J. Chem. Soc., Faraday Trans. I , 1985,81, 1569-1576 Reduction of Carbon Monoxide on a Mediated and Partially Immersed Electrode BY KOTARO OGURA* AND HIROYASU WATANABE Department of Applied Chemistry, Yamaguchi University, Ube 755, Japan Received 1 1 th July, 1984 The reduction of carbon monoxide into methanol has been performed using Everitt's salt {K,Fe"[Fe"(CN),]} as a mediator in the presence of homogeneous catalysts at a three-phase (electrode/solution/gas) interface. These conditions enhance the reduction of CO owing to the rapid transport of CO to the active zone of the electrode. The logarithm of the reduction current is linearly related to log (time) with a slope of - 1/2 and reaches a constant value after an extensive period of polarization. A theoretical equation for the current against time curve is derived based on a surface-diffusion model in which CO from the gas phase is first adsorbed on the electrode surface and then diffuses along the surface toward the intersection of the three phases; the computed plot of current against time fits the experimental one well. The conversion of CO into an organic substance is an important process as it can be used as chemical feedstock.We have previously reported that CO may be converted into methanol using Everitt's salt (K2Fe11[Fe11(CN),], ES} using methanol and penta- cyanoferrate(r1) complex as homogeneous catalysts at room temperature under atmospheric pressure.1.2 In order to obtain a higher conversion efficiency the CO reduction is carried out here at a three-phase boundary at a partially immersed flat electrode in which CO gas is in contact with both solution and metal, as described later.Grove3 found that a marked increase in current results when part of a platinum electrode is exposed to hydrogen gas above the electrolyte. This finding led to the emergence of various gas-diffusion electrodes made up of a large number of cylindrical pores, with each pore consisting of partially immersed flat plates. These electrodes have been widely studied from both and e~perimental~-~ aspects. There are theories relating to the mechanism by which the reaction takes place at the three-phase boundary : (i) the surface-diffusion mechanism and (ii) the electrolyte-film mechanism. In the former chemisorption of the reaction species occurs on the electrode surface above the electrolyte meniscus.The reactive species then diffuse along the electrode surface to the area where the electrochemical reaction takes place, and surface diffusion of the reaction species is the rate-determining step. In the latter mechanism the electrode is considered to be covered with a thin film of electrolyte which separates the gas phase from the metal surface. The gaseous species first dissolves in the outermost layer of the film and then diffuses through the film. In this model it is assumed that the rate-determining step is the diffusion of the reaction species through the liquid meniscus and that the existence of the three-phase boundary is not necessarily related to the electric current. The present work was undertaken using a partially immersed electrode in order to enhance the conversion efficiency of CO into methanol. The reduction of CO in this system is caused by the oxidation of Everitt's salt, coated on a platinum plate, to 1569 F A R 1 521570 REDUCTION OF co INTO METHANOL Prussian blue (PB).However, this redox reaction can only be activated by the presence of methanol and metal complexes.'$ The amount of methanol produced was considerably enhanced under these three-phase conditions, and the experimental plot of current against time was in agreement with the theoretical curve obtained on the assumption of the surface-diffusion model. EXPERIMENTAL The test cell had two compartments separated by a fine-porosity glass frit. The main compartment, in which a platinum plate coated with ES was partially immersed in solution, was connected by a Luggin capillary to a saturated calomel reference electrode.This compartment was constructed so as to permit CO or nitrogen gas to be introduced above the electrolyte. The other compartment contained a platinum-plate counter-electrode. Everitt's salt {K2Fe11[Fe11(CN),]} coated on a platinum plate was prepared by the electro- chemical reduction of Prussian blue {(KFelll[Fell(CN),J}. The preparation of PB on a platinum plate was performed under similar conditions to those in ref. (10) and (1 1). A platinum plate of ca. 4.2 cm2 area was first submerged for 3 min in an equal-volume mixture of concentrated nitric and sulphuric acids and for 2 min in 1 mol dm-3 hydrochloric acid solution in a supersonic syringe.The electrode was then cathodized in 1 mol dm-3 HCl for 15 min at 0.1 mA cm-2 and thoroughly rinsed with distilled water. The platinum plate was cathodically polarized at a constant current density of 0.08 mA cm-2 for 5 min in a mixed solution (1 : 1) of 0.01 mol dm-3 K,Fe(CN), and 0.01 mol dm-3 FeC1;6H20. The electrode was again rinsed, allowed to stand for 1 h in distilled water and dried in air for 24 h. The amount of ES on the platinum plate was evaluated coulometrically in 0.1 mol dm-3 KCl, and an amount of ES corresponding to (1.0-2.0) x lo-' mol cm-2 was formed in each run. The electrolyte meniscus formed on the partially immersed platinum plate coated with ES was observed microscopically. The experi- ments were performed for various electrode positions.The electrolytes were 0.1 mol dm-3 KC1 solution (the pH was adjusted with HCl) containing a given amount of aquapentacyanoferrate(I1) (Na,[Fe(CN),(H,O)]) and methanol, which function as homogeneous catalysts. Aquapentacyanoferrate(I1) was synthesized from penta- cyanonitrosylferrate(Ir1) {Na,[Fe(CN),(NO)]} by the method of Hieber et af.12 The carbon monoxide used in this experiment was supplied by the Seitetsu Kagaku Co. and contained c 500 ppm nitrogen, c 100 ppm carbon dioxide, c 100 ppm oxygen, < 100 ppm hydrogen and < 15 ppm water. 70 cm3 of the prepared solution was transferred to the test cell, the platinum plate coated with ES was partially immersed in the solution and CO was introduced above the solution level. The electrochemical experiments were performed using a Nichia NP-G 1 OOOG potentiostat, a Nichia S-5A function generator, a Nichia N-CR564 coulometer and a Watanabe WX440 X- Y recorder.The reaction products, such as methanol, formic acid and formaldehyde, were analysed by the following methods. A JGC-110 gas chromatograph with a thermal-conductivity detector and a Poropak Q column was used to determine the amount of methanol. The sampling procedure for the gas chromatograph was performed in the following way. 2.5 cm3 of the sample solution was transferred to ace11 which wasconnected to vacuum through a stopcock. A side-port of the cell was fitted with a rubber septum in order to withdraw samples, and the solution was evaporated under 1 Torr pressure at 70 "C. After nitrogen gas was introduced into the cell, gas samples (2 cm3) were taken with a Pressure-Lock air-tight syringe.The calibration curve for this sampling procedure was linear for methanol up to at least 30 mmol dm-3. The amount of methanol was calculated from the difference in the methanol contents of the initial and final solutions. The experimenlal error involved in such a method was within f 2 % . Formic acid and formaldehyde were determined by colorimetric analysis using chromotropic acidI3 but they were present in negligible amounts.K. OGURA AND H. WATANABE 1571 8 0.5 I I I [ n l R I O3 I 2 3 4 S,/cm2 Fig. 1. Relationship between the amount of methanol produced and the electrode area exposed to the gas phase (S,) in 0.1 mol dm-3 KC1 solutions at pH 3.5 containing (A, 0) 5 mmol dm-3 Fe(CN)i-+20 mmol dm-3 CH30H or (a) 1 mmol dm-3 Fe(CN)X-+ 20 mmol dm-3 CH,OH.Electrolysis potential -0.9 V; time 3 h. 0 is the ratio of the area exposed to the gas phase to the total surface area. 0, 0, in an N,-saturated solution under an atmosphere of CO; A, in a CO-saturated solution under an atmosphere of N,. RESULTS AND DISCUSSION A marked enhancement of CO conversion was obtained when the platinum plate coated with ES was partially immersed in the electrolyte. The amount of methanol produced is shown in fig. 1 as a function of the electrode area exposed to the gas phase under an atmosphere of CO or nitrogen gas. In the solution saturated with CO the production of methanol decreases with increasing S, (the area exposed to the gas phase), which is reasonable since the reduction of CO is dependent only on the electrode surface submerged below the level of the electrolyte.In an atmosphere of CO, with the electrolyte saturated with nitrogen gas, the amount of methanol increases with increasing S,. In this case the conversion process is independent of the area below the electrolyte surface, but depends on the area exposed to the gas phase. At 6 = 0.5 the amount of methanol obtained under the three-phase condition is about four times as large as that in the corresponding two-phase state. Using the completely submerged electrodel the formation of methanol was enhanced on raising the concentration of added pentacyanoferrate(r1) complex. A similar result was observed here. In the present case two different concentrations of pentacyanoferrate(I1) were added, and 52-21572 REDUCTION OF co INTO METHANOL I 2 3 4 5 t / h Fig.2. Relationship between the amount of methanol produced and the polarization time at -0.9 V in 0.1 mmol dmP3 KCI solutions at pH 3.5 containing 5 mmol dmP3 Fe(CN)E- and 20 mmol CH30H. 0, Partially immersed electrode in an N,-saturated solution under an atmosphere of CO; a, completely submerged electrode in a CO-saturated solution. methanol formation was always higher in the presence of the higher concentration of pentacyanoferrate(n), as seen from fig. 1. The amount of methanol produced is plotted against the polarization time in fig. 2 for two cases in which the electrode is partially and completely submerged. The three-phase condition always gives more methanol than the two-phase condition.Except for the initial period, the amount of methanol formed increases linearly with time in the three-phase state for the timescale of this study, although the amount obtained in the two-phase state approached a final value over a long time of polarization. In the present system CO is not converted into methanol by direct electrochemical reduction but rather CO reduction is brought about by the oxidation of ES to PB. This redox reaction is activated by homogeneous catalysis in the presence of methanol and pentacyanoferrate(1r). Thecatalytic process has been discussed in detail previous1y.l The net reduction reaction of CO is represented as The external current is consumed in the recovery of the mediator, i.e. the reduction of PB to ES: PB+K++e-+ES.(2)K. OGURA AND H. WATANABE 1573 I I I I J 10 I o2 lo3 I o4 tl s Fig. 3. Plots of current against time at - 0.9 V in solutions containing 5 mmol dm-3 Fe(CN):- and either 20 mmol dmP3 CH,OH (0) or 10 mmol dmP3 CH30H (0). 8 = 0.766. CO must reach the active zone of the ES surface by transport through the surrounding gas and liquid. The above results show an enhancement of the formation of methanol in the three-phase condition, which indicates that the transport of CO to the active zone of the electrode is more favourable through the gas phase than solution. Plots of current against time are shown in fig. 3, where -0.9 V is applied at 8 = 0.766. Log (current) varies at first in a linear way with log (time) with a slope of - 1 /2, and reaches a constant value after a long period of polarization.The current is ascribed to the electrochemical reduction of PB to ES, with PB being provided by chemical reaction.' In such a coupled process the transport of CO to the active zone of the ES surface is the rate-determining step, as discussed below. We will try to understand the physical meaning of the - 1 /2 slope obtained in fig. 3 by a theoretical analysis of the reduction process in the three-phase condition. The surface-diffusion model seems to explain the present experimental results well. The three-phase interface is schematically represented in fig. 4: CO from the gas phase first adsorbs on to the electrode surface, and then CO diffuses along the surface from the point of adsorption towards the intersection of the three phases.In the present treatment the following situation is supposed : before electrolysis CO is adsorbed on the site of ES exposed to the gas phase, and the adsorbed CO is depleted as reaction (1) proceeds, but the concentration of CO at the top of the electrode is always constant; i.e. CO is only supplied from the top (x = 0) of the electrode surface during electrolysis, and the adsorption of CO from the gas phase at various values of x is neglected. Under such an assumption the amount of CO is given by the total number of adsorption sites on the electrode surface multiplied by the ratio (8) of S, to the total area S( = S, + S,): t = 0,o < x < 1: c = K * / S (3)1574 REDUCTION OF co INTO METHANOL x = o - x = l . co solution Fig.4. Schematic representation of the three-phase (metal/solution/gas) interface. S , is the area exposed to gas phase, S, is the area immersed in solution, x is the distance from the top of the electrode to the three-phase interface and rn is the width of the electrode. where C is the concentration (in mol cm-2) of CO at the electrode surface, C* is the amount (in mol) of adsorption sites and S is the electrode surface area (in cm2). We have the following boundary condition: t > O , x = l . . c=o. (4) After a long time of reduction the concentration of CO at the top of electrode (x = 0) reaches a constant value, C# : t + m , x = O : C = C # . ( 5 ) Under these boundary conditions the Fick equation for semi-infinite linear diffusion was solved, and the Laplace transform is given by c c"" C= s+s X (exp {-(s/D):[(2n+l)I+xl}-exp {-(s/o):[(2n+l)f-x]}) (6) n-o where C is the initial concentration of adsorbed CO and D is the surface-diffusion constant of CO.The inverse transformation of c i s given as (7) n-0 The concentration of CO at the three-phase interface can be converted into the diffusion current by the relation i = -zFDm - where z is the number of electrons involved in the CO reduction and F is Faraday's constant. Differentiating eqn (7) with respect to x and substituting the resulting equation in eqn (8) we have I = . zFmOC*Da n:t:s n-0 [exp ( - g ) + e x p (-%)I Dt ' (9)K. OGURA AND H. WATANABE 1575 I O - ~ I d4 d 2 10-5 lo-' 0.012 I I I I I 10 I 0' I o3 I o4 I 0' t l s Fig. 5. Computed current against time curves at various values of 8: C* = 1.036 x mol, S = 10 cm2, m = 1.532 cm, z = 4 and D,, = 1.0 x cm2 s-l.I o - ~ I o - ~ 4 2 lo-s I O-E I I I I I I 10 I o2 103 I o4 IC t/ 5 Fig. 6. Computed current against time curves at various values of Dco: (a) lop3, (b) lo-*, (c) and ( d ) C* = 1.036 x mol, S = 10 cm2, m = 1.532 cm, z = 4 and B = 0.806.1576 REDUCTION OF co INTO METHANOL This equation can be converted into a dimensionless one: where i* = zFmC*D/lS f = (Dt):/l. Fig. 5 shows plots of current against time computed from eqn (9). In this computation CO is assumed to be adsorbed on accessible cations in the ES film and C* to be equal to the amount of ES present on the electrode surface. The number of electrons involved in CO reduction is four per mol of methanol formed.Except for extremely exposed or submerged electrodes the log (current) changes in a linear manner with log (time) with a slope of - 1/2, reaching a constant vaIue after a long period of cathodic polarization, which agrees well with the experimental result shown in fig. 3. Fig. 5 shows that the ratio of the initial to final current is dependent on 8. This is because of the difference in the values of the initial concentration of CO; i.e. if 8 or C* is large, the initial current should be large, but with a small value of C* the initial current cannot be large. However, in both cases the final current should be small because of the depletion of CO. Such a correlation was observed experi- mentally. In fig. 6 the computed plots of current against time are exhibited at various values of D.The curve given for D = 1 .O x lop5 cm2 s-l fits the experimental curve well. Accordingly, in three-phase reduction, CO from the gas phase is first adsorbed on the electrode surface; this is followed by surface diffusion to the three-phase interface. This situation enhances the reduction of CO into methanol, which is based on the rapid transport of CO to the active zone of the electrode. K. Ogura and S. Yamasaki, J . Chem. Soc., Faraday Trans. I , 1985, 81, 267. K. Ogura and M. Kaneko, J . Mol. Catal., in press. W. R. Grove, Philos. Mag., 1893, 14, 127. E. Justi, M. Pilkuhn, W. Sceibe and A. Winsel, High-Drain Hydrogen-Diflusion Electrodes Operating at Ambient Temperature und Low Pressure, (Verl. Akad. Wiss. Lit., Wiesbaden, 1959). R. P. Iczkowski, J . Electrochem. Soc., 1964, 1078. S. Srinivasan and H. D. Hurwitz, Electrochim. Acta, 1967, 12, 495. ' J. O'M. Bockris and B. D. Cahan, J . Chem. Phys., 1969,50, 1307. H. C. Weber, H. P. Meissner and D. A. Sama, J . Electrochem. Soc., 1962, 109, 884. F. G. Will, J . Electrochem. SOC., 1963, 110, 145. l o D. Ellis, M. Eckhoff and V. D. Neff, J . Phys. Chem., 1981, 85, 1225. l 1 K. Itaya, H. Akahoshi and S. Toshima, J . Electrochem. Soc., 1982, 129, 1498. l 2 W. Hieber, R. Nast and C. Bartenstein, Z . Anorg. Chem., 1953, 272, 32. l 3 Colorimetric Analytical Method, ed. L. C. Thomas and G. J . Chamberlin (Tintometer Ltd, Salisbury, 9th edn, 1980). (PAPER 4/ 1206)