Assessment of the Transport of Chemical Warfare Agents in Landfills Part III. Chemical Warfare Agent Bounding Calculations

Transcription

1 Assessment of the Transport of Chemical Warfare Agents in Landfills Part III. Chemical Warfare Agent Bounding Calculations Prepared By: Shannon L. Bartelt-Hunt, Detlef R.U. Knappe, and Morton A. Barlaz Department of Civil, Construction, and Environmental Engineering North Carolina State University, Raleigh, NC September 26, 2005

2 ii Table of Contents Introduction 1 Model Description..2 MOCLA..2 Development of diffusive flux term...4 Strategy for Inclusion of CWAs in MOCLA Analyses...5 Toxic Industrial Chemicals 7 Blister Agents...9 Blood Agents..9 Nerve Agents 10 Ranges for MOCLA input data-physical parameters.. 10 Landfill parameters..11 Landfill cover parameters...14 Landfill liner parameters...15 Ranges for MOCLA input data-chemical parameters. 17 Model Results...29 Time independent equilibrium partitioning...29 Base-case scenario as a function of time...30 Effect of climate..48 Effect of biotic transformation Effect of temperature..55 CWA fate under uncertainty...62 Fate of hydrolysis products.80 Summary..82 References....83

3 1 Introduction Recent events including anthrax attacks on government buildings in Washington D.C. and news organizations in New York and Florida have highlighted the need for a comprehensive plan for the disposal of debris contaminated with chemical and/or biological warfare agents. One alternative for the disposal of contaminated debris is burial in a solid waste landfill. Currently, little information is available on the environmental fate of chemical or biological agents in landfills, and the effectiveness of landfills for containing these types of wastes has never been evaluated. An important first step in evaluating landfills as a means of disposal for building debris contaminated with chemical agents is to estimate the potential fate pathways of these contaminants in a landfill. The objective of this report is to present the results of bounding calculations and subsequent analysis of the potential fate pathways of CWAs in landfills. The purpose of these bounding calculations is to produce reasonable predictions of the quantity of a given CWA that could be transported out of the landfill by a number of different fate pathways. Although the model used to generate the bounding calculations utilizes a relatively simple description of chemical fate in a landfill, the resulting calculations provide a good first approximation of CWA fate in landfills as well as serving to highlight information gaps where additional data is needed to further refine the estimates. First, a description of the model used to describe contaminant fate in a landfill is presented along with justification for the ranges of input values used in the calculations. Then, the results from a number of simulations that were performed to assess the effect of climate (precipitation) and temperature on CWA fate are presented. The primary fate pathway for most CWAs was determined to be abiotic hydrolysis and both climate and temperature were found to have little

4 2 effect on contaminant behavior. Since there is limited data available on biodegradation of CWAs under anaerobic conditions, simulations were performed to assess the impact of the biodegradation rate on CWA transport. Next, in recognition of the fact that there is uncertainty in many of the model input parameters, Monte Carlo simulations were conducted to evaluate the potential for contaminant transport under a wide range of potential input values. Finally, model simulations of the fate of several CWA hydrolysis products are presented. Model Description MOCLA The fate of CWAs was evaluated using MOCLA (Model for Organic Chemicals in Landfills), a model developed by Kjeldsen and Christensen (2001). This model describes both the distribution of organic chemicals between the various phases in a landfill (e.g. leachate, landfill gas, solid waste) as well as the fate of the chemicals in terms of transport via landfill gas, leachate, abiotic transformation or degradation. A schematic of the MOCLA model is presented in Figure 1.

5 3 solid S λ K d K H S a S D Soil Cover Waste gas water J w LCS J L Figure 1. Schematic of the MOCLA model. Adapted from Kjeldsen and Christensen (2001). In MOCLA, the distribution of chemicals between the gaseous, liquid and solid phases is governed by the Henry s law constant for the leachate-gas interface and by the linear partition coefficient for the waste-leachate interface. The fate pathways that are considered in the model include: S a -advective transport of landfill gas which would result in capture in a gas collection system or a release through a cover, S D diffusive transport of landfill gas through the soil cover, S λ transformation of the compound from the leachate phase by biotic or abiotic mechanisms, J w flux of leachate into the leachate collection system, and J L transport of chemicals through the composite liner by diffusion. This final pathway was not included in the original MOCLA formulation, however it was added to the model as part of this study. A number of field and laboratory studies have evaluated the significance of transport by diffusion

6 4 through a composite liner (Johnson et al. 1989; Kim et al. 2001; Foose et al. 2002; Toupiol et al. 2002). Development of diffusive flux term The diffusive flux through the liner is represented by Fick s First law: C, = D (1) x J D liner where J D,liner is the diffusive flux (g/m 2 yr), D is the diffusion coefficient in the composite liner (m 2 /yr), and C x is the concentration gradient (g/m 4 ). The diffusion coefficient in the composite liner is represented by the effective diffusion coefficient through the clay layer of the composite liner. It is assumed that the geomembrane layer will not impede the diffusive flux through the composite liner as volatile organic compounds have been shown to penetrate geomembranes at an appreciable rate (Park and Nibras 1993). Since clay liners are typically compacted at a moisture content less than saturation, diffusion will occur through both the air-filled and water-filled pore spaces within the clay. If it is assumed that no free-phase organic is present in the pore space of the clay liner, then the contaminant concentration in the air and water phases can be related by the Henry s law constant of the contaminant of interest: C a =K H C w (2) where C a and C w are the concentrations of the contaminant in the air phase and water phase in the landfill, respectively and K H is the dimensionless Henry s law constant. In a porous medium, the diffusion coefficient for each phase can be written as an effective diffusion coefficient (D * air or D * water) that is related to the pure component molecular diffusivity (D water or D air ), the total porosity of the porous medium (ε), and the air- or water-filled porosity

7 5 (ε a or ε w ). One commonly-used expression is the Millington (1959) equation, which relates the effective diffusion coefficient to the pure component diffusion coefficient: and 3.33 * ε a air D air 2 D = (3) ε 3.33 * ε w water D water 2 D = (4) ε Combining these expressions along with the Henry s law relation results in the following equation for the diffusive flux through the liner: J D, liner D ε D + K ε = a water w air 2 ε 2 H ε Ca x (5) Further details of this derivation may be found in Johnson and Ettinger (1991). If is assumed that the concentration of the contaminant of interest in the aquifer beneath the liner is equal to zero, then the sink due to diffusion through the composite liner may be written as: S D, liner D ε D + K ε = a water w air 2 ε 2 H ε Ca A L (6) where S D,liner is equal to the sink due to diffusion through the liner (g/yr), L is the thickness of the clay layer (m), and the other terms are as previously defined. Strategy for Inclusion of CWAs in MOCLA Analyses Table 1 lists all of the CWAs considered as part of this study.

8 6 Table 1. Chemical Warfare Agents Military Common Name Designation Chemical Formula CAS number Carbon disulfide CS Toxic Industrial Chemical Formaldehyde CH 2 O Toxic Industrial Chemical Phosphorus trichloride PCl Toxic Industrial Chemical Nitric Acid HNO Toxic Industrial Chemical Sulfuric Acid H 2 SO Toxic Industrial Chemical Tungsten hexafluoride WF Toxic Industrial Chemical Boron trichloride BCl Toxic Industrial Chemical Ethylene oxide C 2 H 4 O Toxic Industrial Chemical Hydrogen fluoride HF Toxic Industrial Chemical Phosgene CCl 2 O Toxic Industrial Chemical Furan C 4 H 4 O Toxic Industrial Chemical H/HD Mustard Gas C 4 H 8 Cl 2 S Blister Agent L Lewisite C 2 H 2 AsCl Blister Agent HN-2 Nitrogen Mustard C 5 H 11 Cl 2 N Blister Agent CX Phosgene Oxime CHCl 2 NO Blister Agent ED Ethyldichloroarsine C 2 H 5 AsCl Blister Agent SA Arsine AsH Blood Agent CK Cyanogen Chloride CNCl Blood Agent AC Hydrogen Cyanide HCN Blood Agent PFIB Perfluoroisobutylene C 4 F Obscurant RP Red Phosphorus PH Obscurant FM Titanium Tetrachloride TiCl Obscurant GF Cyclohexyl Sarin C 7 H 14 FO 2 P Nerve Agent GE C 5 H 12 FO 2 P Nerve Agent GB Sarin C 4 H 10 FO 2 P Nerve Agent GD Soman C 7 H 16 FO 2 P Nerve Agent GA Tabun C 5 H 11 N 2 O 2 P Nerve Agent VE Nerve Agent VG Amiton C 10 H 24 NO 3 PS Nerve Agent VM C 9 H 22 NO 2 PS Nerve Agent VX V-gas C 11 H 26 NO 2 PS Nerve Agent CS Tear Gas C 10 H 5 ClN Tear Agent Use

9 7 Since the various CWAs have very different chemical properties, their environmental fate will vary and it is unlikely that many of these agents would persist in building debris long enough to enter a landfill. The following section discusses the potential fate of some of the CWAs prior to disposal in a landfill. Chemicals that are not likely to be associated with building debris in significant quantities or that are known to rapidly hydrolyze to non-toxic degradation products were not included in the MOCLA simulations. Toxic industrial chemicals Many of the toxic industrial chemicals that could potentially be used as CWAs are inorganic gases that will disperse in the atmosphere after being released or they are inorganic liquids that form non-toxic hydrolysis products when they are in contact with water. For example, phosphorus trichloride is a liquid (b.p. = 76 C) that reacts with water to form phosphoric acid and hydrochloric acid, which would be neutralized prior to disposal in a landfill (Melhem and Reid 1997). Boron trichloride is a gas at room temperature (b.p. = 12.5 C) and it is likely that sorption of this compound to building components would be negligible. Boron trichloride would likely disperse in the atmosphere after an attack, however if boron trichloride was disposed of in a landfill, it would form boric acid (H 3 BO 3 ) and hydrochloric acid on contact with water or moist air. Similarly, hydrogen fluoride is likely to be present as gas or liquid at room temperature (b.p. = C), and therefore it is likely to disperse in the atmosphere if used in an attack on a building. On contact with water, it would dissolve to form hydrofluoric acid. Tungsten hexafluoride, another potential CWA, may be a gas or liquid at room temperature (b.p. = 17.5 C). On contact with water, it reacts to form hydrofluoric acid and tungsten oxyfluorides. Two other compounds that are listed as potential CWAs are nitric and sulfuric acid. Assuming that these compounds were neutralized prior to disposal, the nitrate from nitric acid would likely

10 8 be reduced to nitrogen gas by denitrification and the sulfates from sulfuric acid would be converted to hydrogen sulfide by sulfate-reducing bacteria if active anaerobic refuse decomposition was occurring in the landfill. Thus, these inorganic toxic industrial chemicals were not included in the modeling analysis as they are either gases that will disperse in the atmosphere prior to disposal or liquids that will hydrolyze to form non-toxic products. Several of the toxic industrial chemicals that are listed as potential CWAs are organic gases that are also unlikely to persist in building debris. Formaldehyde (CH 2 O), phosgene (CCl 2 O), and ethylene oxide (C 2 H 4 O) are all gases at room temperature. Their boiling points are C, 8.2 C, and 12 C, respectively. Therefore the primary pathway by which these compounds could enter a landfill is through sorption from the gas phase to the building debris components. A study performed by Singer et al. (2004) examining the sorption of organic gases to various components of a furnished room found that chemicals with Henry s law constants and vapor pressures similar to that of these three compounds did not sorb appreciably to the components of the furnished room (e.g. painted wallboard, carpet, draperies and furnishings) after 12 hours. Therefore, these organic gases would be unlikely to enter a landfill in significant quantities if used in a chemical attack. Two toxic industrial chemicals that are listed as potential CWAs may be disposed of in a landfill in significant quantities. Carbon disulfide (CS 2 ) is an inorganic liquid at room temperature (b.p. = 46.5 C) and it has a relatively high log K ow value. Because of its potential to sorb appreciably to building components, it will be analyzed as part of the bounding calculations. Similarly, furan (C 4 H 4 O) is an organic liquid with high log K ow, making it another candidate for consideration.

11 9 Blister Agents All of the blister agents considered in this study will be included in the bounding calculations. With the exception of phosgene oxime, all blister agents are liquids at room temperature and have the capacity to sorb to building components. Phosgene oxime is a solid that would be used in an attack on a building in an aqueous solution. Although many of the blister agents undergo rapid hydrolysis in contact with water, many of the hydrolysis products retain some degree of toxicity. Therefore, both the blister agents and their hydrolysis products should be examined as part of the bounding calculations. Blood Agents The three CWAs classified as blood agents (e.g. arsine, hydrogen cyanide, and cyanogen chloride) are unlikely to be associated with building debris in any significant quantities. Arsine (AsH 3 ) has a boiling point of C, and therefore would be a gas at room temperature. The likely fate of arsine gas would be atmospheric dispersal, however if some arsine did become associated with the building debris, it would hydrolyze to solid black arsenic (Budavari 1996). Hydrogen cyanide (HCN) has a boiling point of 25.6 C and therefore it could be gas or liquid at room temperature. A study performed by Reucroft and Chiou (1977) examining the sorption of hydrogen cyanide and cyanogen chloride to activated carbon found that the affinity coefficient for hydrogen cyanide relative to chloroform was This indicates that affinity of the organic carbon fraction of the building components for hydrogen cyanide should also be relatively low, especially when considering that volatile organic compounds with properties similar to chloroform did not sorb appreciably to components in a furnished room (Singer et al. 2004). If some hydrogen cyanide were sorbed to building components, it would undergo slow hydrolysis to ammonia and formic acid on contact with water (Wiegand and Tremmelling 1972). Because it

12 10 is likely that a decontamination agent containing chlorine would be used on the building debris prior to disposal in a landfill, the rate of HCN hydrolysis would be increased, further reducing the likelihood of hydrogen cyanide entering a landfill. Cyanogen chloride (CNCl) is a gas at room temperature (b.p. = 13.8 C). Cyanogen chloride is also unlikely to sorb appreciably to building components since the measured affinity coefficient for cyanogen chloride relative to chloroform was only (Reucroft and Chiou 1977). Any cyanogen chloride that would be associated with building components would undergo slow hydrolysis to cyanic acid and chloride ion. If decontamination agents containing chlorine or hydroxide were applied to the building debris prior to disposal, the rate of CNCl hydrolysis would be increased (Na and Olson 2004), which would further reduce the possibility of cyanogen chloride entering the landfill. Nerve Agents All of the CWAs classified as nerve agents are liquids at room temperature, and therefore are likely to become associated with building debris. In addition, the rate of hydrolysis of the nerve agents is much slower than the blister agents, making them more persistent. All nerve agents will be analyzed in the bounding calculations. Physical-chemical property data for all CWAs and their associated hydrolysis products was presented in a previously provided literature review. Ranges for MOCLA input data physical parameters To evaluate the distribution and fate of chemicals in a landfill, data for a number of different input parameters describing the landfill, the landfill cover soil and the landfill liner are required. This section describes each of these parameters and gives values for ranges for these parameters that could be applicable under various disposal scenarios. Ranges were developed for each

13 11 parameter to facilitate uncertainty analysis which is described below. The input parameters required by MOCLA are summarized in Table 2. Landfill parameters Dry bulk density of the waste, ρ b (mt/m 3 ): It is assumed that the building debris will be contained for transport and burial and therefore will not be compacted after placement in the landfill. Therefore, the wet bulk density of the building debris will likely be lower than the average value of wet bulk density for compacted waste, which is 1300 lb/yd 3. The range of wet bulk densities to be used in the bounding calculations will be lb/yd 3, with an average value of 950 lb/yd 3. This corresponds to mt/m 3. The dry bulk density can be determined from the wet bulk density and an estimate of the gravimetric moisture content of the waste. If the gravimetric moisture content (MC) of the waste is assumed to be 10-20% (wet weight basis), then the dry bulk density can be calculated as follows: Using a design basis of 1 m 3 of refuse, the total mass of waste is equal to 0.42 mt, which corresponds to the low end of range. The mass of water in 1 m 3 of refuse is equal to 0.42 times 0.20 (highest possible value of MC) or mt. The total mass of waste minus the mass of water ( ) results in the mass of dry refuse which is 0.34 mt. Therefore, the low value of the range of potential values of dry bulk density of the waste is 0.34 mt/m 3. Similarly, the high end of the range is equal to 0.63 mt/m 3. The average value of dry bulk density is 0.49 mt/m 3.

14 12 Volumetric moisture content of the waste, ε w (m 3 water/m 3 LF): The possible values of volumetric moisture content of the waste can be calculated based on the ranges of the wet bulk density and the gravimetric moisture content, assuming a design basis of 1 m 3 of waste. To estimate the lowest value in the range of volumetric moisture contents, at 10% gravimetric moisture, the mass of water in 1 m 3 of waste is equal to 0.42 mt (total mass) times 0.10 (mass of water/total mass) which is The density of water is equal to 1 mt/m 3, so the volumetric moisture content is equal to m 3 water/m 3 LF. Similarly, the highest value in the range of volumetric moisture content is equal to 0.14 m 3 water/m 3 LF, which is equivalent to 0.70 mt times The average value of the volumetric moisture content is m 3 water/m 3 LF. Volumetric gas content of the waste, ε a (m 3 air/m 3 LF): A range of 0.10 to 0.40 will be assumed for the volumetric gas content of the waste. There is very little data available in the literature on measured values of the effective porosity of waste. Bendz (1998) reported a measured value of 0.12 for a waste sample taken from an operating MSW landfill. Based on the range specified above, the average value for the volumetric gas content of the waste is Fraction of organic carbon, f oc : A range of 0.40 to 0.60 will be used for the fraction of organic carbon in the waste. A value of 0.50 was reported by Barlaz (1998) for a sample of MSW, which will be used as the average value in the bounding calculations. Height of the waste in the landfill, H (m): The height of the waste in the landfill is assumed to vary between m (60 and 200 ft). Based on this range, the average height of waste in the landfill will be 39.7 m.

15 13 The final two terms required to describe the landfill, net precipitation and the gas production rate, are a function of the type of climate in which the landfill is located. As a result, two sets of values for these parameters were used to investigate the effects of an arid or wet climate on the fate of the chemical agents in a landfill. Net Precipitation, N (m/year): The net precipitation, or infiltration, was calculated based on data reported in Camobreco et al. (2000). For the arid climate scenario, data from sites with less than 0.5 m/yr (20 in/year) of total precipitation were used, while the data from sites with greater than or equal to 0.5 m/yr were used for the wet climate scenario. For each site, the total precipitation was multiplied by the percentage of precipitation that becomes leachate given in Camobreco et al. (2000) to determine the net precipitation for each site. Using this method, the net precipitation data for the arid climate was found to range from 0.02 to 0.05 m/yr with an average value of 0.04 m/yr (1.6 in/yr). The net precipitation for the wet climate type ranged from 0.04 to 0.32 m/yr with an average of 0.12 m/yr (4.7 in/yr). Although the data used to generate these estimates encompassed a variety of different percentages of final cover, it should be noted that modeling a dry (fully capped) landfill is potentially a more realistic scenario than a wet landfill since the building debris will likely be encapsulated and the greatest likelihood for leakage from the containers will occur after the final cover has been put in place. Gas production rate, q a (m 3 LFG/m 3 LF yr): The landfill gas production rate can be described by the following equation: q a =2WL o ke -kt (7) where W is the mass of refuse, L o is a term representing the ultimate volume of methane gas produced per unit of wet waste, and k is a rate constant with units of yr -1.

16 14 The mass of refuse (W) is a function of the bulk density of the waste, which as stated above, will be assumed to vary from lb/yd 3, with an average value of 950 lb/yd 3. Assuming 1 yd 3 of waste, W will be equal to 950 lb. L o will be assumed to vary between 85 and 170 L methane/kg wet waste. The value of 170 L methane/kg wet waste is the default used in the New Source Performance Standards of the Clean Air Act amendments. For the arid climate scenario, the rate constant (k) was set equal to 0.02 yr -1. For the wet climate scenario, the rate constant (k) was set equal to 0.05 yr -1. For the arid scenario, q a was found to range from 1.9 to 3.7 m 3 LFG/m 3 LF yr with an average value of 2.8 m 3 LFG/m 3 LF yr. For the wet scenario, q a was found to range from 4.5 to 9.0 m 3 LFG/m 3 LF yr with an average value of 6.75 m 3 LFG/m 3 LF yr. Landfill Cover Parameters Thickness of cover, L (m): The thickness of the landfill cover will be set to 0.45 m (18 in.) which is the minimum thickness specified by Subtitle D regulation for the thickness of the barrier layer. This value will not be varied in the bounding calculations. Total porosity of the soil cover, ε sc (dimensionless): The total porosity of the soil cover will range from 0.3 to 0.5 with an average value of 0.4. A total porosity between 0.3 and 0.5 is common for a wide range of soil types (Sharma and Lewis 1994). Gravimetric Moisture Content (%): The gravimetric moisture content of the soil cover will range from % (dry weight basis) with an average value of 20.0% based on the data presented in Benson (1999).

17 15 Dry bulk density of the cover soil, ρ b (g/cm 3 ): The dry bulk density of the soil cover will range from 1.3 to 2.0 g/cm 3 with an average value of 1.7 g/cm 3 based on the data presented in Benson (1999). It should be noted that the gravimetric moisture content and dry bulk density of the cover soil are not used directly in MOCLA. These parameters are used along with the porosity of the cover soil to calculate the air-filled porosity of the cover soil, which is an input parameter to MOCLA. Landfill Liner Parameters Liner Thickness (m): The thickness of the landfill liner will be set to 0.6 m (24 in.), which is the minimum thickness required by EPA Subtitle D regulations. The liner thickness will not be varied in the bounding calculations. Total Porosity of the liner: The total porosity of the liner will range from 0.3 to 0.5 with an average value of 0.4. A total porosity between 0.3 and 0.5 is common for a wide range of soil types (Sharma and Lewis 1994). Gravimetric Moisture Content (%): Based on the data presented in Benson (1999), the gravimetric moisture content of the liner will range from %, with an average value of 20.0%. Dry bulk density of the liner materials, ρ b (g/cm 3 ): Based on the data presented in Benson (1999), the dry bulk density of the liner will range from 1.3 to 2.0 g/cm 3 with an average value of 1.7 g/cm 3 It should be noted that the gravimetric moisture content and dry bulk density of the liner are not used directly in MOCLA. These

18 16 parameters are used along with the porosity of the liner to calculate the air-filled porosity of the liner, which is an input parameter to MOCLA. Table 2. Physical input parameters for MOCLA. Parameter Units Range Average Source Dry bulk density of the waste (ρ b ) mt/m Assumed values Volumetric moisture content of waste m 3 water/m 3 LF Assumed value Volumetric gas content of waste m 3 air/m 3 LF Assumed values Bendz (1998) Fraction of organic carbon in waste Barlaz (1998) Height of waste m Assumed values Net precipitation (N) arid region (sites with < 20 inch/yr) m/yr (EREF 2000) Net precipitation (N) wet region (sites with > 20 inch/yr) m/yr (EREF 2000) Gas production rate (q a ) arid region m 3 LFG/m 3 LF yr Assumed values Gas production rate (q a ) wet region m 3 LFG/m 3 LF yr Assumed values U.S. EPA Subtitle Thickness of Cover Soil (L) m 0.45 D regulation Total porosity of Cover Soil (ε sc ) Assumed values Gravimetric moisture content Dry bulk density of the cover soil % (dry weight basis) % 20.00% Benson (1999) g/cm Benson (1999) U.S. EPA Subtitle Thickness of liner m 0.6 D regulation Total porosity of liner Assumed values % (dry weight Gravimetric moisture content 20.00% Benson (1999) basis) % Dry bulk density of the liner g/cm Benson (1999)

19 17 Ranges for MOCLA input data chemical parameters There is a lack of experimental data for physical-chemical property data for CWAs. For many CWAs, the only method available to obtain certain physical-chemical parameters required by MOCLA was to estimate parameters using EPIWIN. There is a need to evaluate the uncertainty associated with the physical-chemical data inputs required by MOCLA on the predicted fate pathways for the CWAs, however developing a methodology for assessing the uncertainty in a single point value or of a single estimated value is difficult. In light of the need to assess uncertainty in the absence of experimental data, ranges for solubility, vapor pressure, Henry s law constant and log K ow of the CWAs were developed by determining the maximum and minimum values reported in Mackay et al. (1992) for 37 organic chemicals that were originally modeled using MOCLA (Kjeldsen and Christensen 2001). Using the data found in Mackay et al. (1992), the lowest reported value and highest reported value were determined for each of the organic chemicals. These were compared to the value reported in Kjeldsen and Christensen (2001), which was considered to be the accepted value. A percent difference was calculated for both the low and high values compared to the accepted value, where: high or low accepted Percent difference = *100% (8) accepted The median percent difference determined from the organic chemicals was then applied to each of the CWAs to determine a realistic range for each parameter. The accepted value for each parameter for the CWAs was an experimentally determined value, if available, or an estimate from EPIWIN. The parameter data for the organic chemicals, along with the calculations of percent difference are given in Tables 3-6. The ranges for the CWAs to be modeled using MOCLA are given in Tables 7-10.

20 Table 3. Solubility data from Mackay et al. (1992) for organics originally modeled in MOCLA. Chemical low MOCLA (accepted) Solubility (mg/l) high % difference (low) % difference (high) Benzene, C6 H Toluene, C6 H5 CH Ethylbenzene C2 H5 C6 H Xylene (o), C6 H4 (CH3) Xylene (m/p ), C6 H4 (CH3) , 3, 5 Trimethylbenzene, C6 H3 (CH Phenol, C6 H5 OH Cresol, (o) C7 H7 OH ,4 - Dimethylphenol C8 H9 OH ,2 - Dichlorobenzene, C6 H4 Cl ,4 - Dichlorobenzene C6 H4 Cl Chlorophenol C6 H4 OH Cl ,4 - Dichlorophenol C6 H3 OH Cl Naphthalene, C10 H Methylnaphthalene, C11 H Flourene, C13 H Anthracene, C14 H n-hexane, C6 H n-heptane, C7 H n-octane, C8 H n-nonane, C9 H Cyclohexane, C6 H Chloromethane, CM, methylchloride, CH3 Cl Dichloromethane, DCM, CH2 Cl Trichloromethane, TCM, Chloroform, CH Cl Tetrachloromethane, TeCM, C Cl Chloroethane, ethylchloride, C2 H5 Cl ,1 - Dichloroethane, DCA, CH3 CH Cl ,2 - Dichloroethane, DCA, CH2 Cl CH2 Cl ,1,1 - Trichloroethane, TCA, CH3 C Cl ,1,2 - Trichloroethane, TCA, CH2 Cl CH Cl Chloroethene Vinylchloride, H2C = CH Cl Dichloroethene, 1,1 - DCE, CH2 = C Cl Dichloroethene, cis - 1,2 - DCE, CH Cl = CH Cl Dichloroethene, trans - 1,2 - DCE, Cl CH = CH Cl Trichloroethene, TCE, CH Cl = C Cl Tetrachloroethene, PCE, C Cl2 = C Cl % difference (low) % difference (high) maximum minimum median

21 19 Table 4. Vapor pressure data from Mackay et al. (1992) for organics originally modeled in MOCLA. Chemical low MOCLA (accepted) Vapor Pressure (Pa) high % difference (low) % difference (high) Benzene, C6 H6 1.01E E E Toluene, C6 H5 CH3 8.55E E E Ethylbenzene C2 H5 C6 H5 1.28E E E Xylene (o), C6 H4 (CH3)2 8.80E E E Xylene (m/p ), C6 H4 (CH3)2 1.10E E E , 3, 5 Trimethylbenzene, C6 H3 (CH E E E Phenol, C6 H5 OH 2.67E E E Cresol, (o) C7 H7 OH 2.25E E E ,4 - Dimethylphenol C8 H9 OH 8.27E E E ,2 - Dichlorobenzene, C6 H4 Cl2 1.33E E E ,4 - Dichlorobenzene C6 H4 Cl2 5.33E E E Chlorophenol C6 H4 OH Cl 1.33E E E ,4 - Dichlorophenol C6 H3 OH Cl2 2.40E E E Naphthalene, C10 H8 6.56E E E Methylnaphthalene, C11 H E E E Flourene, C13 H E E E Anthracene, C14 H E E E n-hexane, C6 H E E E n-heptane, C7 H E E E n-octane, C8 H E E E n-nonane, C9 H E E E Cyclohexane, C6 H E E E Chloromethane, CM, methylchloride, CH3 Cl 1.01E E E Dichloromethane, DCM, methylenechloride, CH2 Cl2 4.65E E E Trichloromethane, TCM, Chloroform, CH Cl3 2.00E E E Tetrachloromethane, TeCM, C Cl4 9.16E E E Chloroethane, ethylchloride, C2 H5 Cl 1.01E E E ,1 - Dichloroethane, DCA, CH3 CH Cl2 2.40E E E ,2 - Dichloroethane, DCA, CH2 Cl CH2 Cl 8.13E E E ,1,1 - Trichloroethane, TCA, CH3 C Cl3 1.28E E E ,1,2 - Trichloroethane, TCA, CH2 Cl CH Cl2 2.53E E E Chloroethene Vinylchloride, H2C = CH Cl 3.08E E E Dichloroethene, 1,1 - DCE, CH2 = C Cl2 6.62E E E Dichloroethene, cis - 1,2 - DCE, CH Cl = CH Cl 2.18E E E Dichloroethene, trans - 1,2 - DCE, Cl CH = CH Cl 2.67E E E Trichloroethene, TCE, CH Cl = C Cl2 6.31E E E Tetrachloroethene, PCE, C Cl2 = C Cl2 1.87E E E % difference (low) % difference (high) maximum minimum median

22 20 Table 5. Henry s law constant data from Mackay et al. (1992) for organics originally modeled in MOCLA. Chemical K H (dimensionless) MOCLA % difference % difference low high (accepted) (low) (high) Benzene, C6 H6 2.20E E E Toluene, C6 H5 CH3 2.09E E E Ethylbenzene C2 H5 C6 H5 2.70E E E Xylene (o), C6 H4 (CH3)2 1.76E E E Xylene (m/p ), C6 H4 (CH3)2 2.04E E E , 3, 5 Trimethylbenzene, C6 H3 (CH E E E Phenol, C6 H5 OH 1.61E E E Cresol, (o) C7 H7 OH 2.83E E E ,4 - Dimethylphenol C8 H9 OH 2.79E E E ,2 - Dichlorobenzene, C6 H4 Cl2 4.92E E E ,4 - Dichlorobenzene C6 H4 Cl2 6.46E E E Chlorophenol C6 H4 OH Cl 2.74E E E ,4 - Dichlorophenol C6 H3 OH Cl2 4.44E E E Naphthalene, C10 H8 1.18E E E Methylnaphthalene, C11 H E E E Flourene, C13 H E E E Anthracene, C14 H E E E n-hexane, C6 H E E E n-heptane, C7 H E E E n-octane, C8 H E E E n-nonane, C9 H E E E Cyclohexane, C6 H E E E Chloromethane, CM, methylchloride, CH3 Cl 3.61E E E Dichloromethane, DCM, CH2 Cl2 6.98E E E Trichloromethane, TCM, Chloroform, CH Cl3 7.99E E E Tetrachloromethane, TeCM, carbon tetrachloride 8.07E E E Chloroethane, ethylchloride, C2 H5 Cl 3.47E E E ,1 - Dichloroethane, DCA, CH3 CH Cl2 1.70E E E ,2 - Dichloroethane, DCA, CH2 Cl CH2 Cl 3.74E E E ,1,1 - Trichloroethane, TCA, CH3 C Cl3 1.61E E E ,1,2 - Trichloroethane, TCA, CH2 Cl CH Cl2 3.27E E E Chloroethene Vinylchloride, H2C = CH Cl 9.00E E E Dichloroethene, 1,1 - DCE, CH2 = C Cl2 6.13E E E Dichloroethene, cis - 1,2 - DCE, CH Cl = CH Cl 1.21E E E Dichloroethene, trans - 1,2 - DCE, Cl CH = CH Cl 2.70E E E Trichloroethene, TCE, CH Cl = C Cl2 9.97E E E Tetrachloroethene, PCE, C Cl2 = C Cl2 1.06E E E % difference (low) % difference (high) maximum minimum median

23 Table 6. Log K ow data from Mackay et al. (1992) for organics originally modeled in MOCLA. Chemical K ow (Log K ow ) low MOCLA (accepted) high % difference (low) 21 % difference (high) Benzene, C6 H Toluene, C6 H5 CH Ethylbenzene C2 H5 C6 H Xylene (o), C6 H4 (CH3) Xylene (m/p ), C6 H4 (CH3) , 3, 5 Trimethylbenzene, C6 H3 (CH Phenol, C6 H5 OH Cresol, (o) C7 H7 OH ,4 - Dimethylphenol C8 H9 OH ,2 - Dichlorobenzene, C6 H4 Cl ,4 - Dichlorobenzene C6 H4 Cl Chlorophenol C6 H4 OH Cl ,4 - Dichlorophenol C6 H3 OH Cl Naphthalene, C10 H Methylnaphthalene, C11 H Flourene, C13 H Anthracene, C14 H n-hexane, C6 H n-heptane, C7 H n-octane, C8 H n-nonane, C9 H Cyclohexane, C6 H Chloromethane, CM, methylchloride, CH3 Cl Dichloromethane, DCM, methylenechloride, CH2 Cl Trichloromethane, TCM, Chloroform, CH Cl Tetrachloromethane, TeCM, C Cl Chloroethane, ethylchloride, C2 H5 Cl ,1 - Dichloroethane, DCA, CH3 CH Cl ,2 - Dichloroethane, DCA, CH2 Cl CH2 Cl ,1,1 - Trichloroethane, TCA, CH3 C Cl ,1,2 - Trichloroethane, TCA, CH2 Cl CH Cl Chloroethene Vinylchloride, H2C = CH Cl Dichloroethene, 1,1 - DCE, CH2 = C Cl Dichloroethene, cis - 1,2 - DCE, CH Cl = CH Cl Dichloroethene, trans - 1,2 - DCE, Cl CH = CH Cl Trichloroethene, TCE, CH Cl = C Cl Tetrachloroethene, PCE, C Cl2 = C Cl % difference (low) % difference (high) maximum minimum median

24 22 Table 7. Solubility ranges for CWAs. Solubility (mg/l) Minimum Accepted Maximum Toxic Industrial Compounds Carbon disulfide * Furan * Blister Agents Distilled Mustard (HD) * Nitrogen Mustard * Lewisite * Ethyldichloroarsine Phosgene Oxime Nerve Agents GA (Tabun) * GB (Sarin) GD (Soman) GE GF * VX * VG VM CS * denotes experimentally determined literature value. Accepted values without a * are EPIWIN estimates.

25 23 Table 8. Vapor Pressure ranges for CWAs. Vapor Pressure (Pa) Minimum Accepted Maximum Toxic Industrial Compounds Carbon disulfide * Furan * Blister Agents Distilled Mustard (HD) * Nitrogen Mustard * Lewisite * Ethyldichloroarsine * Phosgene Oxime * Nerve Agents GA (Tabun) * GB (Sarin) * GD (Soman) * GE GF * VX * VG VM CS * * denotes experimentally determined literature value. Accepted values without a * are EPIWIN estimates.

26 24 Table 9. Henry s law constant ranges for CWAs. Henry's law constant (dimensionless) Minimum Accepted Maximum Toxic Industrial Compounds Carbon disulfide * Furan * Blister Agents Distilled Mustard (HD) * Nitrogen Mustard Lewisite * Ethyldichloroarsine Phosgene Oxime Nerve Agents GA (Tabun) * GB (Sarin) GD (Soman) GE GF VX * VG VM CS * denotes experimentally determined literature value. Accepted values without a * are EPIWIN estimates.

27 25 Table 10. Log K ow ranges for CWAs. Log K ow Minimum Accepted Maximum Toxic Industrial Compounds Carbon disulfide * 2.11 Furan * 1.46 Blister Agents Distilled Mustard (HD) Nitrogen Mustard * 0.98 Lewisite Ethyldichloroarsine Phosgene Oxime Nerve Agents GA (Tabun) * 0.43 GB (Sarin) * 0.33 GD (Soman) * 1.94 GE GF VX * 2.27 VG VM CS * denotes experimentally determined literature value. Accepted values without a * are EPIWIN estimates.

28 26 Other input parameters required in MOCLA are the free-air and free-water diffusion coefficients (D air and D water ), and the biotic and abiotic half-lives of the CWAs. Values of D air and D water determined using the Wilke-Lee estimation method (Lyman et al. 1990) were increased and decreased by one order of magnitude in the MOCLA bounding calculations. Values of these parameters for the CWAs are given in Table 11. No information is currently available on the biotic half-lives of the CWAs. Therefore, bounding calculations will be performed with different fixed values for the biotic half-life (λ biotic =, 1000 days, 100 days, and 10 days) to evaluate the sensitivity of the results to the magnitude of this parameter. Information on the abiotic half-lives of many of the chemical agents is available in the literature, however all existing hydrolysis data has been measured under aerobic conditions and may not accurately represent hydrolysis under the anaerobic conditions commonly found in a landfill. These values will be used as the accepted value, with maximum and minimum values determined by increasing and decreasing the accepted value by one order of magnitude. When abiotic half-life information is not available, then the average abiotic half-life for the particular class of compounds (e.g. V-agents, G-agents) will be used as the accepted value for that particular compound. In the case of the toxic industrial chemicals, no abiotic half-life information is available in the literature. In this case, the accepted value will be assumed to be equal to no degradation (infinite half-life). Ranges for the abiotic half-lies for the CWAs are given in Table 12.

29 27 Table 11. Air- and water-diffusion coefficient ranges for CWAs. D * (m 2/ s) D water (m 2/ s) Minimum Accepted Maximum Minimum Accepted Maximum Toxic Industrial Compounds Carbon disulfide Furan Blister Agents Distilled Mustard (HD) Nitrogen Mustard Lewisite Ethyldichloroarsine Phosgene Oxime Nerve Agents GA (Tabun) GB (Sarin) GD (Soman) GE GF VX VG VM CS

30 28 Table 12. Abiotic half-life ranges for CWAs. Abiotic Half-Life (days) Minimum Accepted Maximum Blister Agents Distilled Mustard (HD) Nitrogen Mustard Lewisite Ethyldichloroarsine Phosgene Oxime Nerve Agents GA (Tabun) GB (Sarin) GD (Soman) GE GF VX VG VM CS

31 29 Model Results MOCLA provides two types of output. First, MOCLA estimates the distributions of chemicals between the solid (f s ), liquid (f w ), and gaseous (f a ) phases. These results are presented as the fraction of the target chemical in each phase at equilibrium. Thus, these values are independent of time. MOCLA was also used to calculate the fraction of the target chemical that is remaining, lost to the gas phase by advection (F a ) or diffusion (F gd ), present in the leachate (F w ), transformed by abiotic reaction (F λ, abiotic ) and the fraction lost by diffusion through the liner (F diff ). These values are a function of time and simulations were performed for a number of different variables (Table 13). Table 13. Summary of MOCLA simulations. Simulation Type Climate Variable Base-case scenario as a function arid time = 6 mo,1 yr, 5 yr, 30 yr of time (no biological decay) wet Base-case scenario (1 yr). with biotic degradation Base-case scenario (1 yr) at 40 C wet wet λ biotic = (no degradation), 1000 days, 100 days, 10 days temperature-corrected K H and vapor pressure Monte Carlo simulations wet λ= Fate of hydrolysis products (1 yr) wet λ= Time Independent Equilibrium Partitioning The distribution of the CWAs at equilibrium between the solid, liquid and gaseous phases is independent of time and the results of this analysis are presented in Table 14. For all CWAs evaluated, over 90% of the chemical agent will be associated with the solid fraction (waste) in the landfill. The remainder of the chemical will be associated with the water (leachate) fraction, with little to no chemical in the gas phase. This is due to the relatively large K ow value of most of the chemical agents.

32 30 Table 14. Equilibrium phase fractions for the CWAs. Chemical Agent Phase fractions f a f w f s Toxic Industrial Compounds Carbon disulfide Furan Blister Agents Distilled Mustard (HD) Nitrogen Mustard Lewisite Ethyldichloroarsine Phosgene Oxime Nerve Agents GA (Tabun) GB (Sarin) GD (Soman) GE GF VX VG VM CS Base-case scenario as a function of time The base-case scenario uses the average (accepted) values for all input parameters as given in Tables 2 and 7 to12. Simulations were performed using MOCLA to determine the fraction of the target chemical that is transported by each of the potential fate pathways. These values are a function of time and analyses were conducted for 6 months, 1 year, 5 years, and 30 years. These simulations were performed under the assumption that there would be no biological degradation of the chemical agents as there is no published information documenting anaerobic biological transformation rates for these compounds.

33 31 Predicted fate routes under the wet and arid climate scenarios for each chemical agent after 6 months are given in Tables 15 and 16. These data are also presented graphically in Figures 2 and 3. F a represents the fraction of the chemical transported out of the landfill via advection into the landfill gas collection system and F gd represents the fraction transported via diffusion into the atmosphere through the soil cover. It should be noted that losses via advection into the gas collection system do not necessarily correspond to a release to the environment as this gas is presumably captured by a gas management system; however losses due to diffusion through the soil cover would be directly released into the air above the landfill. Any gas lost by diffusion through the soil cover may undergo some attenuation by aerobic biological processes, especially for compounds that are biodegradable. F w represents the fraction of the chemical that would be transported via leachate and would enter the leachate collection system. F λ represents the fraction of the chemical that would be transformed via abiotic hydrolysis and F diff represents the fraction of the chemical that would diffuse through the composite liner system and enter the unsaturated zone below the landfill.

34 32 Table 15. Predicted fate routes for the CWAs after 6 months under the arid climate scenario. Chemical Agent Fraction remaining Fate routes (after 6 months) F a F gd F w F λ (abiotic) F diff (liner) Toxic Industrial Compounds Carbon disulfide Furan Blister Agents Distilled Mustard Nitrogen Mustard Lewisite Ethyldichloroarsine Phosgene Oxime Nerve Agents GA (Tabun) GB (Sarin) GD (Soman) GE GF VX VG VM CS

35 33 Table 16. Predicted fate routes for the CWAs after 6 months under the wet climate scenario. Chemical Agent Fraction remaining Fate routes (after 6 months) F a F gd F w F λ (abiotic) F diff (liner) Toxic Industrial Compounds Carbon disulfide Furan Blister Agents Distilled Mustard Nitrogen Mustard Lewisite Ethyldichloroarsine Phosgene Oxime Nerve Agents GA (Tabun) GB (Sarin) GD (Soman) GE GF VX VG VM CS

36 Fdiff (liner) Fλ abiotic Fw Fgd Fa Fraction remaining 34 CS VM VG Carbon disulfide Furan Distilled Mustard (HD) Nitrogen Mustard Lewisite Ethyldichloroarsine Phosgene Oxime GA (Tabun) GB (Sarin) GD (Soman) GE GF VX Figure 2. Fate routes for CWAs after six months (arid scenario) Fdiff (liner) Fλ abiotic Fw Fgd Fa Fraction remaining CS VM VG Carbon disulfide Furan Distilled Mustard (HD) Nitrogen Mustard Lewisite Ethyldichloroarsine Phosgene Oxime GA (Tabun) GB (Sarin) GD (Soman) GE GF VX Figure 3. Fate routes for CWAs after six months (wet scenario).

37 35 Analysis of the predicted fate routes for the TICs indicates that the majority of the carbon disulfide and furan disposed of in the landfill will remain after a 6 month period under both climate scenarios. Abiotic hydrolysis appears to be the most significant fate pathway for the majority of the blister agents and a number of the G-agents. Distilled mustard, nitrogen mustard, phosgene oxime, GA, and GB were completely transformed via hydrolysis during the 6 month period. The rates of hydrolysis for Lewisite, ethyldichloroarsine, GD, and GF are slightly slower, resulting in approximately 25-70% of these chemicals being transformed via hydrolysis with the remainder remaining in the landfill. The rest of the nerve agents have relatively slow rates of hydrolysis and therefore a significant fraction will remain in the landfill under either climate scenario. Hydrolysis is the only potential fate route that has an impact on the transport of blister or nerve agents over the six month period as the fractions for all other fate routes are negligible. Results from the one year simulations are presented in Tables 17 and 18 and Figures 4 and 5. Results indicate that the majority (greater than 90%) of the carbon disulfide and furan will still be present in the landfill after one year under the arid scenario and greater than 80% will remain in the landfill under the wet scenario. For these chemicals, the primary fate pathway for the small percentage of the chemical that has left the landfill is advection in the gas phase. Advective losses are higher in the wet scenario due to the increased gas production rate.

38 36 Table 17. Predicted fate routes for the CWAs after 1 year under the arid climate scenario. Fate routes (after 1 year) Chemical Agent Fraction F F λ F diff a F remaining gd F w (abiotic) (liner) Toxic Industrial Compounds Carbon disulfide Furan Blister Agents Distilled Mustard Nitrogen Mustard Lewisite Ethyldichloroarsine Phosgene Oxime Nerve Agents GA (Tabun) GB (Sarin) GD (Soman) GE GF VX VG VM CS

39 37 Table 18. Predicted fate routes for the CWAs after 1 year under the wet climate scenario. Chemical Agent Fraction remaining Fate routes (after 1 year) F a F gd F w F λ (abiotic) F diff (liner) Toxic Industrial Compounds Carbon disulfide Furan Blister Agents Distilled Mustard Nitrogen Mustard Lewisite Ethyldichloroarsine Phosgene Oxime Nerve Agents GA (Tabun) GB (Sarin) GD (Soman) GE GF VX VG VM CS

40 Fdiff (liner) Fλ abiotic Fw Fgd Fa Fraction remainin g 38 CS VM Carbon disulfide Furan Distilled Mustard (HD) Nitrogen Mustard Lewisite Ethyldichloroarsine Phosgene Oxime GA (Tabun) GB (S arin) GD (Soman) GE GF VX VG Figure 4. Fate routes for the CWAs aft er one year (arid scenario) Fdiff (liner) Fλ abiotic Fw Fgd Fa Fraction remaining CS VM Carbon disulfide Furan Distilled Mustard (HD) Nitrogen Mustard Lewisite Ethyldichloroarsine Phosgene Oxime GA (Tab un) GB (S GD arin) (Soman) GE GF VX VG Figure 5. Fate routes for the CWAs after one year (wet scenario).

41 39 Of the two blister agents remaining in significant quantities at the end of the six month simulations, only a small percentage will be left in the landfill at the end of one year. Results from the one year simulations indicate that 15% of the lewisite and 6% of the ethyldichloroarsine will remain, with the rest being transformed via abiotic hydrolysis. The G-agents are completely transformed via abiotic hydrolysis by the end of one year, with the exception of GD (53.1% remaining in the landfill) and GF (23.3% remaining in the landfill). The V-agents and tear gas (CS) are more persistent, with greater than 85% of these chemicals remaining in the landfill, with the remainder transformed via abiotic hydrolysis. Results from the five year simulations are presented in Tables 19 and 20 and Figures 6 and 7. Results indicate that nearly 60% of the carbon disulfide and furan will still be present in the landfill at the end of this time period under the arid climate scenario. Under the wet climate scenario, only 30% of these chemicals will remain in the landfill after 5 years. All blister agents are completely transformed via abiotic hydrolysis after five years. Similarly, all G-agents are transformed via abiotic hydrolysis, with the exception of GD. The simulations indicate that 5% of the initial quantity of GD disposed in the landfill will remain after five years. The V-agents and CS are more persistent, with greater than 50% remaining in the landfill. The dominant fate pathway for these agents is also abiotic hydrolysis, with a very small percentage (less than 1%) being transported via leachate.

42 40 Table 19. Predicted fate routes for the CWAs after 5 years under the arid climate scenario. Chemical Agent Fraction remaining Fate routes (after 5 years) F a F gd F w F λ (abiotic) F diff (liner) Toxic Industrial Compounds Carbon disulfide Furan Blister Agents Distilled Mustard Nitrogen Mustard Lewisite Ethyldichloroarsine Phosgene Oxime Nerve Agents GA (Tabun) GB (Sarin) GD (Soman) GE GF VX VG VM CS

43 41 Table 20. Predicted fate routes for the CWAs after 5 years under the wet climate scenario. Chemical Agent Toxic Industrial Compounds Fate routes (after 5 years) Fraction F F remaining a F gd F λ F diff w (abiotic) (liner) Carbon disulfide Furan Blister Agents Distilled Mustard Nitrogen Mustard Lewisite Ethyldichloroarsine Phosgene Oxime Nerve Agents GA (Tabun) GB (Sarin) GD (Soman) GE GF VX VG VM CS

44 Fdiff (liner) Fλ abiotic Fw Fgd Fa F raction r emain ing 42 CS VM Carbon disulfide Furan Distilled Mustard (HD) Nitrogen Mustard Lewisite Ethyldichloroarsine Phosgene Oxi me GA (Tabun) G B (Sarin) GD (Soman) GE GF VX V G Figure 6. Fate routes for the CWAs a fter five years (arid scenario ) Carbon disulfide Furan Distilled Mustard (HD) Nitrogen Mustard Lewisite Ethyldichloroarsine Phosgene Oxi GA (Ta GB (S GD (Som VG Figure 7. Fate routes for the CWAs after five years (wet scenario). Fdiff (liner) Fλ abiotic Fw Fgd Fa Fraction remaining CS VM me bun) arin) an) GE GF VX

45 43 Results from the thirty year simulations are presented in Tables 21 and 22 and Figures 8 and 9. Results indicate that over this time period, most of the CWAs will have either have been removed from the landfill via advection into the gas collection system or been transformed via abiotic hydrolysis under either climate scenario. The major exceptions are VX and CS (tear gas). VX is persistent relative to the other V-agents since 38% of the VX that was initially disposed of in the landfill will remain after 30 years, with the remainder being transformed via abiotic hydrolysis. Similarly, 45.6% of CS will remain in the landfill, while 54.3% will be transformed via abiotic hydrolysis.

46 44 Table 21. Predicted fate routes for the CWAs after 30 years under the arid climate scenario. Chemical Agent Fraction remaining Fate routes (after 30 years) F a F gd F w F λ (abiotic) F diff ) (liner Toxic Industrial Compounds Carbon disulfide Furan Blister Agents Distilled Mustard Nitrogen Mustard Lewisite Ethyldichloroarsine Phosgene Oxime Nerve Agents GA (Tabun) GB (Sarin) GD (Soman) GE GF VX VG VM CS

47 45 Table 22. Predicted fate routes for the CWAs after 30 years under the wet climate scenario. Chemical Agent Fraction remaining Fate routes (after 30 years) F a F gd F w F λ (abiotic) F diff (liner) Toxic Industrial Compounds Carbon disulfide Furan Blister Agents Distilled Mustard Nitrogen Mustard Lewisite Ethyldichloroarsine Phosgene Oxime Nerve Agents GA (Tabun) GB (Sarin) GD (Soman) GE GF VX VG VM CS

48 Figure 8. Fate routes for the CW As after thirty years (arid sc enar io). Figure 9. Fate routes for the CWAs after thirty years (wet scenario). Fdiff (liner) Fλ abiotic Fw Fgd Fa Fraction remaining 46 CS Carbon disulfide Furan Distilled Mustard (HD) Nitrogen Mustard Lewisite Ethyldichloroarsine Phosgene Oxime GA (Tabun) GB (Sarin) GD ( Soman) GE GF V X VG VM Fdiff (liner) Fλ abiotic Fw Fgd Fa Fr action r emain ing C S Carbon disulfide Furan Distilled Mustard (HD) Nitrogen Mustard Lewisite Ethyldichloroarsine Phosgene Oxim e GA (Tabu n) GB (Sari n) GD (Soman ) GE GF VX VG V M

49 47 Results from the base-case simulations over various time periods indicate that certain CWAs could persist in the landfill, even over long time periods, as presented in Figure 10. Fraction remaining Furan Distilled Mustard (HD) Lewisite GD VX Simula tion time (yr) Figure 10. Fraction of selected CWAs remaining in the landfill as a function of time. Although nearly all of the CWAs have relatively short abiotic hydrolysis rates, the model results predict that some of the agents, primarily the nerve agents that have hydrolysis rates on the order of days, will persist in a landfill for long time periods. This result can be attributed to the model assumption that abiotic hydrolysis takes place only in the aqueous phase. Based on the equilibrium phase fractions predicted by the model, nearly all the chemical is sorbed to the solid phase and unavailable for hydrolysis. The fraction of the chemical that is in the aqueous phase will be quickly hydrolyzed, however th is is only a small fraction of the total amount of the chemical in the landfill.

50 48 The dominant fate route for the majority of the CWAs is transformation via abiotic hydrolysis. Gas phase advection and diffusion were only significant for carbon disulfide and furan. Transport via leachate, diffusion through the soil cover, and diffusion through the composite liner were largely insignificant transport mechanisms for all CWAs. Although transformation of the original agent by hydrolysis does result in significant reductions of the mass of the CWAs, this reaction results in the production of hydrolysis products, some of which may retain some toxicity. Therefore, monitoring the fate of toxic hydrolysis products may be nearly as important as monitoring the fate of the original CWA itself. Because of the importance of abiotic hydrolysis as a transport mechanism for CWAs in landfills, additional work should be done with several of the blister and nerve agents to estimate the fate of the hydrolysis products. Effect of Climate The results for the base-case scenarios presented above indicate that there is very little difference in the predicted fate pathways between the arid and wet climate scenarios. To further evaluate the effect of climate on the fate of the CWAs, a comparison was made between the thirty-year simulation results for the wet climate scenario, the dry climate scenario, and a scenario with infiltration equal to ten times the wet climate scenario (N = 1.2 m/yr) and a gas production rate decay rate of 0.08 yr -1 (q = 10.5 m 3 LFG/m 3 LF yr). Results are presented in Figure 11. a

51 ing Fraction remain Carbon disul fide an Fur V X VG V M CS Arid Scenario Wet scenario 10X Wet Scenario Figure 11. Evaluation of climate scenario on fraction of CWA remaining. Figure 11 shows the fraction remaining for selected CWAs at the end of the thirty-year scenario under three different climate scenarios. In general, there was very little difference among the three scenarios with respect to the fraction of the agent remaining the landfill. The largest differences between the arid and wet scenarios were for the toxic industrial chemicals. This is because the primary transport pathway for these chemicals was advection in the gas phase and this pathway is directly impacted by the change in gas production rate between the wet and arid climate scenarios. For the other agents, the difference between the arid climate scenario and the wet or 10X wet scenario is negligible, which indicates that the fate of chemical agents in landfills located in either wet or arid climates will not be significantly different. As a result, all of the following simulations will use the gas production rate and net precipitation calculated for the wet climate scenario.

52 50 Effect of biotic transformation Because there is little information available on the magnitude of biological transformation of CWAs, it is important to evaluate the potential effects of including biological transformation on the simulation results. In general, biological transformation of CWAs is thought to be limited by the toxicity of these compounds to bacteria. However, since CWAs would likely be present in landfills in relatively low concentrations, biodegradation may occur. To evaluate the impact of biological transformation on the fate of the CWAs, one-year simulations were performed with the following values of biotic half-life (λ biotic ): infinite (no degradation, as in the base-case scenario), 1000 days, 100 days and 10 days. All other MOCLA input parameters remained as in the base-case scenario (wet climate scenario), which are given in Tables 2 and 7 to12. Results of these simulations are presented in Tables 23 to 26.

53 51 Table 23. Predicted fate routes for base-case scenario after one year with no degradation (λ biotic = ). Fate routes (no degradation) Chemical Agent Fraction F λ F λ F F diff remaining a F gd F w (biotic) (abiotic) (liner) Toxic Industrial C ompounds Carbon disulfide Furan B lister Agents Distilled Mustard Nitrogen Mustard Lewisite Ethyldichloroarsine Phosgene Oxime N erve Agents GA (Tabun) GB (Sarin) GD (Soman) GE GF VX VG VM CS

54 52 Table 24. Predicted fate routes for base-case scenario after one year with λ biotic = 1000 days. Chemical Agent Fate routes (λ = 1000 days) biotic Fraction Fλ F F r a F gd F λ F diff emaining w (biotic) (abiotic) ( liner) Toxic Industrial Compounds Furan Blister Agents Distilled Mustard Nitrogen Mustard Lewisite Ethyldichloroarsine Phosgene Oxime Nerve Agents GA (Tabun) GB (Sarin) GD (Soman) GE GF VX VG VM CS

55 53 Table 25. Predicted fate routes for base-case scenario after one year with λ biotic = 100 days. Chemical Agent Fraction remaining F a Fate routes (λ biotic = 100 days) F gd F w F λ (biotic) F λ (abiotic) F diff (liner) Toxic Industrial Compounds Furan Blister Agents Distilled Mustard Nitrogen Mustard Lewisite Ethyldichloroarsine Phosgene Oxime Nerve Agents GA (Tabun) GB (Sarin) GD (Soman) GE GF VX VG VM CS

56 54 Table 26. Predicted fate routes for base-case scenario after one year with λ biotic = 10 days. Chemical Agent Fraction remaining F a Fate routes (λ biotic = 10 days) F gd F w F λ (biotic) F λ (abiotic) F diff (liner) Toxic Industrial Compounds Furan Blister Agents Distilled Mustard Nitrogen Mustard Lewisite Ethyldichloroarsine Phosgene Oxime Nerve Agents GA (Tabun) GB (Sarin) GD (Soman) GE GF VX VG VM CS Results from the MOCLA simulations performed with varying values of biotic half-life indicate that many of the CWAs with short abiotic half-lives are insensitive to changes in biotic half-life over the range of values examined. All of the blister agents and many of the G-agents have abiotic half-lives that are on the order of hours or minutes, therefore, a decrease in biotic half-life from infinity to 10 days does not significantly impact the results of the bounding calculations. Alternatively, decreasing the biotic half-life from infinity to 10 days did have a significant impact on the fate of furan, which was assumed to have an infinite abiotic half-life. When the biotic half-life was decreased to 10 days, the fraction of furan remaining at the end of

57 55 the one-year simulation dropped from 81% to 59%. Carbon disulfide was not included in the simulations with biotic degradation. The carbon in CS 2 is in the +4 oxidation state and is unlikely to be further reduced via anaerobic degradation. The fate of the V-agents and CS (tear gas) was moderately sensitive to the magnitude of the biotic half-life over the range of values tested. Decreasing the biotic half-life from infinity to 10 days decreased the fraction of the V- agents remaining in the landf ill by 3 to 28%. Similarly, the fraction of CS remain ing in the landfill decreased from 97% to 94%. The results of these simulations indicate that the magn itude of the biotic half-life of the CWAs can be significant, especially if the abiotic h alf-life is rel atively long. However, for chemicals with an abiotic half-life on the order of minutes or da ys, any additional transformation due to biotic processes will be negligible over a simulation period of one year. As with abiotic transformation, the model as sumes that biodegradation reactions will take place in the aqueous phase only. Eff ect of Temperature All simulations performed thus far have utilized physical-chemical property data at 25 C, however temperature regimes within landfills with active waste decomposition are often higher than 25 C. To assess the effect of temperature on the fate of CWAs in a landfill, simulations were performed with Henry s law constant and vapor pressure values corrected to 40 C and compared to results from the previous simulations performed at 25 C for a one-year simulation period under wet conditions. Solubility and log K ow values were not temperature corrected. Solubility is not an input parameter to MOCLA because it was assumed that there was no free- phase CWA in the system. Log K ow values were not temperature adjusted because this

58 56 parameter is very insensitive to temperature changes over the ranges of temperature considered (25 C to 40 C) (Sangster 1997). No experimental data for K H and vapor pressure at 40 C could be found in the literature, therefore values of these parameters were estimated using the EPIWIN program. The accuracy of these estimated values was determined by comparing the estimated value of each parameter at 25 C from EPIWIN to the experimental value for each parameter, if available. For most CWAs, there was good agreement between the EPWIN estimate and the experimental value at 25 C, and therefore the EPIWIN estimate at 40 C was used. However, for certain compounds there were significant differences between the EPWIN estimate and the experimental data. In these cases, the EPIWIN estimate at 40 C was not used. Instead, the percent increase between parameter values at 25 C and 40 C was calculated for each compound where the EPIWIN estimate closely matched the experimental data. The average percent increase for each parameter was determined and this value was applied to the available experimental data at 25 C to determine the temperature-corrected parameter value for those compounds for which the EPIWIN prediction was not accurate. The temperature-corrected values for K H and vapor pressure are given in Table 27. Table 28 and Figure 12 present the results of the 25 C simulations. These results were originally presented in Table 18 and Figure 5 and are presented again for ease of comparison. Results of the simulations at 40 C are presented in Table 29 and Figure 13. The equilibrium phase fractions for the CWAs at 40 C are presented in Table 30.

59 57 Table 27 Temperature-corrected values for Henry s Law constant and vapor pressure at 40 C. Chemical Agent K H (unitless) Vapor pressure (Pa) 25 C 40 C 25 C 40 C Toxic Industrial Compounds Carbon disulfide * * Furan * * Blister Agents Distilled Mustard (HD) * Nitrogen Mustard Lewisite * Ethyldichloroarsine * Phosgene Oxime * Nerve Agents GA (Tabun) * GB (Sarin) * GD (Soman) * GE * * GF * * VX VG * * VM * * CS * * *40 C values estimated using EPIWIN.

60 58 Table 28. Predicted fate routes for the CWAs after 1 year under the base-case scenario at 25 C. Fate routes (after 1 year) Chemical Agent Fraction F F F gd F λ F diff a w remaining (abiotic) (liner) Toxic Industrial Compounds Carbon disulfide Furan Blister Agents Distilled Mustard Nitrogen Mu stard Lewisite Ethyldichloroarsine Phosgene Oxime Nerve A gents GA (Tabu n) GB (Sarin) GD (Soman) GE GF VX VG VM CS

61 59 Table 29. Predicted fate routes for the CWAs after 1 year under at 40 C with temperature-corrected chemical parameters. Chemical Agent Fraction remaining Fate routes (after 1 year) F a F gd F w F λ (abiotic) F diff (liner) Toxic Industrial Compounds Carbon disulfide Furan Blister Agents Distilled Mustard Nitrogen Mustard Lewisite Ethyldichloroarsine Phosgene Oxime Nerve Agents GA (Tabun) GB (Sarin) GD (Soman) GE GF VX VG VM CS

62 60 Table 30. Equilibrium phase fractions for the CWAs for the simulations at 25 C and at 40 C with temperature-corrected chemical parameter values. Chemical Agent Phase fractions (25 C ) Phase fractions (40 C ) f a f w f s f a f w f s Toxic Industrial Compounds Carbon disulfide Furan Blister Agents Distilled Mustard (HD) Nitrogen Mustard Lewisite Ethyldichloroarsine Phosgene Oxime Nerve Agents GA (Tabun) GB (Sarin) GD (Soman) GE GF VX VG VM CS

63 Fdiff (liner) Fλ abiotic Fw Fgd Fa Fraction remaining Figure 12. Fate routes for the CWAs after one year under the wet climate scenario at 25 C Figure 13. Fate routes for the CWAs after one year under the wet climate scenario at 40 C 61 Carbon disulfide Furan Distilled Mustard (HD) Nitrogen Mustard Lewisite Ethyldichloroarsine Phosgene Oxime GA (Tabun) GB (Sarin) GD (Soman) GE GF VX VG VM CS Fdiff (liner) Fλ abiotic Fw Fgd Fa Fraction remaining Carbon disulfide Furan Distilled Mustard (HD) Nitrogen Mustard Lewisite Ethyldichloroarsine Phosgene Oxime GA (Tabun) GB (Sarin) GD (Soman) GE GF VX VG VM CS

64 62 Comparison of the simulations performed at 25 C with the simulations performed at 40 C indicate that the temperature-corrected parameter values only impact the fate of carbon disulfide and furan, compounds that are primarily transported via gas-phase advection. The increase in the amount of compound being transported via gas-phase advection at 40 C can be explained by the equilibrium phase fractions given in Table 30. For carbon disulfide and furan, the fraction of the compound in the gas phase at equilibrium (f a ) doubles in the simulations performed at 40 C with temperature-corrected parameter values. An increase in f a will result in more of the compound in the gas phase at equilibrium and therefore, a larger mass will be transported by gas-phase advection over the course of the simulation period. The increased f a value for these two compounds can be directly attributed to the increased value of the Henry s law constant. A comparison of the values of the phase fractions for the other CWAs indicates that these values were not impacted by the increase in Henry s law constant. Although the Henry s law constant increased for all CWAs, the parameter value was several orders of magnitude lower than the values for carbon disulfide and furan. As a result, the relative increase in this value had little effect on the phase fractions for the remaining CWAs. CWA fate simulations under uncertainty All of the bounding calculations performed thus far used point-value estimates for all input variables that represented the average or most-likely value for that parameter. In reality, there is inherent variability in both chemical parameter values as well as the physical parameters describing a landfill. To assess the effect of variability of the input parameter values on the predicted fate pathways for CWAs, Crystal Ball v. 7 was used to perform Monte Carlo simulations on the fate of selected CWAs in a landfill for the base-case scenario under wet conditions after a one year period. Simulations were performed for the following agents: carbon

65 63 disulfide, furan, HD, lewisite, GA, GD and VX. In the Monte Carlo simulations, each input parameter was described by a triangular distribution with a specified maximum, minimum and likeliest (average) value. The range of physical and chemical parameter values used in the simulations are given in Tables 2 and 7 to 13. Latin hypercube sampling was used to select parameter values from the distribution and 1000 realizations were performed for each simulation. Uncertainty in the biotic degradation pathway was not evaluated in the Monte Carlo simulations. In all simulations, the biotic half-life was assumed to be infinite. Based on the lack of experimental data quantifying anerobic biodegradation of CWAs, there was no way to determine reasonable values or ranges for the biotic half-life of the CWAs. In addition, qualitative results from BIOWIN predict that there will be minimal biodegradation of CWAs. Based on the available information, we believe that biodegradation of CWAs will not be a significant fate pathway for CWAs in a landfill. Results of the Monte Carlo simulations for the primary fate pathways for carbon disulfide (Figures 14 and 15), furan (Figures 18 and 19), HD (Figures 22 and 23), lewisite (Figures 26 and 27), GA (Figures 30 and 31), GD (Figures 34 and 35), and VX (Figures 38 and 39) are presented as cumulative probability distributions. In addition, the contribution of MOCLA input parameters to the variance in the results are presented for carbon disulfide in Figures 16 and 17, for furan in Figures 20 and 21, for HD in Figures 24 and 25, for lewisite in Figures 28 and 29, for GA in Figures 32 and 33, for GD in Figures 36 and 37, and for VX in Figures 40 and 41.

66 64 Figure 14. Cumulative probability distribution for the fraction of carbon disulfide remaining in the landfill. Figure 15. Cumulative probability distribution for carbon disulfide transported by gas phase advection.

67 65 Contribution to Variance 80% 60% 40% 20% 0% -20% gas production rate Henry's law constant log Kow bulk density of waste fraction of organic carbon solubility Figure 16. Contribution of MOCLA input parameters to the variance in the fraction of carbon disulfide remaining in the landfill. 20% Contribution to Variance 0% -20% -40% -60% -80% gas production rate Henry's law constant log Kow bulk density of waste fraction of organic carbon air diffusion coefficient Figure 17. Contribution of MOCLA input parameters to the variance in the fraction of carbon disulfide transported via gas phase advection.

68 66 Figure 18. Cumulative probability distribution for the fraction of furan remaining in the landfill. Figure 19. Cumulative probability distribution for fraction of furan transported via gas phase advection.

69 67 20% Contribution to Variance 0% -20% -40% -60% -80% gas production rate Henry's law constant bulk density of waste Log Kow fraction of organic carbon air diffusion coefficient Height of waste Figure 20. Contribution of MOCLA input parameters to the variance in the fraction of furan remaining in the landfill. 80% Contribution to Variance 60% 40% 20% 0% -20% gas production rate Henry's law constant bulk density of waste log Kow fraction of organic carbon air diffusion coefficient Figure 21. Contribution of MOCLA input parameters to the variance in the fraction of furan transported via gas phase advection.

70 68 Figure 22. Cumulative probability distribution for the fraction of distilled mustard (HD) remaining in the landfill. Figure 23. Cumulative probability distribution for the fraction of distilled mustard (HD) transformed via abiotic hydrolysis.

71 69 Contribution to Variance 100% 80% 60% 40% 20% 0% -20% abiotic hydrolysis half-life moisture content of waste log Kow bulk density of waste fraction of organic carbon Figure 24. Contribution of MOCLA input parameters to the variance in the fraction of HD remaining in the landfill. 20% Contribution to Variance 0% -20% -40% -60% -80% abiotic hydrolysis half-life Net precipitation moisture content of waste height of waste gas production rate Figure 25. Contribution of MOCLA input parameters to the variance in the fraction of HD transported via gas phase advection.

72 70 Figure 26. Cumulative probability distribution for the fraction of lewisite remaining in the landfill. Figure 27. Cumulative probability distribution for the fraction of lewisite transformed via abiotic hydrolysis

73 % Contribution to Variance 80% 60% 40% 20% 0% -20% abiotic hydrolysis half-life moisture content of waste log Kow bulk density of waste fraction of organic carbon Figure 28. Contribution of MOCLA input parameters to the variance in the fraction of lewisite remaining in the landfill. 20% Contribution to Variance 0% -20% -40% -60% -80% -100% abiotic hydrolysis half-life moisture content of waste log Kow bulk density of waste fraction of organic carbon Figure 29. Contribution of MOCLA input parameters to the variance in the fraction of lewisite transformed via abiotic hydrolysis.

74 72 Figure 30. Cumulative probability distribution for the fraction of GA remaining in the landfill. Figure 31. Cumulative probability distribution for the fraction of GA transformed via abiotic hydrolysis.

75 73 Contribution to Variance 100% 80% 60% 40% 20% 0% -20% abiotic hydrolysis half-life moisture content of waste bulk density of waste Figure 32. Contribution of MOCLA input parameters to the variance in the fraction of GA remaining in the landfill. 20% Contribution to Variance 0% -20% -40% -60% -80% -100% abiotic hydrolysis half-life moisture content of waste bulk density of waste Figure 33. Contribution of MOCLA input parameters to the variance in the fraction of GA transformed via abiotic hydrolysis.

76 74 Figure 34. Cumulative probability distribution for the fraction of GD remaining in the landfill. Figure 35. Cumulative probability distribution for the fraction of GD transformed via abiotic hydrolysis.

77 75 Contribution to Variance 100% 80% 60% 40% 20% 0% -20% abiotic hydrolysis half-life moisture content of waste log Kow bulk density of waste Figure 36. Contribution of MOCLA input parameters to the variance in the fraction of GD remaining in the landfill. 20% Contribution to Variance 0% -20% -40% -60% -80% -100% abiotic hydrolysis half-life moisture content of waste log Kow bulk density of waste Figure 37. Contribution of MOCLA input parameters to the variance in the fraction of GD transformed via abiotic hydrolysis.

78 76 Figure 38. Cumulative probability distribution for the fraction of VX remaining in the landfill. Figure 39. Cumulative probability distribution for the fraction of VX transformed via abiotic hydrolysis.