Environ. Sci. Technol. 2006, 40, 612-617
Destruction of Gas-Phase Trichloroethylene in a Modified Fuel Cell XIUMIN JU,† MATHIAS HECHT,‡ ROSEMARY A. GALHOTRA,† WENDELL P. ELA,† ERIC A. BETTERTON,§ R O B E R T G . A R N O L D , * ,† A N D A . E D U A R D O S AÄ E Z * , † Departments of Chemical and Environmental Engineering and Atmospheric Sciences, University of Arizona, Tucson, Arizona 85721, and MP Technologies, Science and Technology Park, Tucson, Arizona 85747
A conventional fuel cell was used as a catalytic reactor to treat soil vapor extraction (SVE) gases contaminated with trichloroethylene (TCE). The SVE gases are fed to the cathode side of the fuel cell, where TCE is reduced to ethane and hydrochloric acid. The results obtained suggest that TCE reduction occurs by a catalytic reaction with hydrogen that is re-formed on the cathode’s surface beyond a certain applied cell potential. Substantial conversion of TCE is obtained, even when competing oxygen reduction occurs in the cathode. The process has been modeled successfully by conceptualizing the flow passage in the fuel cell as a plug flow reactor.
Introduction Chlorinated solvents such as trichloroethylene (TCE) and perchloroethylene (PCE) are among the most common contaminants at U.S. Superfund sites.1 Pump-and-treat is a commonly applied method for remediating groundwater contaminated with TCE or PCE, while soil vapor extraction (SVE) is used to recover contaminants from the vadose zone. Once the contaminated fluid streams are brought to the surface, adsorption processes are used to transfer the contaminants to a solid sorbent (e.g., activated carbon). The cost of such operations is influenced by sorbent recovery or replacement costs. Destructive technologies for the treatment of fluid streams containing chlorinated solvents include, among others, heterogeneous catalysis2-4 and electrocatalysis of redox transformations,5-10 and chemical reduction using zerovalent iron (ZVI) or another reductant.11,12 Each method has unique advantages and disadvantages. ZVI provides a passive, economically attractive basis for contaminant removal. Surface reactions leading to passivation of the iron surface have not proven to be as great an obstacle to long-term barriers effectiveness, and the technology has been widely deployed in the United Sates and Canada. However, installation costs can be prohibitive when depth to groundwater is significant. * Address correspondence to either author. Phone: (520)621-5369 (A.E.S.). Fax: (520)621-6048 (A.E.S.). E-mail:
[email protected] (R.G.A.);
[email protected] (A.E.S.). † Department of Chemical and Environmental Engineering, University of Arizona. ‡ MP Technologies. § Department of Atmospheric Sciences, University of Arizona. 612
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 40, NO. 2, 2006
FIGURE 1. Fuel cell reactor experimental setup. Humidified hydrogen is fed to the anode chamber (3). Humidified nitrogen-containing gas-phase TCE flows through the cathode chamber (2). The MEA (1) includes the polymer electrolyte assembly and adjacent gas diffusion layers, exposed to graphite plates with serpentine flow channels. Reactants and products were monitored by gas chromatography with a flame ionization detector. The kinetics and product distribution of chloroalkane and chloroalkene reduction via electrolysis depend on electrode material, temperature, ionic strength, and pH.5-9,13,14 In some situations, the kinetics are first-order in contaminant concentration, suggesting that electrode materials are far from saturated or that reaction rates are governed by mass transport as opposed to electron-transfer kinetics at electrode surfaces. More recently, investigations have turned toward gas diffusion electrodes (GDEs), particularly for treatment of volatile contaminants. Gas diffusion electrodes consist of a gas-diffusion region and a reactive (catalytic) region. The reactive gas must diffuse through a liquid layer that separates the gas phase and solid electrode. This transport step frequently limits the overall transformation rate for the target compound.15 Liu et al.16 studied the reductive dechlorination of TCE in a GDE that was identical in design to a polymer electrolyte membrane (PEM) fuel cell. When it is operated as a fuel cell, hydrogen gas is introduced to the anode chamber, where it is oxidized to produce protons and electrons. The protons are transported through the PEM to the cathode, where they participate in the reduction of molecular oxygen to water. The overall reaction is exergonic, and the resulting current, passed through an external circuit, can perform work. When gas-phase halogenated solvents are supplied to the cathode instead of molecular oxygen, they are reduced to less chlorinated homologues. Dehalogenation rates and product selectivity were functions of cathode material selection, temperature, and cathode potential. Under the most favorable experimental conditions, half-times for TCE destruction were on the order of tenths of a second. However, catalyst surface area and residence time in the cathode chamber could not be determined and only the cell potential was known, so that mechanistically based reactor simulations are not possible on the basis of these experiments. Here we explore the kinetics and mechanism of gas-phase TCE reduction under conditions similar to those employed by Liu et al.16 The objective is to develop a new method to treat SVE gases contaminated with chlorinated hydrocarbons. Among the questions addressed is whether TCE reduction on the porous graphite/Pt cathode occurs primarily through a thermocatalytic (direct reduction of TCE with hydrogen on the catalytic surface) or an electrocatalytic mechanism. The effects of oxygen presence on reactor performance were also investigated.
Materials and Methods Fuel Cell. The PEM fuel cell (Figure 1, MP Technologies) was composed of two gold-plated copper end plates, two graphite 10.1021/es0514895 CCC: $33.50
2006 American Chemical Society Published on Web 12/13/2005
plates to collect current and distribute gas within the cathode and anode chambers, and the membrane electrode assembly (MEA). Serpentine flow channels with a cross-sectional area of 1 mm2 were cut into the graphite plates. Three parallel channels formed one flow path. Thus, each segment of the flow path has a total cross-sectional area of 3 mm2. The length of flow path was 220 mm. The MEA consisted of a Nafion112 proton-conducting membrane onto which the catalyst and gas diffusion layers were mounted on opposite sides. The catalyst layers had a geometric area of 10 cm2 for each electrode. The geometry and structure of the fuel cell do not differ substantially from commercially available products. MEA Fabrication. The gas diffusion layer is composed of Teflon-treated carbon cloth and a layer of mixed acetylene black and Teflon on the top. Teflon-treated carbon cloth was obtained by submerging carbon cloth (Zoltek) into a poly(tetrafluoroethylene) (PTFE) solution (Aldrich). After the excess liquid was drained, the carbon cloth was dried at 120 °C for 30 min and sintered at 350 °C. The mixture of 1:1 carbon/PTFE was loaded onto the Teflon-treated carbon cloth to improve gas diffusion. Acetylene black (Chevron) and PTFE solution were mixed with a small amount of 2-propanol to obtain the mixture paste. The paste was evenly applied to the top of the Teflon-treated carbon cloth and dried under ambient conditions. The GDE was cut to electrode size. The catalyst was prepared by mixing 70 mg of 20% Pt on Vulcan XC-72 (E-TEK) with 5 mL of 2-propanol and sonicating for 10 min. The surface area of the Vulcan XC-72 was 250 m2/g, and the Pt metal had a surface area of 110 m2/g and particle size of 2.5 nm. The catalyst suspension was then brushed on the top of the GDE to obtain a catalyst loading of about 1.4 mg/cm2. The carbon cloth electrodes were placed on both sides of the Nafion-112 membrane, wrapped in aluminum foil, and hot-pressed at the glass transition point of the membrane (140 °C) and 130 psi for 90 s. PEM Fuel Cell Reactor. The reactor configuration and experimental setup are shown in Figure 1. Hydrogen gas was humidified before it was introduced into the cathode chamber to prevent complete drying of the anode chamber. Pressure on the hydrogen side of the reactor was maintained at 10 psig. To test fuel cell performance, humidified oxygen was supplied to the cathode at a rate of 60 mL/min and atmospheric pressure. Electrical current was measured as a function of cell potential, and a reactor performance curve was obtained (current density versus cell potential). The cell potential, defined here as
Ecell ) Ec - Ea - IR
(1)
where IR is the IR drop across the membrane, and Ec and Ea are cathode and anode potentials, respectively, maintained by use of a potentiostat (model 410, Electrosynthesis) and a 24-V power supply (model 6201A, Harrison Laboratories). In the normal mode of operation for TCE conversion (termed here electrochemical operation), the gas stream containing the target contaminant (TCE) replaced the oxygen flow to the cathode, and fractional dechlorination was measured as a function of cell potential. The TCE vapor was obtained by passing a nitrogen side stream through a washing bottle filled with pure TCE liquid. The main stream of humidified nitrogen gas was introduced to dilute gas-phase TCE vapor to the target concentration (500 ppmv) ahead of the cathode chamber. Flow rates of the main stream of nitrogen and TCE-laden nitrogen were set to obtain a constant concentration of TCE [verified by gas chromatographic (GC) measurement] at the reactor inlet in all experiments. All gases were obtained from U.S. Airweld and used without further purification. All experiments were conducted at room temperature.
Analytical.16 TCE and reaction products were measured by gas chromatography [SRI 8610 with flame ionization detector (FID), RTX-5 30 m capillary column, oven and injection temperature 100 °C, carrier gas 14 mL/min He]. A 250-µL sampling loop was used to withdraw gas samples. Data points represent averages of three to five measurements. The lower detection limit was 5-10 ppmv for TCE and reaction products. Products were identified by comparison of retention times with authentic standards. Standards for TCE and less chlorinated homologues (Aldrich) were prepared by injecting known volumes of pure liquid into 70-mL sealed bottles. Standards for ethane and ethylene (Scott gases) were prepared by injecting known volume of pure gas into the sealed bottles.
Mathematical Model In the PEM fuel cell, an electrochemical reaction occurs where carbon particles, polymer electrolyte, and Pt catalyst are in close contact. A carbon substrate must be present to conduct electrons and serve as support material for both electrodes. The polymer electrolyte conducts protons from anode to cathode, thereby completing the electrical circuit. The Pt surfaces catalyze the reactions of interest on both sides of MEA. In the void space of the gas diffusion electrode, a thin film of water forms around the carbon/platinum particles. Due to the production of hydrochloric acid in TCE reduction, this thin film of water is acidic. In acidic solutions on Pt surfaces, hydrogen evolution occurs via the Volmer-Tafel mechanism,17 consisting of the following elementary reactions:
Pt(s) + H(l)+ + e- T Pt-H(s) (Volmer reaction)
(2)
2Pt-H(s) T 2Pt(s) + H2(g) (Tafel reaction)
(3)
Because these are surface reactions, the rates can be expressed in terms of the Pt-H surface coverage, θH, as follows:
ν1 ) k1[H+](1 - θH) - k-1θH
(4)
ν2 ) k2θH2 - k-2PH(1 - θH)2
(5)
where νi are reaction rates, ki and k-i are the forward and backward rate constants, [H+] is the proton concentration in the liquid, and PH is the hydrogen partial pressure. The surface coverage of atomic hydrogen can be controlled by manipulating the electrode potential. When hydrogen desorption and readsorption proceed very fast, reaction 3 can be assumed to be at equilibrium and ν2 ≈ 0. Under these circumstances, eq 5 yields
θH )
(KHPH)0.5
(6)
1 + (KHPH)0.5
where KH ) (k-2/k2) is the hydrogen adsorption equilibrium constant.. Results obtained in this work suggest that TCE reduction is initiated by reaction with atomic hydrogen on the cathode side. The overall reaction is
C2HCl3 + 8Pt-H f C2H6 + 3HCl + 8 Pt
(7)
Assuming that TCE and nascent hydrogen do not compete for adsorption sites on Pt and that TCE adsorption is governed by a linear isotherm, the surface coverage of TCE can be expressed as
θT ) KTPT VOL. 40, NO. 2, 2006 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
(8) 9
613
where θT is the surface coverage of TCE, KT is the adsorption equilibrium constant for TCE, and PT is the partial pressure of TCE. Use of the previous equations allows us to express the rate of TCE disappearance (moles per unit time and surface area) as
r ) kθHθT )
kKH0.5KTPH0.5PT [1 + (KHPH)0.5](1 + KTPT)
(9)
where k is the rate constant. The flow path geometry suggests that the cathode can be treated as a plug flow reactor. We will assume that gas-phase diffusion into the electrode is fast and consider that hydrogen gas will be re-formed in the cathode. Under these conditions, mole balances for TCE and hydrogen gas are
d(QPT/RT) ) -Aavr dz
(10)
d(QPH/RT) ) -4Aavr + qHPa/RT dz
(11)
where R is the gas constant, T is absolute temperature (assumed to be uniform along the flow path), A is crosssectional area of flow path, av is the specific surface area of electrode, qH is the volume of hydrogen re-formed per unit length and time, which will be assumed to be uniform along the flow path, z is the flow direction, and Pa is atmospheric pressure (∼0.92 atm). A total mass balance states that the volumetric flow rate of gas along the flow path will increase due to hydrogen re-formation:
Q ) Q0 + qHz
(12)
where Q0 is the influent volumetric flow rate. Note that changes in total flow due to TCE consumption are negligible because of the low influent TCE concentration (generally 500 ppmv). Subtracting eq 10 from eq 11 and integrating yields
PH ) PH0
(
Q 0 q H P az Q0 + - 4 PT0 - PT Q Q Q
)
(13)
where PH0 and PT0 are the hydrogen and TCE partial pressures in the influent, respectively. Equation 10 can be rearranged to find a differential equation governing the TCE partial pressure profile in the reactor, as follows:
dPT RT qHPT ) -Aavr dz Q Q
(14)
Substitution of eqs 9, 12, and 13 into this equation yields an ordinary differential equation that can be solved numerically, along with the initial condition
PT ) PT0 at z ) 0
(15)
The only unknown parameters in the formulation are A, av, k, KT, and KH. The parameters A, av, k, and KT appear in the same terms and were lumped together as follows:
kf ) AavkKTKH-1/2RT
(16)
Equation 14 was solved numerically and the parameters kf and KH were found by fitting the calculated value of TCE partial pressure at the reactor outlet to experimental data. The fitting procedure was based on the minimization of total squared error between predicted and measured values. Model 614
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 40, NO. 2, 2006
FIGURE 2. Current-potential performance curve for fuel cell operation. The solid line represents a theoretical ideal performance without any overpotential loss due to overpotential nonidealities. The dashed line illustrates the relationship between typical or expected fuel cell performance and several sources of nonideal behavior. Activation losses are due to the kinetics of the electrochemical reactions, ionic resistance refers to losses due to the ohmic resistance of the fuel cell, and mass transfer losses refer to limitation in the fluxes of reactants to the electrode surfaces. simulations were performed to investigate the sensitivity of reactor performance to operational variables.
Results and Discussion Fuel Cell Performance Curve. In the reactions of interest of a polymer electrolyte membrane (PEM) fuel cell, H2 is oxidized on the surface of anode catalyst, and O2 is reduced on the surface of cathode catalyst. The reactions are
H2(g) f 2H+ + 2e1
/2O2(g) + 2e- + 2H+ f H2O
E°a ) 0 V
(17)
E°c ) +1.23 V
(18)
Theoretically, the open circuit potential (OCV) for the H2/O2 fuel cell under standard conditions is OCV ) E°c - E°a ) +1.23 V. In practice, the OCV for PEM fuel cells is around +1 V due to several nonidealities, including hydrogen crossover from the anode side to the cathode side, corrosion of electrode material, and oxygen reduction to hydrogen peroxide instead of water.18 To obtain power from fuel cells, an electrical load is placed in the external circuit between anode and cathode. As current flows, the cell potential will deviate from the OCV due to activation overpotential, resistance loss, and mass transfer overpotential. A plot of cell potential versus current density (current per unit nominal electrode area) is used to describe the performance of the PEM fuel cell. Performance curves depend on pressure, temperature, catalyst type and surface area, and membrane water content. Figure 2 shows the performance curve obtained in this work. There were no obvious mass transfer limitations to the catalytic surface in either electrode chamber, since there was no additional potential drop observed at high current densities (Figure 2). The activation overpotential (difference between performance curve and ideal performance value at zero current density due to the finite kinetics of the electrochemical reactions 17 and 18) was relatively small, indicating that the Pt catalyst was in good condition. TCE Reduction Products and TCE Conversion. The major product of TCE reduction in our experiments was ethane, counting for 95% or more of the carbon in the original TCE. These results are similar to those previously reported by Liu et al.16 A small amount of ethylene (
