For example, voltage drop caused by ohmic resistance in cell components is about 0.02 volt a t 100 ma. per sq. cm. for a parallel plate battery; thus, terminal voltage of the cell can be determined by combining this internal resistance loss with appropiate polarization value from Figure 2. Low temperature, low pressure cells are not subject to electrode attack by electrolyte or oxidation. T h e only lifelimiting factor is wettability of the carbon electrodes, which seems to depend on the potential a t which the electrode operates rather than on current density. Two years' intermittent service has been achieved on 10 ma. per sq. cm. and over 1 year's continuous service on 20 ma. at 0.8 volt. This was at atmospheric pressure, between room temperature and 70" C. In the meantime better repellency treatments and more active catalysts have brought expectations u p to 30 to 50 ma. per sq. cm. over 0.8 volt or at least the same time period. By using high pressures, high currents a t low temperatures can be obtained if more auxiliary equipment is used. Hydrogen is an ideal fuel--'/, pound produces 1 kw.-hr. However for everyd a y purposes, hydrides decomposed by water are more convenient. O n e pound of lithium hydride is equivalent to 1 kw.-hr.
T h e cells developed in these laboratories operate with high current densities on air with only a small potential difference to the pure oxygen-hydrogen cell. With carbonaceous fuels, such as carbon monoxide, alcohols, and aldehydes, good results have been obtained a t low temperatures, but the need for removing carbonate from the alkaline electrolyte complicates these systems. Unfortunately, the present oxygencarbon electrode does not function well in acid. Only current densities of 20 ma. per sq. cm. are obtained a t 0.8 volt. A redox-chemical intermediate such as bromine is necessary for high current outputs. All halogens operate on carbon electrodes with high current densities in acid systems. As a result. hydrogen-chlorine fuel cells can be operated a t high power outputs for extended periods. However, despite higher voltages and current densities, energy output per pound of combined fuel is less than that for the hydrogen-oxygen cell.
References
(4) Hunger, H., Dissertation, Univ. of Vienna, 1954. (5) Hunger, H., Marko, A,, 5th World Power Conf., No. 275 (K-ll), Vienna, 1956.
(6) Justi, E., others, Jahrbuch Akad. Wiss. Mainz (1955). (7) Ibid.,No. 1, (1956). (8) Kordesch, K., Marko, A., Oesterr. Chemiker-Ztg. 52, 125 (1951). (9) Kordesch, K., Martinola, F., Monatsh. Chem. 84, 1, 39 (1953). (10) Marko, A , , Kordesch, K., U. S. Patent 2,615,932 (Oct. 28,1952). (11) Ibid.,2,669,598 (Feb. 16, 1954). (12) Proc. 12th Annual Battery Research and Develon. Conf.. U. S. Armv, Signal . , Research &'Devel. Lab., 1958. 3) Proc. 13th Annual Power Sources Conf. U. S. Army Signal Research and Development Laboratory, Ft. Monmouth, New Jersey, 1959. 4) Spengler, H., Angenw. Chem. 68, 689 (1956). 5) Witherspoon, R. R., Urbach, H. B., Ycager, E., Hovorka, F., Tech. Rept. 4, Western Reserve University, Office of Naval Research Contract Nom 581, (1954). RECEIVED for review July 29, 1959 ACCEPTED December 29, 1959 Division of Gas and Fuel Chemistry, Symposium on Fuel Cells, 136th Meeting, ACS, Atlantic City, N. J., September 13-18, 1959.
(1) Baur, E., Tobler, J., 2. Electrochem.
39, 148-80 (1933). (2) Davtyan, 0.K.. Bull. acad. sci. U.R.S.S. C l a m sci. tech. 1946, p. 107. (3) Ibid.,1 9 4 6 , ~215. .
KARL KORDESCH Union Carbide Consumer Products Co., Parma, Ohio
Catalysis of Fuel Cell Electrode Reactions R E S m R C H ox FUEL cells over the past few years has resulted in the development of commercial prototypes of fuel gas cells operating on such gases as hydrogen, carbon monoxide, and hydrocarbons. Depending on projected applications and power requirements, fuel gas cells have been designed to operate at low and medium temperatures using aqueous electrolytes ( 7 , 2, 6 ) and at higher temperatures using molten salt electrolytes (3, 7). Fuel gas cells, particularly those operating at lower temperatures, are subject to a n irreversible free energy process resulting from the interaction of the reactant gases with the electrode surfaces (3, 8 ) . T h e reactant gases are chemisorbed by the electrode catalyst or the electrode surface and the reaction established is between the chemisorbed species and the electrolyte. T h e potential developed by the cell therefore depends on activity of the chemisorbed species, which is (8) inversely proportional to the heat of chemisorption a t high surface coverages. Thus, the catalyst surface can play a dual role in fuel cell electrode reactions-if chemical
298
kinetics are rate-controlling, it can enhancereaction rate, and it can influence potential of the cell by minimizing- the free energy loss caused by chemisorption.
Fuel Electrode T h e role of the cataIyst a t the anode in a fuel gas cell is twofold: It must rapidly chemisorb the fuel gas in such a manner as to make it more susceptible to oxidation by the active species of the electrolyte and a t the same time it should act to minimize the free energy loss due to chemisorption. Thus, the criteria for a n active catalyst generally will be a weak, but rapid chemisorption of the fuel gas. I n the selection of a catalyst, these conditions must be met in addition to the requirement that the catalyst surface must preferentially chemisorb the fuel gas species over the reaction products so as to limit self poisoning. Hydrogen. Chemisorption of hydrogen, particularly on metal surfaces, has been studied more extensively than other fuel gases. At normal temperatures, chemisorption of the type required for high catalytic activity presumably
INDUSTRIAL AND ENGINEERING CHEMISTRY
involves a partially covalent surface bond between hydrogen atoms and d electrons of the metal. Thus, one requirement for high catalytic activity of a metal in simple gas reactions of hydrogen appears to be that it possesses d band vacancies. This limits the active metal catalysts to the transition elements. T h e early members of the transition series, w-hich have vacancies in both the first and second subbands, chemisorb hydrogen strongly and are in general not as active catalysts in hydrogen reactions as the latter members of the three transition series, which have vacancies only in the second subband. These metals exhibit the lowest heats of chemisorption a t the surface coverages involved in heterogeneous reactions, and are recognized as highly active catalysts for hydrogen reactions. Thus, it appears that the most active metal catalysts for fuel cell electrode reactions where hydrogen is the fuel gas should be selected from those transition metals with d-band vacancies only in the second subband-e.g., group VI11 metals. These views are confirmed by data (Figure 1) obtained with a low-tempera-
FUEL CELLS 90
BO
I
L
30
4
i
3 APPROXIMATE
I
d BAND V A C A N C I E S
Figure 1. When hydrogen is the fuel gas, catalysts should be selected from those transition metals with d-band vacancies only in the second d subband
crease in potential over the pure metals would probably not justify the difficulties of alloying in commercial practice. T h e results given in Table I for Iowtemperature fuel cells, using aqueous hydroxide electrolyte, in general are paralleled by cells operating at higher temperatures, although the free-energy loss on chemisorption decreases with increasing temperature and consequently is of less importance. For example, palladium and platinum catalysts give higher open-circuit potentials with hydrogen as a fuel gas than do such metals as nickel and iron in a variety of molten salt electrolytes in the 200' to 300' C. range. At relatively high temperatures (about 500' to 800' C.) the nature of the hydrogen chemisorption changes for several of the group VI11
ture fuel cell (27' C.) employing an aqueous sodium hydroxide electrolyte and porous graphite electrodes which were impregnated with the metal catalysts. I n general the catalytic activities of the metals parallel their open-circuit potentials-e.g., a small free-energy loss caused by chemisorption generally implies a high catalytic activity. I n Table I, the same trend in activity is observed in the three transition series, the potential reaching a maximum between the last two transition metals and falling sharply for the following I b metal. From several considerations platinum and palladium are probably the best catalysts for hydrogen electrodes in fuel gas cells. Although a maximum in activity is obtained with alloys of certain of these metals, the slight in-
Table I.
Open-circuit Half-Cell Potentials
Half-Cell Potential Mv. Metal
Hydrogen
Ethylene
Acetylene
Carbon monoxide
Fe
575 745 735 365 782 817 785 285 545 645 775 840 305
780 675 647 302 205 365 710 095 405 570 610 380 190
790 715 595 475 475 535 705 294 460 425 625 570 185
440 495 190 475 540 692 225 462 520 510 545 150
co Ni cu Ru Rh Pd Ag W
os
Ir Pt Au
metals. Quite probably, hydrogen forms d s p hybrid bonds with the metal. Acetylene and Ethylene. Catalytic activities of the group VI11 and I b metals in the oxidation of ethylene and acetylene in a fuel cell appear to be similar to their activities with hydrogeni.e., high catalyst activity is favored by vacancies in the d band of the metal. However, these reaction systems are fundamentally more complex than those of the hydrogen cell. T h e reactions which occur at the anode are complicated by two factors: the nature of the chemisorbed complex, which is in doubt, because either carbon-carbon or carbonhydrogen bonds may be broken in chemisorption and oxidation; and amount of self-hydrogenation at the surface. The fuel cell employed for studies of ethylene and acetylene was similar to the cell used for hydrogen except that the electrolyte was a 40y0 aqueous solution of potassium carbonate. T h e most active catalysts among the metals studied are the group VI11 metals of the first transition series along with palladium and iridium (Table I). As in the case for hydrogen, a low heat of chemisorption in either ethylene or acetylene will minimize the irreversible free-energy loss a t the fuel electrode and, hence, produce a higher potentia1 in the fuel cell. There are two factors which determine the heat of chemisorption of ethylene and acetylene: a geometric factor-Le., interatomic distances in the catalyst lattice; and a n electronic factor. T h e slight activity shown by the I b metals in these reactions probably can be attributed either to d-s electron promotion, which causes vacancies in the d band, or to a-bonding by the metals to these molecules which can be achieved by a rearrangement of the metallic orbital together with the formation of a bond by overlap of the filled d orbitals with the antibonding orbitals of the adsorbate (4). Carbon Monoxide. The same type of fuel cell was used for studies on carbon monoxide as with ethylene and acetylene, the electrolyte being a 40y0 aqueous potassium carbonate solution (Table I). The variation in the half-cell potential among the transition metal catalysts is small with the exception of palladium which appears to be the most active catalyst for the reaction. Activities of the I b metals are low as for the fuel gases previously mentioned. This would indicate the necessity of vacancies in the d band of the metal catalysts for a high activity. T h e chemisorption of carbon monoxide may take place by either of two different mechanisms. O n certain metals, such as supported platinum, it VOL. 52, NO. 4
APRIL 1960
299
chemisorbs principally with a one-site attachment forming a surface layer similar in structure to the metal carbonylsLe., M = C = 0 or M - C = 0 (5). T h e second mode of chemisorption is a two-site sorption with the carbon monoxide complexes covering two surface sites. Two-site chemisorption probably takes place on rhodium ( 9 ) and palladium (5)in this manner, although on the latter some one-site chemisorption has been found. All of the transition metals studied probably expose crystal planes suitable for both mechanisms to occur. Oxygen (Air) Electrode
The general requirements of a catalyst at the oxygen electrode of a fuel gas cell are essentially the same as for the fuel electrode catalyst except that negative ion formation is the process under consideration. I n cells employing aqueous hydroxide electrolytes, the oxygen must be chemisorbed in such a manner as to lead to the rapid formation of peroxide and hydroxide ions in the presence of water. A further role of the catalyst in this case is to aid in the decomposition of the peroxide. The open-circuit single electrode potentials of the oxygen electrode in a aqueous hydroxide electrolyte were determined for the group VI11 and I b metal
oxides and oxides of their alloys as catalysts. Typically, the latter two members of the group VI11 metal oxides gave potentials that were only slightly greater than the unactivated graphite and possibly are inactive in this reaction. Alloying the group VI11 metals with the corresponding I b metal to form an alloy oxide did not materially alter the opencircuit potentials until a composition approaching the pure I b metal was reached in which case there was a rapid increase in the potential The most active catalysts, among those investigated, for the electrode reaction of the oxygen half-cell in aqueous hydroxide electrolytes are the oxides of the group I b metals: copper, silver, and gold. Copper and silver oxides are known to be active oxidation catalysts and their presence presumably also promotes decomposition of peroxide ions formed under current drain. Gold films, however, have been reported to be inert toward the chemisorption of oxygen up to 0’ C ( 9 ) . Possibly 0 - 2 ions are formed on the gold surface as an intermediate step in the reduction of oxygen. The activities of the I b metal oxides are in the order: copper > silver > gold. Activity of some oxides as catalysts at the oxygen electrode may be varied considerably by the introduction of a defect structure. I t is well known that
heterogeneous reactions, proceeding by negative ion formation, can be profoundly altered by the defect state of the catalyst surface. literature Cited
(1) Bacon, F. T., BeUma J. 61, 6 (1954). (2) Bacon, F. T., British Patent 677,298 (1952). (3) Broers, G. H. J., Ph.D. thesis, University of Amsterdam, 1958. ( 4 ) Dowden, D. A , , “Chemisorption,” p. 9, Butterworths Scientific Publications, London, 1957. (5) Eischens, R. P., Pliskin, W. A., “Advances in Catalysis,” vol. 10, Academic Press, New York, 1958. (6) Evans, G. E., Proc. Ann. Battery Research Deaelop. 72th Conf.,p. 4,1958. (7) Gorin, E., Recht, H. L., Am. SOC. of Mech. Engrs. 58-A-200,1958. (8) Rozelle, R. B., Young, G. J., J . Phys. Chem. in press. (9) Trapnell, B. M. W., “Chemisorption,” Butterworths Scientific Publications, London, 1955. RECEIVED for review September 15, 1959 ACCEPTED December 16, 1959 Division of Gas and Fuel Chemistry, Symposium on Fuel Cells, 136th Meeting, ACS, Atlantic City, N. J., September 13-18,1959. Work supported by the Office of Naval Research.
G. J. YOUNG and R.
B. ROZELLE
Catalysis Laboratory, Alfred University, Alfred, N. Y.
Electrode Kinetics To
O B T A I x G o O D fuel efficiency, fuel cells must be operated with an internal loss of about 20 to 30% of the open circuit voltage. If the current flowing per square centimeter of electrode area or per pound of cell is small, then the cell is bulky or uneconomic. I n this report, polarization is theoretically analyzed in an attempt to show which factors must be varied to obtain optimum conditions. Voltage loss within a cell is caused by three main factors-activation polarization, mass transport polarization, and ohmic resistance. Activation polarization is caused by energy being converted to heat when chemical reactions in the cell proceed under nonequilibrium conditions. Consider for example, a hydrogen half cell where hydrogen diffuses into a catalyzed porous carbon electrode immersed in an alkaline electrolyte. When no current is flowing, transfer of hydrogen chemisorbed on the carbon surface into the electrolyte is at dynamic equilibrium -Le., hydrogen which moves from the electrolyte to the surface is the same as that which moves from the surface into the electrolyte. Under these equilibrium conditions, free energy does not change during this reaction which occurs across the electrode-electrolyte interface. Consequently, the energy change must
300
result from transfer of electrons across the outside circuit; thus, the well known formula, nTE, = AF, is arrived at, wheIe E, is the theoretical reversible potential. However, when current flows from the cell, the reaction is no longer a t equilibrium, and free energy must be used to drive the reaction over the activation energy for the reaction. Obviously, this free energy cannot also be used in driving electron around the external circuit and therefore electromotive force of the cell falls. T h e relation between current per unit geometric area of electrode i, polarization 7, and activation energy AF* is
where Z = KArSAc(a,)“ (a,)
1-
-AF* aAFo __ e RT e R T
17
=a+blogi
6 =
(3)
2.L--3RT an5
(2)
N , is the number of active sites per unit internal area of the electrode; A , is effective area per unit geometric area of electrode; AF, is the standard state free energy change of the reaction ; a*, a, are activities of the products and reactants, and a is a value between zero and 1. I n many cases a is constant over a wide range of current values. Clearly it is desirable for Z, which is called the ex-
INDUSTRIAL AND ENGINEERING CHEMISTRY
change current, to be as large as possible. The use of a porous high internal area electrode increases A , , while the function of the catalyst is to decrease AF* and increase A‘,. Raising the pressure of reactants increases the activities. I is greatly increased by temperature increase and for high temperature cells activation polarization may be negligible. I t should be noted that polarization may be caused by a slow chemisorption reaction, slow electrochemical reaction, or a combination of the two. When 7 is greater than about 50 to 100 mv. Equation 1 reduces to
Equation 3 is the well known Tafel equation. Because of the exponential nature of this relation, when I is small, small amounts of current will give large polarization, but a point is reached where further large current increases give only slight increases in polarization. I n many cases where the open circuit potential is less than theoretical, it is because I is small and current leakage at open circuit is sufficient to cause polarization.
