J. Chem. SOC., Faraday Trans. I, 1982, 78, 845-854 Mechanism of the Catalytic Oxidation of Hydrogen on Copper BY KAZUNARI DOMEN, SHUICHI NAITO, MITSUYUKI SOMA,? TAKAHARU ONISHI* AND KENZI TAMARU Department of Chemistry, Faculty of Science, The University of Tokyo, Hongo, Bunkyo-ku, Tokyo 11 3, Japan Received 13th April, 198 1 The mechanism of water formation from H, and 0, on copper at 393-423 K was studied using a closed circulating system and X-ray photoelectron spectroscopy. Under the whole pressure range of 0, studied the surface was covered with a Cu,O layer during the reaction. Where the partial pressure of 0, was higher than ca. 9 x lo3 Pa, the reaction rate was expressed as follows: r = k[H,]'[O,]O and H,-D, exchange did not proceed during the reaction. On the other hand, where the partial pressure of 0, was lower than ca.9 x lo3 Pa, the rate of water formation showed a maximum value at 1.3 x lo3 Pa and the H,-D, exchange reaction proceeded gradually. From these results, a possible mechanism of the reaction is proposed. The catalytic oxidation of hydrogen on various metals, such as silver, gold, platinum and palladium,l-' has been extensively investigated because of its apparent simplicity and technical interest. In comparison with these metals, the reaction on copper has not been equally well clarified because of the complexity of the reaction. As was first demonstrated by Bone and Wheelar,* the formation of water from hydrogen and oxygen on reduced copper is always accompanied by the simultaneous oxidation of the copper surface. Pease and Taylor9 and Larson and Smithlo concluded that the oxide thus formed accompanying water formation underwent alternate reduction and oxidation. Under their experimental conditions, however, the reaction seems to be in a transient state and the concentration of oxygen in the gas phase was low.On the other hand, Wilkins and Bastowll attempted to correlate the velocity of oxidation and reduction with the rate of water formation, and suggested that some direct catalytic combination between hydrogen and oxygen was taking place in addition to the cyclic oxidation-reduction process. Cleave and RideaP also observed that the catalytic combination occurred on the oxide, but the detailed reaction mechanism and the nature of adsorbed' species have not yet been clarified. In this study a closed circulating system was used to measure the amounts as well as the reactivity of adsorbed hydrogen and oxygen during the reaction over a wide range of partial pressures of oxygen and hydrogen.This method makes it possible to estimate the extent of oxidation of the catalyst in the course of the water-formation reaction and the reactivity of surface oxygen under the working conditions. By studying the dependences of the rate of water formation upon the amounts of the adsorbed species, as well as the partial pressures of ambient gases, it was possible to elucidate the reaction mechanism of water formation on copper. The H,-D, exchange reaction in the course of water formation was investigated to obtain information about 7' Present address: National Institute for Environmental Studies, Yatabe, Tsukuba, Ibaraki 305, Japan. 845846 CATALYTIC OXIDATION OF H, ON Cu hydrogen activation.X-ray photoelectron spectroscopy (X.P.S.) was also applied to study the state of the catalyst surface during the reaction. On the basis of the results obtained, the mechanism of the catalytic H,-0, reaction over copper is examined. EXPERIMENTAL The copper catalyst employed was prepared by the reduction of Cu(OH), with hydrogen at 423-523 K in a conventional closed-circulation system (ca. 450 cm3). Cu(OH), was precipitated from an aqueous solution of Cu(NO,), (99.9% purity) by NaOH solution, washed thoroughly and dried at 373 K. The copper reduced at lower temperatures (393-403 K) exhibited very high activity for oxidation, probably owing to the higher surface area, and the reaction took place so rapidly that it was impossible to follow the reaction rate exactly in our system.Hence, only the samples reduced at higher temperatures (423-523 K) were studied, where the water-formation reaction occurred at a moderate rate. The amount of catalyst was ca. 5 g, and its surface area was determined as 7 m2 g-l by the B.E.T. method using N, adsorption. By X.P.S., small amounts of C and C1, but no Na, were detected on the surface of the reduced copper. Hydrogen was purified by passing through an Engelhard De-oxo unit and a dry-ice-methanol trap to remove any H,O formed. Oxygen was purified by passing through a dry-ice-methanol trap. The amounts of H, and 0, in the gas phase were analysed by gas chromatography. All experiments were carried out in a closed circulating system, and the H,O formed by the reaction was collected by a dry-ice-methanol trap located just after the catalyst and its weight measured gravimetrically at appropriate intervals in the course of the reaction.To investigate the surface state during the reaction, X.P.S. was applied, using a McPherson ESCA 36 spectrometer. The X.P.S. samples were prepared in a closed circulating apparatus and transferred without contact with air to a dry box, which was purged by nitrogen, attached to the sample chamber of the spectrometer and then mounted on the sample holder of the apparatus. RESULTS H,-0, REACTION ON REDUCED AND OXIDIZED COPPER Before the H,-O, reaction the catalyst was pretreated with the mixture of H, and 0, gas at 423 K until a constant reaction rate was obtained, followed by the reduction with H, at 423 K for 12 h, by which time the consumption of H, had finished completely, and evacuation at the same temperature for 2 h.Then the mixture of H, and 0, (each 1.6 x lo4 Pa) was introduced onto the catalyst at 423 K as shown in fig. 1, which indicates clearly that H,O formation and oxidation of copper occurred at the same time. The amounts of H,O, O,(a) and H,(a) are also represented in fig. 1. Even after the formation of more than ten layers of surface oxide (the surface area was estimated by the B.E.T. method), when metallic copper no longer seems to exist on the surface, the uptake of oxide layer still continued. In comparison with the amount of surface oxygen, that of adsorbed hydrogen was small and its surface coverage (H atom/Cu atom) was < 10%.The H,-O, reaction was also studied on the oxidized copper surface at 393 K, as shown in fig. 2. The oxidized copper was formed by the reaction of H, and 0, and its thickness was ca. 10 layers of C U , ~ . In carrying out the reaction, the system was evacuated for 1 h, then H, alone was admitted onto the catalyst at the same temperature. Immediately after the admittance of H,, the rate of H,O formation or of decrease of H, admitted was equal to that before the evacuation, and the rate of reduction increased gradually. This suggests that the rate of H,O formation was retarded by adsorbed oxygen, which is not in adsorption equilibrium with gaseous 0,.As will be shown later, during the H,-0, reaction, the surface is covered with Cu',K. DOMEN, S. NAITO, M. SOMA, T. ONISHI AND K. TAMARU 847 2.0- 2.0 1 . 5 a" 2 * 1 1 .( 4 0 .! - 8.0 evacuation - (H2 + 0 2 ) (H* 1 8.0 6.0 n c: a W m 4.0 --- - 8 $ 2.0 0 0 100 200 tlmin FIG. 1.-H24, reaction on reduced copper at 423 K: 0, H,(g); A, O,(g); 0 , H,O; 0, H,(a); A, O,(a). FIG. 2.-Effect of O,(g) on the rate of H,O formation. After H , 4 , ( 1 : 1) reaction on the oxidized copper surface ( C U , ~ ) at 393 K for 180 min, the gas phase was evacuated for 1 h, then only H,(g) was reintroduced at 393 K. 0, H,(g); 0, O,(g); 0 , amount of H,O formed.848 CATALYTIC OXIDATION OF H, ON Cu and the increase in the rate of H,O formation corresponds to the appearance of Cu metal.To study the effect of the coexistence of H, gas on the rate of oxidation of copper, 0, alone was first introduced onto the oxidized copper surface (ca. 10 layers of Cu*) which was formed by the H,-O, reaction at 423 K. As shown in fig. 3, the oxidation ? a x ""L I 0, only I I . I I I I n 1 0 30 60 90 tlmin FIG. 3.-Effect of H,(g) on the rate of oxidation of copper. (I) p ( 0 , ) = 1.3 x 104 Pa was introduced on the oxidized surface (ca. 10 Cu,O layers) at 423 K. (11) After evacuation for 1 h at 423 K, p(0,) = 1.3 x 1 O4 Pa and p(H,) = 1.0 x lo4 Pa were reintroduced. 0, Amount of 0, uptake in the catalyst. of copper proceeded at a constant rate, and after 1 h the gas phase was replaced by the mixture of H, and 0,. Then the rate of oxidation of copper was enhanced by the coexistence of H,, in accordance with the results of the conductivity measurements by Palmer.13 Note that, upon oxidation by O,, the surface is covered by Cu", which is not the case for the H,-02 reaction.DEPENDENCE OF THE RATE OF H20 FORMATION ON 0, OR H, PARTIAL PRESSURE The dependence of the rate of H,O formation upon the partial pressure of 0, at 403 K under the constant H, pressure (7.3 x lo3 Pa) was investigated on the oxidized surface (with > 10 oxide layers on average), where the H,O formation reaction proceeds steadily. The measurement began with the highest 0, pressure (2 x lo4 Pa), continuing to the lower partial pressures of 0,. As shown in fig. 4, in the 0, pressure range above ca. 9 x lo3 Pa the rate of H,O formation was independent of the partial pressure of 0,.Below 9 x lo3 Pa 0, the rate increased gradually, showing a maximum at ca. 1.3 x lo3 Pa, and then decreased to zero 0, pressure. However, note that at zero 0, pressure in fig. 4 the rate of formation of H,O by repeated reduction with 7.3 x lo3 Pa of H, increased gradually, as is indicated from the results shown in fig. 2. The effect is considered to be relevant to the results from X.P.S., i.e. Cuo could be detected at this stage for the first time, as will be shown later in detail. On the other hand, even at very low 0, pressure (ca. 6.7 x 10, Pa) oxidation of copper occurred during H,O formation.K. DOMEN, S . NAITO, M. SOMA, T. ONISHI AND K. TAMARU 0.25 h 0.00 0 ." U 2 2 0, 3: - 0 . 2 5 - cr a +-' E! W Bn -. 0 0 0.5 1 .o 1.5 P ( 0 2 ) l l o4 Pa - 849 FIG.4.-Dependence of the rate of H,O formation upon the partial pressure of 0, at 403 K. p(H,) = 7.3 x lo3 Pa: before the sequential measurements, the catalyst surface was oxidized by the H,-0, reaction and > 10 Cu,O layers were formed. Considering the results of the dependence of the rate on 0, pressure mentioned above, two different constant 0, pressures were chosen to study the dependence on H, pressure at 403 K, i.e. a high pressure [p(O,) = 1 .O x lo4 Pa], where the rate of H,O formation was independent of 0, pressure, and a low pressure [p(O,) = 2 x lo3 Pa], where the rate did depend on 0, pressure. Fig. 5 shows that the rate of H,O formation - 0 . 7 5 -O.I0t--- 0.5 0.75 log b ( H 2 ) l 1 o4 1 Pal .o 1.25 FIG.5.-Dependence of the rate of H,O formation upon the partial pressure of H, at 403 K; 0, p ( 0 , ) = 1.0 x lo4 Pa; 0, p ( 0 , ) = 2.0 x lo3 Pa.850 CATALYTIC OXIDATION OF H, ON CU was directly proportional to H, pressure in the higher 0, pressure range. On the other hand, in the lower 0, pressure range it showed a dependence of 0.6 order. The reaction order seems to depend upon the 0, pressure and/or reaction temperature. H2-D2 EXCHANGE REACTION DURING THE H,O FORMATION REACTION A gaseous mixture of H,, D, (1 : 1) and 0, was introduced at 403 K onto a surface having an average of > 10 oxide layers, and the rates of HD and H,O formation were followed by gas chromatography. The experiments were carried out under two different 0, pressures, 1.4 x lo4 and 2.0 x lo3 Pa.The results are shown in fig. 6 and 7, where the rate of decrease of hydrogen pressure (H,+D,) is regarded as the rate 0 ’ 5 ~ 0 0 5 10 15 20 t/min FIG. 6.-H,-D, exchange reaction on Cu,O (ca. 10 layers) at 403 K. Initial pressure of 0, was 1.4 x lo4 Pa. 0, Total pressure of H,, HD and D, (initial ratio of H,:D, was 1: 1); 0, O,(g); A, water; A, HD(g). 1 0.5 0 0 5 10 15 t/min FIG. 7.-H,-D, exchange reaction on Cu,O (ca. 10 layers) at 403 K. Initial 0, pressure was 2.0 x lo3 Pa. 0, Total pressure of H,, HD and D, (initial ratio of H,:D, was 1 : 1); 0, O,(g); A, water; A, HD(g).K. DOMEN, S. NAITO, M. SOMA, T. ONISHI AND K. TAMARU 85 1 of H,O formation. Fig. 6 shows that, in the higher 0, pressure range where H,O formation rate was proportional to H, pressure, the H,-D, exchange reaction proceeded very slowly.For example, after 20 min the decrease of hydrogen amounted to 25% but the amount of HD formed was only ca. 1 % of the total hydrogen. In contrast, as shown in fig. 7 where 0, pressure was low and the dependence of the rate of H,O formation upon H, pressure was less than first order, the H,-D, exchange reaction began to proceed gradually during water formation. Even at this stage, however, the rate of the H,-D, exchange was slow in comparison with the rate of water formation, and after the consumption of most of the gaseous 0, the rate of HD formation increased. Cu(L3M4,,M4,,) AUGER SPECTRA AND 0 (1s) X-RAY PHOTOELECTRON SPECTRA During the reaction the catalyst was investigated by X-ray excited electron spectroscopy to determine the mechanism of this reaction in more detail.Cu(L,M,,,M,,,) Auger spectra and 0 (1s) X-ray photoelectron spectra of the catalyst in the course of H,O formation at 413 K are shown in fig. 8(B) and (C). Fig. 8(B) shows the spectra in case of the lower 0, pressure (1.3 x lo3 Pa) and fig. 8(C) that in case of the higher 0, pressure (1.3 x lo4 Pa). In the Cu(L,M4,,M4,,) Auger spectra 530.4 4 11 cu cul f cuo j I t . # , I t 1 1 1 + ' l " . l # 904 924 536 524 k .e .lev b.e./eV FIG. 8.-Cu(LMM) Auger spectra (kinetic energy, k.e.) and 0 (1s) X.P.S. spectra (binding energy, b.e.) of the catalyst: (A) reduced copper at 423 K for 10 h; (B) after H, (1.3 x 104 Pa) and 0, (1.3 x 10" Pa) reaction for 20 min at 41 3 K; (C) after H, (1.3 x lo4 Pa) and 0, (1.3 x lo3 Pa) reaction for 20 min at 41 3 K; (D) after 0, (1.3 x lo4 Pa) oxidation for 20 min at 413 K; (E) after H, (1.3 x 104 Pa) reduction for 10 min at 413 K; (F) Cu(OH),.852 CATALYTIC OXIDATION OF H, ON Cu peaks at 915.1, 914.4 and 913.3 eV (kinetic energy) have been assigned to Cuo, Cull and Cul, respectively, by Larson.14 Consequently, it is thought that in the case of both higher and lower 0, pressures the catalyst surfaces were covered with Cul in the course of the reaction in the steady state.When the catalyst was reduced by H, only, the new peak, which was assigned to Cuo (915.4 eV), appeared as shown in fig. 8(E). On the other hand, when the surface was oxidized by 0, at ca. 1,3 x lo4 Pa and the same temperature (413 K), only the 914.4 eV peak was detectable, which means that the surface is covered mainly with Cu"; however, the peak is rather broad, which indicates that it might be mixed with Cul and/or Cuo.The 0 (1s) peaks at 530.4 eV (binding energy) during the reaction [fig. 8(B) and (C)] indicate the formation of Cu,0.15 0 (1s) X.P.S. of Cu(OH),, which was precipitated by the same procedure as mentioned in the Experimental section and dried at room temperature, were also examined as shown in fig. 8(F), and gave a broad peak at 53 1.5 eV. Perhaps there exists a small amount of -OH species during H,O formation which could not be detected clearly by our method. DISCUSSION H,O formation from H, and 0, on copper was studied at 403-423 K using a closed circulating system and X-ray excited photoelectron spectroscopy.From the results in fig. 1 it is clear that under the reaction conditions the oxidation of copper proceeds steadily, accompanied by the formation of water. Therefore, various measurements in the steady state were carried out on the oxide layer of known thickness which was formed during the reaction of H, and 0,. It was possible to neglect the effect of gaseous H,O on the reaction rate because H,O formed during the reaction was constantly removed by a dry-ice-methanol cold trap placed immediately after the reactor. As has been clarified in the Results section, to elucidate the reaction mechanism attention should mainly be paid to the partial pressure of 0,. WE define Cuo, Cul and CuI1 as the states of copper in metal, cuprous oxide and cupric oxide, respectively.HIGHER OXYGEN PARTIAL-PRESSURE REGION In the region as follows: of 0, partial pressure above 9 x lo3 Y = k[H,]l[O,]O. Pa, the reaction rate is expressed (1) As shown in fig. 6, the H,-D, exchange reaction does not occur when water formation proceeds, and the amount of adsorbed hydrogen is small. As is demonstrated by X.P.S. and A.e.s. (fig. 8), the surface was fully covered with a Cul layer during the catalytic formation of water, whereas in the absence of hydrogen, copper ions were easily oxidized to Cull by oxygen. The absence of Cull during the catalytic reaction and the rate law (1) suggest that the redox cycle Cul/CulI does not contribute to the catalytic formation of H,O significantly, because if that were the case the zero-order behaviour of oxygen could not be explained.Consequently it seems most probable that in this 0, pressure range the rate-determining step of H,O formation is the activation of the hydrogen molecule on the oxidized surface. The results in fig. 2 also show that the evacuation of gaseous oxygen does not affect the rate of H,O formation immediately, which means that H,O is formed from the adsorbed oxygen on the Cul surface, which does not desorb easily by evacuation at the reaction temperature, 423 K. Further support for the existence of adsorbed oxygen on Cu,O is found in the work of Ostrovsky et al.lS They investigated the heat of adsorption of oxygen on copper and its reactivity with hydrogen at 373-423 K, and found that the heat ofK. DOMEN, S. NAITO, M. SOMA, T.ONISHI A N D K. TAMARU 853 adsorption of 0, on the oxidized surface was lower than that of the formation of Cu,O (334 kJ mol-l). They also reported that oxygen having a heat of adsorption of ca. 293 kJ mol-1 was most reactive with hydrogen, which may be the same species discussed here. As for the activation of hydrogen, it is possible to suppose two active sites where the hydrogen molecule can attack, i.e. on copper and on oxygen. To interpret the data mentioned above, the latter seems to be more probable, and the hydrogen molecule will attack dissociatively adsorbed oxygen on the surface which is 'covered' by adsorbed oxygen. This adsorbed oxygen may be the oxygen atom in the uppermost layer of the Cu,O lattice. In other words, the surface of Cu,O can catalyse H,O formation from H, and 0,.LOWER OXYGEN PARTIAL-PRESSURE REGION In the region of the partial pressure of 0, below 9 x lo3 Pa the reaction kinetics are still complex, but the model of adsorbed oxygen on Cu,O seems able to interpret it. The rate of H,O formation shows a maximum value at a certain oxygen pressure (ca. 1.3 x lo3 Pa) and the reaction order with respect to H, becomes lower. The H,-D, exchange reaction also starts to proceed gradually, although the rate is considerably slower than that of water formation. The surface of the catalyst is still covered with Cul but the 0 (1s) peak [fig. 8(B)] seems to be narrower on the lower binding-energy side in comparison with that of fig. S(C). These phenomena can be understood by considering another reaction mechanism, i.e.the reaction between adsorbed oxygen and adsorbed activated hydrogen on the surface of Cu,O. The amount of adsorbed oxygen on the surface of Cu,O will decrease gradually with the decrease of 0, partial pressure, and if the adsorption of oxygen and hydrogen is competitive the number of sites available for the activation of hydrogen will increase, which will cause the maximum rate shown in fig. 4. The existence of such a maximum rate may also exclude the redox mechanism via the Cuo/Cul cycle in this region. If the reduction of lattice oxygen in Cu,O proceeds and Cuo is produced on the surface during the reaction of H, and O,, Cuo should be oxidized rapidly to CuI. The progress of the oxidation of the catalyst during H, and 0, reaction was confirmed even at the lowest 0, pressure (ca.6.7 x 10, Pa) in fig. 4, and X-ray excited Auger electron spectroscopy could not detect the existence of Cuo on the surface during H,O formation [fig. 8(B)]. However, since the rate of reduction of the lattice oxygen is accelerated when the surface is further reduced and Cuo is clearly detected by X.P.S. (as will be shown later), this redox mechanism (Cuo/Cul) could not predict the existence of the maximum rate in fig. 4. In this region the dependence of the rate of H,O formation on H, is less than first order, which suggests that the rate-determining step changes from the activation of molecular hydrogen to the reaction between H(a) and O(a) on the surface. In view of the fact that the amount of adsorbed hydrogen is not so great (0 < O.l), and the H,-D, exchange reaction-is rather slow in comparison with H,O formation, the probability that H(a) reacts with O(a) must be larger than that of the recombination. The number of active sites for the reaction should be small (probably < 10% of copper atoms on the surface), considering the above results.The H,-D, exchange reaction is supposed to proceed on Cul slowly or on Cuo which is occasionally present on the surface. Although hydroxyl species, which give the 0 (1s) peak at 53 1.4 eV as Cu(OH), [fig. 8 (F)], might be expected to be present on the surface during the reaction, we could not find clear evidence of such a peak.854 CATALYTIC OXIDATION OF H, ON CU REDUCTION OF THE LATTICE OXYGEN The initial rate of the reduction of Cu,O is shown at zero 0, pressure in fig.4, which may correspond to the rate where there is no adsorbed oxygen on the surface. The successive reduction rate is much faster than the initial rate, and at the apparent stationary state which gives constant reaction rate it is faster than the maximum rate in fig. 4. At this stage the H,-D, exchange reaction proceeds rapidly (fig. 7), and Cuo is clearly detected on the surface [fig. 8 (E)]. Accordingly the increase in the reduction rate is attributable to the increase in the number of active sites for the activation of molecular hydrogen, i.e. the number of Cuo atoms on the surface. On the other hand, these phenomena can clearly exclude the possibility that Cuo on the surface is relevant to the phenomena shown in fig. 4, i.e.the rate of water formation has a maximum value at low partial pressures of 0,. The reduction may occur at the boundary of copper metal and copper oxide, as indicated by Pease and Taylor.e CONCLUSION The essential points of the proposed reaction mechanism between hydrogen and oxygen on copper are as follows. Within the range of our experimental conditions for the catalytic formation of water, the reduction of adsorbed oxygen predominates over that of lattice oxygen, and the reaction proceeds on the surface of the Cu,O layer which grows during the reaction. G. C. Bond, Catalysis by Metals (Academic Press, London, 1962). A. T. Larson and P. H. Emett, J. Am. Chem. SOC., 1925, 47, 346. A. F. Benton and P. H. Emett, J. Am. Chem. SOC., 1926,48, 632. A. F. Benton and J. C. Elgin, J. Am. Chem. SOC., 1926, 48, 3027. A. F. Benton and J. C. Elgin, J. Am. Chem. SOC., 1927, 49, 2426. F. E. Smith, J. Phys. Chem., 1928, 32, 719. W. A. Bone and R. V. Wheeler, Philos. Trans. R. SOC. London, Ser. A, 1906, 1, 206. R. N. Pease and H. S. Taylor, J. Am. Chem. SOC., 1922,44, 1637. lo A. T. Larson and F. E. Smith, J. Am. Chem. SOC., 1925, 47, 346. l1 F. J. Wilkins and S. H. Bastow, J. Chem. SOC., 1931, 1525. l9 A. B. Van Cleave and E. K. Rideal, Trans, Faraday SOC., 1937, 33, 635. l3 W. G. Palmer, Proc. R. SOC. London, Ser. A., 1923, 103, 444. l4 P. E. Larson, J. Electron Spectrosc. Relat. Phenom., 1974, 4, 213. l5 G. Schon, Surf. Sci., 1973, 35, 96. l6 V. E. Ostrovsky and N. N. Dobrovolsky, Proc. 4th Int. Congr. Catal., Moscow, 1968 (Akademia Kaido, Budapest, 1970), p. 46. 'I M. Boudart, D. M. Collins, F. V. Hanson and W. E. Spicer, J. Vac. Sci. Technol., 1977, 14, 441. (PAPER 1/596)