Environ. Sci. Technol. 2001, 35, 4320-4326
Reductive Dehalogenation of Gas-Phase Chlorinated Solvents Using a Modified Fuel Cell Z H I J I E L I U , † R O B E R T G . A R N O L D , * ,† ERIC A. BETTERTON,‡ AND EUGENE SMOTKIN§ Department of Chemical and Environmental Engineering, The University of Arizona, Building 72, Room 306, Tucson, Arizona 85721, Department of Atmospheric Sciences, The University of Arizona, Tucson, Arizona, and Department of Chemical and Environmental Engineering, Illinois Institute of Technology
The reductive dehalogenation of gas-phase chlorinated alkanes (CCl4, CHCl3, and 1,1,1-trichloroethane) and alkenes (perchloroethene (PCE) and trichloroethene (TCE)) was conducted in a modified fuel cell. The fuel-cell performance was a function of cathode material, electric potential, temperature, target compound identity and gas-phase concentration, partial pressure of O2 in the cathode chamber, and cathode condition (time in service). TCE conversion was approximately first order in TCE concentration with halflives of fractions of a second. Under the same reactor conditions, CCl4 transformation was faster than CHCl3, and TCE reduction was faster than PCE. Rates of both CCl4 and PCE transformation increased substantially with temperature in the range of 30-70 °C. At 70 °C and a potential (potential of the cathode minus that of the anode) of -0.4 V, single-pass CCl4 conversions were approximately 90%. Mean residence time for gases in the porous cathode was much less than 1 s. The presence of even 5% O2(g) in the influent to the cathode chamber had a deleterious effect on reactor performance. Performance also deteriorated with time in service, perhaps due to the accumulation of HCl on the cathode surface. Conversion efficiency was restored, however, by temporarily eliminating the halogenated target(s) from the influent stream or by briefly reversing fuel-cell polarity.
Introduction Heavily halogenated solvents are energetically poised to participate in reductive reactions of the form
RX + 2e - + H + f RH + X -
(1)
The electronegative character of the chlorine substituents makes hydrogenolysis of CCl4, for example, as energetically favorable as the reduction of molecular oxygen to water (1). Reductant can be supplied by transition metal ions and complexes of those ions (2-4), elemental metals (5-9), hydrogen (often with precious metal catalysis) (10, 11), and * Corresponding author phone: (520)621-2410; fax: (520)621-6048; e-mail:
[email protected]. † Department of Chemical and Environmental Engineering, The University of Arizona. ‡ Department of Atmospheric Sciences, The University of Arizona. § Illinois Institute of Technology. 4320
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 35, NO. 21, 2001
FIGURE 1. Schematic representation of modified fuel cell. RCl represents the target contaminant, and RH is the product of hydrogenolysis. Anode and cathode compartments are separated by a Nafion membrane. In standard operation, H2 was fed to the anode compartment at 10 psig, and the chlorinated target compounds were added to the cathode compartment in a N2 stream at atmospheric pressure. metallic electrodes (12-14) in a variety of abiotic and biochemical settings. Hydrogenolysis of halogenated targets is often initiated by a one-electron (rate-limiting) transfer, which liberates halide ion and produces a radical intermediate, R•. The process is completed by a second electron-transfer reaction and protonation of the resultant carbanion or by hydrogen abstraction from the surrounding medium (1). The kinetics and mechanism of aqueous-phase hydrogenolysis reactions have been studied extensively. Similarly, catalyzed, aqueousphase hydrodehalogenation reactions have received some attention. Palladium was used to catalyze the hydrodehalogenation of chlorinated solvents (15, 16) and 1,3-dibromo3-chloropropane (10, 11). Among the solvents tested, the chlorinated ethenes were most reactive, with half-times on the order of minutes. The primary reaction products were ethane and hydrochloric acid. Dichloroethane isomers were essentially unreactive in this system. These experiments were conducted in dionized water with a total (suspended) catalyst concentration of 0.22 g/L. Hydrodehalogenation of gas-phase contaminants on metal catalysts has also been studied. Palladium and platinum are most commonly used for this purpose. Although high temperatures are usually employed, the reactions proceed at substantial rates even at room temperature. On freshly prepared catalysts, half-times for complete dehalogenation of chlorinated ethenes are typically on the order of seconds. Particularly at low temperatures, however, catalyst poisoning can be problematic. Such poisoning has been variously attributed to acid production/chloride binding, carbon polymerizations (coking), and catalyst sintering (17, 18). Here, we examine an adaptation of fuel-cell technology for the destruction of gas-phase halogenated compounds. Normally, fuel cells generate useful energy from the galvanic oxidation of H2 at the reactor anode and reduction of O2 at the cathodesthe exothermic recombination of H2 and O2 to form water. The circuit is completed by proton transport through a cation-permeable membrane that separates the cathode from the anode (Figure 1). Since the overall process is energetically favorable, the external circuit of the fuel cell can be used to perform work. In our system, however, gasphase chlorinated solvents replace O2, and energy is supplied from an external power source to improve dechlorination kinetics. Experiments are described that establish the general feasibility of reductive conversions; process stoichiometry, 10.1021/es001772y CCC: $20.00
2001 American Chemical Society Published on Web 09/22/2001
kinetics, and efficiency; temperature effects; and sources of cathode poisoning and recovery. Halogenated targets included perchloroethylene (PCE), trichloroethylene (TCE), carbon tetrachloride (CT), chloroform (CF), and 1,1,1trichloroethane (1,1,1-TCA). The modified fuel-cell technology could be used to treat gas-phase contaminants mobilized by soil vapor extraction. The process may also be useful for the reductive destruction of freons (19, 20). Metal-supported gas-diffusion electrodes (GDEs), similar in concept to the cathode used here, have been extensively investigated with respect to CO2 (21, 22) and NO (23, 24) reductions. Electrochemical reduction of chlorofluoromethanes using GDEs has also been reported (2, 26). Interest in this area has generally focused on catalyst selection. For example, Ag-, Cu-, In-, and Pb-supported GDEs promoted CFC-12 decomposition.
Experimental Section The membrane electrode assembly (MEA) consisted of a cation-permeable Nafion 117 (DuPont) membrane and two catalyst-impregnated porous carbon electrodes that were hot pressed to both faces of the membrane (Figure 1). Platinumloaded carbon black (30% w/w; Aldrich) typically served as the anode. Hydrogen gas (10 psig) flowed through the anode chamber, where it was oxidized to protons. Porous carbon black (Fisher Scientific) impregnated with 30% catalyst (w/ w) served as the cathode. Halogenated VOCs diluted in nitrogen or air (depending on experimental objectives) were continuously fed to the cathode chamber at atmospheric pressure. A potentiostat (model 410, Electrosynthesis) and a 24-V power supply (model 6201A, Harrison Laboratories) were used to establish the potential difference between cathode and anode (reported here as cathode potential (Ec) minus anode potential (Ea), in volts). Each experiment was initiated using a freshly prepared or a cleaned (N2-purged) cathode. Reactant gases were transported to the anode and cathode chambers via channels that were machined into copper current collectors. Gases reached the chamber exits by passage through the porous electrodes (27). Reactor chambers were designed to enhance reactant contact with catalytic sites, circumventing kinetic limitations due to mass transport through a gas-diffusion layer. Unfortunately, it was not possible to measure electrode-reactive surface area or the distribution of gas detention times in the porous electrodes. Electrode Preparation. MEAs were prepared by the method of Wilson (28). Palladium/carbon black ink was prepared by suspending 0.15 g of Pd/C (30% w/w, Aldrich) in 1 g of 5% Nafion solution (Aldrich) and stirring for 24 h. Platinum-black ink was prepared in a similar fashion by suspending 0.10 g of Pt-black (30% w/w, Aldrich) in 0.35 g of the same Nafion solution. To obtain catalyst membranes, inks were repeatedly painted onto 5 cm2 sheets of Teflon, dried, and weighed until the desired catalyst loading (30 mg of Pd/C and 30 mg of Pt-black) was achieved. Membranes were then hot-pressed onto opposite faces of a Nafion 117 sheet (130 °C, 100 psi). After 30 s, the pressure was increased to 1500 psi for an additional 40 s. When the press had returned to room temperature, the Teflon backings were peeled off, leaving porous catalyst films on both sides of the Nafion sheet. The loaded Nafion was overlaid with porous carbon cloth (Electrosynthesis) and inserted into the MEA. MEAs containing other metallic catalysts on the cathode side were prepared similarly. In each case, a 30% w/w metal/C mixture was the starting material, so that catalyst mass was approximately the same in all cathodes tested. Catalyst Preparation. When metal/C mixtures were not commercially available, they were prepared from analyticalgrade reagents by impregnating metal-free carbon black. Metal salts (AgNO3, Cu(NO3)2‚6H2O, Cu(NO3)2‚2.5H2O, Pb-
FIGURE 2. Extent of TCE dehalogenation and product formation as functions of potential (cathode minus anode) in the modified fuelcell reactor. Reactor conditions: influent TCE, 2560 ( 210 ppmv; cathode compartment flow rate, 0.30 mL s-1; Pd/C (30% w/w) cathode; 10 psig H2 (30 °C). Data points represent mean values; error bars represent ( 1 SD, from four to six measurements. Although both trans- and cis-DCE were detected in these experiments, they did not contribute materially to mass balances at the potentials investigated. Consequently, both are omitted from the figure. (NO3)2, Ni(NO3)2‚6H2O, or Zn(NO3)2) were dissolved in 10 mL of deionized water (Milli-Q; >18 MΩ‚cm resistance) at levels designed to yield 30% w/w metal/C mixtures. Methanol (1 mL) and carbon black (5.0 g) were added to the solution and mixed for 10 min. The suspension was then evaporated to dryness. When dry, the contents were reduced at 300 °C for 3 h under hydrogen (1 atm). After the contents were cooled, the metal-impregnated carbon was ground to powder and stored at room temperature until used. Fuel-Cell Reactor Configuration. Deionized water saturated with the halogenated target compound was sealed in a 300-mL flask. A pool of halogenated compound was maintained in the flask to ensure saturation. N2 was then bubbled through the saturated solution to produce a humidified, solvent-containing gas stream. A second N2 stream was humidified by bubbling through pure water. The two gas streams were mixed to achieve desired concentrations of target compounds in a single influent stream to the cathode chamber of the MEA. In this manner, total gas flow rate and contaminant concentration could be regulated independently. Hydrogen was also humidified prior to its introduction into the anode compartment. Operational pressures in the reactor anode and cathode were 10 and 0 psig, respectively. Mass flow controllers in both influent gas lines regulated the gas flows. The cathode chamber influent and effluent gas flows were analyzed for VOCs. Most experiments were carried out at ambient temperatures, although the system was capable of operating at up to 90 °C with additional temperature-control equipment. Influent TCE concentrations were in the range of 2200-2800 ppmv in all TCE conversion experiments. Influent levels of other targets were similar. See relevant figure captions for experiment-dependent influent concentrations. Analytical. Halogenated target compounds and reaction products were measured by gas chromatography (HewlettPackard 5890 with FID; 50 °C; DB-624 capillary column; He, 8 psi; injector 150 °C, detector 275 °C). A 25-µL sampling loop was used to withdraw gas samples from the reactor inlet and outlet. Data points represent averages of four to six measurements. Error bars are included in Figure 2 to illustrate variability. Approximate lower detection limits were 5-10 ppmv for PCE, TCE, CCl4, and CHCl3; 2 ppmv for vinyl chloride, DCE isomers, and CH2Cl2; and 99.5%) commercially available gases into sealed glass vessels.
Results and Discussion Decomposition of TCE at a Pd-Impregnated Carbon Cathode. At a reactor temperature of 30 °C, 10% TCE degradation was observed when cathode and anode were directly connected (short-circuit condition). The extent of reaction increased with cell potential (Figure 2), reaching nearly 50% at -0.35 V when the current limit (1.0 A) of the potentiostat was reached. The major reaction products were ethane and ethene. DCE isomers accounted for less than 5% of the total TCE transformed. Vinyl chloride was not detected. A theoretical average detention time of 1.7 s was calculated for these experiments based on the entire internal volume of the cathode chamber (0.5 mL) and the gas flow rate (0.30 mL s-1). The actual reaction timesthe time spent by gases within the porous carbon-Pd electrodeswas much lower, probably only a fraction of a second. There was a linear relationship between ln(C/Co), where C and Co are influent and effluent concentrations, and the reciprocal of the volumetric flow rate through the cathode (Figure 3). That is
ln R ) - kθ
(2)
where R ) C/Co; k is an effective, first-order reaction rate constant (s-1); and θ is the reactor detention time (s). The linear dependence suggests that TCE reduction was first order with respect to TCE concentration under the conditions of the experiment. Regression analysis was used to estimate the rate constant (0.36 s-1). It is emphasized, however, that since the surface area per unit volume and detention time in the porous cathode are unknown, k represents a conditional rate constant. Comparison of Cathode Materials. Among the seven metals tested, Pd proved the most effective, requiring the lowest potential difference for the same degree of TCE conversion in one detention period (Figure 4). TCE transformation activity decreased in the following order: Pd > Ni > Co > Ag > Cu > Zn. The activity of the lead-supported electrode, while appreciable at the short-circuit potential, 4322
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 35, NO. 21, 2001
increased only slowly with cell potential. At sufficiently large potential differences, both zinc- and lead-containing electrodes were outperformed by a metal-free, carbon electrode (Figure 4). Explanations for the apparent dependence of TCE reaction rate on the identity of the metal catalyst are speculative. Results are, however, in general agreement with the findings of others. Schreier and Reinhard (10) reported that Pd promotes the dechlorination of PCE in aqueous media in the presence of molecular hydrogen. Hydrogen absorption by Pd may have contributed to its effectiveness. Comparative Arrhenius parameters for the hydrogenolysis of alkyl chloride and bromide by a variety of zero-valent metals highlighted the efficiencies of Pt and Pd as catalysts for dehalogenation reactions (29). Although chlorofluoromethanes were effectively reduced using Ag-, Cu- and Pb-impregnated electrodes (25, 26), none was as effective as Pd for the electroreduction of TCE. Differences in the relative electroactivities of alternative cathode materials were attributed to cathode surface charge and reactant affinity for the electrode material (25). The order of effectiveness among zero-valent metals for electrolytic reduction of CCl4 in aqueous media differed substantially from that reported here (14). Metal-dependent rates in aqueous-phase experiments were poorly correlated with material-dependent hydrogen overpotentials (30); interference due to hydrogen evolution was evidently not a source of observed differences. Here there was an inverse relationship between metal-dependent overpotential for hydrogen evolution and metal-dependent rate constants for reductive dehalogenation of TCE (above). That is, metaldependent overpotentials for hydrogen formation from water can be ordered as follows: Pd , Ag < Co < Ni < Pb , Zn. The qualitative relationship between the apparent utility of materials for promoting TCE reductive dehalogenation (Figure 4) and (low) hydrogen overpotential suggests that nascent hydrogen is an important species in the dehalogenation mechanism. Target Identity. The results of electrolysis experiments conducted under similar experimental conditions (30 °C, Pd/C cathode, gas flow rate ) 0.30 mL s-1) indicate that, although rates of reductive dehalogenation were similarly dependent on the difference in potential between cathode and anode, relative rates of dehalogenation at a single cell potential were sensitive to compound identity with TCE ∼ CT > 1,1,1-TCA ∼ CF ∼ PCE (Figure 5). A similar reaction order was observed previously for the electrochemical reduction of chlorinated aliphatics in water at a Ni cathode at a constant potential (31). The greatest difference between the aqueous- and gas-phase experiments arises from the high gas-phase reactivity of TCE as compared to its relatively sluggish decomposition rate in the aqueous phase. Vogel et al. (1) indicated that relative rates of aqueousphase hydrogenolysis reactions are related to their reduction potentials. Others have suggested that aqueous-phase dehalogenation rates are more closely related to the lowest unoccupied molecular orbital energies (32) or to carbonchlorine bond strength (31). An adequate explanation for material- and target-dependent variations in gas-phase conversion rates, however, is not currently available. Product Identification and Implications for Mechanism. TCE was the only target compound for which a detailed mass balance was attempted. For other chlorinated targets, yield information is based on semiquantitative measurements of reaction products. In experiments with PCE, the predominant products were hydrocarbonssethene, ethane, 1-butene, trans- and cis-2-butene. Chlorine-containing products (TCE, cis- and trans-DCE) accounted for less than 3% of the PCE transformed (Pd/C) (Table 1). Larger cell potentials and (presumably) more negative cathode potentials minimized the production of chlorine-containing reaction intermediates
FIGURE 4. Fractional TCE dehalogenated as a function of reactor potential difference and cathode material. Theoretical detention time was 1.4-1.7 s. Influent TCE concentration was 2200-2800 ppmv.
TABLE 1. Distribution of Reaction Products as a Function of Target Compound Identitya target compound PCE TCE CT CF 1,1,1-TCA
majorb TCE CF 1,1-DCA
chlorinated product minorc
trans-DCE cis-DCE trans-DCE, cis-DCE, 1,1-DCE DCM DCM 1,1-DCE
majord
non-chlorinated product minore
ethane, ethene
1-butene, trans-2-butene, cis-2-butene
ethane, ethene methane methane ethane
acetylene, 1-butene, 2-butenes ethene
Reactor conditions were 30 °C, -0.4 V; 0.3 mL 30% (w/w) palladium-impregnated carbon black. b Accounting for 0.5-5.0% of target compound transformed. c Accounting for
