J. Chem. SOC., Faraday Trans. I, 1986,82, 2025-2041 Catalytic Reactions on Metal-supported Semiconductors Oxidation of CO over ZnO Films on Silver Eliezer Weisst and Mordechai Fohnan* Department of Chemistry and Solid State Institute, Technion I.I. T., Haifa 32000, Israel The catalytic activity of silver-supported zinc oxide films in CO oxidation has been studied. Attention has been directed to the influence of the semiconductor-metal contact on the catalytic behaviour of the semicon- ductor. Auger electron spectroscopy and electron microscopy have shown the presence of zinc oxide films uniform both in thickness and composition, which were free of contaminants. The kinetics of the catalytic reaction have been studied in a gas-flow system. The results fit two rate equations leading to two possible mechanisms: one with a rate-determining step which is a reaction between adsorbed CO and adsorbed oxygen, without mutual displacement; the other which corresponds to the Eley-Rideal mechanism.Both mechanisms assume a dissociative, asymmetric, acceptor-type adsorp- tion of the oxygen molecule, and a donor-type adsorption of the CO molecule. It has been found that the elementary constants in the rate equations for CO oxidation depend on the semiconductor film thckness. Good agreement between the experimental and the theoretical results has been found. The Schottky model, together with the rigid band model, have been used in the theoretical considerations. The Schottky barrier height for the ZnO-Ag system has been calculated using the kinetic results, and is in very good agreement with the measored values reported in the literature.According to the so-called ‘electron theory of catalysis’1-3 or the ‘ rigid band the concentration and distribution of electrons in the catalyst play a predominant role in catalytic processes and their interpretation. To determine the elementary steps of a catalytic reaction one has to change the distribution of the electrons over the quantum states of the bands at the surface of the solid, i.e. to change the Fermi level position in the energy gap at the surface. In a semiconductor (s.c.) catalyst supported by a metal the large number of electrons in the support can modify the spatial distribution and the average free energy of the relatively few electrons in the S.C.This takes place only near the interface. In fig. 1 (a) the phases of the metal and S.C. are separated. We choose the case where the metal has a larger work function (&J than the n-type S.C. (&). In fig. 1 (b) the two phases are in electric contact. As was suggested by Schottky,g electrons diffuse from the S.C. into the metal until their Fermi levels are equalized. This results in build-up of a space charge in the boundary layer, extending to a depth 1 in the s.c.; E, and E, have been bent upwards, and the distance of the Fermi level from Ec has increased. Within the boundary layer, this effect favours donor-type catalytic reactions, i.e. reactions where the reactant donates electrons to the catalyst to form the activated state. It is clear from this explanation that electronic interaction effects can change the catalytic activity only if the surface of the S.C.catalyst lies within the width of the boundary layer ( I ) from the metal support [fig. 1 (c)]. One can determine the dependence of the catalytic energy of activation (EaJ on the thickness ( d ) of the S.C. catalyst layer7 using the Poisson equation and the Schottky t Based in part on a D.Sc. dissertation presented to the Senate of the I.I.T. by E. Weiss.2026 CO Oxidation over ZnO Films on Silver C F F %a= &I-x C C I z I I 1 O d 1 Fig. 1. Energy bands in an n-type semiconductor in contact with a metal (&, > # s c ) : (a) before contact, (b) on contact, with S.C. layer thicker than the screening length ( I ) , (c) on contact, with S.C. layer thickness ( d ) smaller than the screening length.F-F: Fermi level. V-V: Top of valence band. C-C: Bottom of conduction band. barrier (+SB). For a donor-type reaction on a rectifying contact and an n-type S.C. (q4m > q4sc) one obtains Eat = aD-N +SB--d2 (1) ( E 2ne2N ) where a, is a constant, N is Avogadro’s number, N is the number of charge carriers within the s.c., e is the elementary charge and E is the S.C. dielectric constant. Metal-supported semiconductor catalysts were studied by Schwab et and Steinbach.15 As catalysts they used S.C. films on metal foils, S.C. powders distributed on metal foils or mixtures of powders of the two components. According to these authors, in some cases considerable differences in the apparent energy of activation were obtained.In one case only7y (p-type S.C. with q4sc > &J the results were compared with the theory and some correlation with an equation similar to eqn (1) was found. These former studies lack chemical and morphological characterization of the S.C. surfaces and the bulk, as well as the s.c.-metal interfaces. In those instances in which the S.C. layer was prepared by oxidizing a metal film supported by metal foil8? l3 there was no indication of the termination of the oxidation in the film-foil interface. There was no direct determination of the S.C. layer thickness, though this is the independent variable in eqn (1). In the studies where one or both phases were powders9? lo, l2 there was the possibility of bifunctional catalytic activity of the mixtures. The influence of the s.c.-metal interface on the S.C.catalytic activity was examined via the change in the apparent energy of activation. In those studies no detailed analysis of the energies of activation of the elementary steps was undertaken. The main objective of the present work was the study of the influence of the metal-s.c. interface on the rates of elementary reactions in a redox catalytic system. To achieve that, the catalyst was prepared under well controlled conditions and was characterized carefully to verify the presence of a continuous S.C. film of uniform thickness and composition. Special stress was put on the absence of contaminants, on the surface and in the bulk of the catalyst as well as at the metal-s.c. interface. The experiments were designed to enable the finding of the detailed rate equation and the influence of the metal-s.c.contact on the elementary rate constants.E. Weiss and M. Folman 2027 Experiment a1 Catalyst Preparation Thin zinc oxide films deposited on silver foils (Fine, Holland-Israel Ltd) served as catalysts. The films were produced by r.f. sputtering from a 99.99 wt % ZnO target, 6 in? in diameter, using a Perkin-Elmer model 2400 sputtering system. The distance between the target and the substrate was fixed at 1.5 in. After pumping the bell-jar down to < Torrt pure argon was admitted and sputtering was performed at 0.022 Torr Ar. Prior to deposition, the substrate was sputter-etched at 250 W for 5.5 min. Thin films were produced at a deposition rate of ca. 40 A min-l at 100 W and 500 V bias (for deposition times exceeding 30 min the deposition rate increases up to 80 A min-l at 120 min sputter-deposition). Catalyst Characterization Catalyst Composition Catalyst surfaces and chemical composition beneath the surface of the as-deposited films following the catalytic reaction were characterized by Auger electron spectroscopy (AES) in a PHI 590A system using a primary electron beam of 3 keV and 1 PA.Depth profiles were obtained by monitoring the peak-to-peak signal heights of the Auger transitions 0 KLL (503-510 eV), Zn LMM(994 eV) and Ag MNN (351 eV) during 4 keV Art ion-sputter etching. Catalyst Morphology The morphology of the catalysts was established by scanning electron microscopy (SEM) using JEOL JSM C-35 and JEOL JEM T-200 instruments.The electron beam energy was 25 keV. Electrical Conductivity The d.c. conductivity of ZnO films deposited on a microscope slide glass (6 x 6 mm) was measured by the van der Pauw method in a specially designed cell which was pumped down and then filled with gas up to 1 atm.$ As contacts served films of gold 3000 A thick deposited by electron gun evapQration. The sample was heated in the selected atmosphere for a pre-set period of time and was allowed to cool to room temperature for conductivity measurements. Film Thickness Step-height measurements were performed by means of a stylus, using a Sloan Dektak profilometer, and interference (wavelength 2950 A), using Reicher interferometer. For interference measurements the samples were coated with 1000 A gold layers.t 1 in = 2.54cm. $ 1 Torr M 133.3 Pa. 0 1 atm = 101 325 Pa.2028 CO Oxidation over ZnO Films on Silver +15 Fig. 2. Diagram of apparatus for kinetic experiments. 1, 0,-CO-He mixture cylinder; 2, He cylinder; 3, pressure control unit; 4, four-port isolation valve; 5, needle valve; 6, microcombustion furnace; 7, Pyrex reactor; 8, micro filter; 9, metering valve; 10, gas sampling valve; 11, gas chromatograph; 12, 13, external separation columns; 14, cooling bath; 15, line to flow meters and vent. Catalytic Experiments Apparatus The kinetics of the catalytic reaction were studied in a flow system operating at atmospheric pressure and described in fig. 2. The reactor was a Pyrex tube with Kovar ends 240 mm long (1 80 mm in the oven) and 14 mm i.d. The reactor ends were covered by asbestos paper to prevent light penetration into the reactor.The temperature of the reactor was controlled within f 1 K. Gas-flow velocity was varied from 5 to 30 cm3 min-l. No detectable reaction was found on the walls even at the lowest flow rate at temperatures below 400 "C. Detection of Reactants and Products The compositions of the gas mixtures were established by gas chromatography (g.c.) analysis. Separation was achieved using four columns, 1 m long and 1/4 in in diameter, filled with Porapak N, Porapak T (Waters Assoc.), Chromosorb 102 (Johns Manville) and Molecular Sieve 5A (B.D.H.). The grain size was 8&100 mesh. The first two columns were kept at 150 "C; the two others were at 14.5 "C. Helium served as the carrier gas. Detection was by means of the thermal conductivity method.Fig. 3 shows a representative chromatogram. The chromatograms were standardized by admitting pure gases or mixtures of known composition to the gas-sampling valve at various pressures directly from the vacuum system. A linear relationship between the chromatograms' heights and the gas pressures was found. Catalytic Procedures Reaction mixtures were prepared in an evacuated gas cylinder in the following manner : the cylinder was filled with CO (oxygen) up to a partial pressure of 670 Torr, oxygen (CO) was then admitted up to the preselected pressure (1-4 atm). Finally helium was added to the mixture up to a pressure of 4 atm.E. Weiss and M. Folman 2029 co c x cos AL I co il A c x z + c I1 i \ h I I I 0 1 2 3 4 5 6 7 8 9 1011 121314151617181920 tlmin Fig.3. Gas chromatograph separation for CO oxidation. (a) Superimposed chromatograms of pure gases simulating product mixture. (b) Typical chromatogram of product mixture. Flow through external columns (see fig. 2): I both, 11 12, 111 13. The catalyst was activated (in the reactor) for at least 12 h at 390 "C by passing the reaction mixture over it. The influence of activation on the catalytic activity is shown in fig. 4. Catalyst activity was determined by the composition of the product mixture as a function of flow rate, feed mixture composition and reactor temperature. Steady-state composition was established after attaining reproducibility of 1-274 . The system reached the steady-state composition in under 15 min.The steady-state stability was checked for hours and sometimes even for days.2030 CO Oxidation over ZnO Films on Silver 1.9 1 . 8 " I ; l a 7 E E ;=; 1.6 0" g1.5 1.4 '4 --. 1 . 3 1 0 5 10 15 20 25 tlh Fig. 4. Change of catalytic activity with time of activation (Po,/Pco = 6). Rate Calculations The rate calculations were based on the rate equation of a mixed-flow reactor.16 Since the feed mixture does not contain the product (CO,) the steady-state equation is given by where Pco, is the CO, partial pressure within the reactor and at its outlet (Torr), U is the flow velocity of the reaction products (cm3 min-l), V is the reactor volume (cm3), P is the reaction rate (Torr min-l) and t, is the residence time (t, = V/U min). Results and Discussion Catalyst Characterization Auger Electron Spectroscopy AES surveys of two 'as prepared' ZnO films on silver foils are shown in fig.5. It may be seen from the spectrum in fig. 5(a) that no detectable contaminants are present on the film surface. The thin film has the same composition as the pure zinc oxide powder which served as standard [fig. 5 ( c ) and (d)]. The ZnO-Ag interface is also free of contaminants and there is no additional layer of different composition at the interface. This can be seen from the AES surveys taken at the interface itself [fig. 5(6) and (e)]. In films of thickness < 100 A, holes (ca. 1 pm in diameter) in the ZnO layer were found, as judged from the AES line scans (fig. 6). The amount of Ag detected in those cases was ca. 5% . In films > 200 A the amount of silver detected was < 1 % (which is the limit of detection of the Auger system).The thickest film produced (ca. 1 pm thick) showed cracks after its use in the reactor. Catalysts, which were employed in the reactor for a number of days up to few weeks and maintained at 300-390 "C and 1 atm of 0,-CO mixture, were analysed by AES. In films of thicknesses > 100 A and < 10000 A, no changes in surface composition or depthE. Weiss and M . Folman 203 1 2 Ag Zn I , I l l , l I l 110 330 550 770 990 kinetic energyleV z6 .3 s 5 7 r 1 6 1 0 5 c I I l J l \ I I I 1 0 2 4 6 8 10 12 14 16 sputter time/min 110 330 550 770 990 kinetic energy/eV Fig. 5. AES of ZnO films on silver, prepared by r.f. sputtering. (a)-(c) Sample 315 (thickness: ca. 1200 A); (a) spectrum of the surface; (b) spectrum of the surface obtained after 8 min sputtering; (c) depth profile; sputtering rate of Ta,O, 200 A min-l; (d)-(e) sample 316 (thickness: ca.200 A); (d) depth profile; (e) spectrum of the surface obtained after 4 min sputtering; sputtering rate of Ta,O, 50 8, min-l. Zn 0 1 2 3 4 5 6 7 8 Erm Fig. 6. Auger line scan of zinc oxide film (100 8, thick) on silver. profiles were observed and no interdiffusion of the components of the two phases was noticed. Also, no traces of Ag were found on the films' surfaces. To avoid any possibility of catalytic activity of the Ag substrate, only films in the range 200-1 200 A were employed.Plate 1. SEM analysis of thin zinc oxide films on silver. Film thickness: (a) 1 pm; (b) 600 A.E. Weiss and M. Folman (Facing p . 2032)2032 CO Oxidation over ZnO Films on Silver Scanning Electron Micrc zopy SEM studies of the morphc ogy of the catalysts showed a granular structure. The grain size could not be determin i. The step structure [plate l(a)] indicates that the growth of the film shows no prefei zd crystallographic direction. Results for the thinner fi IC are LQC A.*A--- ,- - - instance, for a 600 A film, although a step was prepuLwu, IL Luuiu nave been detected only after the film w a s purposely damaged [plate 1 (b)]. Film Thickness Results of film-thickness variation within the deposition system are described in fig. 7. The longest dimension of the catalyst (its diagonal) is equal to the target radius, therefore the film-thickness variation along the catalyst is < 5 % .Very high reproducibility in the film thickness was obtained. Electrical Conductivity The room-temperature specific conductivity of a 600 A thick zinc oxide film as a function of oxidation time at 400 "C is shown in fig. 8. The high conductivities of the untreated film [point (a)] and after 15 min heating (at 400 "C) in helium lpoint (b)] are due to interstitial zinc, resulting from film deposition in an oxygen-free atm0~phere.l~ Oxidation of the film caused a monotonic decrease in conductivity, owing to oxidation of the interstitial zinc. This was noticed for the first 7.5 h, after which the conductivity reached a constant value of ca. s2-l cm-l. CO Oxidation over ZnO Films The understanding of the influence of the s.c.-metal interface on the catalytic behaviour of the S.C.requires a detailed understanding of the kinetic mechanism of the reaction. For that reason CO oxidation over the zinc oxide films had to be characterized kinetically. Inhibition by CO, It was found previously18 that the CO, produced may inhibit the CO oxidation. Therefore, a study of the kinetics of this reaction should start by determining the extent of the inhibition by CO,. The influence of flow velocity on the product pressure is described in fig. 9. The linear relationship means that in our system inhibition by CO, can be neglected. CO Oxidation Rate Equation To investigate the influence of the thickness of the S.C. catalyst on the reaction rate a suitable rate equation had to be found. Fig. 10 and 11 show some representative results in which the reciprocals of the reaction rates show a linear dependence on the reciprocals of the oxygen and CO pressures. These results obey the following rate equations :19 (3) r = kPo,Pco/[(1 +KO,PO,)(l + ~ c o ~ c o ) 1 which corresponds to a reaction between two adsorbed molecules with no mutual displacement; and: (4) = kP0, PCO/(l +KO, Po, + Kc0 PCO).E.Weiss and M . Folman 2033 10 000 8 000 2 3 v) 2 6000 % 4 000 2000 F t -- ---I I I ! I I I I 3 2 1 0 1 2 3 distance/in Fig. 7. Film thickness variation as a function of distance from the centre of the cathode in the r.f. sputtering system. 1 o3 lo2 1 0' - '5 loo 2 lo-' - I 10-2 1 d3 loq4 I I I I I I 0 2.5 5 7.5 10 12-5 oxidation timelh Fig. 8. tl s Fig. 9. Fig. 8. Room-temperature specific conductivity of a ZnO film (thickness ca.600 A) as a function of oxidation time at 400 "C. (a) Sample as deposited; (b) after 15 min heating (at 400 "C) in helium. Fig. 9. The influence of flow velocity on the product pressure.2034 CO Oxidation over ZnO Films on Silver 3 I 0.6 kz E E 2 0 . 4 4 -. 6 9 0.2 0 100 200 300 400 500 600 700 1 . 2 1 . o - I v) g O . 8 E .-O. 6 --... 4 n c 0.4 0.2 t / 0 100 200 300 400 500 600 700 Po, /mmHg Fig. 10. CO oxidation rates as a function of reactant pressure for several catalysts at two temperatures (in "C). Points are experimental values. The lines are calculated curves according to either eqn (3) or eqn (4). Eqn (4) corresponds to the Eley-Rideal reaction mechanism between two adsorbed molecules, of which one is weakly adsorbed.Although the latter is adsorbed on the adsorption sites of the second reactant, it does not react from this state. The two possible mechanisms differ only in the adsorption strength of the two reactants. The linear dependence in fig. 1 1 indicates that the adsorptions are first order. In accordance with the results presented above, inhibition by the product has not been included in eqn (3) and (4). (Additional mechanisms, such as Langmuir-Hinshelwood or two-step 'redox mechanism'20 were tested, but were not found to be applicable to our system). In order to find the elementary constants in the rate equations we shall designate by i, the intercept of the straight line obtained from plotting t - l us. P,: for a constant pressure of CO (Pc0,,), and by S, its slope.Likewise, the intercept and slope of r-l us. P& for a constant pressure of oxygen (Poz,y) are iy and S,, respectively. The rate constants in eqn (3) were calculated using the following expressions:E. Weiss and M. Folman 2035 7 I 25 E E 0 1 2 3 4 5 6 7 0 1 2 3 4 5 6 7 0 lo3 mmHg/Po2 1 O3 mmHg/Pc. Fig. 11. Inverse CO oxidation rates as a function of inverse reactant pressures for several catalysts at two temperatures (in "C). (a) Constant CO pressure. (b) Constant 0, pressure. The terms for k calculated by the last equation using the experimental results are practically equal. The rate constants in eqn (4) were calculated by k = Po,, yes, - cr p,O,,~l-l = [P,O,z(~& Po2, y)I-l KO, = i, kP,,,, = S, k- P& K,, = i, kPo2,, = Sx k- Pcb,,.(6) The two right-hand terms in each of eqn (6) are equal within experimental error. In a few cases differences between the two terns for KO, were found. The lines in fig. 10 are calculated plots of eqn (3) and (4) using the parameters obtained from either eqn ( 5 ) or (6), respectively. Both equations fit the experimental results well, therefore it is difficult to decide which of the two proposed rate equations describes better the oxidation of CO over ZnO. Mechanism of CO Oxidation over ZnO Films The detailed mechanism which gives the above rate equations will be outlined below. The mechanism will be based on the assumptionla* 21-23 that the rate-determining step is the reaction between adsorbed oxygen and adsorbed CO. that in the temperature range 373-453 K oxygen molecules adsorb on ZnO as 0; ions.On the other hand, at temperatures above 503 K the adsorption is dissociative and 0- is formed. Our experiments showed that the oxidation rate is linearly proportional to Po, (and not to Pg,"). Owing to this we have to assume that on adsorption a dissociation of the oxygen molecule into two different species takes place. Of the two species only one is charged. Therefore the adsorption of oxygen proceeds according to : Kl It has been 0 2 , g =O,,a, (7)2036 CO Oxidation over ZnO Films on Silver where ad stands for adsorbed, s for a surface atom, e- is a free electron and V& stands for a neutral oxygen vacancy at the surface. Of the two oxygen atoms obtained as a result of the reaction between 02,ad, a free electron and an oxygen vacancy, one is being adsorbed as 0-, while the other is incorporated into the surface.As was proposed by Doerfler and Hauffer21 we assume a donor-type adsorption of CO, and like Goepel and we designate it by K3 CO + 0, + C02 -V& +e-. (9) Here V$,s stands for an ionized oxygen vacancy at the surface. The main difference between the two mechanisms lies in the energy of the adsorption in the stage described by eqn (9). For the Eley-Rideal mechanism this adsorption is weak and the adsorption of CO on oxygen sites has to be taken into account: K 4 CO, + toad (on Old adsorption sites). co,d does not participate in the oxidation mechanism. The rate-determining step is (1 1) k co, ’ v:,~ + o k d -$ C02,ad. Since no inhibition by CO, was observed, the rate-determining step is followed by fast desorption of the product: C02,ad c02,g (fast)m (12) The surface-state energy of adsorbed oxygen atom lies very low with regard to the Fermi level of Zn0,23 therefore this adsorbed oxygen atom is completely ionized. Since the equilibria (7) and (8) are fast, the coverage is given (according to the Langmuir isotherm) by and for the Eley-Rideal mechanism by The surface-state energy of the donor V& (ED,s) is very close to the Fermi-level energy,23 therefore its degree of ionization is given by the Fermi-Dirac statistics.l. 2* 23 The surface coverage of the active species is given by and for weak adsorption KO2 - v;,s1 = F(E) K3 PCO.F(E), the Fermi-Dirac distribution function, is given byE. Weiss and M.Folman 2037 TI0C 390 370 350 330 I I I I I I I lo3 KIT Fig. 12. Arrhenius plots for the elementary constant k for various catalysts: (a) according to eqn (3), (b) according to eqn (4). Equating eqn (16a) and (16b) with eqn (3) and (4) one obtains k = k, Kl K, K3 F(E) KO, = 4 K,, = K3 or Kc, = K4. The Influence of the ZnO Film Thickness on its Catalytic Activity The influence of the ZnO film thickness on its catalytic activity can be studied in view of the rate equations for the oxidation of CO over ZnO films proposed in the former sections. This was done by studying carefully the elementary constants .of eqn (3) and (4) * The energies of activation (EaJ and the frequency factors ( A ) of the elementary constants k were determined from Arhenius plots for films of different thicknesses.Representative Arhenius plots are shown in fig. 12. Fig. 13 shows the change of E,, and A with the thickness ( d ) for the two rate equations. Inspection of fig. 13 reveals a pronounced influence of the thickness of the oxide layer on both the energy of activation for the rate constant k and its frequency factor. However there is no such influence for films >600 A thick. The results for K,, and KO, of eqn (3) and (4) are given in table 1. Inspection of the results for Kco according to eqn (4) shows that for all films (except for the thinnest2038 CO Oxidation over ZnO Films on Silver d l a 0200400 600 ,, 1200 I I I I / I .-. - I 0 E 3 30 3 \ w" 2 5 2 0 14 12 10 8 T c 3 6 4 2 0 1 2 3 " 9 10 li / 1 O2 minZ Fig. 13. Activation energy and frequency factor as a function of the semiconductor film thickness (d), according to eqn (3) (0) and eqn (4) (A).Table 1. Values of K,, and K02 in the rate equations of CO oxidation ... 4 n . 1 r, I , V T 1-1 m ,*a I, I , -1 \--1 aeposi tion iu- ~co/(mmng) A iu- n02/(mmng) A no. /min /102A T/"C: 340 355 370 385 340 355 370 385 catalyst time, t, d 316 31 1 304 315 nrrnrdino tn Pnn (A1 \ '/ ---"'-"'a c v .,-A* 5.0 2.0 1.8 1.6 2.4 2.1 11.2 4.5 0.25 0.66 0.91 0.98 15.0 6.0 1.8 2.7 3.2 3.7 30.5 12.0 3.5 3.8 4.7 5.6 316 311 304 315 according to eqn (3) 5.0 2.0 4.7 4.2 5.4 4.2 11.2 4.5 0.51 2.4 2.8 3.3 15.0 6.0 2.0 3.0 4.0 4.0 30.0 12.0 4.4 5.0 6.0 7.0 7.0 7.0 12.0 9.2 0.14 0.56 1.1 1.2 _- 12.0 11.0 13.0 13.0 0.20 - 2.3 - 5.2 9.0 0.22 1.1 5.0 11.0 0.24 1.1 9.2 12.0 0.34 2.2 8.0 16.0 0.36 2.1 catalyst where the results are scattered) E,, = 1Of 1 kcal mol-l and 1nA = 1.5f0.2.When Kco is calculated by means of eqn ( 3 ) one has to use values for k calculated by eqn (17). This results in a large error. Therefore we were not able to characterize the temperature dependence of the Kco constant. However, inspection of the results in table 1 reveals no dependence of Kco on the ZnO film thickness. KO, evaluated for both mechanisms does not show any clear dependence on temperature for all thicknesses. Its value differs from one catalyst to another with no correlation with its thickness.E. Weiss and M. Folman 2039 In light of the results presented above we conclude that both K,, and KO, do not depend on the thickness of the oxide film, while k does depend on it.Reduction of the film thickness causes a decrease in the activation energy as well as in the frequency factor. 26 Our catalyst was an n-type S.C. with dSc < dm. According to eqn (1) the energy of activation should decrease on reduction of the S.C. thickness. Inspection of eqn (16a), (16b) and (17) reveals that k depends on the Fermi-level position, while K,, and KO, do not, which is in accordance with our results. The dependence is via F(E). According to Goepel and ED,,-Ec = 0.2 eV (where, as above, ED,, is the energy of the donor surface state CO.V,,,); therefore, ED,s- E, > 0.2 eV. In the temperature range of our experiments (610-670 K) the Boltzmann approximation can be applied to eqn (17) (with an error < 5%). CO oxidation over ZnO is a donor-type k = k , Kl K2 K3 F(E) Only the last exponent in eqn (18) depends on the ZnO film thickness, and the changes in its value are the changes observed in Eac, shown in fig.13. As is expected from eqn (l), the energy of activation is linearly dependent on the square of the S.C. layer thickness up to a certain value above which the energy of activation remains constant. Schottky Barrier Height Calculation The foregoing results will be taken into account in calculating the Schottky barrier height in our catalytic system. The difference between the activation energy at thickness approaching zero and the maximum activation energy (of the free oxide) is ca. 12 kcal mol-1 (fig. 13). This is the extent to which the ZnO bands are bent due to the Ag support: Yo x 12 kcal mol-1 x 0.5 eV.Since the thickness of the space charge layer ( I ) is ca. 600 A (fig. 13) and assuming for the ZnO film a dielectric constant between 10 and 20,27 a charge carrier concentration of 1017cm-3 is obtained. Therefore the difference between the Fermi level and the conduction band edge (ECF) at 650 K (taking the electron effective mass for ZnO as 0.27 of the free electron mass)28 is:29 ECF x 0.25 eV. The Schottky barrier height should, therefore, be: $SB = Yo + E,, M 0.5 eV + 0.25 eV = 0.75 eV. This value agrees well with Mead’s results for the Schottky barrier height for ZnO-Ag (0.68 eV).30 Change of the Frequency Factor ( A ) Fig. 13 reveals also that the frequency factor ( A ) depends on the thickness of the S.C. layer.Although the catalysts had similar geometric areas their specific areas may depend on the film thickness. Generally for sputtered films the specific area, as well as the number of active sites, increase with their thickness. Therefore A also increases with the film thickness. The number of active sites is the dominant factor acting when activating a fresh catalyst. As shown in fig. 4 there was a continuous decrease in catalytic activity during the activation, until a constant value was reached. Heating the catalyst in an oxygen-rich2040 14 CO Oxidation over ZnO Films on Silver ' 12 10 T c 4 8 - 6 - - - -10 - 8, - - 6 - 1 1' 10 20 30 €,,/kcal mol-' Fig. 14. Compensation plots for the elementary constant k in the rate equation of CO oxidation over ZnO films of different thickness, according to eqn (3) (0) and eqn (4) (e).atmosphere caused an oxidation of the donor dopants, which caused an increase in ECF. Therefore, as explained above, this would have caused a decrease in Eat, i.e. an acceleration of the reaction. On the other hand the results in fig. 4 reveal that during the activation the decrease of the number of active sites plays the dominant role. As already suggested [eqn (9) and (1 l)] the active site is an oxygen vacancy at the surface of the ZnO. Under the conditions of the activation process the probability for the formation of this active site decreases drastically. The change of A in fig. 1.3 may also be a manifestation of the compensation effect,31 as can be seen in fig. 14. If this is the case, the compensation here is non-linear. Though this fact is not clear from fig.14, it is revealed by the absence of an isocatalytic temperature (a temperature where all Arrhenius plots for the different catalysts cross each other). This can be seen in fig. 12. Conclusions It has been shown that a ZnO thin film prepared by r.f.-sputtering yields an oxide, free of contaminants on its surface as well as in its bulk and at the s.c.-metal interface. This indicates that the supported ZnO films are suitable for the study of the influence of the s.c.-metal contact on the catalytic activity of the S.C. It was found that the rate-determining step in the oxidation of CO over ZnO is either a surface reaction between two adsorbed reactants without mutual displacement or a reaction between the strongly adsorbed oxygen with weakly adsorbed CO (Eley-Rideal mechanism).The proposed mechanism includes dissociative, asymmetric, acceptor-type adsorption of an oxygen molecule. While the adsorbed oxygen is completely ionized, in the donor-type adsorption of the CO molecule the degree of ionization is given by the Fermi-Dirac distribution, It was found that the activation energies of the elementary constants in the rate equations increase with the square of the thickness of the S.C. film. This agrees well with the theoretical results derived from coupling of the Schottky model, describing the s.c.-metal contact, with the rigid band model, describing catalysis on semiconductors. These models do not take into account surface parameters such as the energy spectrumE.Weiss and M . Folman 204 1 of donor and acceptor surface states and their influence on the position of the Fermi level at the surface of the s.c.; nevertheless one can apply the two models in the case of ZnO. The authors thank I. Lior, Dr R. Brener, C. Cytermann, A. Shai and A. Friedman for assistance with various aspects of the study. E. W. gratefully acknowledges the Wolf Fund for their generous support. References 1 Th. Wolkenstein, Adv. Catal., 1960, 12, 189. 2 F. F. Volkenstein, The Electron Theory of Catalysis on Semiconductors (Macmillan, New York, 1963). 3 B. Claudel, Chem. Phys. Aspects Catal. Oxid. (Proc. Spring Sch. CNRS Catal. Oxid., 1978) (CNRS, 4 S. R. Morrison, CHEMTECH, 1977,7, 570. 5 S. R. Morrison, The Chemical Physics of Surfaces (Plenum Press, New York, 1977).6 W. Schottky, Naturwissenschaften, 1938, 26, 843. 7 V. L. Vinetskii, I. V. Kel’man and D. V. Sokolskii, Russ. J. Phys. Chem., 1977, 51, 834. 8 G. M. Schwab and R. Sieged, 2. Phys. Chem. N.F., 1966,50, 191. 9 G. M. Schwab and H. Derleth, 2. Phys. Chem. N.F., 1967, 53, 1. 10 G. M. Schwab and K. Koller, J. Am. Chem. SOC., 1968,90, 3078. 11 G. M. Schwab and A. Kritikos, Helv. Phys. Acta, 1968,41, 1166. 12 G. M. Schwab and H. Zettler, Chimia, 1969,23,489. 13 G. M. Schwab and B. Matthes, 2. Phys. Chem. N.F., 1975, 94, 243. 14 G. M. Schwab, Adv. Catal., 1978, 27, 1. 15 F. Steinbach, Angew. Chem., Int. Ed. Engl., 1967, 6, 999; Nature (London), 1967, 215, 152; 2. Phys. 16 W. J. Moore, Physical Chemistry (Longmann, London, 5th Edn, 1975), pp. 350-352. 17 Handbook of Thin Film Technology, ed. L. I. Maissel and R. Glang (McGraw-Hill, New York, 1970), 18 W. R. Murphy, T. F. Veerkamp and T. W. Leland, J. Catal., 1976, 43, 304. 19 K. J. Laidler, Catalysis, ed. P. H. Emmett (Reinhold, New York, 1954), vol. 1, chap. 4. 20 P. Jim, B. Wichterlova and J. Tichy, Proc. 3rd Int. Congr. Catal., Amsterdam, 1964 (North 21 W. Doerfler and K. Hauffer, J. Catal., 1964, 3, 171. 22 I. Kobal, M. Senengancik and H. Kobal, J. Chem. Phys., 1983,78, 1815. 23 P. Esser, R. Feierabend and W. Goepel, Ber. Bunsenges. Phys. Chem., 1981, 85, 447, and references 24 P. Bonasewicz, R. Littbarski and M. Grunze, Curr. Top. Muter. Sci., 1981, 7, 371. 25 G. M. Schwab and J. Block, 2. Phys. Chem. N.F., 1954, 1, 42. 26 J. M. Thomas and W. J. Thomas, Introduction to the Principles of Heterogenous Catalysis (Academic 27 G. Heiland, E. Mollow and F. Soeckmann, Ado. Solid State Phys., 1959, 8, 191. 28 S. M. Sze, Physics of Semiconductor Devices (John Wiley, New York, 2nd edn, 1981). 29 W. Shockley, Electrons and Holes in Semiconductors (Van Nostrand, New York, 1953), sec. 16.3. 30 C. A. Mead, Solid State Elec., 1966, 9, 1023. 31 A. K. Galwey, Adv. Catal., 1977, 26, 247. Pans, 1980), p. 397. Chem. N.F., 1968,60, 126. chap. 3, p. 30. Holland, Amsterdam, 1965), vol. 1, p. 199. therein. Press, London, 1967). Paper 51493; Received 25th March, 1985 68 FAR 1