The Effect of Temperature and Pressure on the Performance of a PEMFC Exposed to Transient CO Concentrations

Transcription

1 Journal of The Electrochemical Society, A29-A /2002/150 1 /A29/6/$7.00 The Electrochemical Society, Inc. The Effect of Temperature and Pressure on the Performance of a PEMFC Exposed to Transient CO Concentrations Mahesh Murthy, a, * Manuel Esayian, a,c, * Woo-kum Lee, b * and J. W. Van Zee b, *,z a W. L. Gore & Associates, Incorporated, Elkton, Maryland, , USA b Department of Chemical Engineering, University of South Carolina, Columbia, South Carolina 29208, USA A29 Data are reported for Gore s advanced PRIMEA membrane electrode assembly MEA series 5561 exposed to relatively high concentrations 500, 3,000 and 10,000 ppm of CO in hydrogen at 202 and at 70 and 90 C. The steady-state and transient measurements obtained in this study at low reactant stoichiometry and 202 are compared with earlier results 1 for atmospheric conditions to show the effect of temperature and pressure on the poisoning and recovery rates. All data are reported for a 25 cm 2 laboratory-scale proton exchange membrane fuel cell PEMFC using CARBEL GDM CL gas diffusion media GDM for conditions with and without air-bleed treatments. For 500 ppm CO/H 2 mixtures without air-bleed, the performance at 202 and 0.6 V provides a steady-state current density of 1.0 A/cm 2 at 90 C but only 0.4 A/cm 2 at 70 C. At 101 and 70 C, exposure to 500 ppm CO/H 2 mixtures requires 5% air-bleed to obtain this performance. Transient experiments with these CO levels indicate that there is up to a four time decrease in the poisoning rates at 202 vs Further at 202, increasing the cell temperature from 70 to 90 C results in approximately a fourteen time decrease in the poisoning rate for 3,000 ppm CO/H 2 mixtures and approximately four time decrease for 10,000 ppm CO/H 2 mixtures. The data discussed in this paper are suitable for verifying numerical models of a PEMFC and establishing a baseline for new recovery schemes using new MEAs with enhanced CO tolerance. In addition, the results have implications for the design of reformate fuel-processing systems and the use of effective control schemes to prevent CO transients The Electrochemical Society. DOI: / All rights reserved. Manuscript received March 28, Available electronically November 15, The production of electricity from a hydrocarbon-fuel reformer and a proton exchange membrane fuel cell PEMFC connected in series may result in exposure of the membrane electrode assembly MEA to transient pulses of relatively high concentration up to 1% of CO. It is well known that these concentrations may create substantial poisoning of the anode electrocatalyst in the fuel cell even with advanced Pt alloy catalysts. Recently, the poisoning and recovery of fuel cell performance in response to step changes in the level of CO in the anode fuel stream have been studied 1,2 for the PRIMEA MEA series Here we expand that data to include the effects of the cell temperature and reactant back pressure on these transient responses. The data obtained in this study may be useful for commercial PEM fuel cell systems operating with high CO levels. Also, the data discussed in this paper are specifically relevant to fuel cell systems that may be operating with a reformer under dynamic load demands thus resulting in transient CO levels. Data are reported for cell temperatures of 70 and 90 C, low reactant stoichiometries, and a back pressure of psig. Poisoning and recovery rates are reported for surface-saturated CO conditions. CARBEL CL gas diffusion media are used. Also included are data that show the interaction of air-bleed treatments and temperature and pressure on the fuel cell performance. In a recent paper 1 we showed steady-state polarization and transient data for 500 and 3,000 ppm CO in H 2 at low stoichiometric flows, atmospheric pressure, and 70 C. The transient data showed recovery of the performance 0.6 A/cm 2 at 0.69 V from a steadystate CO coverage during cycles between neat hydrogen and 3,000 ppm CO in H 2 and cycles between 50 ppm CO/H 2 and 3,000 ppm CO/H 2 mixtures. The steady-state data showed substantial poisoning of the electrode at 3,000 ppm CO and that 15% air-bleed did not improve cell performance. In another publication 2 we emphasized that air-bleed alone was not the preferred recovery scheme when the cell was dosed with very high levels of CO such as 10,000 ppm. On the other hand, injection of 5% air-bleed in the anode stream provides complete recovery in performance at 500 ppm CO for current * Electrochemical Society Active Member. c Present address: Department of Chemical Engineering, University of South Carolina, Colombia, SC z vanzee@engr.sc.edu densities less than 0.6 A/cm 2, and a current density of 1.0 A/cm 2 is obtained at 0.6 V for a PRIMEA MEA series 5561 using CARBEL GDM CL. Equation 31 in a recent study 3 indicates that steady-state performance in the presence of CO may decrease as operation pressures increase. This may be a result of a relative increase in the partial pressure of CO or a decrease in the partial pressure of water, which is a required reactant for hydration of the alloy during the electrochemical oxidation of CO. For example, for a 1% CO/H 2 mixture, if the feed is humidified at a dew point of 70 C, the partial pressure of CO and H 2 will increase from 0.7 and 69 to 1.7 and 169 when the operating pressure is increased from 101 to 202. Our experience, as shown below with data, indicates that increasing the pressure does not necessarily result in decreased performance in the presence of CO. One source for this discrepancy may be associated with differences in the type of the MEA used here and in Ref. 3. For instance the experimental data reported in Fig. 8 of Ref. 3 was obtained with a Nafion 1135 membrane and with Pt electrodes 0.2 mg/cm 2 on anode and cathode sides. Further, it may be worth distinguishing other differences between the two sets of experiments. Our data was generated in H 2 /CO mixtures instead of reformate mixtures with 40% H 2 ) as used in Ref. 3. It is also important to point out that our data were obtained with very high levels of CO as compared to data in Ref. 3. Therefore, the relationship between pressure and the critical or limiting current density as defined by Eq. 31 will vary under these circumstances. For example, b fc ratio of the rate of backward to forward rate constants for the CO chemisorption reaction on the catalyst surface maybemuch greater that x CO P A, i.e., the partial pressure of CO. This is possible if the dependency of increasing b fc with increase in CO coverage is assumed as per the discussion in Ref. 3. Therefore, for 3,000 ppm CO used in our experiments, b fc may be significantly higher than the partial pressure of CO especially for an alloy anode. If this is true then the critical current density, j hl, will be proportional to the total pressure and will be independent of the value of n. Second, if k ec and h are not equal to zero which is possible for a highly active alloy catalyst, then Eq. 31 is not strictly applicable to the experimental conditions and parameters used in this paper. The aim of this paper is not to refute the model results discussed in Ref. 3 but rather to contrast the differences between the two sets of experimental data. The key goal is to provide reproducible data suitable for verification of numerical models of poisoning and recovery behavior, as

2 A30 Journal of The Electrochemical Society, A29-A Table I. Partial pressure operating conditions for experimental data. Data at 70 C, 101 from Ref. 1 are shown for comparison. Cell temperature, cell pressure Dewpoints anode/cathode P W anode/cathode P H, P CO, at 1% P CO, at 0.3% P CO,at 0.05% 70 C, C/70 C 47/ C, C/60 C 31/ C, C/90 C 58/ P Oxygen, well as expand the base of available steady-state and transient data useful to designers of control schemes and processing systems. The reviews in the literature concerning electrocatalysis for CO oxidation found Ref. 4-7 are applicable. Additional discussion of CO poisoning can be found in Ref. 3. Our papers 1,2 provide summaries of CO poisoning studies in laboratory scale PEMFCs with focus on experiments using extremely high CO levels. In Ref. 1 we show that the rate at which the cell poisons and recovers varies due to the type of gas diffusion media for a particular cell hardware and MEA. We also discuss the effects of GDM on the effectiveness of air-bleed as a recovery scheme. In Ref. 2 we provide data to discuss the dependence of poisoning and recovery rates on the exposure time of the MEA to different levels of CO in hydrogen. In addition, Carrette et al. 6 see Ref. 7 also report a transient study and a recovery scheme for 1% CO based on electrochemical oxidation. Springer et al. 3 presented a two-dimensional model and a set of rate expressions for the poisoning of a MEA. Experimental The studies focused on quantifying the change in performance of PEM fuel cells at 70 and 90 C in the presence of relatively high concentrations of CO up to 10,000 ppm in the anode stream. In addition, the effect of air-bleed on the performance was examined in the experiments. The MEAs used were PRIMEA MEA series 5561, consisting of GORE-SELECT membranes 25 m nominal membrane thickness and catalyst loadings of 0.45 mg/cm 2 Pt alloy on the anode and 0.4 mg/cm 2 Pt on the cathode. The active area of the membrane was 20 cm 2 and the triple-serpentine flow fields consisted of 30 equally spaced channels of height 0.1 cm and width 0.08 cm. The GDM used in the experiment was CARBEL GDM CL 16 mils m. Compressible gaskets of appropriate thickness were employed so that the internal pressure on the gas diffusion media was approximately at 150 psi clamp torque 50 in. lb f /bolt). Other details have been discussed extensively in Ref. 1. The experiments consisted of both steady-state and transient measurements. For steady-state experiments, the current-voltage polarization behavior was measured in the presence of neat hydrogen and mixtures of CO in H 2 at a stoichiometry of 1.2 for the anode and 2.0 for the cathode. d In these polarization experiments, the gases used were 500, 3,000, and 10,000 ppm CO in H 2. After recording a polarization curve with neat H 2, polarizations in CO/H 2 mixtures with and without air-bleed were obtained. The technique increased the total anode flow by adding dry, room-temperature air to the required stoichiometric flow until the air comprised the desired fraction based on the H 2 flow rate. Three levels of air-bleed were used, 5, 10, and 15% air for the 500, 3,000, and 10,000 ppm CO, respectively. In order to obtain a polarization curve, the cell voltage was randomly set from 0.45 V to open-circuit voltage, and the resulting steady-state currents were measured. The randomization gave reproducible results and accounted for any hysteresis in the measured current densities. During polarization curve measurements, the flow rates were adjusted manually in an iterative manner to maintain the desired stoichiometry. d Our anode flowmeters were limited to 50 standard cm 3 /min, and this resulted in stoichiometries greater than 1.2 for current densities less than 0.35 A/cm 2. After establishing steady-state behavior, the experiments with transient feed conditions for the CO/H 2 mixtures were conducted at a fixed current density of 600 ma/cm 2. These transient conditions corresponded to repetitive introductions of a high CO concentration followed by periods with neat H 2 or a low CO concentration. The objective of these experiments was to determine the relative rates of poisoning and recovery when the surface was at pseudo steady-state coverage. For the first set of experiments a baseline with neat hydrogen was established and a cycle of 500, 3,000, or 10,000 ppm CO/H 2 was introduced for 5 min, followed by neat hydrogen for 25 min. This cycle was repeated six times. Then, a 50 ppm CO/H 2 gas mixture was introduced as a baseline and a repetitive test for a CO/H 2 mixture was performed e.g., 500 ppm CO/H 2 was introduced for 5 min and then 50 ppm CO/H 2 for 25 min. Our assumption is that the latter condition probably mimics the real dynamics of the reformer. Each cell was held at 0.6 V for 50 h before the polarization data were obtained. Table I shows the values of the partial pressures of H 2,O 2,H 2 O, and CO used in the experiments. These values were calculated from the known system back pressure and data obtained from calibration experiments using our humidity bottles. 8 That is, the humidity bottle temperatures used in the experiments were 75/ 65 C for a cell temperature of 70 C and 90/95 C for a cell temperature of 90 C. We calculated the dew point to be 70/60 C and 85/ 90 C, respectively, from the humidity calibration data obtained at 101. These humidification temperatures produced optimum polarization curves during preliminary experiments in neat hydrogen as described in Ref. 1. For current densities lower than 0.6 A/cm 2, changes in the dew points did not affect the performance in neat hydrogen indicating a well-humidified membrane. Since the data presented below for high CO concentrations show current densities less than 0.6 A/cm 2, we do not expect the inlet humidity to contribute to the differences in the performance with CO/H 2 mixtures as discussed below. Note that we chose to study the MEA at these optimal humidity conditions because these conditions insured a well-conditioned MEA/GDM combination i.e., a membrane that is not dry and a GDM that is not flooded with liquid water and that if we had fixed the humidity, we may not have compared the best performance at each temperature. Resistance measurements from either current interrupt or high frequency methods were not available with the test stations used in this study, but assuming that the resistance is controlled by the anode relative humidity, we estimate the areal resistance of the membrane to be and cm 2 at 70 and 90 C at the optimum humidity for each cell temperature. These values were obtained by interpolation from ex situ experiments 9 using high frequency resistance HFR measurements at 7 khz at various cell temperatures and relative humidity. Multiple MEAs were used during the experiments and the data are reproducible to within 5% of the current density at a given voltage. The cells are operated for 200 to 300 h during these measurements and therefore the data can be considered to correspond to beginning of life BOL. Also for a given MEA no significant differences between the initial neat H 2 and the final neat H 2 curves were observed. The reproducibility between MEAs was confirmed with polarization data recorded using neat hydrogen for various samples.

3 Journal of The Electrochemical Society, A29-A A31 Figure 1. Effect of temperature on the performance at 202. Open symbols correspond to 70 C and filled symbols correspond to 90 C. Concentration of CO is indicated: neat hydrogen and ; 500 ppm and ; 3,000 ppm and ; and 10,000 ppm and. Figure 3. Comparison of anode overpotentials calculated by difference due to CO poisoning at 202 for 70 and 90 C. Symbols correspond to the operating temperatures and concentrations as described in the caption for Fig. 1. Results and Discussion Effect of temperature on polarization results. Figure 1 shows the effect of temperature on the steady-state polarization behavior with various CO concentrations at 202. For neat hydrogen, there is no difference between the two temperatures until 2.0 A/cm 2. Note that at 101, other measurements not presented here indicate that raising the cell temperature from 70 to 90 C yields a significant decrease in the cell performance for an air cathode. This decrease is probably related to the decrease in the partial pressure of oxygen caused by an increase in the partial pressure of water vapor under these conditions rather than membrane or MEA dehydration. Therefore, since the effect of raising the pressure from 101 to 202 had a significant effect on the cell performance at 90 C and 101, these experiments were excluded from this study. From Fig. 1, the fact that the 90 C performance is higher than 2.1 A/cm 2 even though the partial pressure of oxygen is less see Table I may indicate a slight increase in the membrane conductivity with the 20 C increase in temperature. Most remarkable in Fig. 1 is the increase in performance with 500 ppm CO at 90 C. This sharp increase at voltages less than 0.85 V may be due to a temperature dependence of the isotherm and the fact that this MEA has a large number of available catalytic sites. The increased performance is not as dramatic for 3,000 or 10,000 ppm CO but in general, increasing the temperature has the effect of lowering the surface coverage sufficiently so that the 3,000 ppm CO at 90 C yields similar performance to the 500 ppm at 70 C. Figure 2 shows the effect of temperature when air-bleed is used at 202. The electrochemical oxidation of CO may be shown with graphs of anode polarization in Fig. 3 and 4. This overpotential is calculated from the difference between the cell potential with neat hydrogen and the cell potential with CO/H 2 at the same current density. We assume that the hydrogen over potential with neat hydrogen is negligible, and that the ohmic contribution to the cell voltage and the cathodic overpotential depend only on the current density. Both Fig. 3 and 4 clearly indicate the benefits in reducing the anode overvoltage by raising the cell temperature with these CO levels. Effect of temperature and pressure on transient results. For the transient experiments, the poisoning and recovery rates were measured repetitively. An example is shown in Fig. 5 when 10,000 ppm CO/H 2 gas mixture was introduced. Tables II and III and Fig. 6 summarize the rates of poisoning and recovery of the MEA. These data correspond to values when the surface has reached an equilibrium coverage indicated by the values of the slopes becoming constant. For example, the poisoning rate in Fig. 5 changes from an initial value of 0.42 V/min to a final average rate of 0.40 V/min. Table II shows, for a neat hydrogen baseline, that the poisoning and recovery rates for 500 and 3,000 ppm CO at a cell temperature of 70 C decrease as the back pressure is increased from 101 to 202. For 500 ppm CO, the poisoning rate changes from 0.46 to Figure 2. Effect of temperature on the performance during air-bleed at 202. Symbols correspond to the operating temperatures and concentrations as described in the caption for Fig. 1. Figure 4. Anode overpotentials due to CO poisoning during air-bleed at 202. Symbols correspond to the operating temperatures and concentrations as described in the caption for Fig. 1.

4 A32 Journal of The Electrochemical Society, A29-A Figure 5. Transient performance with neat hydrogen and 1% CO in H 2 with 600 ma/cm 2 at 202 and 90 C. Figure 6. Poisoning and recovery rates data from Table II as a function of temperature and pressure for the neat H 2 baseline. Poisoning and recovery rates data from Table III as a function of temperature and pressure for the 50 ppm CO/H 2 baseline V/min and recovery rate drops from 0.14 to 0.03 V/min. The procedure for obtaining these slopes is discussed in Ref. 2. Tables II and III also indicate that for 500 ppm CO/H 2 at 90 C with 202, the rates are so fast that they cannot be measured not available, N/A with our equipment. This is consistent with the expectation that temperature at low coverage has a dramatic influence on improving the CO tolerance. This table also shows the cell temperature effects on the poisoning and recovery rates when 3,000 ppm CO is introduced. The poisoning rate is decreased from 1.1 to 0.08 V/min decrease by approximately a factor of fourteen times when the cell temperature is increased from 70 to 90 C at 202. Likewise, the recovery rate is increased from 0.04 to 0.06 V/min as the cell temperature is increased from 70 to 90 C. A similar trend is shown for 10,000 ppm CO as shown by the data in Table II. Note that the data for 10,000 ppm CO at 101 are from Ref. 2 and that they were collected after a 90 s exposure instead of the 300 s exposure as in the other cases. Therefore, one could argue that longer exposures for this CO concentration might yield a higher poisoning rate. However, Ref. 2 indicates that exposure time had a minor influence on these rates, and therefore it is reasonable to assume this value shown in Table II for comparative purposes. When the cell temperature changes from 70 to 90 C at 202, the poisoning rate is decreased from 1.56 to 0.40 V/min i.e., a decrease by a factor of four times and the recovery rate is increased from 0.03 to 0.04 V/min. Comparison of Table III and Table II shows that the poisoning rates and the recovery rates are slower with a 50 ppm CO baseline mixture than with a neat H 2 baseline. The slower recovery rates can be explained since there is less driving force from the surface to the bulk with the 50 ppm CO baseline. The slower poisoning rates Table II. Dependence of poisoning and recovery rates on COÕH 2 mixture composition at 600 maõcm 2 with neat-hydrogen as the baseline. Exposure to CO and baseline level was 300 and 1500 s respectively. CO/H 2 ppm T cell 70 C, P 101 T cell 70 C, P 202 T cell 90 C, P N/A N/A 3, , a 0.10 a a Data from Ref. 2 obtained for 90 s exposure to 10,000 ppm CO. Table III. Dependence of poisoning and recovery rates on COÕH 2 mixture composition at 600 maõcm 2 with 50 ppm COÕH 2 as the baseline. Exposure to CO and baseline level was 300 and 1500 s, respectively. CO/H 2 ppm T cell 70 C, P 101 T cell 70 C, P 202 T cell 90 C, P N/A N/A 3,

5 Journal of The Electrochemical Society, A29-A A33 Figure 7. Effect of pressure and CO concentration on the performance at 70 C. Data for 0 psig 101 is from Ref. 1. Open symbols correspond to 0 psig 101 and filled symbols correspond to 15 psig 202. Concentration of CO is indicated: neat hydrogen and ; 500 ppm and ; and 3,000 ppm and. Figure 8. Effect of pressure and CO concentration on the performance during air-bleed at 70 C. Data for zero psig 101 is from Ref. 1. Symbols correspond to the operating pressures and concentrations as described in the caption for Fig. 7. cannot be explained fully at this time but one may speculate that they are related to a difference in the relative number and activity of free catalyst sites in the MEA conditioned with the 50 ppm CO mixture compared to the MEA conditioned with neat hydrogen 100% free catalyst sites. The number of free sites diminishes with an increase in the temperature and the pressure and the difference in rates is also less as the pressure and temperature are increased. It is worthwhile to briefly discuss the procedure for obtaining these rates. A detailed discussion is provided in Ref. 1 and 2. The sample rates for the transient data were 5 and 10 s per datum for the 3,000 and 500 ppm CO concentrations, respectively. The rates of poisoning were obtained by linear regression of a minimum of four points from the voltage/time data that span the maximum and minimum voltages. Thus, the rates were determined over a period of 20 for the 3,000 ppm data and 40 s for the 500 ppm data. Some judgment was necessary to select the points for this span for rapid changes such as those observed with 3,000 ppm CO. A R 2 of 0.90 was typical for these data points. One should note that faster sampling rates may yield different values for the slopes, and that the signal-to-noise ratio was not quantified for the data shown in Tables II and III. It should be noted that faster sampling rates were attempted, but the data seemed to correlate with the voltage resolution of the load cell 5 mv. In contrast, using more points at a given sampling rate resulted in a large variation in the calculated rates. For the 500 ppm CO concentrations, the poisoning rates were slower, more points could be used, and thus one can have more confidence in the precision of these measurements. For the recovery rates, the R 2 was typically 0.95 because the changes were slower and more points could be used in the regression. Effect of pressure on polarization results. Figures 7 through nine show the effect of pressure on the polarization performance. Data at 101 from Ref. 1 are included for comparison. Figure 7 shows that increasing the pressure from 101 to 202 results in a significant increase in the cell performance at 0.6 V, i.e., from approximately 1200 to 1600 ma/cm 2. This increase in cell performance in pure H 2 is mainly attributed to the cathode due to increase in P Oxygen from 15 to 38 as listed in Table I. It is interesting to note that when the anode is switched to either 500 or 3,000 ppm CO, the data at both 101 and 202 are nearly identical. Figure 8 shows a similar comparison with air-bleed for both 500 and 3,000 ppm CO. In order to uncouple the anode effects from the overall MEA performance it is useful to plot the anode overvoltage similar to Fig. 3 and 4 as shown by Fig. 9. Figure 9 compares the anode overvoltage at these two pressures for these two CO concentrations as a function of current density for operation with and without air-bleed. For the case without air-bleed, increasing the pressure seems to be beneficial in lowering the anode polarization for 3,000 ppm CO. For example, at a current density of 100 ma/cm 2, the anode overvoltage is lowered by approximately 100 mv by increasing the pressure from 101 to 202. However with 500 ppm CO, the difference in anode polarization is insignificant without air-bleed. It is interesting to note that when air-bleed is used, the anode polarization at 202 is significantly lower with 3,000 ppm CO. It is important to note that an increase in pressure may result in an increase in the permeability of both H 2 and O 2 across the membrane. If O 2 crossover were the critical factor for improved performance, then substantial enhancement in cell performance can be expected without air-bleed. The data from Fig. 9 indicates that significant enhancement is observed mainly for the case with air-bleed with 3,000 ppm CO. Conclusions Steady-state and transient performance of advanced MEAs PRIMEA MEA series 5561 from W. L. Gore & Associates, Inc. during exposure to relatively high concentrations of CO has been presented to establish a well-documented baseline for designers of Figure 9. Interactions of the effect of pressure, air-bleed, and CO concentration on the overpotential at 70 C. Data for zero psig 101 is from Ref. 1. Solid lines correspond to air-bleed and dotted lines correspond to without air-bleed. Symbols correspond to the operating pressures and concentrations as described in the caption for Fig. 7.

6 A34 Journal of The Electrochemical Society, A29-A new MEAs and PEMFC control systems. Rates of poisoning and recovery are reported for equilibrium coverage of CO at 70 and 90 C cell temperature, low reactant stoichiometry, and 202. Steady-state polarization data also obtained at the above conditions is reported for 500 ppm, 3,000 ppm, and 1% CO with and without air-bleed. Injection of 5% air is shown to completely recover the performance of the MEA at 500 ppm CO for current densities less than 0.6 A/cm 2. With 3,000 ppm CO, some recovery is achievable albeit not complete even with 15% air-bleed and current densities at 0.6 V averaged 1.0 A/cm 2. The effect of temperature seems to indicate a strong dependence of the absorption isotherm on temperature between 70 and 90 C. Transient measurements at 202 indicate that poisoning rates are substantially lowered with an increase in pressure up to four times and temperature up to fourteen times. Acknowledgments The authors gratefully acknowledge that W. L. Gore & Associates, Inc., supported this work. The authors also acknowledge Steve MacKenzie from W. L. Gore & Associates, Inc., for providing the necessary materials to conduct this study. CARBEL, GORE- SELECT, PRIMEA, and GORE and designs are trademarks of W. L. Gore & Associates, Inc. W. L. Gore & Associates, Inc., assisted in meeting the publication costs of this article. References 1. M. Murthy, M. Esayian, A. Hobson, S. MacKenzie, W.-k. Lee, and J. W. Van Zee, J. Electrochem. Soc., 148, A M. Murthy, M. Esayian, A. Hobson, S. MacKenzie, W.-k. Lee, and J. W. Van Zee, J. New Mater. Electrochem. Syst., To be published. 3. T. E. Springer, T. Rockward, T. A. Zawodzinski, and S. Gottesfeld, J. Electrochem. Soc., 148, A S. J. Lee, S. Mukerjee, E. A. Ticianelli, and J. McBreen, Electrochim. Acta, 44, J. J. Baschuk and X. Li, Int. J. Energy Res., 25, L. P. L. Carrette, K. A. Freidrich, M. Huber, and U. Stimming, Phys. Chem. Chem. Phys., 3, K. A. Freidrich, M. Huber, L. Carrette, and U. Stimming, in Proceedings of the 2000 Fuel Cell Seminar, Fuel Cells-Powering the 21st Century, Portland, OR, Oct 30, 2000, pp W.-k. Lee, Ph.D. Thesis, Department of Chemical Engineering, University of South Carolina, Columbia, SC W. Liu, Personal communication, Gore Internal Report no. MA1097 Oct 2000.