 mg of VX on each surface. c Time elapsed from the beginning of the experiment. d Time elapsed from the removal of the source. e Average value obtained by dividing the amount of VX in the sampler of the last collection by the volume of air passed through the sampler. f Total amount collected from the removal of the source to the end of the experiment. g Stainless steel.

air for 8 h under similar conditions. It should be pointed out, however, that this correction was required only toward the end of the experiment when the air concentration of VX decreased to below 10 pg/L. The total amount of VX vapors collected during the first four cycles of sampling amounted to 10.177 mg. The amount accumulated from cycle 5 to the end of the experiment was 0.0273 mg, which represents the VX desorbed from the system’s walls (0.27% of the volatilized VX). Thus, together with the residual amount found on the plate, a total of 13.0 mg was recovered. Given the estimate of an average 5% loss during sampling and the relative error of the assay procedures, this value is within reasonable agreement with the 15 mg of VX applied to the inert surface (>95% recovery after correction for 5% impurities in VX). The time dependence of volatilization of VX from the inert surface and the rate of its desorption off the walls after removal of the plate are illustrated in Figure 2. Calculation of the air concentration of VX was based on the amount of vapors collected, divided by the air volume passed through the samplers. The rate of VX desorption changed over time, suggesting the existence of at least two populations of adsorbed VX. Thus, 96% of the total desorbed VX vapors were released within the first 7 h. Yet, it took 40 h to reduce the average concentration of VX to less than 0.005 ng/L. Notably, the total time to decrease VX concentration to below 0.6 pg/L was 130 h (Table 2). It is possible that the nonhomogeneous desorption is the result of the presence of adsorbing components such as metal oxides. However, more experiments are required to address this question properly. 2220

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 7, 2004

FIGURE 3. Disappearance of VX with increasing time as measured by using methanol extraction (static experiment). 9 mg of VX was applied to each asphalt sample. Data points were fitted to a biexponential decay curve. When VX volatilization from an inert surface was monitored at 2% relative humidity (not shown), patterns essentially similar to those shown in Table 1 and Figure 2 were observed, with a total recovery of 17.05 mg of VX. This value is within reasonable agreement with the 15.8 mg of VX applied to the inert surface. A period of 75 h was needed to bring the VX concentration to below 0.002 ng/L after removal of the contaminated plate. The total amount of desorbed vapors was 0.0346 mg. Extraction of VX from Asphalt (Static Experiments). To obtain some idea of the fate of VX on asphalt, small samples (3.5 × 3.5 cm) that had been exposed to VX for several days in closed glass vials were extracted with methanol and analyzed for VX and its commonly known degradation products by use of 31P NMR spectroscopy. The time course of “disappearance” of VX from the small asphalt pieces is shown in Figure 3. When benzene was used to extract VX after 8 days, the amount of recovered material was similar to that obtained with methanol. Further, only a negligible amount of VX could be obtained by use of benzene following methanol extraction of a VX-contaminated asphalt sample. These observations validate the suitability of methanol for the extraction of intact and presumably unbound VX. The data points in Figure 3 reasonably fit a biexponential decay curve; it is suggested that the reduction of the amount of VX extracted by methanol from small asphalt samples is a combination of a time-dependent binding process of VX to asphalt components together with its slow decomposition. This is supported by the finding that, only 20 days after loading VX on asphalt (static experiments), methanol extraction clearly revealed the presence of a considerable amount of three known hydrolysis products of VX (3), namely, Oethyl methylphosphonothioic acid (1), S-(2-N,N-diisopropylaminoethyl) methylphosphonothioic acid (2), and O-ethyl methylphosphonic acid (3) (Figure 4). Thus, despite the fact that t1/2 of VX “disappearance” from asphalt was 4 days, it took about 5t1/2 to clearly identify the presence of the hydrolysis products. These uncertainties precluded the full mass balance of VX in the dynamic experiments (air flows above the contaminated asphalt). It should be noted that the loss of VX due to evaporation in the closed container is negligible (on the basis of the volatility of approximately 10 µg/L at room temperature). Dynamic Experiments. a. Volatilization of VX from Asphalt by Dry Air at 40 °C (Relative Humidity 2%, Air Velocity 0.7 m/s). Panels A and B of Figure 5 show the time course of VX volatilization and the ratio of cumulative vapors to the applied amount of VX (VXT/VX0), respectively. The data points

FIGURE 4. 31P NMR spectra of methanol extracts of VX-contaminated asphalt (9 mg/sample): (upper panel) immediately after application of VX; (lower panel) after 20 days at room temperature. The following degradation products were identified: 1, CH3P(O)(OC2H5)SH; 2, CH3P(O)(OH)SCH2CH2N(isoPr)2; 3, CH3P(O)(OC2H5)OH. The percent molar decomposition of VX appears to be 72%.

TABLE 3. Rate Constants of VX Volatilization from I

II

Asphalta III

experiment

k1, h-1

%b

k2, h-1

%b

k3, h-1

%b

dryc wetd

2.07 2.22

17 20

0.58 0.21

44 61

0.028 0.042

39 19

a Data were fitted to a triexponential association curve. b Percent of accumulated vapors of each fraction, at the specified k value. c 2% relative humidity, 40 °C. d 70% relative humidity, 40 °C.

of Figure 5B were fitted to biexponential (dotted line) and triexponential (solid line; r 2 ) 0.9982) association curves, and the corresponding rate constants are summarized in Table 3. As seen, the rate constants decreased over time, suggesting dynamic changes in the partitioning of the VX on asphalt. The experiment was terminated after 11 days when the VX concentration in air decreased to below 0.1 ng/L. A total of 4.4 mg of VX (29% of the original load) was collected from the air. An additional 0.85 mg (5.7%) of VX was extracted by methanol at the end of the experiment. Lack of quantitative data with respect to the amount of asphalt-induced hydrolysis products precluded mass balance in this (and the next) run. Yet, from the static experiments with small samples of asphalt it may be inferred that after 11 days 4-5 mg of VX was hydrolyzed. On the whole, it is estimated that, after correction for sampler efficiency and the predicted amount of hydrolyzed VX, a total of 10.2 mg (68%) of the original amount could be accounted for.

After removal of the asphalt block from chamber B the collection of VX vapors continued for approximately 58 h. A total of 195 ng of VX was desorbed from the interior walls with the following distribution: In the first 173 min, 30 ng was desorbed, in the next 220 min, 75 ng. The next 1069 min released 60 ng, and during the last 1993 min 30 ng was collected, a value that corresponds to an average of 0.003 ng/L VX in air. Notably, when the humidity increased to 50%, an additional 230 ng of VX was collected over 811 min, and 45 ng in the next 945 min. This indicates that water molecules in air can replace VX and facilitate desorption. This assumption implies competition between sorption of VX and that of water on polar surface sites. Such sorptive sites may include metal oxides and silanol groups of various parts of the system. b. Volatilization of VX from Asphalt by Wet Air at 40 °C (Relative Humidity 70%, Air Velocity 0.7 m/s). This experiment was also terminated after 11 days, when the concentration of VX was 0.033 ng/L. Panels C and D of Figure 5 show the time course of VX volatilization and the ratio of cumulative vapors (VXT) to the applied amount of VX (VX0). The data points of Figure 5D were fitted to biexponential (dotted line) and triexponential (solid line; r 2 ) 0.9999) association curves. The total amount of collected VX vapors was 5.0 mg, and an additional 0.13 mg was recovered from the asphalt block by methanol extraction at the end of the experiment. As explained above for the dry air experiment, the total amount of recovered VX was estimated at 10.2 mg (68%) of the applied material. The observation that the amount of VX recovered from the asphalt (by methanol extraction) at the end of the experiment was approximately 6-fold less than that obtained in the dry experiment indicates that water may expedite the hydrolysis of VX on the asphalt surface. After removal of the asphalt block, the next 60 h of vapor collection gave the following results: In the first 196 min, 35 ng of VX was collected; in the next 238 min, 35 ng. The following 1220 min gave 90 ng. During the last 1946 min 110 ng was collected from the air (corresponding to an average of 0.01 ng/L). A total of 270 ng of VX was desorbed.

Discussion Volatilization of certain OPs from urban surfaces cannot be readily obtained under environmental conditions because of safety restrictions. Thus, the primary goal of this work was to develop a reliable setup for measurements of rates of volatilization of hazardous compounds in the laboratory, and provide the first insight into VX behavior on asphalt. To this end, a key requirement for the evaporation system is the ability to maintain constant temperature, humidity, and air velocity over at least 2 weeks of continuous operation. Another objective was to enable at least 8 h of differential vapor collection cycles, under conditions that would guarantee the stability of VX in the sampled solution. Results show that these goals were satisfactorily achieved with respect to VX and the evaporation setup can amply be used to characterize its rate of volatilization from various surfaces. The climatic chamber, together with a suitable analytical procedure, allowed the quantitative and meaningful determination of as low as 1 pg/L VX vapors in air. The design of the system allows use of extrapolated data to environmental conditions. For example, the size and geometry of the dispensed VX kept the drops apart after spreading was completed. The height of the air tunnel above the sampled surface is 10 mm. Since the laminar boundary layer above ground is around 0.5 mm (17), it was assumed that the vapor pathway from the surface to the flowing air is similar to what one would expect to find 1 cm above the same surface under environmental conditions. To enable mass balance of the applied liquid, all the air that passed above the sample was collected, a requirement that dictated VOL. 38, NO. 7, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

2221

FIGURE 5. Volatilization of VX from asphalt road samples. Upper panels (40 °C at 2% relative humidity, air velocity 0.7 m/s): (A) time course of VX released into the air from the contaminated surface; (B) Fraction of VX released vs time. Data points were fitted to biexponential (dotted line) and triexponential (solid line, r 2 ) 0.9982) curves. Lower panels (40 °C at 70% relative humidity, air velocity 0.7 m/s): (C) time course of VX released into the air from the contaminated surface; (D) Fraction of VX released vs time. Data points were fitted to biexponential (dotted line) and triexponential (solid line, r 2 ) 0.9999) curves. airflow of 40 lpm (0.7 m/s). Calculations show that this velocity corresponds to a Reynolds number of approximately 470, suggesting laminar flow conditions. Yet, the setup features of the climatic chamber will allow in future experiments the increase of the air velocity to a value that represents turbulent flow. Calibration was based, among other tests, on experiments with stainless steel surfaces. A total of 91-114% of the applied VX could be accounted for, by use of the combined cholinesterase assay and 31P NMR spectroscopy. The capacity of the interior walls to adsorb and release VX vapors was estimated at a time point where VX on the inert plate was still available in sufficient quantities before the plate was removed to enable measurement of VX desorption from the walls of the system (Figure 2). An average of approximately 30 µg of VX vapors was desorbed (Table 2), less than 0.3% of the total volatilized VX collected from the original source in the first 4 h of the experiment. This small value is attributed to the design of the system, the heating of the pipelines, and the nature of the materials used to build the evaporation unit. However, it is important to note that low levels of VX vapors continued to desorb from the walls for 2-3 days at the detectable range. This observation was made possible due to the low dynamic range of the enzymatic assay. In general, the desorption was biphasic. The fast phase of vapor release (t1/2 ) 1-2 h) was observed for 96% of the adsorbed VX, followed by a slow terminal decay characterized by a half-life of about 10 h. Since the plots of Figure 2B,C represent desorption in the absence of the source, it appears that about 5% of the VX vapors that originated from the interior walls differ from the main desorbed bulk. It is possible that metal oxides or silanol groups of the system constitute an adsorbing layer that contributes to the nonhomogeneity of the desorption curve. Similar results were obtained for the 2% and 0.1-0.2% relative humidity experiments. The influence of water molecules on the desorption phenomenon is an open question that warrants further clarification; however, it is possible that they 2222

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 7, 2004

can compete with VX and enhance the desorption of the latter from the inner walls of the evaporation system. The reproducibility of the desorption profiles established a “cleansing” procedure of 4 days of continuous fresh air ventilation between two consecutive runs. This assured the decrease of the VX concentration in air to below 1 pg/L. Further, the observed desorption behavior could now be used as a baseline of the system for the interpretation of subsequent experiments. Results provide for the first time information on the rate of volatilization of VX from asphalt. A plot of the fraction of vaporized VX (VXT/VX0, Figure 5B,D) vs time is best described by a triexponential association curve (Table 3). The size and individual rate constants of the three exponents differed slightly between the two experiments; however, the portions that evaporated with the greatest rate in each run were approximately the same (17% and 20% of the total accumulated mass). In addition, regardless of the humidity, the ratio of the total amount of volatilized VX (VXT)∞) to the original quantity applied on the asphalt (VX0) was approximately the same (∼0.3). In general, these results are consistent with the hypothesis of spreading and diffusion of the organic liquid droplets from the surface into deeper layers of the asphalt. Visual observation of VX spots on asphalt samples suggests rapid spreading, and the spot size remains unchanged even after 11 days, although the moist appearance diminished gradually over time. This behavior is essentially similar to results of a recent photographic study that followed the fate of VX droplets on an asphalt surface for 3.5 h from deposition (18). Similar behavior of VX on concrete where rapid penetration of VX produced a sorbed phase was recently described (7). Thus, the findings that 70% humidity gave results similar to those recorded with 2% humidity indicate that a considerable amount of VX was rapidly shielded by the inner asphalt layers from direct contact with water in air. The fact that drops of VX in a closed glass vial decomposed much faster than the same amount of VX on asphalt supports this conclusion (not shown).

Estimates of mass balance of VX on asphalt were at 68% for the dynamic (wet and dry) experiments. The incomplete recovery is probably due to difficulties in extraction and quantitation of degradation products of VX (19, 20). Similar behavior was reported with VX on concrete (6). 31P NMR and GC-MS spectroscopy together with the cholinesterase-based assay will provide tools to achieve complete mass balance in future investigations. The primary hydrolysis products of VX in aqueous solutions are compounds 2 and 3 (3). Therefore, the finding that the concentration of 1 exceeded significantly the expected amount is an indication of an enhancement of a specific pathway of VX degradation on an asphalt surface. In view of the variability between asphalt samples, and the desorption patterns observed with the inert surface (Figure 2), more experiments will be needed to better determine when contaminated asphalt no longer presents environmental hazards. These future experiments are also expected to gain better understanding of the relationship between the rate of volatilization and the nature of surface and weather conditions. On the whole, data from the dynamic experiments suggest that approximately 30% of the applied VX was collected from the air and 1-6% of intact VX was recovered from the asphalt sample at the end of the experiment. Results from the static experiments suggest that an additional 30% of VX was hydrolyzed by asphalt constituents after 11 days (from static experiment estimates). The lack of a suitable analytical procedure prevented complete mass balance determinations, and thus, the remaining VX (about 35%) is unaccounted for.

Acknowledgments We are in debt to Dr. Y. Alexander for critically reviewing the manuscript.

Literature Cited (1) For an extensive review and references see: Jansma, J. W.; Linders, J. B. H. J. Report No. 679102030; National Institute of Public Health and Environmental Protection: Bilthoven, The Netherlands, 1995. (2) Munro, N. B.; Ambrose, K. R.; Watson, A. P. Environ. Health Perspect. 1994, 102, 18-38.

(3) Munro, N. B.; Talmage, S. S.; Griffin, G. D.; Waters, L. C.; Watson, A. P.; King, J. F. Environ. Health Perspect. 1999, 107, 933-973. (4) Compton, J. A. F. Military Chemical and Biological Agents; The Telford Press: Caldwell, NJ, 1987; p 9. (5) Hartmann, H. M. Regul. Toxicol. Pharmacol. 2002, 35, 347356. (6) Groenewold, G. S.; Appelhans, A. D.; Gresham, G. L.; Olson, E.; Jeffery, M.; Wiebel, M. J. Am. Soc. Mass Spectrom. 2000, 11, 69-77. (7) Wagner, G. W.; O’Connor, R. J.; Procell, L. R. Langmuir 2001, 17, 4336-4341. (8) Yang, Y.-C.; Szafraniec, L. L.; Beaudry, W. T.; Rohrbaugh, D. K. J. Am. Chem. Soc. 1990, 112, 6621-6627. (9) Yang, Y.-C.; Baker, J. A.; Ward, J. R. Chem. Rev. 1992, 92, 17291743. (10) Kingery, A. F.; Allen, H. E. Toxicol. Environ. Chem. 1995, 47, 155-184. (11) Verweij, A.; Boter, H. L. Pestic. Sci. 1976, 7, 355-362. (12) Kuitunen, M.-L.; Hartonen, K.; Riekkoia, M.-L. J. Microcolumn Sep. 1991, 3, 505-512. (13) D’Agostino, P. A.; Provost, L. R.; Visentini, J. J. J. Chromatogr. 1987, 402, 221-232. (14) Mangino, D. J.; Pilie, R. J.; Reynolds, R. L.; Czarneeki, J. A. Proceedings of the 1995 ERDEC Scientific Conference on Chemical and Biological Defense Research, Edgewood, US Army Chemical and Biological Defense Command: Aberdeen Proving Ground, MD, November 1995; pp 275-278. (15) Grunwald, J.; Marcus, D.; Papier, Y.; Raveh, L.; Pittel, Z.; Ashani, Y. J. Biochem. Biophys. Methods 1997, 34, 123-135. (16) Ellman G. L.; Courtney, K. D.; Andres, V. Featherstone R. M. Biochem. Pharmacol. 1961, 7, 88-95. (17) McIlveen, R. Basic Meteorologysa physical outline; Van Nostrand Reinhold (UK): Berkshire, England, 1986. (18) Harvey, S. P.; Guelta, M. A. Report ECBC-TR-015; Edgewood Chemical Biological Center: Aberdeen Proving Ground, MD, February 1999. (19) Verwejj, A.; Van Liempt-van Houten, M. A.; Boter, H. L. Int. J. Environ. Anal. Chem. 1985, 21, 63-77. (20) Griest, W. H.; Rasey, R. S.; Ho, C.-H.; Caldwell, W. M. J. Chromatogr. 1992, 600, 273-277.

Received for review March 3, 2003. Revised manuscript received November 7, 2003. Accepted November 14, 2003. ES030382J

VOL. 38, NO. 7, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

2223