Disposal of Activated Carbon from Chemical Agent Disposal Facilities (2009)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

4 Interactions of Chemical Agents with Activated Carbon Application of activated carbon and other adsorbents can also be attracted by functional groups on the sur- for removal of chemical constituents from commercial rounding adsorbent pore walls. These functional groups and industrial gas streams has been a widespread prac- are formed on the surface during the activation process tice for many decades. Carbon use in chemical agent and greatly enhance physical adsorption. They also disposal facilities was discussed in detail in a previous contribute to pore filling with adsorbate molecules at National Research Council report (NRC, 1999). This liquid-like densities. In contrast to physical adsorption, chapter explains some adsorption fundamentals and chemical adsorption involves the formation of chemi- the known reactions of agents on activated carbon. cal bonds between adsorbate molecules and functional These are prerequisites for understanding the chemi- groups on the adsorbent surfaces, interactions that cal fates and levels of agent loadings on carbon. Also often lead to dissociation of the adsorbate molecules. examined in this chapter is the ability to analyze such Such interactions—for example, hydrolysis by water agent loadings. adsorbed on the carbon—are important in determin- ing and understanding the ultimate fate of adsorbed agents. FUNDAMENTALS OF ADSORPTION Both equilibrium processes and rate processes must Adsorption processes generally involve the par- be considered to understand adsorption processes titioning of a chemical solute (such as the agents of (Mattson and Mark, 1971; Weber and DiGiano, 1996). concern here) between the bulk fluid phase (e.g., water Adsorption isotherms quantitatively describe equilib- or air) and the surface of the solid adsorbent material. rium loadings of solutes on solid adsorbents in liquid- Gas-phase applications of adsorption typically involve and gas-phase applications, respectively, as functions physical adsorption (physisorption) and chemical of their liquid-phase concentrations or partial pres- adsorption (chemisorption). Physical adsorption entails sure at a fixed temperature. Equilibrium is a dynamic the attraction of molecules to surfaces via dispersion- phenomenon, involving molecules adsorbing and repulsion forces, termed London–van der Waal forces, desorbing simultaneously at equal rates. Adsorption of and hydrogen bonding. Gas-phase molecules condense mixtures of different chemical vapors is complicated in these force fields and adhesion to the surfaces is by different molecular species competing for available described in terms of Lennard-Jones and electrostatic adsorbent surface sites and the possible replacement potentials (Mattson and Mark, 1971). When the forces of some adsorbed molecules by others that are more involved are relatively weak, the adsorbate (e.g., agent) strongly adsorbed. In some cases, the adsorption of molecules remain intact and are held in close proximity molecules of one chemical species can enhance the to adsorbent surfaces. In microporous adsorbents such adsorption of another—for example, the adsorption of as activated carbon, molecules entering the micropores low molecular weight alcohols is enhanced by adsorbed 33 

34 DISPOSAL OF ACTIVATED CARBON FROM CHEMICAL AGENT DISPOSAL FACILITIES water vapor. Such cooperative mechanisms involving Activated carbon surfaces are generally populated water generally occur only for water-soluble chemicals by oxygen-containing functional groups (e.g., –OH, adsorbed at low loadings. =O, and –COOH). These groups are formed during the Overall rates of adsorption of agents by activated activation process and by exposure to air afterwards, carbon involve both mass transfer and chemical reac- and they are instrumental in both chemisorption-driven tion rates. Mass transfer mechanisms influencing adsorption and adsorbate transformation. This is par- process performance in adsorption beds include (1) ticularly the case for small polar molecule adsorption external mass transfer from the bulk fluid phase pass- at ambient temperatures in moist air. The transforma- ing through the bed to exterior surfaces of adsorbent tion reactions, which include hydrolysis, dissociation, particles contained in the bed, (2) intraparticle mass oxidation, complexation, and acid-base reactions, transfer by fluid-phase diffusion within pore fluids and/ depend in large measure on the molecular properties or adsorbed-phase diffusion along pore wall surfaces, of the adsorbates and the properties of the adsorbent and (3) hydrodynamic axial dispersion of adsorbate precursor and its activation conditions (Bandosz and through the bed within the external fluid phase. The Ania, 2006). length of an adsorption wave front, which is the adsor- These functional groups cause the activated car- bate fluid-phase concentration profile, passing through bon surfaces to exhibit some polarity, which plays a a bed of carbon in a fixed-bed adsorber is characterized specific role in attracting chemical agents containing as the active mass transfer zone (Figure 4-1). The mass oxygen, sulfur, nitrogen, halogens, and phosphorus transfer zone by definition extends from a performance- and in enabling the retention of water. The presence designed maximum allowable fluid-phase concentra- of water in carbon pore systems is crucial for achiev- tion at its furthest depth of penetration into the bed to ing the hydrolysis reactions that occur at pore wall a concentration slightly less than the feed concentration surfaces. Significant quantities of water—loading up near the influent end of the bed (Weber and DiGiano, to 70 weight percent—are adsorbed on virgin carbon 1996). When the wave has passed through the bed to the surfaces when the relative humidity exceeds 50 percent point that the maximum allowable effluent concentra- (McCallum et al., 1999). The principal centers for tion has reached the end of the bed, “breakthrough” of water adsorption are the micropores, in which water the bed with respect to prespecified effluent constraints is attracted to functional groups and/or forms clusters is said to have occurred. by hydrogen bonding, which results in condensa- Mass Transfer Zone Agent Front Unused Bed Inlet Concentration Agent Direction of Gaseous Flow Concentration Breakthrough Concentration Carbon Bed Depth Inlet Outlet FIGURE 4-1  Mass transfer zone in a carbon adsorption bed. The three curves represent the agent front after progressive pe- riods of time. The breakthrough concentration typically represents the maximum acceptable effluent concentration; once the breakthrough concentration has been achieved, the filter has reached the end of its useful protective life and requires changeout. SOURCE: Adapted from Holgate et al., 1993. FIGURE 4-1.eps 

INTERACTIONS OF CHEMICAL AGENTS WITH ACTIVATED CARBON 35 tion (McCallum et al., 1999). Figure 4-2 illustrates pits, and shells of various nuts. The choice of precur- the S-shaped isotherm commonly exhibited by water sor and of activation process determines pore size adsorbed on wood- and coal-based carbons. distributions, surface areas, and surface chemistries of the activated carbon product. The carbon used in Finding 4-1.  Because moisture is always present in the adsorption-based filters at the Chemical Materials the air that continually flows through the carbon beds Agency (CMA) chemical agent disposal facilities is at chemical agent disposal facilities, water is always made from coconut shells. Activated carbons made available on the carbon to hydrolyze adsorbed chemi- from this material typically have more micropores per cal agents. unit mass (volumes of <2 nm3), greater surface areas, and greater crush strengths than carbons produced from The presence of ash in activated carbons is also more common materials such as bituminous coal and known to enhance surface reactivity by causing cata- wood. Moreover, they often have greater adsorption lytic reactions (Hsu and Teng, 2001). Before they capacities for specific adsorbates. are used, activated carbons usually contain from 2 to As discussed in Chapter 2, the large carbon adsorption- 15 percent inorganic matter such as oxides of alkali and based filter units installed at the chemical agent disposal alkaline earth elements, other oxides, aluminum, iron, facilities were manufactured by IONEX Research Cor- and silicon. For many applications, a low level of inor- poration. The units contain several banks of filter trays ganic impurities in activated carbon is desirable. For in series, each containing IONEX 03-001 (formerly other applications, however, higher ash content may be C-800) 8 × 16 mesh coconut shell carbon, trade named beneficial because certain ash constituents may selec- Cocoanut. Each filter tray contains two thin beds of tively chemisorb specific types of metals, inorganic carbon in series. IONEX carbons possess surface areas species, and some synthetic organics, as well as play a of 1,150 m2/g and a bulk density of 520 kg/m3. As beneficial role by catalyzing surface reactivity. designed, agent in the first carbon bank of a multiple- bank system will break through into the next carbon bank in the series after the first bank has been exposed to ADSORPTION OF CHEMICAL AGENTS a quantity of adsorbate sufficient to exceed its adsorption ON HEATING, VENTILATION, AND capacity. In this case, the first bank provides the bulk of AIR CONDITIONING CARBON agent removal and the next bank is said to “polish” the Activated carbons are produced from various precur- effluent. sors, including petroleum residues, coal, wood, fruit As shown in Table 4-1, all three chemical agents— GB, VX, and HD—are adsorbed effectively by coconut shell activated carbon up to about 30 weight percent. However, as discussed below, the agents all react with the moisture on the carbon to form the expected hydrolysis products. In 2007, several carbon samples from the Anniston Chemical Agent Disposal Facility (ANCDF) were analyzed for residual nerve agents GB and VX at government and contractor surety laborato- TABLE 4-1  Agent Loadings on Cocoanut Activated Carbon Relative Humidity Maximum Loading Agent (%) (g agent/g carbon) FIGURE 4-2  Water adsorption isotherms on activated GB Dry 0.318 carbons made from different types of wood (W, W1, and GB 66 0.383 W2) and coals (N, N1, and N2) at 25°C. Relative pressure VX Dry 0.298 is the ratio of the actual pressure over the vapor pressure HD Dry 0.379 at the temperature of measurement and ranges from 0 to 1. SOURCE: Adapted from material from by Susan Ankrom, SAIC Task SOURCE: Reprinted with permission from Bandosz et al., Manager, ANCDF, “Published values for agent loading capacity of MDB 1996. Copyright 1996 American Chemical Society. and PFS carbon,” Presentation to the committee, June 6, 2008. 

36 DISPOSAL OF ACTIVATED CARBON FROM CHEMICAL AGENT DISPOSAL FACILITIES ries. They included (1) carbon samples taken from both pressures and volatilities of the agents are given in Bank 1 and Bank 2 during the changeout on December Table 2-5.) Because the vapor pressure of VX is much 1, 2006, following completion of all GB agent and VX lower than that of GB, only a comparatively smaller rocket and projectile campaigns, (2) a carbon sample amount of VX was ever in the gas phase and available from the pollution abatement system (PAS) filtration to be transported through the heating, ventilation, and system (PFS), and (3) a sample of unused carbon as a air conditioning (HVAC) system to the HVAC filters control. No agent was detected on Bank 2 carbon. and, finally, adsorbed on the carbon filters in any given The analytical results from the U.S. Army Edgewood time interval. The chemistry of the adsorbed agents Chemical and Biological Center (ECBC) are shown in with the water on the carbon is discussed in detail in Table 4-2 (Buettner et al., 2008). The amount of GB the next section. that must have been adsorbed on Bank 1 carbon during The carbon samples from ANCDF discussed above processing of GB munitions in the munitions demilitar- are from a chemical agent disposal facility that uses ization building (MDB) is indicated by the 13 weight incineration technology. At disposal facilities such as percent of its hydrolysis product, isopropyl methyl- the Newport Chemical Disposal Facility (NECDF), phosphonic acid (IMPA), that was found on the carbon VX was instead destroyed by a chemical neutraliza- by solid-state magic angle spinning (MAS) nuclear tion (hydrolysis) process. When a chemical agent is magnetic resonance (NMR). Because the molecular destroyed by neutralization (as was the case at NECDF) mass of GB and IMPA are approximately the same and instead of incineration, hydrolysis products are formed one mole of IMPA is produced for each mole of GB, (none are formed by incineration). At NECDF, the the mass percent of IMPA is approximately equivalent HVAC carbon had been exposed to the neutralization to the amount of GB to which the carbon was exposed. reactor venting system for the duration of NECDF The MAS NMR method was not sensitive enough to disposal operations, so volatile hydrolysis and thermal detect trace amounts of GB if any remained. degradation (90°C) products of VX were adsorbed on In comparison, only a trace amount of the VX hydro- the HVAC carbon in addition to VX itself. lysis product, ethyl methylphosphonic acid (EMPA), Extractive analysis of HVAC carbon samples from was found on carbon from Bank 1. This small amount NECDF by a Southwest Research Institute (SwRI) lab- of the hydrolysis product is attributed to the low oratory indicated the presence of volatile VX impuri- volatility of VX in the ambient air stream. (The vapor ties, hydrolysis by-products, and degradation products TABLE 4-2  Analytical Results of HVAC and PFS Carbon Samples Collected from ANCDF in January 2007 HVAC Bank 1 HVAC Bank 2 PFS Carbon New Carbon Headspace vapor analysis GB < 1.5 × 10–5 mg/m3 Same as Bank 1 Not analyzed Not analyzed VX < 5.1 × 10–7 mg/m3 Thermal desorption followed GB < 1.0 × 10–4 mg/m3 Same as Bank 1 Not analyzed Not analyzed by GC/MS/FPDa VX < 1.0 × 10–4 mg/m3 Solid-phase NMR, MAS 31P or GB < 1,500 ppm No phosphorus compound Relatively large Relatively small MAS 1H VX < 1,500 ppm was detected at a detection water peaks water peaks IMPA = 13 wt percentb limit of 1,500 ppm EMPA = Nondetectb Solvent extraction followed by GBc Not analyzed Not analyzed Not analyzed GC/MS VX < 20 ppb Remaining filter capacityd 12 percent 100 percent Not analyzed 100 percent (control) aThermal desorption up to 100°C and 1 L processing volume. FPD, flame photometric detection. bThe sample was also extracted in CD3CN to resolve the IMPA peak in the 31P NMR MAS spectra. NMR analysis of the liquid extract gave 92 percent IMPA, 7 percent MPA (methyl phosphonic acid, CH3P(O)(OH)2), and a trace of EMPA. cAnalytical procedures and results for GB are being revised and validated. dAdsorption capacity was determined by conducting DMMP (dimethyl methylphosphonate, CH3P(O)(OCH3)2) breakthrough tests at 3,000 mg/m3 DMMP concentration and 0.016 m3/min flow rate. SOURCE: Adapted from Buettner et al., 2008. 

INTERACTIONS OF CHEMICAL AGENTS WITH ACTIVATED CARBON 37 of the aminothiol group. These VX-related adsorbates Recommendation 4-1.  Banks 3-6 at the Anniston were extracted into solvent and became potential inter- Chemical Agent Disposal Facility have not been ferences for the gas chromatography (GC)/mass spec- exposed to agent and can be disposed of using proce- trometry (MS) analysis that was conducted. However, dures for unexposed carbon. analysis of the extracts by GC/MS indicated the VX to be below 95 parts per billion (ppb). REACTIONS OF CHEMICAL AGENTS At the time this report was being prepared, no MDB ON ACTIVATED CARBON HVAC carbon sample exposed to distilled mustard agent HD from a chemical agent disposal facility had As mentioned above, it has been widely reported that been available for laboratory analysis. Of the currently adsorbed chemical agents GB, VX, and HD degrade operating facilities, only the Tooele Chemical Agent on activated carbon with time (Brevett et al., 1998; Disposal Facility (TOCDF) had destroyed mustard Karwacki et al., 1999; Wagner et al., 2001; McGarvey munitions by incineration, but the Bank 1 and Bank 2 et al., 2003; Columbus et al., 2006). Degradation gen- carbon filter units had not yet been removed. erally increases as the relative humidity of the vapor phase increases. The time to reach 50 percent degrada- Finding 4-2.  The level of GB degradation product tion ranges from days to weeks at ambient tempera- found on the Anniston Chemical Agent Disposal tures. Most of the studies used indirect thermal desorp- Facility munitions demilitarization building heating, tion GC/MS methods for measuring the desorbed agent ventilation, and air conditioning system Bank 1 filter concentrations in the vapor phase (Karwacki et al., samples demonstrates that the filter had been exposed 1999). However, hydrolysis products and intermedi- to high levels of volatile GB, had adsorbed the GB, ates of the reactions of these agents on carbon are and that all or most of the GB had hydrolyzed to its usually ionic compounds, which are not detectable by degradation products. GC. Starting in the 1990s, the reactions of agents on carbon have also been investigated by solid-state MAS Finding 4-3.  Based on the analytical results of the NMR techniques that can identify and quantify agents changeout of the Anniston Chemical Agent Disposal and agent reaction products on the surfaces of carbon Facility munitions demilitarization building heating, directly and simultaneously. The following sections ventilation, and air conditioning system Bank 1 carbon summarize these direct MAS NMR observations of filter, very little VX degradation product, ethyl methyl- agent reactions on wet carbon. phosphonic acid, was detected on the carbon sample. Therefore, it can be concluded that very little of the GB Reactions low-volatility VX was transported by the heating, ven- tilation, and air conditioning system and adsorbed on Figure 4-3 shows 31P MAS NMR spectra from a the carbon during the processing of the VX munitions recent study. A reaction-time profile is revealed for at Anniston Chemical Agent Disposal Facility. 10 weight percent GB on wet coconut shell carbon containing 13 weight percent water at room tem- Finding 4-4.  Chemical agent has not been observed perature. In the initial spectrum, only the doublet GB beyond the munitions demilitarization building heat- peaks (δP = 27.5 and 18.8 ppm due to P-F splitting, ing, ventilation, and air conditioning system Bank 2 JPF = 1046 Hz) were observed. Spectra taken at 6, 13, filter units at the Anniston Chemical Agent Disposal and 16 days show a decrease in the GB peaks and the Facility. Furthermore, the changeout of Bank 2 on emergence of an IMPA peak at δP = 20.5 ppm (IMPA December 1, 2006, showed that it had retained adsorp- is the main hydrolysis product of GB). At 16 days, only tion capacity equivalent to new (unused) carbon, indi- a small amount of GB (the shoulder on the main IMPA cating that Bank 2 was not exposed to any significant amount of agent. Leonard Buettner, John Mahle, George Wagner, Tara Sewell, and Nicole Fletcher, all of the U.S. Army Edgewood Chemical and Biological Center, and David Friday, Houston Advanced Research Brian O’Donnell, Chief, Secondary Waste, Closure Compliance, Center, “Adsorbent analysis of Anniston Chemical Agent Disposal and Assessments, CMA, “NECDF carbon shipment decision,” Facility MDB Bank 1 and Bank 2 filter samples,” Presentation to Presentation to the committee, July 24, 2008. the committee, July 23, 2008. 

38 DISPOSAL OF ACTIVATED CARBON FROM CHEMICAL AGENT DISPOSAL FACILITIES FIGURE 4-3  Phosphorus-31 magic angle spinning (MAS) nuclear magnetic resonance (NMR) spectra of 10 wt percent sarin (GB) on humidified (13 weight percent water) activated carbon over time: initial and after 6, 13, and 16 days, left to right. FIGURE 4-3.eps SOURCE: Leonard Buettner, John Mahle, George Wagner, Tara Sewell, and Nicole Fletcher, all of the U.S. Army Edgewood bitmap Chemical and Biological Center, and David Friday, Houston Advanced Research Center, “Adsorbent analysis of Anniston Chemical Agent Disposal Facility MDB Bank 1 and Bank 2 filter samples,” Presentation to the committee, July 23, 2008. IMPA, isopropyl methylphosphonic acid. peak) remained. GB is known to react with water to In developing the current solvent extraction method form IMPA by the following equation: for the Anniston carbon samples, the detection of GB at a concentration in excess of the amount that was CH3P(O)(OiC3H7)F + H2O → expected on the carbon—in effect, a false positive—was GB similar to what Rohrbaugh et al. (2006) had observed in analyzing aqueous acidic samples composed of GB CH3P(O)(OiC3H7)OH + HF (1) hydrolysis products. Furthermore, with an improved IMPA analytical method, GB was detected in the GB hydro- lysate when the pH of the hydrolysate was adjusted It should be noted that the initial GB hydrolysis reac- to below 5 (Malloy et al., 2007)). Accordingly, when tion was relatively rapid because it was base-catalyzed developing analytical methods for exposed carbon that (samples of new unused carbon added to deionized would allow it to be cleared at the waste control limit water gave pH readings of around 10). As more GB (WCL) or the permit compliance concentration (PCC) was adsorbed and hydrolyzed, the pH of the adsorbed level, it is important to avoid conditions that can cause phase on the carbon was reduced and the acid-catalyzed the false positive detection of GB. As GB on the carbon hydrolysis rate was slower than the base-catalyzed hydrolyzes, the pH decreases because the degradation hydrolysis rate. Consequently, the hydrolysis of GB on products are acids, and the rate of hydrolysis becomes carbon decreased as the carbon became more acidic. slower. The Army interprets this as a sign of re-forma- According to analyses by SwRI, the Anniston exposed tion. At the time this report was being written, it was Bank 1 samples gave pH readings around 3.0 when the not at all clear to the committee from the available data carbon was added to deionized water. whether re-formation was in fact occurring. Finding 4-5.  The degradation of GB on carbon pro- The pH measurements were provided in a personal communi- duces isopropyl methylphosphonic acid and hydro- cation between Matthew Blais, SwRI, and the committee, March fluoric acid. In aqueous solutions and at a pH of less 17, 2009. than 5, these compounds may slow the rate at which 

INTERACTIONS OF CHEMICAL AGENTS WITH ACTIVATED CARBON 39 GB hydrolyzes, as previously reported by Malloy et al. close to 0 and 24 days clearly demonstrated that VX (2007). Under such conditions, GB on carbon may not reacted rapidly on the wet carbon. degrade completely. The degradation of VX occured through an autocata- lytic chain. After a small amount of the initial hydro- lysis product, EMPA, was produced, the VX primarily VX Reactions reacted with the EMPA to give the diphosphonate As illustrated in Figure 4-4, the reaction of 10 weight compound VX-pyro (equation 2) as the only observed percent VX on wet coconut shell carbon containing 13 initial product. The VX-pyro, an anhydride of EMPA, weight percent water was also monitored by 31P MAS subsequently reacted with water adsorbed on carbon NMR at room temperature (Karwacki et al., 1999; to produce more EMPA (equation 3), which reacted Wagner et al., 2001). In the initial spectrum, as VX with the remaining VX to form more VX-pyro. Thus, was added to the carbon, the broad VX peak at δP = an autocatalytic chain reaction was propagated. Also, 49.4 ppm was reduced and replaced by a sharp major because the hydrolysis of VX-pyro (equation 3) was product peak at δP = 16.4 ppm, which was identified much slower than its production, VX-pyro accumu- as diethyl dimethylpyrophosphonate (VX-pyro). The lated as the main product during the 24-day monitor- minor broad shoulder peak at δP = 20 ppm was pro- ing period. These observations are consistent with the duced by the hydrolysis product EMPA (see equations rates and mechanisms of VX reacting with less than 2 and 3). In the final spectra, taken at 24 days, VX 10 weight percent water in the bulk organic VX phase disappeared and the VX-pyro peak increased signifi- (Yang et al., 1996). It can therefore be concluded that cantly. The toxic hydrolysis product EA-2192 was not on carbon, the main reaction occurred in the adsorbed detected. This pair of spectra recorded at reaction times VX phase, in which only a small amount of water was present. FIGURE 4-4  MAS NMR spectra for 10 weight percent nerve agent VX absorbed on humidified (13 weight percent water) carbon, left to right: initial and at 24 days showingFIGURE 4-4.eps heterogeneous autocatalytic hydrolysis of VX over 24 days, left to right. SOURCE: Leonard Buettner, John Mahle, George Wagner, Tara Sewell, and Nicole Fletcher, all of the U.S. Army Edgewood bitmap Chemical and Biological Center, and David Friday, Houston Advanced Research Center, “Adsorbent analysis of Anniston Chemical Agent Disposal Facility MDB Bank 1 and Bank 2 filter samples,” Presentation to the committee, July 23, 2008. EMPA, ethyl methylphosphonic acid. 

40 DISPOSAL OF ACTIVATED CARBON FROM CHEMICAL AGENT DISPOSAL FACILITIES CH3P(O)(OC2H5)(SCH2CH2N(iC3H7)2) + for the VX concentration to degrade below the waste VX control limit or permit compliance concentration. CH3P(O)(OC2H5)OH EMPA Mustard Agent Reactions → [CH3P(O)(OC2H5)]2O + The 13C MAS NMR from carbon in normal isotopic VX-pyro abundance is not sufficiently sensitive to detect mus- tard agent HD and its degradation products on carbon HSCH2CH2N(iC3H7)2 (2) directly. However, as reported by Karwacki et al. (1999), 13C MAS NMR has been used to examine the reaction [CH3P(O)(OC2H5)]2O + H2O → of a 13C-enriched HD compound (0.1 g/g loading) on VX-pyro wet coconut shell carbon containing 13 weight percent water. As the 13C-enriched HD peaks decreased, the 2 CH3P(O)(OC2H5)OH (3) hydrolysis products mustard chlorohydrin (CH), thiodi- EMPA glycol (TG), and sulfonium ion (CH-TG) were detected (see Table 4-3). The reported reaction was slow, with After most of the VX is converted to VX-pyro, 71 percent of the HD remaining on the carbon after 115 which is soluble in water, this product will continue days at 30°C. In comparison, reaction of the same 13C- to react with the water adsorbed on carbon surfaces enriched HD on BPL carbon containing 38.8 weight to eventually give EMPA (equation 3) as the final percent water was complete in less than 24 hours at product. This was confirmed in 2006 by a study of the 50°C (McGarvey et al., 2003). The rate of degradation reaction of VX adsorbed on a range of carbon samples thus appears to be markedly dependent on the type of (Columbus et al., 2006). The authors reported that the carbon used and the temperature. VX reaction was complete in less than 20 days. When When adsorbed HD reacts on wet carbon, the main the final carbon sample was extracted in ethanol, EMPA products are TG, hydrochloric acid, and a range of was detected by 31P NMR as the only phosphorus- branched sulfonium ions produced from the reaction containing product. of HD with TG. The production of the sulfonium ions It should be noted that VX dissolves in acidic water indicates that HD degradation reactions can occur in but the protonated VX does not react with water under mixtures with high HD to water ratios. acidic pH (Yang, 1999). The above VX degradation With the same 13C-enriched HD loaded onto wet reaction with water occurs in pH ranges close to neu- carbon fibers, Brevett et al. (1998) identified another tral—from weakly acidic to weakly basic. If the water linear sulfonium ion, H-2TG, produced from the reac- adsorbed on the carbon sample becomes strongly tions of HD and TG (see Table 4-2). These sulfonium acidic, adsorbed VX may dissolve readily in the acidic ions do not contain 2-chloroethyl groups and therefore water in the pores, and the protonated VX in the water do not have vesicant properties. The authors reported phase will neither hydrolyze nor react via the above that the rate of HD degradation on these carbon samples autocatalytic reaction mechanism (Yang, 1999). was much faster than the degradation times reported by Karwacki et al. (1999). The Brevett results showed that Finding 4-6.  Although VX is barely soluble in water, after about 6 weeks at room temperature, most of the it reacts to form water-soluble VX-pyro, which then HD was converted to TG and H-2TG. reacts in the water phase to produce ethyl methylphos- The formation of sulfonium ions CH-TG and H-2TG phonic acid. The initial degradation of VX on carbon indicates that TG was present in the adsorbed water follows an autocatalytic hydrolysis mechanism that phase and was able to compete with water in reacting occurs exclusively in the bulk VX phase. Ethyl methyl- phosphonic acid is the only phosphorus-containing final product. The toxic EA-2192 hydrolysis product The CH-TG was detected only after the carbon sample had been has not been detected. extracted into a solvent. BPL, a trademarked product of Calgon Carbon Corporation, is a bituminous coal-based granular activated carbon. Recommendation 4-2.  The Chemical Materials Note that 50°C would not be a practical temperature for HVAC Agency should determine the length of time required air at a chemical agent disposal facility. 

INTERACTIONS OF CHEMICAL AGENTS WITH ACTIVATED CARBON 41 TABLE 4-3  Chemical Formulas for Mustard Agent and Its Finding 4-8.  Analytical studies have shown that sulfo- Hydrolysis Products nium ions interfere with the analysis of trace amounts Abbreviation Chemical Formula of mustard in aqueous solutions. HD ClCH2CH2SCH2CH2Cl CH ClCH2CH2SCH2CH2OH Recommendation 4-4.  The Army should establish TG HOCH2CH2SCH2CH2OH definitively whether or not the presence of sulfo- CH-TG HOCH2CH2SCH2CH2S+(CH2CH2OH)2 H-TG ClCH2CH2SCH2CH2S+(CH2CH2OH)2 nium ions interferes with the analysis of mustard on H-2TG (HOCH2CH2)2S+CH2CH2SCH2CH2S+(CH2CH2OH)2 carbon. Summary of Studies of Agent Reactions on Carbon As discussed above, direct MAS NMR measurements with insoluble HD at the HD-water interface. Sulfo- have shown that all three agents (GB, VX, and mustard) nium ions have frequently been observed in two-phase are unstable on wet carbon surfaces and degrade with liquid mixtures of HD and water at relatively high HD time. In the case of water-soluble GB, IMPA was the to water ratios (Yang et al., 1988). Once produced, only phosphorus-containing compound detected. For they are soluble and relatively stable in water but react the sparingly soluble VX and practically insoluble more rapidly if NaOH is present (Yang et al., 1988). In HD, more complicated products were obtained from previous studies sponsored by the Aberdeen Chemical reactions with water and the initial hydrolysis prod- Agent Disposal Facility, these sulfonium ions present ucts. Given the relatively high concentrations needed in acidic aqueous solutions made it difficult to prove for detection (approximately 0.1 to 1.0 percent g/g) that batches of HD hydrolysate had the required levels by solid-state MAS NMR, it is uncertain whether the of agent destruction. It is believed that if sufficient HCl reactions continue to completion or trace amounts of is present during solvent extraction and subsequent agent remain on the carbon samples. To determine GC/MS analysis, some of the major sulfonium ions ppb levels of detection of residual agents on carbon, such as CH-TG may decompose and react with chloride other analytical approaches, such as extraction of ion to form HD. The hydrolysate batches were found to the adsorbed phase from the carbon sample, possibly contain less than 20 ppb HD only after aqueous NaOH followed by GC/MS detector analysis, are required. solutions were added to the acidic hydrolysates. It may During this type of analysis, caution must be exercised therefore be that these sulfonium ions and hydrochloric to verify that agent is not re-formed during the analyti- acid also cause problems in clearing carbon samples to cal process, particularly when the carbon samples are the required WCL or PCC level. acidic. Based on the MAS NMR measurements of the initial stages of degradation, the committee estimates, Finding 4-7.  Experimental data indicate that it takes by extrapolation, that at room temperature it will take weeks to months at room temperature for mustard on approximately 1 month following removal of the car- carbon to degrade. bon from the filter units for GB and VX to degrade to minimal levels and several months for mustard to do the Recommendation 4-3.  The Chemical Materials same. Of course, these reactions proceed even while the Agency should determine the rate of degradation of carbon is in place during disposal operations. mustard on carbon under controlled constant conditions with greater accuracy in order to predict if and when the Finding 4-9.  The chemical agents VX, GB, and concentration of mustard on carbon will be minimal, mustard all degrade on wet carbon, with VX and GB or below the waste control limit or permit compliance degrading faster than mustard. Increasingly rapid rates concentration. For example, any acceptable analytical of agent decomposition occur with increasing tem- method needs to be verified by a high level (about 80 perature and humidity. Also, the final concentrations percent or more) of agent spike recovery. of agents on the carbon are dependent on the pH of the water adsorbed on the carbon. Under acidic conditions, the hydrolysis of GB is equilibrium controlled, so the Although the WCL is 200 ppb, a value of 20 ppb was used to GB may not degrade to the waste control limit or permit ensure destruction to a 99.9999 percent level. compliance concentration. 

42 DISPOSAL OF ACTIVATED CARBON FROM CHEMICAL AGENT DISPOSAL FACILITIES Recommendation 4-5.  The Chemical Materials isotherm is known, then the adsorbed-phase concentra- Agency should determine the time necessary for each tion or loading can be determined from the gas-phase of the three agents (GB, VX, and mustard) to degrade concentration. Three issues associated with the use of on carbon to minimum values and determine if that headspace analysis must be considered to achieve a value is below the respective waste control limits or reliable analysis of agent loading on carbon. First, the permit compliance concentrations. Given that mustard gas-phase concentration of agent that would be in equi- is the last agent scheduled for processing at chemical librium with an agent loading of 20 ppb at ambient, and agent disposal facilities, it should be determined if the even moderately elevated temperatures could be unde- slow rate of degradation of HD on carbon will impact tectable by headspace analysis. Second, the adsorp- the schedule for facility closure. tion isotherm would be needed to correlate loadings with gas-phase concentrations at agent loadings near 20 ppb. Third, a pure-component adsorption isotherm METHODS FOR DETERMINING CHEMICAL would not even apply to the real system, which would AGENT LOADING ON ACTIVATED CARBON contain coadsorbed amounts of other components, such The shipping of agent-exposed carbon to off-site as water and degradation products. disposal facilities will require a determination of the As mentioned earlier, several carbon samples from loading of agent(s) on the carbon on a mass basis (mass ANCDF were analyzed for residual GB and VX at of agent per total mass of carbon and all adsorbates). both government and contractor surety laboratories in The levels adopted in the Resource Conservation 2007 by extracting agent into a solvent. The samples and Recovery Act permits for the Pine Bluff Chemi- included (1) a carbon sample from the changeout cal Agent Disposal Facility (PBCDF), ANCDF, and on December 1, 2006, of Banks 1 and 2 of HVAC TOCDF are 20 ppb for GB and VX and 200 ppb for filter unit 102 following completion of all GB agent mustard, where the total mass considered is the adsor- and VX rocket and projectile campaigns, (2) a PFS bent plus all adsorbates, including water and hydroly- carbon sample, and (3) an unused carbon sample as sis products. The levels that have been adopted at the a control. Again, no detectable agent was found on Umatilla Chemical Agent Disposal Facility (UMCDF) Bank 2 carbon. are 16 ppb for GB and 13 ppb for VX, which are the The HVAC Banks 1 and 2 carbon samples were practical quantitative limits. analyzed at SwRI, first by solvent extraction (Envi- Because each bank in the MDB HVAC filter units ronmental Protection Agency [EPA] SW-846 Method contains 48 filter trays with two layers of carbon in 3571) and then by GC/MS (EPA SW-846 Method each tray, a statistically reliable method is needed 8271) to determine the levels of GB and VX remain- for the selection of a few trays from the location of ing on the carbon. This method showed the results for highest flow in each bank. This sampling approach VX to be valid and below the WCL or PCC, but the would be sufficient to determine the maximum load- results for GB on the carbon were unexpectedly high. ing of agent on the carbon in each bank. At ANCDF As described earlier in this chapter, GB can react with it was decided to sample at several different locations the water on the carbon to form the hydrolysis products within the trays and then mix the samples to produce IMPA and hydrofluoric acid (HF). In a previous study a homogenized standard sample, an approach found to (Malloy et al., 2007), these products were assumed be acceptable. to be able to react to re-form GB in GB hydrolysate Headspace vapor analysis provides a valid measure when the pH of the hydrolysate samples was adjusted of the inhalation threat from agents. However, to use to below 5. For carbon samples containing these prod- this method to accurately measure agent loading on ucts, the re-formation is believed to occur in the solvent carbon requires measurement of the gas-phase concen- after extraction. Thus, without modification, EPA SW- tration in equilibrium with the carbon and also requires 846 Method 3571 was not able to determine the actual knowledge of the adsorption isotherm for that agent concentration of GB on the carbon at the 20 ppb level. under relevant conditions. In principle, if the adsorption The PFS carbon sample was analyzed not by SwRI but rather Matthew Blais, SwRI, “Carbon analysis for GB,” Presentation by ECBC; however, SwRI did analyze a sample from HVAC Banks to Robert Beaudet and Yu Chu Yang, committee members, January 1 and 2 via NMR, with results indicating less than 1,500 ppm for 13, 2008. GB and VX. 

INTERACTIONS OF CHEMICAL AGENTS WITH ACTIVATED CARBON 43 Since August 2008, CMA has undertaken to enhance TABLE 4-4  Analysis of GB and VX on Carbon and this method to prevent or minimize GB re-formation Method Detection Limits (MDLs)a by modifying the conditions under which the extrac- Type of Agent tion is performed. A working group consisting of staff Carbon Detected Agent Sample Analytical Method MDL (ppb) (ppb) from SwRI, Battelle, ECBC, and the laboratories at ANCDF, UMCDF, and PBCDF was formed to address GBb IMPA- and SwRI modified 5 (SwRI) — HF-loaded extraction method 4 (ECBC) this problem, and the experimental work to modify new, unused 3571 (being Method 3571 was performed at SwRI. Because the carbon validated) work was not complete when this report was being ANCDF Same as above — 129 written and the progress report was made available to Bank 1, the committee only after it had completed its fact find- standard carbon batch ing, the committee was not able to evaluate whether VX New, unused SwRI extraction 4 — the modified method could validate that the carbon carbon method 3571 for VX was below the WCL for GB. However, the committee for ANCDF reports the preliminary findings of the SwRI working ANCDF Same as above — 17 group below. Bank 1 Unlike previous analyses of Bank 1 samples, a stan- carbon dard sample from the ANCDF Bank 1 filter was pre- New, unused SwRI extraction 14c — pared at SwRI. This standard sample was a well-mixed carbon method 3571 for VX for NECDF composite sample of carbon samples taken from the Bank 1 filter at different locations. Three replicates of NECDF Same as above — 80 Bank 1 this sample were analyzed for GB, and the results were carbon closely similar, indicating that the standard sample aNo MDL for mustard on carbon had been established at the time this was indeed a homogeneous sample and representative report was prepared. of the Bank 1 filter. This step ensures experimental bThe new, unused, loaded carbon was spiked with GB. reproducibility. cA different carbon mass was used for the NECDF MDL determina- tion. Re-formation was minimized by careful choice of SOURCE: Personal communication between Michael MacNaughton, SwRI, solvent (dichloromethane), pH control (use of a pH = 7 and Robert Beaudet, committee chair, January 15, 2009. buffer), a longer extraction time (30 minutes), and the addition of 1.0 M calcium nitrate, which sequesters fluoride ions. results indicate that the MDL is about 5 ppb, but this SwRI has determined the method detection limit value has not yet been validated.10 (MDL) for this procedure to be 5 ppb. It did this by Other methods could be investigated to determine loading unused carbon with IMPA/HF to simulate the concentration of agents on carbon if the extrac- Bank 1 carbon, spiking it with GB, and then analyz- tion method cannot be validated to show acceptable ing it. recovery of agent. However, direct methods involving measuring concentrations in the adsorbed phase are Finding 4-10.  The method detection limit of 5 ppb not sensitive to low levels of agent. Indirect methods for GB obtained by Southwest Research Institute from involving the removal of the adsorbates from the carbon new unused carbon samples loaded with isopropyl and analyzing them separately could be investigated: methylphosphonic acid and hydrofluoric acid is the best These include chromatography, solid-liquid extraction, indication of the extent of re-formation for the current or thermal/vacuum desorption (LeVan and Carta, 2007). recommended analytical procedures. Of the chromatographic methods—specifically, elution and displacement chromatography—displacement The modified Method 3571 is being validated at ECBC and the laboratories at ANCDF, UMCDF, and PBCDF. When this modified method is applied to the 10Matthew Blais, SwRI, “Carbon analysis for GB,” Presentation ANCDF Bank 1 sample, it shows that the concentration to Robert Beaudet and Yu Chu Yang, committee members, January of GB on the carbon is 129 ppb (see Table 4-4). Early 13, 2008. 

44 DISPOSAL OF ACTIVATED CARBON FROM CHEMICAL AGENT DISPOSAL FACILITIES chromatography would provide the most conclusive Brevett, C., B. MacIver, K. Sumpter, and D. Rohrbaugh. 1998. SSMAS NMR Study of HD, GD, and VX in Carbon Filter Textiles for Wipes, measurement of total residual agent. In displacement ECBC-TN-035. Aberdeen Proving Ground, Md.: Edgewood Chemical chromatography, a sample of the exposed carbon could and Biological Center. be placed in a small column, probably after grind- Buettner, L., J. Mahle, G. Wagner, T. Sewell, and N. Fletcher. 2008. Ad- sorbent Analysis of Anniston Chemical Agent Disposal Facility MDB ing, and a strongly adsorbed solvent (the “displacer”) Bank-1 and Bank-2 Filter Samples Following Completion of the GB passed through the column. The displacer is adsorbed Agent and VX Rocket Campaign, ECBC-TR-XXX. Aberdeen Proving more strongly than any of the adsorbed agent and will Ground, Md.: Edgewood Chemical and Biological Center. displace the agent from the carbon, so that the entire Columbus, I., D. Waysbort, L. Shmueli, I. Nir, and D. Kaplan. 2006. De- composition of adsorbed VX on activated carbons studied by 31P MAS adsorbed phase (all desorbed agent plus some of the NMR. Environmental Science and Technology 40(12): 3952-3958. displacer solvent) is eluted from the column. A liquid Holgate, H., L. Scherer, and A. Talib. 1993. Assessment of Carbon Filter chromatography apparatus can be used for this purpose System Performance. MTR 93W0000034. McLean, Va.: MITRE Cor- poration. to provide the desired low flow rates. The extract from Hsu, L., and H. Teng. 2001. Catalytic NO reduction with NH3 over carbons the column (adsorbates plus displacer solvent) could modified by acid oxidation and by metal impregnation and its kinetic be analyzed to determine the agent concentration on studies. Applied Catalysis B: Environmental 35(1): 21-30. Karwacki, C., J. Buchanan, J. Mahle, L. Buettner, and G. Wagner. 1999. the exposed carbon. This can be done either temporally Effect of temperature on the desorption and decomposition of mustard during the chromatographic displacement process by from activated carbon. Langmuir 15: 8645-8650. measuring time-dependent concentrations as the agent LeVan, M., and G. Carta. 2007. Adsorption and ion exchange. Perry’s is eluted, provided that the detector is sufficiently Chemical Engineers’ Handbook, 8th edition. D.W. Green and R.H. Perry, eds. New York, N.Y.: McGraw-Hill Professional. sensitive, or it can be performed on aliquot portions of Malloy, T.A., L. Dejarme, C. Fricker, J. Guinan, G. Lecakes, and A. Shaffer. the extract (or on the entire extract) after completion 2007. Bench-Scale Evaluation of GB Hydrolysis, TRRP #02a Phase II, of the process. Thus, with the measured agent effluent Test report, Rev. 0. Aberdeen Proving Ground, Md.: Program Manager for Assembled Chemical Weapons Alternatives. concentration passing through a maximum and declin- Mattson, J., and H.B. Mark, Jr. 1971. Activated Carbon: Surface Chemistry ing to zero, it can be determined positively by displace- and Adsorption from Solution. New York: Marcel Dekker, Inc. ment chromatography that the agent has been entirely McCallum, C., T. Bandosz, S. McGrother, E. Muller, and K. Gubbins. 1999. A molecular model for adsorption of water on activated carbon: Com- removed from the exposed carbon. parison of simulation and experiment. Langmuir 15(2): 533-544. Another alternative for indirect analysis, solid-liquid McGarvey, D., J. Mahle, and G. Wagner. 2003. Chemical Agent Hydrolysis extraction (also referred to as leaching), requires care- on Dry and Humidified Adsorbents, ECBC-TR-334. Aberdeen Proving ful application. As with liquid-liquid extraction, the Ground, Md.: Edgewood Chemical and Biological Center. NRC (National Research Council). 1999. Carbon Filtration for Reducing process would normally be performed with multiple Emissions from Chemical Agent Incineration. Washington, D.C.: Na- contacts in a batchwise mode to make sure that essen- tional Academy Press. tially all residual agent has been extracted from the Rohrbaugh, D., G. Hondrogiannis, and Y.C. Yang. 2006. Analytical Method and Detection Limit Studies for Detection of GB in GB Hydrolysate, carbon for analysis. As with displacement chromatog- ECBC-TR-509. Aberdeen Proving Ground, Md.: Edgewood Chemical raphy, the best solvent would be one that is adsorbed and Biological Center. more strongly than the adsorbates, so that essentially Wagner, G., B. MacIver, C. Karwacki, J. Buchanan, and D. Rohrbaugh. 2001. Fate of Mustard on Activated Carbons, Part 2: 13C MAS NMR the entire adsorbed phase is removed efficiently in no Studies, ADE491596. Aberdeen Proving Ground, Md.: Edgewood Re- more than a few batches. Thermal desorption is another search, Development and Engineering Center. indirect method, one that has been used by ECBC as Weber, W., Jr., and F. DiGiano. 1996. Process Dynamics in Environmental mentioned above. However, heating a sample may Systems. New York: Wiley InterScience. Yang, Y. 1999. Chemical detoxification of nerve agent VX. Accounts of cause thermal decomposition, and a method using a Chemical Research 32(2): 109-115. vacuum and capture in liquid nitrogen may prove more Yang, Y., L. Szafraniec, W. Beaudry, D. Rohrbaugh, L. Procell, and J. promising. Samuel. 1996. Autocatalytic hydrolysis of V-type nerve agents. Journal of Organic Chemistry 61(24): 8407-8413. Yang, Y., L. Szafraniec, W. Beaudry, and J. Ward. 1988. Kinetics and mecha- REFERENCES nism of the hydrolysis of 2-chloroethyl sulfides. Journal of Organic Chemistry 53(14): 3293-3297. Bandosz, T., and C. Ania. 2006. Surface chemistry of activated carbons and its characterization. Pp. 159-230 in Activated Carbon Surfaces in Envi- ronmental Remediation, 7, T.J. Bandosz, ed. Oxford, U.K.: Elsevier. Bandosz, T., J. Jagiello, A. Krzyzanowski, and J. Schwarz. 1996. Effect of surface chemistry on sorption of water and methanol on activated carbons. Langmuir 12(26): 6480-6486.