J. CHEM. SOC. FARADAY TRANS., 1994, 90(9), 1301-1306 I8O Tracer Studies of CO Oxidation with 0, on MOO, Part 1.-Diffusion of *O Atoms from Active Sites during the Catalysis and the Determination of the Number of Active Sites Yasuo lizuka Department of Chemistry and Materials Technology, Kyoto Institute of Technology, Matsugasaki, Sakyo-ku, Kyoto, 606,Japan The catalytic oxidation of CO with l80,has been carried out on MOO, at 753 K. The isotopic composition of the product CO, was examined as a function of time. The amount of C '*02was always negligible. The percentage of C'80'60 increased rapidly at the beginning of the reaction then gradually reached a plateau value. Inter- ruption of the oxidation did not significantly affect the percentage of C'80160in the CO, produced.Oxygen vacancies produced by the reduction of MOO, with CO apparently remained unchanged on the surface after evacuation for 120 min at 718 K. The incorporation of 0, into oxygen vacancies took place rapidly. A small fraction of "0 atoms taken up into oxygen vacancies were recovered in subsequent CO reduction processes, while the greater part of the l80diffused into the bulk of the MOO,. The catalytic oxidation of CO with l80,op MOO, could be explained in terms of mobile and immobile sites with regard to ''0 atoms taken up into vacancies of active oxygen. This model enables calculation of the time dependence of the yield of C l80l60in the product CO, and estimation of the number of immobile sites. "0 tracer techniques have frequently been used for eluci- dation of the role of surface oxide ions in the catalytic oxida- tion of CO and hydrocarbons on metal 0~ides.l-l~ Keulks found that only a small fraction of oxygen in oxidized pro- ducts is "0 in the oxidation of propene with "0, over a bismuth molybdate catalyst.' He inferred that oxide ions in several subsurface layers participate in the oxidation, and the diffusion of oxygen from the surface into the bulk and from the bulk to the surface must be rapid.Wragg et al. showed, using '0-enriched gaseous oxygen and "0-enriched bismuth molybdate catalysts, that the oxygen atoms in acrol- ein are derived from the metal oxide in the catalytic oxida- tion of propene., Similarly, the participation of lattice oxide ions in the formation of acrolein was demonstrated in the selective oxidation of propene on multicomponent metal oxidesg and on multicomponent bismuth molybdate cata- lyst~.",'~ The participation of lattice oxide ions was also shown in catalytic oxidation of CO on V205,4 V205 doped with Mo6+,, y-Fe,O, and Pr6011." cussion of the isotope exchange is reported in the following paper. Experimental Analytical-grade MOO, (99.5%) from East Merck Company was used as the catalyst.According to the manufacturer's data, its principal impurities are C1, SO,, Pb, Fe and NH,; each is present at levels < 100 ppm. The specific surface area of MOO, was determined to be 0.40 m2 g-' by the BET met hod. CO, 0, and CO, , obtained commercially, were purified by fractional distillation at low temperature."0, ("0-atom fraction, 98.6%) from Commissariat a L'Energie Atomique was used without further purification. All reaction mixtures of CO and 0, or ''0, had a molar ratio of CO : 0, of 2 : 1. The details of the procedures employed have been described previou~ly.'~.' However, the apparatus was modi- fied slightly in order to combine the isotope tracer technique Krenzke and Keulk~,~ with kinetic measurements, as shown in Fig. 1. The circula- and Moro-oka and co-worker~~.~~ determined the amount of lattice oxide ions entering the oxi- dized products from variations in the "0 content in the oxi- dized products with the progress of the catalysis. However, the number of surface oxide ions that are active sites in the course of the catalysis has rarely been determined to date.The kinetics and mechanism of CO oxidation with 0, on MOO, were elucidated in our previous paper by using a usual kinetic method combined with electrical conductivity measurements of the catalyst during the The object of this paper is to determine the number of surface oxide ions that are active sites during the catalysis of CO oxidation with 0, on MOO,. An l80tracer technique was combined with the kinetic method for this purpose. Two problems are encountered when estimating the number of active sites from the variation in the "0 content in C02 with the progress of the catalytic oxidation of CO with "0,. One problem is the diffusion of l80atoms from active sites into the bulk during the catalysis and the other is the occurrence of oxygen exchange between "0 atoms in the product CO, and surface lattice l60oxide ions of the catalyst. The diffu- sion of ''0 from active sites into the bulk is considered in the present study.Oxygen isotope exchange between CO, and surface lattice oxide ions is briefly described. A detailed dis- tion loop of a reaction mixture had two pathways. One passed through metal cocks B and C, and trap TI; another passed through metal cocks F and G, and trap T, . Hereafter, the former is referred to as the T, pathway and the latter as the T, pathway. A quadrupole mass spectrometer (ULVAC MSQ-150) was connected to the T, passway through a Granville-Phillips variable-leak valve.MOO, powders (1.0-9.9 g) were pretreated at 800 K with circulating 0, (ca. 7.0 kPa) for ca. 3 h. Subsequently, the catalyst was cooled to the reaction temperature and evac- uated. Then a reaction mixture of CO and 0, (ca. 5 kPa) was admitted into the loop (TI pathway). The product CO, was always condensed in trap T,, which was cooled with liquid nitrogen. Having confirmed the establishment of steady cata- lytic activity, the mixture was removed from the loop. Fresh MOO, was usually subjected to this pretreatment prior to catalytic studies. In a typical isotopic experiment, a mixture of CO and I8O2 (1.33 kPa) was introduced into the loop. T, was always cooled with liquid nitrogen, while T, was cooled, if necessary. The variation of the isotopic composition of CO, with reac- tion time was examined at regular intervals (usually every 20 min) by switching the circulation loop from TI to T, for a J.CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 Fig. 1 Schematic representation of the reactor portion of the recir- culatory reaction apparatus. The circulation loop has two circulation pathways. A-H, metal stop cocks; QR, quartz tube reactor; EF, elec-tric furnaces; PG Pirani vacuum gauge; CP, circulation pump; M, U-shaped mercury manometer; TI-T, ,liquid-nitrogen-cooled traps ; V, variable-leak valve; IP ion pump; QM, quadrupole mass spec- trometer. fixed period (usually, 2 min) and analysing the CO, con-densed in trap T, mass spectrometrically. During the cata- lytic reaction, the pressure was maintained at 1.33 kPa (_+3%) by supplying an adequate amount of the mixture through cock E. Before the next isotopic experiment, the catalyst was evacuated at 800 K and exposed to a mixture of CO and 0, for 5 h until the l80content in the product CO, became negligible.The reaction system was changed into a static one for the examination of isotope exchange between CO, and MOO,, where the static system was enclosed with metal cocks A, B, G and H (Fig. 1). The catalyst was exposed to CO, contain-ing l80. The CO, used was obtained from the oxidation of CO with “0, in the course of the catalysis. The isotopic composition of CO, was followed at intervals of 15 min. The consumption of CO, in the mass analyses was negligible compared to the total CO, in the system. Electrical conductivities of the catalyst were measured as described previ0us1y.l~ Rates of uptake of 0, into the reduced catalyst were determined from pressure changes mea- sured with a Pirani gauge.Traps T, and T, were cooled con- tinuously by using liquid nitrogen or a dry ice-acetone mixture to protect the catalyst from contamination due to grease and mercury vapour. Results CO Oxidation with ‘*O,on MOO, and Oxygen Isotope Exchange between CO, and MOO, The catalytic CO oxidation with 1802on MOO, at 753 K was performed for 122 min at a total pressure of 1.33 kPa. The yield of C 1802 in the product CO, was always negligi- ble in comparison to those of C l80l60and C 1602.The yield of C l80l60is plotted against reaction time in Fig. 2. Almost all CO, was C 1602at the beginning of the catalysis. The percentage of C “0 l60increased rapidly with time in the early stages of the catalysis but showed a tendency to saturation, with a plateau value of ca. 13%. The time depen- dence of the yield of C l80l60on the left-hand side of Fig. 2 I I I 0 30 60 90 120 150 180 210 240 t/min Fig. 2 Variation of the yield of C l8Ol6O in CO, as a function of time. The catalysis between CO and “0,on MOO, was performed at 753 K, maintaining the pressure of the reaction mixture at cu. 1.33 kPa for 122 min. CO, collected for 2 rnin at the end of the reaction was exposed to the catalyst after evacuation at the same temperature.was independent of the amount of catalyst (1.0-9.9 g) as long as the pressure of the mixture was maintained at 1.33 kPa. The product CO, was collected in trap T, for 2 rnin at the end of the catalysis, i.e., from 120 to 122 min. The amount of oxygen in the CO, collected corresponded to ca. 1% of the total surface lattice oxide ions. The total number of surface lattice oxide ions was calculated from the surface area of MOO, and the number of oxide ions per unit surface area of MOO,, i.e. 1.7 x 10’’ rn-,.14 The fraction of C l80l60in the CO, was ca. 8%. After evacuation of the catalyst, the CO, in trap T, was vaporized and passed over the catalyst at 753 K. The variation of C l80l60yield with time is shown on the right-hand side of Fig.2. The isotopic composition of CO, barely changed with time during its contact with the catalyst. Time Dependence of the Number of Oxygen Vacancies Previously we reported that the rate of incorporation of oxygen into the reduced MOO, catalyst can be expressed by the equation -d(O,)/dt = k,, Po,(l -O), (1) where k,, is the rate constant for incorporation of oxygen into the catalyst and Po, is the pressure of oxygen. The term (1 -0) denotes the ratio of the amount of deficient active oxygen sites to the total amount of active oxygen atoms on the ~ata1yst.l~ The oxidized catalyst was first reduced at 718 K with circu- lating CO (ca. 2.0 kPa for 30 min) and then evacuated. The amount of oxygen vacancies was 3.6 x mol (3.4 x 10” m-,) from the amount of CO, collected in trap T,.However, the total amount of active oxygen atoms on the catalyst was estimated to be 1.3 x 10l8 m-,, assuming that the total amount agreed with the limiting amount of CO, produced in the reduction of the catalyst with CO for a sufficiently long time.14 The estimated value corresponded to 7.6% of the total surface lattice oxide ions on MOO,. Therefore, (1 -8) was 0.26. The catalyst was then exposed to oxygen (0.54 Pa for 3 min) 14 rnin after the beginning of evacuation in the static system. The pressure of 0, , which decreased rapidly owing to 0, uptake, was measured every 30 s. After the 0, had been completely removed, the catalyst was evacuated for 43 rnin and exposed to oxygen once more (0.54 Pa for 3 min).The decrease in 0, pressure was measured. The same pro- cedures and measurements were repeated after a further evac- uation for 57 min. Fig. 3 shows the decrease in 0, pressure J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 0123456 t/min Fig. 3 Decrease of the oxygen pressure due to incorporation into MOO, that had been reduced by CO at 718 K. Oxygen (0.54 Pa) was admitted for 3 min (n) 14, (b) 60 and (c) 120 rnin after evacuating the co. with time for the three cycles of oxygen dosing. The total times elapsed after starting the first evacuation were 14, 60 and 120 min. Eqn. (1) and the slopes of the straight lines in Fig. 3 gave (1 -6) values at 60 rnin and 120 min. The two (1 -6) values obtained were corrected for oxygen incorpor- ation.Fig. 4 shows the corrected values of (1 -0) at 14, 60 and 120 min. There is no appreciable change in (1 -6) with time, implying that the oxygen vacancies produced by CO reduction are not diminished in number even after 120 min. Electrical Conductivity Changes due to Oxygen Uptake MOO, , previously oxidized with circulating oxygen, was reduced at 713 K with CO (6.6 kPa for 30 min). After evac- uating the reduced catalyst for 10 min, oxygen (0.13 Pa) was introduced into the reactor for 4 min. Then the catalyst was evacuated again for 75 rnin at the same temperature, and sub- sequently exposed to 0, (0.13 Pa for 4 min) again. The same exposure of the catalyst to oxygen was repeated after a further evacuation for 60 min.Fig. 5 shows that the electrical conductivity of MOO, increased significantly after reduction. Evacuation did not alter the electrical conductivity, but introduction of oxygen (0.13 Pa) caused a slight decrease. At the end of the experi- ment, i.e., at 205 min, 0, (2.6 kPa) was introduced into the system. The electrical conductivity of the catalyst returned immediately to the value before reduction. The reduction of active oxide ions on the surface with CO leaves behind charges in the catalyst. The charge left behind may be responsible for the electrical conductivity. However, the decrease in the electrical conductivity is caused by the incorporation of 0, into the oxygen vacancies. 0 30 60 90 120 150 t/min Fig.4 Variation of the deficient ratio of active oxygen, (1 -O), with time -evac evac io 60 90 1;o 150 180 t/m in Fig. 5 Variation of the electrical conductivity of the catalyst during reduction, evacuation and reoxidation of MOO, at 713 K Effect of Interruption on the Catalytic Oxidation of CO with 1802 CO oxidation with "0,was carried out on MOO, at 753 K for 120 min. The pressure of the mixture was 1.33 kPa. After the first evacuation (60 min), a mixture was introduced again into the catalytic system, and the oxidation of CO was per- formed for 70 rnin at the same temperature and pressure as before. The second evacuation was for 110 min, after which catalysis was carried out for 60 rnin under the same condi- tions. Fig. 6 shows the variation of the C "0 l6O yield in the product CO, throughout these cycles.The decrease in the percentage of C'80160 in the product CO, caused by the two interruptions is relatively small and independent of the length of interruption. The percentage of C "0 l60rapidly returned to the level it had attained before the interruption and then increased gradually with time. Recovery of "0 from the Active Sites by CO PulseReduction MOO, was first reduced for 30 min with CO (2.64 kPa) at 753 K. This yielded many vacancies of active oxygen on the surface. After removal of CO, the reduced catalyst was oxi- dized with "0, (2.64 kPa) for 5 min. The results in Fig. 5 show that "0atoms are instantaneously taken up by almost all the vacancies. After continuous evacuation of the sample for 10 min, the catalyst was again reduced with CO (27 Pa) for 5 min to recover "0 from the active sites.The amount of l8O recovered was determined from the amount of CO, col-lected in trap T, and the proportion of C "0 l6O in it. This CO pulse reduction for 5 rnin was repeated twice more at intervals of 30 min. The "0 recovery experiment was per- formed twice on catalysts that had previously been oxidized 30.0 1 1 -CO+'802 co + 1 802 co + 1e02 c 3 evac e--+ evac ( 3 n ( 3 ( ) 20.0-v 0 -.-0 ..---------_--.-. 10.0-9 1 -0.oJ 1 I I b 6.0 E 0 30 60 -90 120 150 180 tlmin Fig. 7 Variation of the amount of l80recovered in each CO pulse reduction with time.Successive CO pulse reduction experiments were started 10,60 and 120 rnin after evacuating the ''0,at 753 K. with 180, and continuously evacuated for 60 rnin and 120 min before starting the CO pulse reduction. Fig. 7 shows the amount of l80recovered in each CO pulse reduction process as a function of the time elapsed after the evacuation of . The amount of l8O recovered decreased markedly with the repetition of the CO pulse reduction. However, the amount recovered after the first reduction was almost inde- pendent of the elapsed time. The total amount of l80recov-ered was only a fraction of the total amount of taken up and was almost independent of evacuation time in each case. Total Amount of Recoverable "0on the Active Sites The CO oxidation with 180, was carried out for 240 rnin on MOO, at 753 K and 1.33 kPa.After the yield of C l80l60 reached saturation, the reaction mixture was removed com- pletely and the CO, in trap T, was pumped off. A mixture of CO and 1602 was introduced into the circulation loop and the catalytic oxidation of CO was performed at 1.33 kPa for >600 min. All the product CO, was collected in trap T,. The catalysis was continued until the amount of C "0 l6O was negligible. The total amount of l80recovered from the active sites was 1.6 x lop6mol (4.0 x 1017 m-') from the amount of CO, in trap T, and its C l80 l60content. Discussion Oxygen Isotope Exchange between CO, and MOO,during Catalysis Miura et al. investigated the catalytic oxidation of CO with 1802 on Bi, Ca, Ni and La molybdate catalysts by measuring l80-atom fractions in CO, .6 The distribution of in CO, was in good agreement with the value calculated on the assumption that isotope mixing between CO, and several layers of surface lattice oxygens of the catalyst was rapidly completed.6 The formation of C l80,was always negligible in the present experiment, suggesting that isotope mixing is slow on MOO,.Our previous study showed that the rate of CO oxidation with 0, on MOO, was independent of the pressure of COZ.l4 This means that the adsorption of CO, on the active sites is extremely weak. The weak adsorption may be responsible for the slow isotope mixing between CO, and the surface oxide ions of MOO,. We confirmed that the time dependence of the C l80l6O yield shown on the left-hand side of Fig.2 was independent of the amount of catalyst. The contact time of the product GO, with the catalyst bed during the catalysis varied with the amount of catalyst. The right-hand side of Fig. 2 shows that the exposure of the CO, produced at the end of the catalysis to the catalyst surface does not change the percentage of J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 C l60in it appreciably. This suggests that the isotopic composition of the product CO, is hardly altered by contact with the surface lattice oxide ions of MOO, in the course of the catalysis. Therefore the observed isotope composition must be that of the CO, just produced at active sites and it reflects the isotopic composition of the active sites.Since the yield of C "0, is always negligible compared to those of C "0 l60and C 1602, we conclude that the percentage of C l80l6O in the product CO, reflects the fraction of l8O in total active oxygen. Diffusion of l8OAtoms from the Active Sites during Catalysis It was reported previously that the catalysis of CO oxidation with 0, on MOO, proceeds by the cyclic repetition of the following two steps: CO(g) + O(S)+CO,(g) + (s) (1) 2(s) + O,(g) 20(s) (11)+ where O(s) refers to active oxygen on the surface of MOO, and (s) refers to an active oxygen mcancy on the ~urface.'~ A CO molecule in the gas phase reacts with abtive oxygen on the catalyst to produce CO,, which is desorbed, leaving an oxygen defect on the surface in step (I).An oxygen molecule in the gas phase is dissociatively and irreversibly adsorbed on two paired defects in step (11). Oxygen adsorption on paired defects is supported by the quantitative relation expressed by eqn. (1). The rate constant of step (11)is much larger than that of step (I). Accordingly, almost all the active sites are covered with oxygen during catalysis [the (1 -0) value for active oxygen is only ca. l-2%].14 The rate of reduction of the active oxygen by CO is dynamically balanced with the rate of uptake of oxygen by the paired active oxygen vacancies at the steady state.I4 The I8O2used was 98.6%pure; however, Fig. 2 shows that hardly any CO, molecules produced at the beginning of the catalytic oxidation contained atoms.Thus the active oxygen with which CO molecules react must be lattice oxide ions on the surface of MOO,. Since the pressure of the C0-180, mixture was always constant during the catalysis, the amount of l80incorpor-ated into MOO, must increase in proportion to the reaction time. Previously we reported that the rate constant for the production of CO, at 753 K is 6.0 x lo-' mmol kPa-' min-' rn-,.14 The amount of l8O atoms taken up during catalysis up to 122 rnin corresponds to 3.6 times the number of active oxide ions on MOO,. The amount of active oxide ions, determined by the CO reduction method, is 1.6 x 10l8 m-' at this temperat~re.'~ If we assume that all active oxide ions work equally well as active sites and do not migrate, the percentage of C l8Ol60in the product CO, should reach 96% after 120 rnin from the beginning of the catalysis.If all surface lattice oxide ions work equally well as active sites and do not migrate, the yield of C l80l6O should be 28% at 120 min. Fig. 2 shows that the yield of C180160 increases rapidly at the beginning of catalysis and levels off gradually. The yield of C l80 l60 is 8% at 120 min, showing that the diffusion of l80atoms from active sites into the bulk must occur simultaneously with the diffusion of l6O oxide ions from the bulk to the active sites. However, the oxide ions at the active sites cannot be in equilibrium with the oxide ions in several subsurface layers in the bulk because (1) the yield of C'80160 shows a tendency to saturation at a value much lower than 98.6%(Fig.2) and (2) the amount of l80in CO, produced in the CO pulse reduction decreased markedly with repetition of the pulse (Fig. 7). If oxide ions in several sub- surface layers equilibrate with those at the active sites and J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 participate equally in the catalysis, the percentage of C l80 l60should reach 98.6% after a sufficiently long reac- tion time and the amount of l80recovered should be inde- pendent of the number of CO pulse reductions. Oxidation and Diffusion of Vacancies of Active Oxide Ions Keulks suggested that in the oxidation of propene with l80, over bismuth molybdate catalysts the adsorption of 1802 occurs elsewhere than the site for propene adsorption.2 Bielanski and Haber stated that oxygen vacancies produced during the catalytic oxidation of hydrocarbon molecules on surfaces of oxysalts such as molybdates and tungstates are reoxidized not only by oxygen molecules from the gas phase but also by oxide ions diffusing from the bulk.16 However, Fig.4 shows that oxygen vacancies produced by CO reduction remain without any diminution on the surface even after 120 min. On the other hand, the adsorption of 0, on reduced MOO, occurs rapidly at 718 K independent of the elapsed time under evacuation (Fig. 3). The adsorption of 0, probably occurs at active oxygen vacancies produced by CO reduction. Fig. 5 shows that a sharp decrease in the conductivity is accompanied by the adsorption of 0, on oxygen vacancies.Eqn. (1) and the decrease of the conductivity suggest that the oxidation of paired vacancies of active oxygen by an 0, mol-ecule takes place according to 02(g)+ 2(s) + 4e --+ 2o2-(s) where the excess charges in the reduced MOO, may be fixed on oxide ions on active sites. On the other hand, Fig. 5 also shows that the electrical conductivity of the reduced MOO, does not change when the catalyst is evacuated. The decrease in (1 -0) is small, as shown in Fig. 4. This probably means that the reoxidation of oxygen vacancies on the surface by oxide ions diffusing from the bulk must be accompanied by the diffusion of oxygen vacancies into the bulk of MOO,.Fig. 4 suggests that this might be an extremely slow process at this temperature. Diffusion of Oxide Ions and Characteristics of Active Sites We can conclude from these results that oxygen vacancies produced during the catalytic oxidation of CO with 1802 are oxidized by 1802 molecules in the gas phase and not by l60 oxide ions diffusing from the bulk. Consequently, Fig. 2 shows that a large fraction of the l80atoms taken up into vacancies of active oxygen must subsequently diffuse from active sites into the bulk. The diffusion of l80must occur prior to reaction with CO and must be accompanied by replacement with l60oxide ions diffusing from the bulk, because a large part of the CO, produced is C 1602 even at 120 min. The diffusion of l80oxide ions into the bulk and their replacement with l60oxide ions does not need to produce vacancies in the bulk.Fig. 6 shows that the interruption of the catalysis does not affect the percentage of C l80 l60in CO, . This suggests that the fraction of l80 on active sites does not change during the interruption. Accordingly, a fraction of atoms taken up into active sites during catalysis must remain at active sites. A greater part of CO, produced in the CO pulse reduction pre- sented in Fig. 7 is C 1602, indicating that a large part of "0 atoms taken up into oxygen vacancies diffuse into the bulk and are replaced by l60oxide ions diffusing from the bulk. However, Fig. 7 also shows that there is a small amount of "0 atoms remaining on the active sites and these can be recovered by the CO pulse reduction. // "0 mobile sites Q c1602is constantly Vtotal < produced "0immobile sites The percentage of C'80'60 varies with time from 0 to 98.6% Fig.8 Reaction model of CO oxidation with "0,on MOO, based on immobile and mobile active sites. The reaction rate is composed of two parts which proceed independently on two kinds of active sites. All the present results can be explained in terms of two kinds of active sites: mobile sites and immobile sites. Fig. 8 shows a schematic diagram for mobile and immobile sites in the catalytic CO oxidation with 180, on MOO,. Both sites contribute independently to the catalytic oxidation of CO. The atoms taken up on mobile sites during the catalysis immediately diffuse into a huge amount of l60oxide ions in the bulk prior to their reaction with CO and, therefore, cannot be recovered during the CO reduction step.In other words, l80 atoms at mobile sites are immediately replaced by l60oxide ions supplied from the bulk. As a result, the mobile sites are constantly covered with l60oxide ions and only Cl60, is produced there. On the other hand, l8O atoms taken up into immobile sites during the catalysis do not diffuse and can be recovered by the CO reduction step, resulting in the formation of C l80l60.Although all immo- bile sites are covered with l60before the beginning of the catalytic oxidation of CO with the percentage of l80 at the immobile sites gradually increases with time and finally reaches 98.6%.Concomitantly the percentage of C l8O l60 in CO, produced on the immobile sites also increases grad- ually and finally reaches 98.6%. The percentage of C l80l60 in the CO, produced on all the active sites increases grad- ually and shows a tendency to saturation at a value much lower than 100% (Fig. 2). Time Dependence of the Yield of C "0 l60and Determination of the Number of Immobile Sites According to the proposed reaction model, the yield of C l8O l60in CO, produced over all the active sites can be expressed as follows : C l80 l60(Yo)= (ratio of the rate of production of CO, at immobile sites to the total rate of production of CO,) x (percentage of in active oxide ions at immobile sites) =RxC (2) The ratio R is estimated to be 0.13 from the percentage of C l80 l60at saturation in Fig.2. The "0 percentage at the immobile sites, C,varies with time. The rate of uptake of l80atoms by immobile sites is equal to the rate of production of CO, at the immobile sites at steady state.14 The rate of consumption of l80atoms at the immobile sites is equal to the rate of production of C l80 l60.The difference between uptake and consumption corresponds to the change in the fraction of l80at the immobile sites. The material balance gives 98.6V dt -VC dt = N dC (3) 15001 0 20 40 60 80 100 120 140 t/min Fig. 9 Time dependence of the C '*O l60yield in CO, calculated according to the proposed reaction model. Various numbers of immobile sites N were assumed and calculated theoretical curves were compared with the observed data (0).(a) 1.0%,(b) 1.8%,(c) 2.0%,(d)5.0%.where N is the number of immobile sites, 98.6 is the percent- age of l80in and V is the rate of production of C02 at the immobile sites. Integration of eqn. (3) yields C = 98.6[ 1 -exp(-Vt/N)] (4) where t is in min and the initial condition of integration is C = 0 at t = 0. The combination of eqn. (2) with eqn. (4) gives the time dependence of the yield of C l80l60 in CO, . The rate of production of CO, at immobile sites, i.e. V can be obtained from the product of the total rate of production of CO, and the ratio R. However, the number of immobile sites, i.e. N, is unknown. We attempted to calculate the time dependence of the yield of C l80l60in CO,, assuming N to be 1.0, 1.8, 2.0 and 5.0% of the total surface lattice oxide ions.The results are com- pared to typical experimental results in Fig. 9. The number of immobile sites N can be estimated accurately because theo- retical curve changes sensitively with N. The best fit was obtained with N = 1.8%. The number of immobile sites is quite small on the surface of MOO,. After the oxidation of CO with "0, has been performed for a sufficient time, e.g. for 240 min on the catalyst, all immobile sites are expected to completely covered with l80 atoms. These l80atoms may be recovered by subsequent oxidation of CO with 1602 for sufficiently long, e.g. >600 min. The total amount of "0 atoms recovered corresponds to the number of immobile sites.The observed amount of l80 atoms recovered was 1.6 x mol (4.0 x 1017 rn-,). On the other hand, the number of immobile sites calculated J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 from the time dependence of the yield of C l80 l60using the method presented in Fig. 9 was 1.7 x mol (4.3 x 1017 rn-,). This agreement not only confirms the model of immo-bile and mobile sites presented in Fig. 8 but also means that "0 atoms incorporated into mobile sites do not participate in the catalytic reaction at immobile sites. The percentage of l80 at immobile sites does not change during the interruption of the catalysis. The solid curve in Fig. 6 shows the calculated result according to the proposed reaction model. The curve reproduces the experimental results fairly well.The percentage of C "0l60 just after each interruption is slightly lower than the calculated curve. However, the extent of lowering is independent of the length of interruption. A small limited portion of l80atoms at immobile sites might be replaced very slowly with l60oxide ions from the bulk during the interruption. The author thanks Professor Shigeyoshi Arai and Plenary Professor Takuya Hamamura of Kyoto Institute of Tech- nology for their helpful discussions and advice. References 1 A. Roiter, Kinet. Catal., 1960, 1, 63. 2 G.W. Keulks, J. Catal., 1970, 19,232. 3 R. D. Wragg, P. G. Ashmore and J. A. Hockey, J. Catal., 1971, 22, 49. 4 H. Kakioka, V. Ducarme and S. J. Teichner, Kinet. Catal., 1973, 14, 78. 5 T. Otsubo, H. Miura, Y. Morikawa and T. Shirasaki, J. Catal., 1975,36,240. 6 H. Miura, Y. Morikawa and T. Shirasaki, Nippon Kagaku Kaishi ,1975, 11, 1875. 7 L. D. Krenzke and G. W. Keulks, J. Catal., 1980,61,316. 8 K. Sakata, F. Ueda, M. Misono and Y. Yoneda, Bull. Chem. SOC. Jpn., 1980, 53, 324. 9 Y. Moro-oka, W. Ueda, S. Tanaka and T. Ikawa, Proc. 7th Znt. Congr. Catal., Tokyo, 1980, ed. T. Seiyama and K. Tanabe, Kodansha, Tokyo, part B, p. 1086. 10 W. Ueda, Y. Moro-oka and T. Ikawa, J. Catal., 1981,70,409. 11 Y. Takasu, M. Matsui and Y. Matsuda, J. Catal., 1982,76,61. 12 I. Brown and W. R. Patterson, J. Chem. SOC., Faraday Trans. I, 1983,79,1431. 13 I. Matsuura, H. Hashiba and I. Kanesaka, Chem. Lett., 1986, 533. 14 Y. Iizuka, Y. Onishi, T. Tamura and T. Hamamura, J. Catal., 1980,64,437. 15 Y. Iizuka, Y. Onishi, T. Tamura and T. Hamamura, J. Catal., 1981,70,264. 16 A. Bielanski and J. Harber, Catal. Rev. Sci.Eng., 1979,19, 1. Paper 3/02846A; Received 19th May, 1993