Hydrogen Oxidation on Gas Diffusion Electrodes for Phosphoric Acid Fuel Cells in the Presence of Carbon Monoxide and Oxygen

Hydrogen oxidation has been studied on a carbon-supported plat inum gas diffusion electrode in a phosphoric acid electrolyte in the presence of carbon monoxide and oxygen in the feed gas. The poisoning effect of carbon monoxide present in the feed gas was measured in the temperature range from 80 to 150~ It was found that throughout the temperature range, the potential loss due to the CO poisoning can be reduced to a great extent by the injection of small amounts of gaseous oxygen into the hydrogen gas containing carbon monoxide. By adding 5 volume percent (v/o) oxygen, an almost CO-free performance can be obtained for carbon monoxide concentrations up to 0.5 v/o CO at 130~ 0.2 v/o CO at 100~ and 0.1 v/o CO at 80~ respectively. Introduction The hydrogen electrode reaction is very reversible in acid electrolytes. Its electrochemical performance is generally excellent in fuel cells. The overpotential at the operating current densities in fuel cells, e.g., 200 to 400 mA/cm 2, is only ca. 20 mV for either phosphoric acid_fuel cells (PAFC) at 190~ or for solid polymer electrolyte fuel cells (SPEFC) at 80~ In the fuel cells using molten carbonate (MCFC) or solid oxide electrolyte (SOFC), the overpotential at the hydrogen electrode is even smaller due to the extremely high operating temperature (650~ for MCFC and 1000~ for SOFC). The major problem at the hydrogen electrode, however, is to eliminate the poisoning effect due to impurities such as CO, H2S, and SO2. These impurities are generally present more or less in the fuel hydrogen produced either by reforming of natural gas or by coal gasification. Carbon monoxide, among these impurities, is the most challenging one for noble metal eleetrocatalysts in acid electrolytes. The poisoning effect of platinum catalysts by carbon monoxide has been intensively studied by many investigators in the field of fuel cells and electrocatalysts. 1-n These studies have mainly been concentrated on the mechanism of adsorption and oxidation and the nature of the absorbed species on the noble metal electrodes in various acid electrolytes. I-~ Two kinds of adsorption mechanisms have been proposed, i.e., the linearly and bridge or multibonded CO species. 2-8 As assumed, the linearly absorbed carbon monoxide species, --CO, involve one adsorption site per CO particle, while the bridge or multibonded carbon monoxide species, =CO, requires two or more adjacent platinum surface sites. In the case of hydrogen oxidation on a carbon-supported plat inum gas-diffusion electrode in the presence of carbon monoxide, the poisoning effect is found to depend logarithmically on the ratio of carbon monoxide concentration to the hydrogen concentration, [CO]/[H~]. This indicates that the poisoning is a simple competition with hydrogen for active sites, ~'~~ since the strong ehemisorptive bond of the carbon monoxide molecule may lead to a surface blockage from hydrogen oxidation. The fact that the potential loss of the hydrogen electrode is due to the CO blocking coverage of the active sites accessible for hydrogen led a number of investigators to corre* Electrochemical Society Active Member. late the potential loss with the surface coverage of carbon monoxide, 0co. By measuring the ratio of the current density in the presence of CO (/co) to that of pure hydrogen (ill2) at a given potential in the linear portion of the polarization plots, 9'1~ the surface coverage of carbon monoxide can be calculated. The fractional coverage thus obtained was found to follow a Temkin isotherm relationship. By comparing the coverage-concentration relationship of the CO adsorption with the Temkin isotherm expression, the free energy change and the standard entropy of the CO adsorption were calculated and found to have large negative values, indicating the strong adsorption and the strong favoring at lower temperature. 1~ From the fuel cell point of view, the strong dependence of CO poisoning on temperature provides a possibility for solving the problem, i.e., raising the operating temperature of the fuel cells. Studies of the CO tolerance for acid fuel cells at different temperatures showed that operating temperatures above 135~ are needed to avoid significant potential losses due to carbon monoxide poisoning. One of the main reasons for operating PAFCs at 180 to 190~ is to tolerate 1 to 2% carbon monoxide without significant potential losses or other serious difficulties for the hydrogen oxidation reaction, n-13 However, increasing the temperature represents a compromise between lower poisoning and higher rates of corrosion and component degradation, particularly at the cathode. 14 In the case of the SPEFCs, the operating temperature is typically below 100~ since the membrane must not be dehydrated to maintain the conductivity. At such low temperatures, the poisoning problem of carbon monoxide is even more challenging: even very small amounts of carbon monoxide, e.g., 10 to 50 ppm, in the hydrogen feeding gas may cause significant potential loss in connection with the hydrogen electrode. Removal of the low level carbon monoxide is challenging also for the methanol reforming process, because as an alternative solution the CO content in the feed gas can be reduced before it enters the fuel cell. The reforming product gas normally contains carbon monoxide of the level of i volume percent (v/o). It has been found ~s that this CO level can be reduced by injecting a small amount of oxygen directly into the reformer where noble catalysts are present, since the CO oxidation by molecular oxygen occurs. An extension of the idea has been made by Gottesfeld et al. 16 to SPEFCs. By injecting low levels of 02 into the hydrogen feeding stream of a SPEFC, the cell voltage loss Downloaded 28 Jun 2010 to 192.38.67.112. Redistribution subject to ECS license or copyright; see http://www.ecsdl.org/terms_use.jsp J. Eloctrochem. Soc., Vol. 142, No. 9, September 1995 9 The Electrochemical Society, Inc. 2891 0.5 caused by the poisoning of I00 ppm CO was sufficiently recovered. In the present work the hydrogen oxidation on carben-supported plat inum gas-diffusion electrodes has been ~= 0.4 studied in the presence of both carbon monoxide and oxygen in 100% H3PO4 in the temperatures ranging from 80 to ~ 0.5 150~ > Experimental -~ The gas-diffusion electrodes consist of a layer of electro~ 0.2 chemically inert carbon paper and an active layer. The active layer consists of a mixture of 60 weight percent (w/o) >~ pla t inum supported on Vulcan XC-72 carbon powder (Pt/ o C~10 w/o) and 40 w/o PTFE as binder. The plat inum loading is 0.5 mg/cm 2. The electrodes were made by a tape-casting method in this laboratory. ~7 Phosphoric acid (85 w/o, analytical grade, Riedel-de Hahn) was purified by treatment with hydrogen peroxide and concentrated to 100 w/o by heating. The concentration of the acid was checked by density measurements. An electrochemical half-cell made of PTFE was used to perform the polarization measurements. It consists of a gas-diffusion working electrode (active geometric area was 0.78 cm2), a reversible hydrogen reference electrode, and a plat inum plate as counterelectrode. Pure gas or gas mixtures were passed through the gas chamber located on the carbon paper side of the electrode, as described elsewhere./8 The cell was placed in an oven maintained at the desired temperature (~ 1~ An electrochemical interface (SI 1286, Schlumberger Technologies Ltd.) and a potentiostat/galvanostat (Model 273, EG&G Princeton Applied Research) were used to perform the measurements. The polarization curves were obtained by the current step potentiometry. The chronopotentiometric curves were recorded by computer sampling at each current density applied. The polarization potentials at various current density settings were then taken from these curves when a steady state was reached. The duration for reaching such a steady state varies mainly due to different feeding gas compositions, from 30 s for pure hydrogen at various current densities to more than half an hour when carbon monoxide is present. A current interruption techuJ nique was used for the IR correction. By means of a digital 7{12 oscilloscope (DSO 1602, Gould Electronic, Ltd.), the IR part of the potential was obtained 30 ~s after the current was interrupted. > Purified hydrogen (>99.998 v/o, Hede Nielsen A/S) and E carbon monoxide (>99.98 v/o, Dansk Ilt & Brintfabrik) were used in the measurements. The gas flow was controlled by means of a mass flow controller (5850TR, Brooks Instrument B. V.). O Results and Discussion 9 The hydrogen oxidation process was found to be almost independent of temperature in the presently studied temperature range. For example, at a current density of 400 mA/cm 2, the overpotentia] difference is less than 5 mV w for a temperature variation from 80 to 150~ This is due to T Qf the fact that the hydrogen electrode is very reversible in the acid electrolyte. When carbon monoxide, however, is present in the hydrogen feeding gas, the overpotential is > very temperature dependent. Figure 1 shows polarization E curves for the oxidation reaction of hydrogen containing _: 0.i v/o carbon monoxide at different temperatures. In this o '~ and the following figures, each point represents the averc age value of at least three measurements, and different points at the same current densities are from independent measurements in which different pieces of a produced electrode are examined. At temperatures higher than 130~ the increase in overpotential due to the presence of 0.1 v/o carbon monoxide is between 10 and 15 mV at the operating current densities (200 to 400 mA/cm ~) for the fuel cells, as seen from Fig. i. When the operating temperature is lower than II0~ the poisoning effect is dramatically higher. At a temperature of 80~ the typical operating temperature of SPEFCs, for ini 'l / ~' 'r [CO]/[H2] Temp. I 0.1/99.g 80~ + 0,1/99.9 9OoC r 0.1/99.9 100~ & 0.1/99_9 110~ X Q.1/99,9 130~ rr 0.1/99.9 1500C O 0 /100 150~