THE UNIVERSITY OF MICHIGAN INDUSTRY PROGRAM OF THE COLLEGE OF ENGINEERING OXYGEN TRANSFER AT COBALT FERRITE SURFACE Ching-Rong Huang A dissertation submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy in the University of Michigan Department of Chemical and Metallurgical Engineering 1966 October, 1966 IP-750

ACKNOWLEDGEMENTS The author wishes to express his appreciation to the members of his doctoral committee, Professors G. Parravano, J. Fo Verdieck, J. D. Goddard, E. Eo Hucke and Go B. Williams, for their guidance during the course of his work. The author is especially grateful to his Chairman, Professor Parravano who suggested the topic for this research, for his numerous helpful suggestions and invaluable criticismo The author is also grateful to Professor Hucke who served as the acting Chairman in the latter part of this work while Professor Parravano was on sabbatical leaveo The author is indebted to the National Science Foundation for its financial support for four years. Finally the author would like to thank Mr. Wo C. Gates, Jr. for reading the manuscript, to thank the staff of the Industry Program of the College of Engineering for typing and printing this manuscript, and to express my deep gratitude to my devoted wife for her constant encouragement and for the typing of the rough draft of this manuscript. ii

TABLE OF CONTENTS Page ACK IOWLEDGMENTS.................................. ii LIST OF TABLES................................................. iv LIST OF FIGURES................................................. v LIST OF APPENDICES...................................... vii NOMENCLATURE.......................................... viii ABSTRACT....................................................... ix I. INTRODUCTION............................................... 1 II. THEORY..................................................... 4 III. SURVEY OF LITERATURE..................................... 19 TV. EXPERIMENTAL APPARATUS AND PROCEDURE...................... 26 V. EXPERIMENTAL RESULTS....................................... 35 VI. DISCUSSION OF RESULTS...................................... 59 VII. CONCLUSIONS......6.9............................. 69 APPENDICES....................................................... 71 BIBLIOGRAPHY...................................................... 109 iii

LIST OF TABLES Table Page 1 Summary of Experimental Results of the Exchange Reaction of C02 and CO on Co3_ Fex 04............................ 37 2 Activation Energy of the Exchange Reaction of C02 and CO on Co3_X Fex 04................................... 37 3 Summary of Experimental Results of the Adsorption of Oxygen on Co3_x Fex 04..................................... 44 4 Summary of Experimental Results of the Desorption of Oxygen on Co3_x Fex 04................................... 5 Experimental Data on the Formation of 1CO catalyzed by 1.000 gram of Co3_x Fex 04................................. 71 6 Experimental Data on the Adsorption of Oxygen by 50.000 grams of Cobalt Ferrite.............................. 82 7 Experimental Data of Oxygen Desorption from 50.000 grams of Cobalt Ferrite.............................. 90 8 The Result of Run 17-21 of the Exchange Reaction of CO2 and CO............................................. 98 iv

LIST OF FIGURES Figure Page 1 Apparatus for the Exchange Reaction of CO2 and CO.., 28 2 Geigen Counter and Accessaries...................... 30 5 Apparatus for the Adsorption and Desorption of Oxygen.. 32 4 Formation of 14CO Versus Times on Co. g0lFe 2.09904 with A=2.48 x 104 cm2 and T=350~C....36 5 The Value m from k(ao) Versus a................. 39 6 Activation Energy of the Exchange Reaction from k(ao) Versus 1 a0.464 45 T 7 Adsorption of Oxygen on Co. 994Fe2.00604............. 46 8 Activation Energy of Oxygen Adsorption from Kads.Versus 1...................................................... 48 T 9 Rate Constant k(ao) of the Exchange Reaction Versus Composition x, T=350~C and a0=0.464............... 51 10 Activation Energy of the Exchange Reaction Versus Composition x, ao=0.464.............................. 52 11 Amount of Oxygen Adsorbed in 24 Hours Versus Composition x, t=300 C............................. 53 12 Initial Rate of Oxygen Adsorption on Co Fe 0 Versus Composition x, T=300~C and Pai=100 mirxonsXo Hg. 55 13 Initial Rate Constant of Adsorption Kaqs Versus Composition x on Co3_xFex04 at T=300 C 56 14 Initial Rate of Desorption from Co3_xFexO4 Versus Composition x at 300~C.......................57 15 Amount of Oxygen Desorbed from Co3 xFex04 in 24 Hours Versus Composition x at 300~C.............58 16 pl4C /P14 o Versus t for Run 17-21 and initial slope................................................. v

LIST OF FIGURES (CONT D) Figure Page 17 P14 /P14 Versus t for Run 17-21 and results P14CO 14CO2 from the Computer Simulating Method............... 1 18 Pa Versus t for Run 10-14 and Results from the Computer Simulating Method........................ 103 19 Composition of (100), (110) and (110) Planes in CoFe2 4........................................... 7 vi

LIST OF APPENDICES Appendix Page I Experimental Data.............................. 71 II Sample Calculation............................... 96 III Structure of Cobalt Ferrite................... 106 vii

NOMENCLATURE a lattice parameter of cobalt ferrite ao ratio of PC02 PCO A surface area of the catalyst | A tetrahedral site in the spinel structure IBI octahedral site in the spinel structure C1, C2 constants C 2 IBI C+2 ion on the tetrahedral site 0 0o 03 IB C+3 ion on the tetrahedral site o I i o Co 2 -BI Fe+3 I|B the ion pair, the proposed active center on the catalyst surface e free electron Ea activation energy F Gibbs free energy Fe+2 J[B Fe+2 ion on the tetrahedral site Fe+3 IBI +3 Fe+3 IBJ Fe 3 ion on the tetrahedral site h free hole k(ao) forward rate constant of the exchange reaction defined by equation (16) k'(ao) backward rate constant of the exchange reaction defined by equation (16) kads. adsorption rate constant defined by equation (40) k dedesorption rate constant defined by equation (49) des. "kp pre-exponential factor defined by equation (28) kf rate constant defined by equation (30) K1 equilibrium constant viii

Kads overall rate constant of adsorption defined in equation (39) m a constant defined in equation (25) n n-type semiconductor "n1C4 gm-mole of 14C0 na gm-mole of oxygen in the adsorption reservoir nd gm-mole of oxygen in the desorption reservoir N Avogadrots number p partial pressure R gas constant t time T temperature V volume of reactor or reservoir x a parameter which defines the composition of C03-x Fex 04 z concentration of CO 3 IBI in cobalt ferrite Oi~a ~a constant defined in equation (25) aPB ~ number of the active center CO+2 [BI - Fe+3 JBj per unit surface area 7ry~ ~a constant defined by equation (47) a constant defined by equation (54) Subscripts and Superscripts a refer to adsorption b refer to the bare site d refer to desorption f denote the final condition ix

i denote the initial condition o refer to the occupied site r indicate the reference state of adsorption t refer to total x

ABSTRACT Semiconductors are commonly used catalysts. Cobalt ferrite, C03_xFexO4, was very suitable to be chosen for studying catalysis. It can be made a n-type or p-type semiconductor by changing slightly the ratio of iron and cobalt. Therefore it gives an opportunity to study the effect of composition on the catalytic activity without introducing impurities into the catalyst. The exchange reaction between carbon dioxide and carbon monoxide was investigated on cobalt ferrite catalysts, C03xFex04 with four different compositions x ranging from 1.903 to 2.099. The reaction occurred in a constant volume reactor. The reaction rate was measured at temperatures ranging from 250~ to 4100C with carbonl14 dioxide as the tracer. One feature of the catalyzed exchange reaction is that the reaction rate is studied under equilibrium condition. There is no net increase nor decrease of carbon dioxide and carbon monoxide in the gas phase. The thermodynamic activity of the catalytic intermediate on the surface remains constant during the reaction. The adsorption and desorption of oxygen was investigated on five different cobalt ferrite catalysts with the same range of x as the exchange reaction. The adsorption occurred in a constant volume reactor with initial pressure of oxygen near 0.1 mm of mercury. The desorption of oxygen was under high vacuum with pressure less than 1 x 10-5 mm of mercury. The rate was measured by the pressure change of the system with the ionization gauge at temperatures ranging from 100~ to 500~C. xi

The experimental results show that the rate constant of the exchange reaction and the initial rate of oxygen adsorption are at their maximum when x is near two. This can not be explained by the conventional electronic defect mechanism which is based on the electronic defect, electron donor for the n-type semiconductor or electron acceptor for the p-type semiconductor, as the active center of the reactions, But the result can be explained by the proposed mechanism of cyclic electron-hole transfer of the cation pair which can supply an electron and anelectron hole to the catalytic intermediate and act as the active center of the reactions. This study has not only furnished basic data for the exchange reaction of carbon dioxide and carbon monoxide and for the oxygen adsorption - desorption on cobalt ferrite, but also has proposed a new concept for studying the mechanism of heterogeneous catalysis. xii

I. INTRODUCTION Catalysis plays an important role in many chemical processing industries. Let us imagine that a solid - the catalyst - is introduced into a mixture of reacting gases. The rate of reaction increase- by hundreds or thousands of times. In the absence of a catalyst the reaction either hardly occurs or proceeds very slowly. Catalysis provides a new path of stepwise reactions which are associated with the interaction between reactants and catalyst. The catalyst participates in the formation of an catalytic intermediate on its surface and returns to its original state after the completion of the chemical reaction. The increase in the reaction rate, caused by the fact that the reaction follows the path of intermediate steps with the catalyst, has been made possible by the decrease in activation energy due to a more favorable form of the bonds between the reactants. Semiconductors, like metals, are commonly used catalysts. The catalytic action of semiconductors was discovered and used in the chemical processing industry long before the concept of a semiconductor itself appeared. It is now obvious that the catalytic activity of semiconductors is very closely connected with the electronic properties inside and on the surface of semiconductors. For examples, the influence of impurities of a semiconductor on its catalytic activity has been investigated; the correlation between the electrical conductivity or thermolelectric power of a semiconductor and its catalytic activity has been discovered. In order to investigate the mechanism of the catalytic reaction, it is necessary to understand the solid state reactions of the semiconductor. -1

-2The object of this research was to study the catalytic activity of cobalt ferrite as a function of its composition. The first part of the research was the study of catalytic exchange reaction of carbon monoxide and carbon dioxide by using carbon-14 as the tracer. The second phase of the research was the study of the chemisorption and desorption of oxygen on cobalt ferrite. Cobalt ferrite, Co- Fex04, was chosen as the catalyst for the following six reasons. (1) It can be made a n-type or p-type semiconductor by changing slightly the ratio Of iron and cobalto Therefore it gives an opportunity to study the effect of composition on the catalytic activity without introducing impurities into the catlyst. (2) The crystal structure and electronic properties of cobalt ferrite has been investigated extensively. (3) It is chemically stable up to 1100~C in the atmosphere due to its close-packed, face-centered cubic arrangement of the spinel structure. With composition ranging from 1.84 < x < 2.34, the spinel is found stable in one phase. (4) It does not contain an appreciable number of lattice vacancies - Schottky type defect, or of interstitial ions - Frenkel type defect; the ratio of cations to anions can be regarded as a constant of 3:4. The characteristic defects of cobalt +2 +2 ferrite are Fe, which replaces Co in case of x > 2, and Co which replaces Fe+ in case of x - 2. (5) The thickness of the space charge layer is smaller than the interatomic distance between lattice ions. The small thickness gives the advantage of permitting the electronic boundary phenomena to be neglected and appropriate equilibrium assumptions to be used in later derivations, With experimental results, we can compare the theoretical explanation on the mechanism

-3of the catalytic reaction by the electronic defect and by the cyclic electron transfer of the cation pairs. The exchange reaction between carbon dioxide and carbon monoxide was investigated on cobalt ferrite catalysts, Co 3_Fex04, with four different compositions x ranging from 1.954 to 2.099. The reaction occurred in a constant volume reactor.. The reaction rate was measured at temperatures ranging from 250~ to 410~C with carbon-14 dioxide as the tracer. One feature of the catalyzed exchange reaction is that the reaction rate is studied under equilibrium condition. There is no net increase nor decrease of carbon dioxide and carbon monoxide in the gas phase. The thermodynamic activity of the catalytic intermediate on the surface remains constant during the reaction. The adsopption and desorption of oxygen was investigated on five different cobalt ferrite catalysts with the same range of x as the exchange reaction. The adsorption occurred in a constant volume reactor with initial pressure of oxygen near 0.1 mm of mercury. The desorption of oxygen was under high vacuum with pressure less than -5 1 x 10 mm of mercury. The rate was measured by the pressure change in the system with the ionization gauge at temperatures ranging from 100~ to 500~C.

II. THEORY A, Properties of Cobalt Ferrite (a) Structure Cobalt ferrite e3o Fe 04 is one of many composite oxides having the gereral formula AB204 which crystallize with the crystal structure called spinel [73]. The spinel structure is characterized by face-centered cubic close packing of 0- ions and A+2 and B+ metallic ions in certain interstices [9] [55]. A unit cell of spinel crystal contains eight molecules of AB204 and, therefore, thirty-two 0-o ions. The close packed unit cell contains sixty-four interstices surrounded by four 0-2 ions (coordination number 4, tetrahedral) and thirty-two interstices surrounded by six 0-2 ions (coordination number 6, octahedral). In the spinel unit cell, eight of these tetrahedral sites, denoted Jas A, and sixteen of the octahedral sites, denoted as IBI, are occupied by these metallic ions. +2 +3 The metallic ions A and B are distributed among the cation sites in different ways [4]. In "normal" spinel, all sixteen B3 ions occupy the sixteen octahedral sites and all eight A+2 ions occupy the tetrahedral sites. In "inverse" spinel, the +2 +3 sixteen octahedral sites are occupied half by A+2 and half by B+3 This has been determined by x-ray diffraction. Cobalt ferrite exists in a structure [48] which is very close to the structure of the "inverse" spinel, Fe+31AI[Co+21BIFe+31BI] 04 (see Appendix III). -4

-5(b) Composition Stoichiometric cobalt ferrite, CoFe204, hardly exists. The non-stoichiometric cobalt ferrite, Co xFe x04 has been investigated [47] with x ranging from 1.84 to 2.34. When x is larger than 2.34, a new phase of Fe203 will co-exist with the spinel phase. Also when x is less than 1.84, a new phase of wustite will co-exist with the spinel phase. When x is in the vicinity of 2, if x < 2, indicating an excess of cobalt, Fe+3 on the octahedral sites is essentially replaced by Co+3 ion. If, on the other hand, x > 2, indicating an excess of iron, Co+2 on the octahedral sites is essentially replaced by Fe+2 ions. Using this disorder model and appropriate assumptions, it is possible to obtain the concentration of each cation on different sites as a function of the composition x [48]. There are eight possibilities of cations which may occupy the lattice sited IAJ and JBI. These are Fe3 |AI|, Fe+2 I|A, Co+3 |A|, Co2 |A|, Fe3 IBI, Fe+2 BI, Co+31BI and Co+2 IBI. (i) Fe and Co balance: [Te+3lAI] + [Fe+2A^l] + [Fe+31Bj] + [Fe+2BlI] = x (1) [CO+31AI] + [CO+2eIA] + [Co+31BI] + [Co+2BI] = 3-x (2) (ii) tetrahedral and octahedral sites balance: [Fe+31AI] + [Fe+21AI] + [Co+31AI] + [Co+21AI] = 1 (5) [Fe+31BI] + [Fe+21BI] + [Co+31BI] + [Co+21BI] = 2 (4) (iii) charge ratio balance: [Fe+31AI] + [Fe+31Bl] + [Co+1A'1] + [Co+31B ] = 2 {[Fe+2}AlI + [Fe+2^Bl] + [Co+IAI] + [Co+2jB\]} (5)

-6[Fe+3A|], denoted as concentration of Fe+ in the tetrahedral sites, is assumed to be unity and independent of x, if we designate the concentration [Co+31BI] by z. Solving equations (1) to (5), yields [C+2 IBj] = 3-x-z (6) [Fe+21BI] = x-2+z (7) [Fe+31BI] = 1-z (8) (c) Electronic Property The semiconducting properties of Co 3xFex04 have been investigated with x ranging from lo9 to 2.1[35]. Electrons or electron holes can be introduced into cobalt ferrite to obtain n-type or p-type semiconductivity by varying x values. When x > 2, the excess Fe is added to CoFe204 to replace Co, it enters the crystal structure as Fe+2 BI, which acts as an electron donor and causes the ferrite to become a n-type semiconductoro In a similar way, when x < 2, the Co is added to CoFe204 to replace Fe, it enteres the crystal structure as Co +31B, which asts as an electron acceptor, causing the ferrite to become a p-type semiconductor. Therefore it gives an opportunity to study the catalytic activity of an oxide catalyst having a single substrate with either electrons (n-type catalyst) or electon holes (p-type catalyst) in excess without adding foreign impurities to the ferrite. Also the validity of the electronic defect explanation to catalysis can be investigated. Since catalytic reaction may involve the electron transfer from the catalyst surface to the catalytic intermediate, so it is necessary to look into the defect reaction of cobalt ferrite from its characteristic defects.

-7Let us first define the characteristic defect of cobalt ferrite. The characteristic defect of cobalt ferrite is defined as Fe+2 JB or Co+3J1B in the lattice structure. In the range of 1.9 < x < 2.1, Jonker [35] has shown that the characteristic defects Fe+21BI and Co+31BI are completely ionized. This means that the activition energy of ionization is very close to zero. Therefore, we may write the predominate solid phase reactions from the characteristic defects as following: For n-type, Fe+2JBI - Fe+31Bli+e- (9) For p-type, Co+3IBI - Co+2lBji+h+ (10) where Fe+3|B.i is the ionized donor in n-type cobalt ferrite. It is important to note that the electron can also be generated by the normal cations Co+3|B| and Fe+3 on the solid surface, Co+2IBI - Co+31BIi+e- (11) also Fe+3BI -> Fe+2 lBi+h+ (12) if the solid is in contact with gas molecules with high electron or hole affinity. Jonker has found that the activation energies for equations (11) and (12) are 10.94 to 11.75 Kcalo and 4.4 to 4.73 Kcalo respectively [55]. The next sections are the analysis of the catalyzed exchange reaction of CO2 and CO and of the adsorption and desorption of oxygen on cobalt ferrites based on the'solidd'sta'te:,reac.tion of the solid phase. B. CO CO Exchange Reaction (a) Description of Model The kinetics of the catalyzed exchange reaction on cobalt ferrite between carbon dioxide and carbon monoxide was investigated in

-8a constant volume reactor. Radioactive carbon-14 dioxide was employed as the tracer. An important feature of the exchange reaction of CO2 and CO is that the gas phase, composed of a fixed ratio of CO2 and CO, is in equilibrium with the catalyst surface. The following reaction was studied, 14 14 c02(g)+' CO(g) - CO(g)+ C02(g) (13) 14 with a small amount of CO2 introduced in the gas phase. It is assumed that reaction (13) follows the sequence of steps: 14Co2(g) 14co(g)+ Oads. (14) CO(g)+ 0ads. C02(g) (15) Since there is no net formation of either CO2 or CO, the activity of oxygen atom adsorbed on the surface is constant. It is obvious that the activity of oxygen on the catalyst surface depends on the ratio of CO2 and CO in the gas phase. The next three sections are the analysis of the correlation of this dependence with the properties of cobalt ferrite. (b) Evaluation of Rate Constants Since the activity of oxygen on the catalyst surface is a constant for a prefixed C02/CO ratio, the rate of formation of 14 -CO can be written as 14.' dn CO ka At k (ao) P - k (ao) p (16) Adt P C02 C14 where a = CO2 (17) o PCO and A = surface area of the catalyst. The forward and backward rate constants k(ao) and k' (ao) are defined by equation (16). Note that at constant temperature for a

-9catalyst with fixed x, k(ao) or k (ao) is a function of ao only. The net rate of formation of 14CO becomes zero, when equation (16) reaches equilibrium. It yields ( 14C2 C02 k (ao) 14 P CO = (CO k ()8) CO eq. Eliminating k'(ao) from equation (16) with equation (18), it follows that Ad = k(a) (p14 - ao p ) (19) A d t. CO2 P14CO In a constant volume reactor, the total amount of radioactive 14 C is constant, so + 14 = Pi (20) 14C02 14CO 1CO2 i 14 where p 14 is the initial partial pressure of C02 in the C02 14 reactor at t=O. There is no CO in the gas mixture at t=O. Combining equations (19) and (20) to eliminate P14co and applying ideal gas law; V dP= k(a0) [p14 - (1 + a0) P14cI (21) RTA dt c2CO In a constant volume reactor at constant temperature with prefixed CO2/CO ratio, ao and k(ao) are constants, so that equation (21) can be integrated with the initial condition P14c= 0 at t =0.

10P 14C p CO d P 14CO i d P 14 C02 p C i02 P14O = k(aO) RTA(1 + ao) ftdt 0 1 - % V 1 + ao P 14 C02 (22) In [1- (+ a) P14CO k(ao)RTA(l + aQ)t (23) In [ 1 - (1 + ao) pi I P 14 V C02 or P14C _ 1 { - exp.[-k(ao) RTA (1 + ao) t]} (24) P 14o 1 + ao 1+a At a given temperature and C02/CO ratio, the rate constant k(ao) can be calculated by equation (24) from the experimental data of the percentage of 14CO formation: and time. (c) Dependence of the Rate Constant as a Function of C02/CO ratio and Temperature The most commonly postulated mechanism to explain the catalytic reaction is based on the assumption that reacting molecules rearange themselves on the active centers of the catalyst surface. The adsorbed fragment of the reacting molecule on the active center is called the catalytic intermediate. Although there are widely different opinions on the exact catalytic intermediate and on the active center in a reacting system, this does not disprove their existence. The proposed catalytic intermediate for this exchange reaction of C2 and CO is ads. The employing of Oads as the catalytic intermediate was postulated by Wagner [75] [76] for the CO2 - CO exchange reaction on Wustite. And it was applied by Grabke [26] [27] to study the rate of oxygen transfer from CO2 to the surface of the

-11oxides Fex 0, Fe204, CoO, ZnO and MgO. There are various states in which 0ads may exist. It may exist as chemisorbed oxygen atom, chemisorbed 0 ion or chemisorbed 02 ion. The state of the catalytic intermediate is dependent on the properties of the oxide. Under equilibrium condition and assuming that the mass action law holds, we may express, in general, the thermodynamic activity of the active center as a linear function of (Pc 0 /Pco)- m which is equal to ao as defined in equation (17). m is a positive constant which is dependent on the electronic property of the active center and on the existing state of 0ads Consider the phenomenological derivation of equation (16), it is desirable to correlate the dependence of k(aO) on a with the mechanistic expression of the thermodynamic activity of the active center. The dependence can be described by k(ao) = a a(25) where a is a function of temperature, The constant m will be discussed in detail in the next section. The value of m can be obtained experimentally from the known values of k(ao) as following, fin [k(ao)] = - m S n ao + n a (26) (i n [k(ao]) = (27) n a (27) at fixed ao, the temperature dependence of k(ao) follows the relation k(a) = kpe-Ea /RT (28) where k is the pre-exponential factor, and Ea is the activation energy of the reaction which can be obtained by plotting log k(ao) 1 versus m

-12(an [k(ao)] _ Ea a n 1 R T (d) The Proposed Active Center and Catalytic Intermediate The proposed active center for the CO2 - CO exchange reaction is the ion pair Co BI - Fe+3IBI. The ion pair Co+21BIFe+1BI appears on plane (100) of the "inversed" spinel structure of cobalt ferrite. The average inter-ionic distance of the pair is 2o9(A. (see Appendix III). The characteristic feature of the pair is that they can supply to the catalytic intermediate Oads. an free electron and an electron hole with which the simultaneous catalyzed oxidation of CO and reduction of CO? in the exchange reaction can proceed. An electron donor itself, for example, the characteristic defect Fe +2BI is not qualified as an effective active center. The reason is that without an hole supplied from the active center the Oads is very strongly bound to the active center and usually hinder the course of catalysis. The hindered effect by the strongly bound catalytic intermediate was also shown in many other catalyzed reactions [2]. It was established long time ago that molecules possessing an electron donor group and a hole donor group may act as catalysts in homogeneous catalysis. It was known as the dual theory of catalysis [7]. Taylor [68] [69] bases his support of the dual theory on the rate of hydrolysis of various esters in presence of acids with and without the addition of salts. He has found that the addition of IN KC1 to 0.1N HC1 causes a 24% increase of reaction rate. From conductivity data, the degree of dissociation of O.1N HC1 is

-13about 90% and this is reduced to 75% by the addition of iN KC1. Taylor, therefore, concludes that the increase in the reaction rate caused by the addition of KC1 is due to the formation of more undissociated HC1 which acts as a catalyst by having an electron donor group H and a hole donor group Clo Recently, the mechanism of alkylations of diazoalkanes catalyzed with fluoroboric acid was based on the complex formation of the diazoalkane and HBF4 with H as the electron donor group and BF4 as the hole donor group [5] [46]. There are other pairs, eg. free electron - free electron hole pair, pairs of the characteristic defect and the normal lattice cations, Fe+21BI - Fe+31BI or Co+2BI - Co 31BI and pair of two different types of characteristic defects Fe +2B - Co+31BI which may supply an electron and a hole to the catalytic intermediate. But they are rejected as the active center in the following discussion. The free electron - free electron hole pair are difficult to be found on the surface of semiconductor with the inter-particle distance in the order of magnitute of angstrom. The two opposite charged particles of the pair tend to collide with each other at that distance. The pairs Fe+21BI - Fe+31BI and Co+21BI - Co+31BI are rejected because of their unbalanced ability of gererating a free electron and a free hole, which was discussed in Section A(c). the electronic property. The overall electronic characteristics of the pairs are an electron donor andi.a hole donor respectively. The pair of characteristic defects Fe+ BI - Co+31BI are extremely unstable because each defect is completely ionized to form a free electron or a free holeo After the reaction of the electron and the hole, the pair become Fe+3- Co+2 which is the proposed active centero

-14It is worth noting that although the characteristic defects Fe+21BI and Co+31BI are rejected as active centers, they may effect the activation energy of ionization of the proposed active center. Any characteristic defect located next to the proposed active center (see Appendix III) may alter the polarity of the center so that the potential field, which effects the activation energy of ionization, is changed. With the proposed Co I+BI - Fe |B|I as the active center for the exchange reaction, we may write the rate expression of the forward reaction of equation (16) as the following: d n 4CO =-kf [Co+2 BI - Fe BI] P140) A d t CO( where kf is the forward reaction rate constant, and [Co+2BI - Fe +3BI] is the activity of the bare active center Co2 |BI - Fe+3IBI on the catalyst surface. Recall that the CO2 - CO gas mixture is in equilibrium with the catalyst surface. This means that the activity of the occupied active center and of the bare center remains constant during the exchange reaction. We may write the equilibrium reaction as CO2(g) + CO +IB - Fe+31BI - (Co+3IBI- Fe+3BI) - O-ads + CO(g) (31) it follows k= [(Co+31BI - Fe+3IB() -0- o ] (P 2)- 1 1 [Co+ IBI - Fe"'Bi ] P,,I (32)

-15where [(Co+31BI - Fe+31B- 0 ads] is the activity of the occupied center. Ratterman (49) has found that the conductivity of cobalt ferrite catalyst is nearly independant of the ratio of CO and CO 2 in a flow reactor. From the material balance of free electrons, [e]total = [e-]solid +dl[e ]trans (33) where [e-] is concentration of the free electron. The subscripts "total" refers to the total amount, "solid" refers to the solid phase and "trans" refers to transfer to catalytic intermediate. c1 is a proportional constant. In order to find the effect of changing ao on the free electron concentration, differentiate equation (33) with respect to a d [e-]total d [e ]solid d [e"] tran (34) + C d a d 1 d aa The left hand side of equation (34) is zero, because [e-]total is a constant. The first term of right hand side of equation (34) is zero, which was experimentally found by Rattermano Therefore [e ]trans is a constant independent of a o Since [e-]trans = [(Co+31BI - Fe+3JBI) 0 - 0- ] we may rearrange equation (32) and yield ads [Co+2BI - Fe+31BI] = c2 ao 1 (35) where c2 is a constant independent of ao. The next task is to compare the different expressions of the forward reaction of equation (14). From equations (16), (25), (30) and (35), we have

-16d n14io -m A- d t-= k(ao) P14c a 14 A d t CO 0 C02 2 = kf[Co+BI - Fe+3IBI] p14CO P C02 -1 = c2 x kf P14C f P CO CO2 (36) Since by definition ao= PCo2/PCO, so it is concluded from equation (36) that m=l, if the state of oxygen on the active center is 0-. By the same analysis, if we have the state of ads' 2 -1 oxygen ads instead of ads on the catalyst surface, we would have the equilibrium reaction COg)2 Co+2IBI - Fe5+31BI (Co+3.: I - Fe+31B )2 - O-ad+ CO(g) (37) and [Co+53BI - Fe+31Bi = constant x (P 0-) 5 (38) CO The value of m would be 0.5. C. Chemisorption and Desorption of Oxygen Chemisorption of oxygen on cobalt ferrite occurs when an electron prossessing the necessary activation to pass through the potential barrier, reacts with the colliding oxygen molecule on the ferrite surface. Therefore, the rate of adsorption is strongly effected by the number of oxygen molecules colliding with a unit area of the surface per unit time, by the type and number of active centers which donate electrons and by the activation energy for each type of active center.

-17The number of collisions between oxygen molecules and the surface can be obtained from the kinetic theory of gases. It is directly proportional to the pressure of oxygen of the system and inversely proportional to the product of one-half power of the temperature and the molecular weight of oxygen. The activation energy of chemisorption is a function of the electron affinity of the oxygen molecule, the activation energy needed for the jump of an electron from the active center and the work function to transfer an electron to the oxygen molecule. If the surface coverage of the adsorbed ions is high, the interaction of the adsorbed ions is also important. Hill [31] has treated models with nearest neighbor interactions from the viewpoint of lattice statisticso The last and the most important factor which effects the rate of adsorption is the type and number of active centers to be considered in the chemisorption of oxygen on cobalt ferriteo It has been postulated by numerous authors that the electronic defects on the surface of the semiconductor are the active centers for chemisorption. If a semiconductor has defects to the extent of 0.1% to 1% of defects of the normal lattice ions than the surface coverage will be between 0.1% and 1%. But experiments [3] [32] show that the percentage of the surface coverage is far more than 1%. In the chemisorption of oxygen on cobalt ferrite, it is postulated that two types of active centers are available. Type one is the "strong" chemisorption center which is composed of the characteristic defects of Fe +2B|. Type two is the "weak" chemisorption center which is composed of the normal lattice cation pair Co+2|B| - Fe + 3BI.

Both types have the ability to donate free electrons. As is discussed in the Section B (c), Fe+21BI is completely ionized with the +2 activation energy of ionization equal zero. The Co IBI of the weak chemisorption center Co+2JBI - Fe+31BI is partially ionized with the activation energy of ionization from 4.4 to 4.73 Kcal. The oxygen molecule is adsorbed more strongly.on the strong chemisorption center than on the weak chemisorption center, because the latter has the ability to generate a free hole. Since the cobalt ferrite was prepared by firing under the oxygen in the atmosphere and was exposed to the air for more than one year, it is appropriate to assume that all the strong chemisorption centers are covered by oxygen molecules and that the weak chemisorption centers are the ones which participates in the adsorption and desorption of oxygen in this study. After hypothesizing the active centers, namely Co+2 BI - Fe+3BI pairs, for the adsorption and desorption study, it is necessary to choose a reference state for the surface, this state must have a constant number of bare active centers per unit surface area before adsorption experiment.

III. SURVEY OF LITERATURE In this chapter, a brief review of the recent literature related to the catalytic activity of cobalt ferrite is reported. The literature is divided into three sections, the first of which is concerned with the study of the catalyst, cobalt ferrite. The second section emphasizes the catalyzed exchange reaction between CO2 and CO. The third concerns the adsorption and desorption of oxygen on semiconductors. A. Study of Cobalt Ferrite The crystal structure of spinel was determined to be.face-centered cubic by Bragg [9] for MgA1204. Barth and Posnjak [4] pointed out the two possibilities of distributing the cations while retaining the cubic symmetry of spinel. The electronic conductivity and cation arrangement of a large number of spinel oxides were studied by Verwey and co-workers [72] [73]. The relations between the electronic conductivity of certain spinels and the arrangement of the cations in the crystal structure were studied. Gorter [25] summarized the experimental and theoretical data from literature on cation distribution of spinels and carried out measurements of the saturation magnetization against temperature for a number of mixed crystal oxides with spinel structure. The phase diagram for the Fe-Co-O system was constructed by Robin and Benard [54] based on X-ray diffraction data at temperatures up to 1000~C. Smiltens [62] studied the isotherms of the same system at 1200~C, 1400~C and 1626~C. He also reported that the non-stoichimetric cobalt ferrite with a spinel structure has the metal to oxygen ratio of -19

-20Let the activity of the occupied centers at the reference state be [Co+31BI - Fe+31BI] or and integrate the above equation to obtain [Co+31BI - Fe+31BI] - [Co+31BI - Fe+31BI] or Va N ApRT ( Pai - Pa (4) where p a is the initial pressure of oxygen before the adsorption Rearrange the above equation and combine with equation (41): [Co+21BI - Fe+3lBI]b = [CoI+2BI - Fe 35Bl]t - [CO+31BI - Fe+3BI Va Np,) (44) - [C~o+3B - Fe+35BI ]or - ART (Pai - Pa) Put equations (42) and (44) into the rate equation (40): Va d. pa _ [Co +2) i +53 i C+3 +3 Va dpa kads. {[Co IBI - Fe BI]t - [Co |BI-Fe |BI]or ART dt _Va N \ (45) ABRT (ai- P Pa The above equation can be simplified by defining [Co+2JB - Fe+3B]br = [Co+2IBI - Fe+31BI]t_[Co+31BI-Fe+31BI]or (46) which is the activity of the bare centers at the reference state of the surface and Va N =ART (47) T dt kadsBC BI - Fe+31B ]brY (Pai- Pa)J P (48) dt adsL 0 Ja 4

-21oxidation of CO proceeded on transition metal oxides by means of oxygen extraction reactions, the oxide surface being alternately reduced by CO and oxidized by 02. Wolkenstein [81] [82] [83] [84] treated the electronic phenomena in catalysis by quantum mechanics. He defined "weak" or "strong" chemisorption by the formation of covalent bond or ionic bond between the adsorbed species and the conduction electrons or electron holes of the semiconductor catalyst. Dowden [14] [15] approached the catalytic activity from the 3d-electrons of the metal ions of the transition metal oxide catalysts. The boundary-layer theory, developed by Hauffe [29] and Weisz [77] [78], emphasized the electron transfer at the interface and the electron density at the boundary layer. When the interaction between the adsorbate molecules and a solid surface involves transfer of electron, it varies the boundary layer depth and the electrical potential on the surface. The change in the surface electrical potential caused corresponding changes in the catalytic activity of the absorbate molecules. Boudart [8] outlined a qualitative picture involving changes of Fermi level of the surface, which he considered as a quasiisolated entity, with chemisorbed species equivalent to added impurities and applying analogies with behavior of bulk Fermi level in semiconductors. Wagner [74], followed by Schwab [60] [61] and Parravano [44], studied the catalytic activity by doping the semiconductor catalyst with impurity. The use of isotopes as tracers has been demonstrated as a powerful tool to study reaction kinetics. The most commonly used isotopes are 180, 13 and 14C The exchange reaction between CO and CO2 using 13C was studied by Hayakawa [30]. The CO - CO2 exchange reaction was investigated

-22by Garner [22], Winter [79] and Hauffe [18] on zinc oxides, cuprous oxides and nickel oxide. Recently, Wagner [75] [76] postulated a mechanism for the CO2 - CO exchange reaction on Wustite. Grabke [26] [27] followed Wagner's treatment to study the CO2 - CO exchange reaction on different oxides. C. Chemisorption and Desorption of Oxygen Excellent reviews of chemisorption have been provided by Low [43], Trapnell [32], Wolkenstein [82] [83], Winter [80], Parravano and Boudart [50], Hauffe [28] and Morrison [45]. The quantitative treatment of chemisorption can be divided into the following approaches. The approach of chemisorption from the kinetic theory of gases [32] emphasizes the sticking probability of a collision between the gas molecule and the unoccupied site. The lack of direct application of this approach is due to the difficulty of expressing the sticking probability analytically and to the neglect of reactions at the gas-solid interface and in the solid phase. The absolute rate theory was developed by Glasstone, Laidler and Eyring [24] [37]. The theory is based on the assumption that the gas molecule, moving from the gas phase to the adsorbed phase, passes over a potential energy barrier. An activated complex is formed when the molecule is at the top of the barrier. The activated complex is in statistical equilibrium with the molecules in the gas phase and with the vacant surface sites. If the activated complex is immobile, the rate of adsorption is u =Cg Cj 5 ) e - Ea/RT (56) Il F g. fs

-23where Cg is the number of gas molecules per cm3, Cs is the number of bare sites per cm2, Fg is the partition function of the gas per cm3, f [ is the partition function of the activated complex, fs is the partition function of the sites, h is the Planck's constant, k is the Baltzmann's constant, and u is the rate of adsorption. The rate of desorption is given by v = Ca ( e-Ea/RT (57) h ha where Ca is the surface concentration of the adsorbed molecule, fa is the partition function of the adsorbed molecule, and v is the rate of desorption. The adsorption of oxygen on cuprous oxides was studied by Stone [34] [56]. The experimental results were interpretated by the absolute rate theory. For the initial stage of adsorption, the rate follows the simple theoretical equation for dissociative adsorption. The activation energy of adsorption is constant at 7 Kcal/mole. But by and large the absolute rate theory does not usually hold for chemisorption. For a wide variety of chemisorption systems, the rate of adsorption obeys the Elovich equation [17] which maintains that the rate of adsorption decreases exponentially with increase in the amount absorbed on the solid surface. The Elovich equation can be derived for a uniform or a non-uniform surface on the basis of a variation of activation energy with the amount of absorbate on the surface [11]. Taylar and the Thon [67] have shown that, for a large number of systems, plotting the volume adsorbed against time in a semilog paper gives a straight line. The systems included the adsorption of H2 on Cr203 gel [12], on 2MnO' Cr203 [70],on

-24ZnO - MoO3 [71] and others. The Elovich equation has found wide application in chemisorption kinetics, the following being just a few studies concerning chemisorption of oxygen: 02 on CoO [65], 02 on V205 [16], 02 on Si [41], 02 on Ge [42], 02 on CoO Cr203 [67] and 02 on NiO2 [18]. The recent approach to the mechanism of chemisorption on semiconductors emphasizes the electronic defect of the semiconductor. The electronic defect is strongly effected by chemical stoichiometry. Defects, which act as electron donors or as electron acceptors, are generated by the metal - excess or the oxygen - excess in these oxides. Chemisorption was treated quantitatively by the boundary layer theory [1] [29] [77] [78]. It is assumed that the electron transfer is taking place across the interface during chemisorption until the potential energy of the electrons is the same in the semiconductor and on the other side of the interface Wolkenstein [83] [84] has suggested that weak chemisorption occurs on the normal lattice ions and does not involve defects. He regards as strong chemisorptions those which involve interactions between absorbates and defects, and which may involve electron transfer to the absorbate. His idea of weak chemisorption, which does not involve transfer to electrons falls outside the boundary layer theory receives further from the work of Dowden, Mackenzie and Trapnell [15], who found no correlation between the conductivity of an oxide and its activity in H2/D2 exchange. Desorption may take place from the occupied sites, provided the absorbed particle possesses the necessary activation energy. Thus the rate of desorption is a function of surface coverage and activation energy. Langmuir [39] has found that the rate of thorium evaporation from tungsten increases exponentially with increase in adsorbed amount. The desorption of nitrogen from iron has been investigated by Scholten [59]

-25and co-workers, who found a linear dependence of activation energy on absorbed amount. There are several theoretical treatments on the rate of desorption. A simple rate equation, the Polanyi - Wigner [22] equation, was obtained by assuming that any particle possessing the requisite activation energy desorbs within the period of one vibration perpendicular to the surface. Langmuir [40] has derived the lifetime of an adsorbed particle on the surface using an empirical vapor pressure equation. The rate of desorption is inversely proportional to the lifetime and is proportional to the number of absorbed particles per unit area. Lennard-Jones and Devonshire [23] have calculated the lifetime of an absorbed particle by quantum mechanics using the probability of transfer of a single quantum of energy from the solid to the adsorbed particle. It may be valid only for physically adsorbed particles however. Desorption rate according to absolute rate theory [24] [37] is proportional to the frequency of vibration of the activated complexes perpendicular to the surface. Both theoretical and experimental investigations of desorption kinetics are much less numerous than those of adsorption kinetics.

IV. EXPERIMENTAL APPARATUS AND PROCEDURE A. Preparation of Cobalt Ferrite The unsupported cobalt ferrite samples were prepared by Pietrzak and Gates [52] at the University of Michigan. Reagent Grade cobalt carbonate and iron oxide were weighed and mixed. The mixture was ball milled in acetone for twenty-four hours in a stainless-steel ball mill. The slurry was then dried in a large beaker. The dried cake was crushed into powder, loaded into a crucible and fired in air at 1950~F in a furnace for twelve and one half hours. The resulting cobalt ferrite was in the form of a black powder. Five samples with different Fe/Co ratios were prepared. The ratio of iron and cobalt in the cobalt ferrite samples was determined by a Norelco X-ray Fluorescent Spectrometer. The composition shown from the spectrometer was slightly different from the composition calculated from the original weighed amounts of reactants. This difference can be explained by the loss of cobalt in the sample during the firing process. The specific surface area of the samples was determined by the B.E.T. method. The amount of nitrogen adsorbed on the sample at liquid nitrogen temperature was measured as a function of pressure. The detailed apparatus and calculations for this determination are in a Technical Bulletin [19] of the Mellon Institute of Industrial Research. Two samples, with x = 1.903 and x = 2.099, were measured. The measured specific surface area of the two is 2.422 and 2.545 square meters per gram respectively. The specific surface area of samples was taken as the average value 2.48 square meters per gram. -26

-27B. The C02 - CO Exchange Reaction A diagram of the apparatus for the exchange reaction of C02 and CO is shown in Figure 1, in which all the major components are identified. A Geiger Counter (Model No. FD-1 Gas Flow Counter by Tracerlab Inc.) was used for the measurement of soft beta radiation from 14C in the gas mixture. A SC-90 Utility Scaler with a SC-42A Dual Timer also by Tracerlab was employed to amplify and register the radiation (Figure 12). The carbon dioxide and carbon monoxide in the cylinders (by Matheson Company Inc.) were introduced into the storage bulbs after removing the moisture by passing the gases through a drying column packed with Drierite (W. A. Hammond Drierite Company). The gas samples were analyzed on the mass spectrometer (Type 21-013B, Modified to Type 21-103C specifications by Consolidated Engineering Corporation). The oxygen content was beyond the limits of detection of the mass spectrometer. The radioactive 14C02 with 1.0 me radiation was supplied in a sealed glass tube by New England Nuclear Corporation. The sealed glass tube was placed next to an iron rod inside a 10 mmOD glass tubing closed at one end. The other end of the tubing was connected to the apparatus for the exchange reaction. After evacuating the system to 1 x 10-5 mm of Hg, a magnet was used outside the tubing to move the iron rod which broke the tip of the sealed glass tube which re14 14 leased the C02 gas into a Toepler pump. The C02 was mixed with C02 - CO gas mixture in the Toepler pump and stored. A weighed amount of 1.0000 gram of cobalt ferrite catalyst was loaded into a ceramic boat and placed in the center of the reactor. The catalyst was outgassed at 400~C for twenty-four hours under a pressure of 1 x 10-5 mm Hg. The temperature of reactor was controlled by a Model JP

Hg Diffusion Pump stora e bulbs rt~ flr f —--- To Vacuum -- mp reactor Geiger Counter o "^. 0"^ —- sample reservoir ma o-m er n4360 (J McLeod Gauge manometer,; McLeod A 1''| C02 trap MGauge o JJ McLeod Gauge C __j C; - -to Vacuum Pump Figure 1. Apparatus For The Exchange Reactor of C02 and CO

-29Temperature Controller by West Instrument Corporation with a Iron-Constantan thermocouple. After outgassing the temperature was lowered to the temperature of the next experiment. A gas mixture of CO2 and CO without the radioactive CO2 was introduced to the reactor to treat the catalyst surface. The time of the pretreatment was twenty-four hours. The surface pretreatment was repeated to make sure the gas mixture was in equilibrium with the solid surface. Then the reactor was evacuated at 1 x 105 mm Hg for three minutes, and the experiment started by introducing the gas mixture having the same CO2/CO ratio as the pretreatment but with the tracer14CO2 in it. A gas sample was withdrawn to the Sampling Toepler Pump for analysis after a certain time. The time interval depended on the rate of the exchange reaction. The sample gas was pumped to a reservoir with a mica window below the Geiger Counter to count the total C02 and CO in the gas. The sample reservoir (see Figure 2) with a glass flange on the top end was made of a piece of 30 mm OD glass tubing about 1.5in. long rounded at the bottom end. The top end was sealed by a mica window with density of 6.0 mg/cm2. The sealand was an expoxy cement supplied by Sears, Roebuck and Co. The sample gas was pumped back and forth three times through the CO2 trap at liquid nitrogen temperature to condense 14 14 the CO2 and C02 from the gas mixture. Then the CO was counted in 14 14 the reservoir. The total count of C02 and 14CO was corrected from the background count. The 14CO count was corrected from both the back14 ground count and the residue CO2 count which were obtained before the experiment. The sample gas was put back to the reactor after the analysis.

cable Geiger Counter / - / Shielding flow meter Scaler timer T sample sampleQ gas reservoir gas cylinder o Figure 2. Geiger Counter and Accesoaries.

-3114 The initial amount of radioactive C02 was the same for each run. The total pressure of the reactor which was constant in each run, varied from 3 to 7 cm of Hg between runs depending on the ratio of C02 and CO. The rate constant was found experimentally independent upon the total pressure in this range. The temperature ranged from 250~ to 390~C. The volume of the reactor was 507 cc. Approximately 55 cc of the gas sample was taken from the reactor for analysis. C, Adsorption - Desorption of Oxygen A sketch of the apparatus for the adsorption and desorption study is presented in Figure 3. The rate of adsorption was obtained from the pressure drop of the oxygen reservoir versus time. The rate of desorption was obtained from the pressure increase of the oxygen collecting reservoir versus time. The pressure was measured by two calibrated thermocouple gauges attached to the reservoirs and was checked occasionally with Mcleod Gauge. The thermocouple gauges (Televac, Model II), supplied by The Fredericks Company, can measure the range of pressure between 0.001 to 0.500 mm of Hg. The calibration of the sample container, adsorption oxygen reservoir and the desorption oxygen collecting reservoir was carried out at room temperature via the expansion method assuming ideal gas behavior. The attached gas burette and the mercury manometer were used for the calibration. The volume of the dead-space of the sample container, of the adsorption reservoir plus the manifold and of the desorption reservoir was 158.6 cc, 5302 cc and 5205 cc respectively, A 30 mm diameter Vycor tubing closed at the bottom was used as the sample container, A weighed sample of 50.000 grams of cobalt ferrite was placed in the sample container which was then connected via a gradient

storage bulbs Hg diffusion _ —Dorto pump - > > - manifoldrsvo manometer thermocot ple pp gauge cold trap ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~~~~~~~~~~~~~~~~r F e 5. A u sample r D container Adsorption J: n flcLeod oxygen 7auge ~ reservoir burette to atmosphere Desorption oxygen - to vacuum pump I reservoir to vacuum pum~p cold trap Figure 3. Apparatus for the Adsorption and Desorption of Oxygen

-33seal to the 15 mm Pyrex tubing of the system. After elimination of all leaks, the system could consistently be evacuated to a dynamic vacuum of 1 x 10-5 mm of.Hg by the mercury diffusion pump and the mechanical pump connected in series. The system was frequently tested under static vacuum for leaks and the leak rate was less than 1 micron per day. The arbitrary reference state of the catalyst surface was chosen by evacuating the catalyst for twenty-four hours under 1 x 10O'5mm Hg pressure at 400~C. The temperature of the sample container was controlled by a Model JP Temperature Controller (West Instrument Corporation) with a Iron-Constantan thermocouple attached to the outer wall of the container. Oxygen for the adsorption studies was generated by the thermal decomposition of potassium permanganate. The gas was first passed through a trap packed with glass beads and cooled by dry ice and isopropyl alcohol before storage. Oxygen was introduced to the adsorption reservior and the sample manifold after the system was evacuated to 1 x 10-5 mm Hg and the samples was in its reference state. Adsorption began when the 15 mm stop-cock connecting the sample container and the sample manifold was opened. The normal duration of an adsorption run was two days. After the completion of an adsorption run, the stopcock to the sample container was closed, and the sample manifold and adsorption and desorption reservoirs were evacuated. Desorption was always followed at the temperature at which oxygen was adsorbed and hence the sample temperature was not changed. At zero-time, the stopcock to the sample container would be opened, and the subsequent pressure increase in the desorption reservoir with time followed. The amount of oxygen retained in the sample container at the end of adsorption would be subtracted from the oxygen desorbed. The sample manifold was maintained at pressure 1 x 10-5 mm Hg. The desorbed oxygen was pumped to the desorption reservior by the

-34mercury diffusion pump. The amount of oxygen desorbed was smaller than the amount adsorbed at the same temperature (see Table 4). After the completion of a desorption run, the temperature of the sample container would be raised to 4000C or even 450~C for outgassing till the amount of oxygen adsorbed was outgassed as determined by a material balance. The surface of the sample would return to its reference state and the next adsorption run might be followed. The error of the Geiger Counter was within 1 per cent for 1,000 countsfmin. The errors for both temperature controllers for the reactors were within + 3~C at 350~C. The ionization gauge had an error less than 1 micron at 100 microns. Thus, the maximum percentage error of rate constants is + 9 per cent.

V. EXPERIMENTAL RESULTS A. C02 - CO Exchange Reaction The CO2 - CO exchange reaction was catalyzed by four cobalt ferrite samples, Co3_xFex 04, having compositions x = 1.903, 1.954, 2.006 and 2.009. Table 1 summarizes results of experiments performed at different CO2 - CO ratios and different temperatures on each sample. All the results were obtained on 1.000 gm of cobalt ferrite catalyst with surface area 2.48 square meter which had been outgassed for one day at 400~C and at a pressure of 1 x 10 5 mm Hg and pretreated twice with the C02 - CO gas mixture before each run. The percentage of 14C02 in C02 was 0.00266%. Two or three runs were performed under the same experimental condition to insure the reproducibility of results 14 within 1.5% of CO formation. Experimental data are tabulated in Appendix I. The results of 14 the percentage formation of CO versus time for Co0901 Fe2 099 04 at 3500C with different C02/CO ratios are plotted in Figure 4 as an example plot of raw data. The rate constant k(a')was obtained from the rate of formation of 14C0 by applying Equation (21), V dPt4 = k(ao)[4 - (1 + ao)P14] (21) Two methods were employed to calculate the value of k(ao), the initial rate method and the digital computer simulating method. The detailed calculation is in Appendix II. The values of k(ao) are presented in Table 1. -35

50-. 284 40 k / ^C U V^- / o = 0.464 40 - 40 2 3 4 5 6 7 t(hrs.) Figure 4. Formulation of 14Co Versus Time on CO 901Fe2 0g-OL With A = 2.48 x 104 cm2 10 and T 350 I E 3 4 5 6 7 t(hrs.) Figure 4. Formulation of 14C0 Versus Time on CO0.901Fe2.09904 With A = 2.48 x 104 cm2 and T = 350~C.

-37TABLE 1 SUMMARY OF EXPERIMENTAL RESULTS OF THE EXCHANGE REACTION OF CO2 AND CO ON Co3_xFex04 Te. k(ao) x 107 x Pco2/Pco (c) [ mole Ratio (hr.atm.cm2 0.464 350~C 0.064 1.903 0.464 380~C 0.735 0o464 410~C 3.030 0.464 3100C 0.094 1.954 0.464 3500C 0.760 0.464 3900~ 2.670 0.464 250~C 0.0524 2.006 0.464 300~C 0. 500 0.464 350~C 3.920 0.464 3100C 0.299 2.099 0.464 3500C 2.55 0.464 390~C 6.37 0.218 350~C 0.122 03 0. 464 350~C 0.064 1903 1.442 3500~ 0.0184 2.190 350~C 0.0132 0.218 250~C 0.100 2.006 0.464 2500~ 0.0524 1.442 2500C 0.0145 0.284 350~C 4.560 2.099 o.464 3500C 2.550 1.122 350~C 1.140 2.565 350~C 0.464 TABLE 2 ACTIVATION ENERGY OF THE EXCHANGE REACTION OF C02 AND CO ON Co3-xFexO4 x 1.903 1.954 2.006 2.009 kcal 30.4 E [kiial]; 54.8 32 4 27.7 30 4

-38Knowing the values of k(ao), the constant m which characterizes the catalytic intermediate on the catalyst surface was obtained from Equation (27), ain[k(ao)]] 1 (I n[k(ao)T) = -m (27) \ bn ao Figure 5 is the plot of k(ao) versus ao for x = 1.903, 2.006 and 2,099. The results show that m = 1 for all three samples. The activation energy Ea of four samples with ao = 0.464 was calculated from Equation (29), (/)n[k(ao),o Ea 1 Figure 6 presents the plot of k(ao) versus - for four samples. The values of Ea are tabulated in Table 2. B. Adsorption and Desorption of Oxygen The generalized rate expression of oxygen adsorption can be written as d na - A dnt Kads Pa = Z s [active center i]b Pa (39) A d t a s a i kt ads i where k a = the rate constant for i type active center ads

-39x = 2.099, T = 350C o-7 x = 1.903, T =350~C x = 2.006, T = 250~C 0.1 1.0 10 Oo Figure 5. The Value m From k(ao) Versus a.

-40[active center i]b= the activity of the bare i type active center na = moles of 02 in the gas phase during the adsorption pa = pressure of O2 in the reactor during the adsorption and Kad is the overall rate constant which is a function of catalyst composition, temperature and surface coverage Based on the model described in the previous section, the rate of adsorption of oxygen on cobalt ferrite, provided the desorption rate is small, can be written as d +2 3 - At = kads [Co IBI -Fe BI]b P (40) where [Co +BI - Fe+31B|]b activity of the bare active center Co2IBI. Fe lIBI Since the total number of bare and occupied active centers is constant for a catalyst, we have [Co+21BI - Fe+31BI]b + [Co+31BI - Fe+31Bl]0 = [Co+2j1B - Fe+ 3BI]t (41) where the subscripts o and t stand for occupied and total respectively. From the material balance of oxygen, we have A P +51+3 Va d[Co+3BI - Fe+3BI] = RT d p = dn (42) N 0 d a - dna where A = surface area of cobalt ferrite = number of active centers per unit area N = Avogadro number V = volume of the adsorption oxygen reservoir a

-41Let the activity of the occupied centers at the reference state be [Co+31Bj - Fe+31BI] or and integrate the above equation to obtain [Co+31BI - Fe+31BI] - [Co+31BI - Fe+31BI] or Va N ApRT ( Pai -Pa ) where Pai is the initial pressure of oxygen-before the adsorption Rearrange the above equation and combine with equation (41): [Co+21BI - Fe+3BI ]b = [Co+2BI - Fe +3lB]t - [Co+3JBI - Fe+3BIor Va N (ai - Pa)(4) Put equations (42) and (44) into the rate equation (40): Va d pa Co+2 +3 +3 +3 -- PA = kads. [C IBI - Fe IBl]t - [Co J |BI-Fe IBI]or ART dt Va N a (Pai- Pa)Pa (45) ApRT The above equation can be simplified by defining [Co+2 BI - Fe+31BI]br = [Co+2 BI - Fe+3lBI]t-[Co+31 B-Fe+31BI]or (46) which is the activity of the bare centers at the reference state of the surface and Va N = a (47) A-RT Va dpat = kad[Co BI - Fe+31BI]br7- (Pai- Pa)J Pa (48) i dt a~l

-42By knowing the rate of pressure drop in the system, the rate constant can be calculated from equation (48). The rate of desorption can be expressed by dnd = k des. [Co IBI - Fe+3BlI] (49) Adt where nd= moles of 02 desorbed Since the desorption experiment starts right after an adsorption experiment, the activity of the occupied center at the beginning of the desorption is [CO~+IBI Fe+ BI ]or + 7 (Pai - Paf) where Pf is the final pressure or equilibrium pressure of adsorption. The material balance of oxygen during the desorption be comes # d[Co 3BI - Fe+lBI] = - d d pd = dnd (50) where Pd = the pressure of the bulb collecting the desorbed oxygen nd = moles of 02 desorbed V = the volume of the oxygen collecting reservoir d The integrated form of the above equation is [CO+31BI - Fe+31BI]or + 7 (Pai - Paf) - [CO 3IB - Fe +3BI] Vd N \ =Vd (Pd - Pdi) (51) APRT where di= the initial pressure of the bulb collecting the desorbed oxygen

-43Put equations (50) and (52) into equation (49), V dPd I- = kd es. {[C3B - FeIB +]or + (Pai - Paf) - Vd (Pd Pdi)} (52) ApRT Define [Co+3BI -Fe+IBI]oid = [Co+31BI - Fe]31BI]or + Y(Pai - Paf) (53) = the initial activity of occupied center at the beginning of the desorption and Vd N -= ADRT (54) The desorption rate equation (52) becomes V d dpd B ART = kdes. {[Co+ IBI - Fe+31B]id - (Pd - Pi)} (55) By measuring the rate of pressure increase in the collecting bulb, the desorption rate constant is obtained from equation (55). The adsorption and desorption of oxygen were performed on five cobalt ferrite samples with x = 1.903, 1.954, 2o006, 2.058 and 2.099. All the experiments were obtained on 50.000 gram of samples, corresponding to 1.24 x 10 cm surface area, placed in the sample container. The reproducibility of results of adsorption runs was within 2 microns for Pa after the surface was brought back to its reference state by checking the material balance of oxygen. Without checking the material balance of oxygen, the reproducibility of Pa was as poor as 20 to 30 microns. Table 3 summarizes the results of the adsorption runs. Experimental data of adsorption are tabulated in Appendix I, Figure 7 is a sample plot of the pressure of oxygen Pa versus time t at 300~C, 200~C and 100~C on cobalt ferrite with x=2.006.

TABLE 3 SUMMARY OF EXPERIMENTAL RESULTS OF ADSORPTION OF OXYGEN ON Co3xFeO4 Ma x 1011 (dna) 1010 Initial rate Temp. Pai Paf A Adt t=o const. Kads. x 104 mole mole mole Ekcal x (C) (microns) (microns) (cm -m) ( Ea(.... cm ) hr.cm(.a t 300 101.7 54.0 1.09 1.01 0.755 1.903 200 104.1 62.5 0.952 0.64 0.468 2.81 100 100.0 70.1 0.684 0.273 0.204 500 103.3 39.0 1.47 0.95 0.70, 1.954 400 105.7 55.0 1.16 0.54 0.394 7.32 300 102.0 85.6 0.376 0.174 0.131 300 102.3 2.15 2.29 4.66 3.6 2.006 200 101.2 6.2 2.18 2.96 2.22 2.18 100 100.0 19.1 1.85 1.581.20 400 100.9 28.0 1.67 0.66 0.514 2.058 350 99.4 44.5 1.23 0.342 0.262 9.83 300 103.2 69.0 0.794 0.163 0.120 400 103.3 21.0 1.88 0.507 0.373 2.099 300 102.5 52.0 1.06 0.203 0.151 7.59 200 105.2 99.2 0.137 0.0538 0.039

-45410~C 380 C 350~C 300 ~C 250'C x = 1.903 x = 2.006 \ x = 2.099 I0 if 1 tN' t \ \ \ 10-' I I I I I I 1.4 1.5 1.6 1.7 1.8 1.9 T..LOs (OK-') Figure 6. Activation Energy of the Exchange Reaction From k(ao) Versus 1- = 0.464. T

t2 clP so~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ to 40 60 Fu7Aotn0~ 2.0~~~~2 ~~O tc~~on f xyg o n C0.94 F2. oo6 04 Fig ~Adsorptio fOYe,iure 7

-47From Figure 7, it shows that Pa changes very little after the first hour of adsorption. The final pressure of adsorption paf is then arbitrarily chosen as the pressure at twenty-four hours of adsorption. The amount of oxygen adsorbed per square centimeter surface was calculated from the integrated form of Equation (42), Lnua V a A A-J (Pai - Paf) (58) The experimental data follow very well to the theoretical rate expression of Equation (48), VRT dta = kads {[C2 B- - Fe+3B ]br - (Pai - Pa)} Pa (48) The constants kads and [Co+21BI - Fe+31B]br were calculated by simulating method with digital computer. The detailed procedure is in Appendix IIo The values of initial rate of adsorption and initial rate constant K-ads were obtained from Equation (40) at t = 0, (A t = o KadsPai = kads[Co 21B - Fe+3BI]br Pai (59) The activation energy of adsorption of initial rates was calculated from 6,raads] Ea )] = -f (60) T(T) ref.state R by knowing the K-ads at reference state for three different temperatures. Figure 8 is the plot of log Kads versus - for five samples. T The experimental data on desorption were poor. The derivation of Pd for runs under the same condition was as high as 50% because of the following reasons: (i) The rate of desorption was so slow that the pressure change with respect to time could not be read accurately from

-48500 400 350 300 200 I0 I |^^o^.^^ /-x = 2.006 x =1.954 A 1.903 x 2.o58 i0 15 IcT" iCy- \X\ /~ x ~ 2 099,.I I I I 1.5 1.7 1.9 2.1 2.3 2.5 2.7 T x 10' (~K-') Figure 8. Activation Energy of Oxygen Adsorption From Kads. Versus ~.

-49the thermocouple gauge. (ii) The desorbed oxygen was collected by the mercury diffusion pump into the desorption reservoir. The efficiency, the percentage of the desorbed oxygen collected by the pump at the same short period of time, could not be very high because of the extremely low pressure of the sample container. (iii) There was a certain amount of oxygen trapped in the dead-space of the sample container. The trapped oxygen may also contribute to the error of the desorption data. The results presented for the discussion were the amount of oxygen desorbed per unit surface area in twenty-four hours and the initial rate of desorption. The former was obtain by integrating Equation (50), which gave Vd (Pdf Pdi) (61) A ART The latter was obtained by taking the average rate of desorption in the first five minutes. Table 4 summarizes the results of desorption. Co Summary of Results in Terms of Catalyst Composition The purpose of this research was to investigate the catalytic activity of cobalt ferrite in terms of its composition. The following figures summarize the results plotted against the composition of the catalyst so that it is convenient for the discussion. Figure 9 is the plot of the rate constant of the exchange reaction versus the composition x at 350~C and with C02/C0 ratio of 0.464. Figure 10 was plotted with the activation energy Ea of the exchange reaction versus the composition x, C02/C0 ratio = 0.464 and the temperature ranging from 250 to 410~C. Figure 11 shows the plot of the amount of oxygen adsorbed per unit surface area versus the composition x at 300~C.

TABLE 4 SUMMARY OF EXPERIMENTAL RESULTS OF THE DESORPTION OF OXYGEN ON CO3_xFexO4 TemperaTempera- df -Pd' Ara Imole Ard 12 dnd 12 mole x ture(~C) (microns) A 10 mo x 1012 x 1012 m X cm i A cm2 Adt hr. cmo 1.093 300 6.80 1.09 1.59 1.25 200 2.1 0.952 0.49 1.05 100 0.31 0.684 0.072 0.122 1.954 500 4.72 1.47 11.3 9.8 400 14.5 1.16 3.38 3.5 300 7.3 0.376 1.70 1.4 2.006 300 6.60 2.29 1.55 0.725 200 1.20 2.18 0.28 0.312 100 2.05 1.85 0.47 0.221 2.058 400 81.5 1.67 19.0 11.6 350 38.3 1.23 8.94 2.80 300 16.0 0.794 3.74 1.68 2.099 400 31.1 1.88 7.3 2.80 300 13.5 1.06 3.16 1.40 200 0.2 0.137 0.047 0.056

4.0 I E I I 3.0 I 0 O~~~~~~~~~~~~~~~~~ N / 0' S / 2.0 / 0 1.88 1.72 1.96 2.00 2.04 2.08 2.12 Figure 9. The Rate Constant k(a0) of the Exchange Reaction Versus Composition x, T = 3500C, ao = 0.464.

-5260 50 40 0M 30 z W 20 z 0 0I I I I 1.88 1-92 1.96 2.00 2.04 2.08 2.12 X Figure 10. Activation Energy of the Exchange Reaction Versus Composition x, a = 0.464.

-532.5 I 2.0 / \ 1.5 I =o1^~ I I. I IC 0.5- / I 0 I I X 0)O I 0 0 0.I I I I I I 1.88 1.92 1.96 2.00 2.04 2.08 2.12 Figure 11. Amount of Oxygen Adsorbed in 24 Hours Versus Composition x, T = 300~C.

-54Figure 12 shows the initial rate of adsorption versus the composition x at 300~C. Figure 14 represents the initial adsorption rate constant kads versus the composition x at 300~C. Figure 14 indicates the initial rate of desorption versus the composition x at 300~C. Figure 15 presents the amount of oxygen desorbed per unit area versus the composition x at 3000C.

-555 I I 4 - I I I I 2 3 I 0 2 o I I / s18 1.92 1.96 2.00 2.04 2.08 2.12 Figure 12. Initial Rate of Oxygen Adsorption on C03_xFex04 Versus Composition x, T = 300~C Pai l100 Micron of Hg.

-56-.4 I 0 j3I V I.!0 - 2 I 0 / O I I 1.88 1.92 1.96 2.00 2.04 2.08 2.12 x Figure 13. Initial Rate Constant of Adsorption Kads Versus Composition x on Co3 Fe 04 T = 300 o~-x T = 300~C

-571.5 \ ~u \ / iEn 1.0 _ 0.5 1.88 1.92 1.96 2.00 2.04 2.08 2.12 x Figure 14. Initial Rate of Desorption from C03x Fe. 04 Versus Composition x, T = 300~C.

-583.6 3.2 2.8 Q / I..' 0 Versus 2.Composition x 0 1.6 1.2 1.88 1.92 1.96 2.00 2.04 2.08 2.12 x Figure 15 Amount of Oxygen Desorbed. From CO _x Fex 04 in 24 Hours Versus Composition x T = 300 C.

VI. DISCUSSION OF RESULTS The results shown in Figures 9, 12 and 13 give a uniform pattern of curves which has a maximum when x is in the vicinity of two. These extremes indicate that the rate constant of the CO2 - CO exchange reaction, the rate constant of adsorption of oxygen, and the initial rate of adsorption for the intrinsic sample, which is an insulator at x = 2, are higher than the extrinsic samples with x A 2, The same conclusion was obtained in some other works [51] [33]. The conclusion is, however, contrary to other statements in the literature [13] [66]. These claim that the catalytic activity of oxides is according to the order p-type oxides > insulator > n-types. The reason for the claim is from a specific experimental result. That is, the p-type oxides, the oxides of copper, nickel, cobalt and iron have higher catalytic activity in the decomposition of N20 than the n-type oxides, the oxide of zinc, chromium and gallium. These two contradictory orders of activity are discussed in the following paragraphs. The catalytic activity, as well as other properties, of a metal oxide is determined by the metallic element of the oxide, the crystal structure and the defects. The metallic element and its electronic structure determine the valence of the cation. The crystal structure and the metallic element determine the type of chemical bonds between the metallic element and the oxygen of the oxide. The metallic element of the oxide is undoubtedly the predominant factor in effecting the chemical properties of the oxide. Defects of the metallic oxide may act as electron donors or electron acceptors which effect predominately the electronic properties of the oxide. -59

-60The results in this research are obtained in which the metallic cations and the crystal structure of the oxide catalyst are invariant. The only parameter is the defect of the oxide with which the oxide is made into n-type, intrinsic or p-type by varying the composition of the cations. The other statement is based on the comparison of oxides with different metallic cations and with different crystal structures. It may well be that copper oxide has higher catalytic activity than zinc oxide, not because they are p- or n-type but because they are different materials. In order to understand the mechanism of a catalytic reaction, it is necessary to know the active center and the catalytic intermediate of a reacting system. In the exchange reaction of CO2 and CO, the result in Figure 5 gives the value of m = 1 for three samples with x = 1.903, 2,006 and 2.099. This result indicates that the catalytic intermediate on all three types -- n-type, p-type and intrinsic -- of cobalt ferrite is Cads (see Chapter 2, Section C). The extra electron of the catalytic intermediate Oads is obtained from the active center on the catalyst surface. The proposed active center is the cation pairs Co+2 BI - Fe+31BI -- the normal lattice cation pairs. Electronic defects have been considered as the active centers for catalytic reactions in the literature [10][62]. If the active center for the exchange reaction of C02 and CO are defects, the catalytic activity, which may be represented by the rate constant, for n-type or p-type cobalt ferrite would be higher than the intrinsic cobalt ferrite. But the results in Figures 9 and 13 show that it is not so. This indicates that the active center for the exchange reaction is not the characteristic defect. Experimental results have verified that Co+2J1B - Fe+31BI is the active center. For the intrinsic cobalt ferrite

-61catalyst, the concentration of the active center Co+2 JBJ - Fe+3iBI is higher than in either the n-type or the p-type cobalt ferrite, as is the catalytic activity. The next item to be discussed is the dependence of the magnitude of the rate constants k(ao) and Kads on catalyst compositions. The theory which is employed to explain the catalytic activity difference among catalysts with different composition is based on the activity of the active center Co+2| BI - Fe+31Bi. For n-type cobalt ferrite, which has a certain amount Co +2BI replaced by Fe+21BI, and also for the p-type which has a certain amount of Fe+31Bt replaced by Co +2JB; so that the surface concentration of the active center Co+21B1 - Fe+31B1 for the n-type or the p-type is lower than the activity of the intrinsic. The catalytic activity difference is in the order of magnitude of eighty, but the concentration difference among samples is within 10% in the value of x, This can be explained by the distribution of defects in a single crystal and by the activation energy. The sample was prepared by firing cobalt carbonate and iron oxide at 1950~C. At this high temperature, the defects are ionized to form free electrons or electron holes depending on whether the sample is iron-excess or cobalt-excess in its composition. We may expect that the defect concentration on the surface layer is higher than the defect concentration in the bulk. The activation energy difference (cfo Tables 2 and 3) also contribute to the difference of the catalytic activity. It is worth noting that the plot of k(ao) or K-ads versus x in Figure 9 or 13 would pass through a maxi um at x = 2 if the curve is continuous. Unfortunately there is no way of knowing experimentally what the exact maximum value of k(ao) or KLads. is, since the stoichiometric cobalt ferrite, COFe204 is seldom obtained,

-62With the catalytic intermediate and the active center defined, it is possible to postulate the reaction mechanism of the exchange reaction as following: 0 — (i) C02(g) + Co+2BI - Fe +3 B - Co3JB - Fe3JBf + co(g) (62) 0 — 0(ii) CO+3IBI - Fe+31Bl - CO+31B - Fe+31BI (63) 0(iii) CO+3BI - Fe+31BI + CO(g) - Co+31BI - Fe+2BI + C02(g) (64) (iv) CO+3IBI - Fe+21B - Co+2IBI - Fe+3 1B (65) Because of lack of thermodynamic data for surface reactions on cobalt ferrite, it is rather difficult to discuss each step quantitatively. But we can discuss the steps in a qualitative manner. Step (i) is the reduction reaction of C02 which has a possible positive AH and AF. So the activity of the products CO and C +3jBI - Fe+2IB1 is smaller than the activity of the reactants C02 and Co+21BI - Fe+31BI in an equilibrium system. Since aco and aco2 are fixed by the partial pressure of CO and C02O we may expect to have [Co+2IB - FeI31B] [C 31B| - Fe+31] BI. This was verified by the conductivity measurement by Ratterman [49]. Step (ii) involves a change of bonds from O — Co+31B to O- - Fe+31BI. Because of the same charge and almost the same ionic radius (r F+3 = 0.67 A, rco+3 = 0o65 A) of Co+31BI and Fe+3jBI, we may expect very little change in the vibration frequency, and likewise in the vibration partition function and in the free energy. Therefore the activities of Coi+31B - Fe+31BI and Co+3IB - Fe+3IB1 are almost the same. Step (iii) is a possible exothermic reaction of the oxidation of CO,.so the AH and AF are negative. We may expect [Co+3IBI - Fe+2IBI] >> C+31BI - Fe+3lBI. Step (v) involves

-63an electron transfer between the ion pair. In cobalt ferrite, the Co+21BI - Fe+31BI with lower potential energy is more stable than Co+31BI - Fe+21BI. Jonker [35] has shown that the potential energy of Fe+31BI is 0.025 eV lower than Fe+2J1B and the potential energy of Co+2IBI is 0.15 eV lower than Co+3IBI. Therefore [Co+2 BI-Fe+31BI] is higher than [Co+31 BI - Fe+2 J1B]. We may conclude that the activity of the bare and covered active center is in the order of [Co+21BI - Fe+31B ] > [Co+3IBI - Fe+2lBj ] [Co+31B - F+31Bj] = [C +3IBI - Fe3Bj]. The experimental result of the oxygen adsorption in Figures 11, 12 and 13 also show that the intrinsic cobalt ferrite has a higher activity than both the n-type and the p-type. With the same proposed active center as the exchange reaction it is postulated that the mechanism of oxygen adsorption and desorption follows the steps: 0 0-. (i) 02(g) + Co+2IBI - Fe+3BI -_Co+31BI - Fe+3IB (66) 0 0I I 0- 0 (ii) Co+31BI - Fe+31BI - Co+31BI - Fe+31BI (67) 0I 0 - 0- - - - - 0 (iv) C+3I Fe+31 BI Co+31B! - Fe+21BI (69) O - 0 - O (v) Co+31BI - Fe+2|B| _ Co+31BI Fe+2|B + 02(g) (70) (vi) Co+3BI - Fe+2jBj - Co+2IBI - Fe+3BI (71)

-64Step (i) is the step of oxygen adsorption in which an electron is transferred from the active center to the oxygen molecule. Step (ii) (iii) and (iv) are the electron transfer steps between two oxygen atoms and between the oxygen atom and Fe+3IB1. Step (v) is the step of oxygen desorption. Step (vi) is the internal electron transfer of the active center which returns to the more stable form. The next few paragraphs are to discuss separately the effect of the defects on the rate constant and on the activation energy of the exchange reaction, of the oxygen adsorption and of the oxygen desorption. The rate constant k(ao) of the exchange reaction decreases with an increase of the defect concentration for both n-type and p-type samples as shown in Figure 9. There are two ways that k(ao) is effected by defects. The first and the direct way which has been discussed is that defects reduce the number of active centers. The second and the indirect way is that the defect located next to the active center (see Appendix III) may alter the polarity of the active center so that the potential field, which effects the activation energy, is changed. From quantum mechanics, it has been shown that the formation of ionic bonds and nonpolar covalent bonds represents extreme cases of the chemical bond formation. The intermediate case is the formation of polar covalent bonds. For n-type samples, it is assumed that the predominate defect Fe+2IBI is covered by ~ads. (5-) (+) (see Section IIoB, (d)) with the reaction Fe+21BJ + Oads -,Fe+31Bli + Oads. The ionized defect Fe+31BI, has a positive polarity 6+ because of the donation of an electron. If the active center of the exchange reaction, Co+21BI - Fe+31BI, is next to the defect Fe+3Bli the positive polarity 5+ will contribute a certain attraction to the outer-shell 3d electron of the Co+21BI ion of the active center, to make

-65the 3d electron more difficult to ionize, Thus the activation energy of the exchange reaction should be higher for the n-type sample than for the intrinsic. For p-type, the predominate defect is Co+31B!, an electron acceptor. If the active center Co+2 BI - Fe+3IB1 is next to Co+31BI with a strong positive polarity, the active center will need an even higher energy than the n-type to become ionized. Therefore the activation energy of the exchange reaction for the p-type should be higher than both the intrinsic and the n-type. The result shown in Table 2 is so. The rate constant Kads. and activation energy Ea of oxygen adsorption as shown in Figure 13 and Table 3 are influenced by the defect in the same way as is the exchange reaction. The primary effect of the defect is the influence on the number of active centers from which the intrinsic sample has a larger Kads. than both the n-type and the p-type. The secondary effect is the alteration of the polarity of the active center from which the activation energy for the n-type or the p-type sample is higher than the intrinsic. These effects are shown in Figure 13 and Table 3. Table 3 also shows that the activation energy tends to drop for both n-type and p-type when the defect concentration goes beyond a certain valueo This can be explained by the bare "strong" chemisorption center (see Sec. IIoC) of defects Fe+2IBI and Co+21B1i participating in the oxygen adsorption. If the oxygen covered defect Fe+21BI or Co+2|BIi concentration on the surface is close enough to a certain value, there will be an appreciable amount of oxygen molecules desorbed from the "strong" chemisorption center when the sample is brought to its reference state by heating it at 400~C under vacuum. The bare "strong" chemisorption center has a much lower activation energy than the "weak'" active center Co+2j1B - Fe+31BI in the adsorption of oxygen. The experimental

-66value of the activation energy at x = 1.903 and x = 2.009 is the overall activation energy of adsorption of the two types of active centers. It is worth noting that there is no activation energy drop of the exchange reaction at high defect concentrations because the "strong" chemisorption centers are covered by Oads. which do not participate in the exchange reaction. The desorption of oxygen molecules involves the breaking of 0 - Co+31|B and 0 - Fe+21BI bonds (see page 63 Step (v)). It is not surprising that the activation energy of desorption is much higher than the activation energy of adsorption. The high activation energy of desorption results that the effect of the activation energy outweighs the effect of the surface coverage in the determination of the rate of desorption. The effect of the polarity of the defect toward the active center is the same as in the exchange reaction and in the oxygen adsorption. The positive polarity of the defects next to the active centers of the n-type and p-type samples makes it easier for the active center to take the electron back from the adsorbed oxygen molecule in the n- or p-type than in the intrinsic (see page 63 Step (iv)). Therefore the activation of desorption is lowered for both n-type and p-type samples. This results in the higher desorption rate for both n-type and p-type samples than for the intrinsic as shown in Figure 14. The desorption rate is reduced again when the defect concentration is higher than a certain value. This is due to the desorption of oxygen molecules which were adsorbed on the "strong" active centers. The explanation is the same as in the adsorption of oxygen at high defect concentrations. It is important to investigate the possible secondary reactions which may effect the rate of formation of 14CO in the exchange reaction

-67of C02 and CO. The adsorption of CO has been observed on both n-type and p-type semiconductors [32]. There is a possible exchange reaction between the 1CO in the gas phase and the COa. The activaads. tion energy of this reaction is high because COads is strongly adsorbed on oxides [23]. The second possible reaction is 14CO(g) + CO(g) - C + C02(g) which is thermodynamically possible when the p2c/p ratio is higher than 3 x 10-5 at 350~C. Brandner [10] reported the formation of carbon on gold strips used as the catalyst in the exchange reaction CO2 and CO. These two possible secondary reactions contribute to the disappearance of C14C from the gas phase when there is an appreciable amount of 14O formed in the gas phase during the later half of an experiment. But the rate constant k(ao) which is obtained from data of the very early part of the experiment, is not effected by these secondary reactions. Figure 4 shows that these secondary reactions did occur. Finally, it is shown that the rate of oxygen adsorption does not follow the absolute rate theory. If the adsorbed oxygen molecule is immobile on the surface, Equation (56) derived from the theory can be simplified to Ea/RT h4 ( u e C (72) a s 8,I(2=mkT)3/2 If the adsorbed oxygen molecule is mobile, Equation (56) becomes Ea/RT kT h (hT u e a = Cg h ) (73) 6 h (2=mkT)1/2 For the sample with x = 2.006 at T = 300~C, the experimentally obtained value of u eEa/RT is 1.053 x 1012 molecules per sec. From Equations (72) and (73), the calculated values of u eEa/RT are 3.25 x 1020 and

-682.61 x 1017 molecule per sec. respectively. Comparing these values, it is obvious that the absolute rate theory does not apply to our model of the rate of oxygen adsorption.

VII. CONCLUSIONS The following conclusions have been obtained from results of the study of the catalyzed exchange reaction of CO2 and CO, and the adsorption and desorption of oxygen on cobalt ferrite Co3_xFexO4, with x ranging from 1.903 to 2.099. (1) For the exchange reaction and the oxygen adsorption, the catalytic activity of the intrinsic cobalt ferrite is higher than the catalytic activity of the n-type or the p-type. This is contrary to what was claimed in literature. For the oxygen desorption, the rate of desorption of the n-type or the p-type is higher than the rate of desorption of the intrinsic sample. We may generalize the results by stating that: for reactions involving the electron transfer from cobalt ferrite to the catalytic intermediate, the catalytic activity follows intrinsic > n-type or p-type, for reactions involving the electron transfer from the catalytic intermediate to the solid, the catalytic activity follows n-type or p-type > intrinsic. (2) The catalytic intermediate for the exchange reaction is 0ads. The catalytic intermediate of the oxygen adsorption and desorption is in different forms of 02ads. The active center for both reactions is the ion pair, Co+21B -_ Fe+3BI. This is also contrary to some literature in which the defects are being considered as active centers for the catalytic reaction. (3) The defects Co+31BI and Fe+2IBI effect the catalytic activity of the exchange reaction in two ways. The predominant effect is that the defects reduce the number of the active center Co+21|B| - Fe+3IBI on the catalyst surface. The second effect is that the -69

-70defects alter the polarity of the active center, so that the activation energy of the exchange reaction increases with the increase of the defect concentration. (4) For the adsorption of oxygen, the defects effect the rate of adsorption in three ways. The first two are the same as in the exchange reaction, namely the effects of the number of the active center and the polarity. The third way is that the defect causes a decrease of the activation energy when the surface concentration of defects is higher than a certain value. This is due to the defect itself acting as the active center, at the high defect concentration. (5) The defects give the same three effects in the desorption of oxygen as in the oxygen adsorption. Because of the high activation energy of desorption, the polarity effect outweighs the effect of the number of active centers. This results that the rate of desorption for the n-type or the p-type is higher than the intrinsic sample. The third effect at a high defect concentration results in the increase of the activation energy. This effect causes the decrease of the desorption rate at high defect concentrations. (6) There are two possible secondary reactions occurring with the exchange reaction. They are (i) 14CO(g) + COds. > ads + CO(g) and (ii) CO(g) + CO(g) - 14C(s) + C02(g). They may cause the decrease of 4CO in the gas phase.

APPENDICES APPENDIX I EXPERIMENTAL DATA Ao Experimental Data of the Exchange Reaction of CO2 and CO TABLE 5 EXPERIMENTAL DATA ON THE FORMATION OF 4CO CATALYZED BY 1.000 GRAM OF Co 3Fe 04 Corrected Corrected Total Counts Total Counts 1 CO Counts 14CO Counts Time per Two per Two per Two per Two 14 (hours) Minutes Minutes Minutes Minutes CO% Run No. 18-2, x=1o903, ao0.464, T=350~C 1.5 21514 21466 705 553 2o57 10.0 20727 20699 3118 2966 14.32 22.67 20542 20494 5181 5030 24.6 34.67 20388 20324 5980 5828 28.7 47533 19842 19794 6331 6180 31o2 59.67 19276 19228 6061 5910 30.8 70.67 18540 18492 5451 5300 28.6 95.17 17896 17858 4346 4194 23.5 108.0 17590 17542 3947 3796 21 6 143.17 16752 16704 2923 2772 16.5 Run No. 18-3,x=1.90, ao=O.464, T=380~C 1.0 20814 20766 3222 3070 14.8 3.0 20327 20279 5412 5260 25.9 5.0 19751 19703 6115 5964 30 2 10,17 18766 18718 5442 5390 28.8 24.67 17422 17374 3578 3426 19.7 -71

-72 - TABLE 5 (CONT'D) Corrected 14 14 Corrected Total Counts Total Counts CO Counts CO Counts Time per Two per Two per Two per Two 14 (hours) Minutes Minutes Minutes Minutes C0% Run No. 18-4, x=1.903, a =o.464, T=380~C 0.5 21375 21327 2430 2278 10.67o 2.0 20849 20801 4867 4716 22.6 4.0 20195 20144 5972 5820 28.8 6.0 20086 20038 6242 6092 30.4 25.5 17842 17794 3365 3214 18.1 Run No. 18-5, x=1.903, a =0.464, T=410~C 0 1.0 23217 23169 6349 6198 25.7 2.0 22676 22628 6931 6781 29.9 3.5 21726 21678 6543 6392 29.4 5.5 20937 20889 5289 5138 24.6 11.0 16404 16365 3620 3468 21.2 Run No. 18-6, x=1.903, a =0.464, T=410~C 0.5 21899 21851 5043 4892 22.4 1.5 20786 20728 6103 5952 28.8 2.5 19718 19670 5758 5606 28. 5 Run No. 14-6, x=1.954, a =0.464, T=350~C 2.0 19568 19520 5290 5138 26.3 5.33 18302 18254 6838 6676 35.6 10.33 17581 17533 6146 5994 34.2 25.0 16187 16139 3838 3686 22.6 29.67 15582 17534 3298 3146 18.0

-73TABLE 5 (CONT'D) Corrected 1 Corrected Total Counts Total Counts CO Counts CO Counts Time per Two per Two per Two per Two 14 (hours) Minutes Minutes Minutes Minutes CO% Run No. 14-7, x=1.954, a =0.464, T=3500C 1.0 19322 18274 3183 3032 16.6 4.09 18752 18704 6462 6310 33.8 6.0 18422 18374 6673 6522 3555 12.5 17436 17388 5431 5278 30.4 24.5 16oi0 15962 3545 3392 21.3 Run No. 14-8, x=1.954, a=0.464, T=310~C 2.0 19905 19857 1888 1736 8.75 4.0 19811 19763 3214 3062 15.5 8.84 19623 19575 5341 5188 26.6 23.0 18912 18864 7212 7060 37.4 Run No. 14-9, x=1.954, a =o.464, T=310~C 14.o 18959 18911 6112 5960 31.4 19.0 18608 18560 6493 6342 34.1 27.0 18530 18482 6844 6692 36.3 36.67 17568 17520 6698 6546 37.3 49.0 17286 17238 628 6 6056 35.1 66.67 16552 16504 5279 5128 31.1 86.0 16o68 16020 4662 4510 28.1

-74TABLE 5 (CONT'D) Corrected 14 14Corrected Total Counts Total Counts CO Counts CO Counts Time per Two per Two, per Two per Two 14 (hours) Minutes Minutes Minutes Minutes CO% Run No. 14-10, x=1.954, a=0.464, T=390~C 1.0 20198 20150 6710 6658 32.6 2.0 19236 19188 5859 5708 29.8 5.0 18563 18515 5048 4896 26.4 4.0 17894 17846 4245 4094 22.9 Run No. 14-11, x=1.954, ao=0.464, T=390~C 0.67 20676 20628 6030 5878 28.5 1.50 20208 20160 6285 6134 30.4 5.50 17951 17903 3530 3218 18.0 Run No. 14-12, x=1.954, a=0.464, T=390~C 0.33 20848 20800 4130 3818 15.9 Run No. 10-8, x=2.006, ao=0.464, T=3000C 1.0 18707 18659 1802 1650 o8.85 3.5 16606 16558 3939 3788 22.8 Run No. 10-9, x=2.006, ao=0.464, T=300~C 2.0 19440 19392 3211 3058 15.8 3.67 19377 19329 4441 4289 22.2 10.84 18734 18686 7046 6894 36.9 22.67 17873 17825 6642 6290 35.2 33.33 16918 16870 5772 5620 33.3 48.67 16057 16009 4658 4506 28.1

-75TABLE 5 (CONT'D) Corrected 14 14Corrected Total Counts Total Counts CO Counts CO Counts Time per Two per Two per Two per Two 14 (hours) Minutes Minutes Minutes Minutes CO% Run No. 10-10, x=2.006, ao=0.464, T=300~C 1.5 19197 19149 2365 2213 11.6 6.25 18808 18760 6211 6059 32.3 17.92 19298 19250 7135 683 36.3 24.84 17452 17404 6410 6249 35.9 47.84 14719 15671 4679 4527 28.9 Run No. 10-12, x=2.006, aO=O.464, T=250~C 1.5 20945 20897 439 287 1.37 6.17 18610 18562 1093 941 5.07 11.5 18624 18576 1775 1623 8.75 27.0 19510 19462 3187 3035 15.6 35.0 18495 18447 3714 3562 19.3 48.17 18998 18950 4602 4450 24.0 75.17 17684 17636 5712 5560 31.5 97.17 17661 17613 6254 6102 34,7 106.67 17414 17366 6218 6066 34.9 118.67 16859 16811 6410 6258 37.2 123.0 16682 16634 6351 6199 37.3 144.o 18042 17994 6671 6520 36.2 168.67 16382 16334 6591 6440 39o 4

-76TABLE 5 (CONTID) Corrected 14Corrected Total Counts Total Counts CO Couns CO Counts Time per Two per Two per Two per Two (hours) Minutes Minutes Minutes Minutes CO% 192.00 15885 15837 6585 6433 40.6 216.67 15667 15619 6284 6152 39.2 241.0 15845 15797 6099 5947 37.6 Run No. 10-13, x=2.006, ao=0.464, T=350~C 1.0 20775 20727 7106 6954 33.6 2.5 19121 19073 7107 6955 36.5 4.5 17904 17856 5628 5476 30.7 11.5 16244 16196 3014 2862 17.7 Run No. 10-14, x=2.006, ao=464, T=350~C 0.5 19518 19470 5073 4921 25.3 1.5 19373 19325 6946 6594 34.1 3.5 17598 17550 5965 5813 33.1 10.5 16344 16296 3086 2934 18.0 25.0 14502 14454 1198 1046 7.3 Run No. 17-1, x=2.099, a=0.464, T=350~C 1.0 21456 21408 5540 5388 25.2 2.0 20346 20298 6993 6841 33.7 3.5 19929 19881 7333 7181 36.1 9.5 19135 19087 5874 5722 30.0 24.5 18133 18085 3689 3537 19.5

-77TABLE 5 (CONT'D) Corrected 14Corrected Total Counts Total Counts CO Counts C Counts Time per Two per Two per Two per Two 14 (hours) Minutes Minutes Minutes Minutes CO% Run No. 17-2, x=2.699, a=0.464, T=350~C 0.5 19526 19478 2969 2817 14.5 1.5 19093 19045 5630 5478 28.8 3.0 18543 18495 6712 6560 35.5 4.5 17781 17733 6450 6298 35.5 23.0 15951 15903 3060 2908 18.3 29.0 15332 15284 2533 2381 15.6 Run No. 17-3, x=2.099, a=0.464, T=310~C 2.0 19005 18957 1828 1676 8.9 4.5 18613 18565 3321 3170 17.1 10.5 18411 18363 5780 5628 30.7 19.0 18209 18161 6981 6830 37.6 24.0 17643 17595 6692 6540 37.2 28.5 17187 17139 6566 6414 37.4 36.5 17107 17059 6273 6121 35.9 52.5 16590 16542 5548 5396 32.6 72.0 15821 15773 4811 4660 29.6 95.3 15431 15381 3898 3746 24.4 Run No. 17-5, x=2.099, a=0.464, T=390~C 1.0 20547 20499 6944 6792 33.1 2.0 19595 19547 5768 5616 28.7

-78TABLE 5 (CONT'D) CorrecteCorrected Total Counts Total Counts 1CO Counts CO Counts Time per Two per Two per Two per Two 14 (hours) Minutes Minutes......Minutes...Minutes CO% 3.0 18643 18595 4826 4674 25.2 4.0 17936 17888 4193 4041 22.6 5.0 17389 17341 3508 3356 19.3 Run No. 17-6, x=2.099, ao=0.464, T=390~C 0.67 21749 21701 7353 7201 33.2 1.67 19673 19625 6271 6120 31.2 Run No. 17-7, x=2.099, aO=0.464, T=390~C 0.33 20944 20896 6037 5885 28.2 1.33 19849 19801 6419 6267 31.7 Run No. 18-11, x=1.903, a =o.464, T=350~C 1.0 9001 8953 209 133 1.49 3.0 8952 8904 544 468 5.26 9.75 9029 8981 1352 1276 14.2 23.67 8673 8625 2617 2541 29.4 34.0 8728 8680 2866 2790 32.1 48.0 8456 8408 2835 2759 32.8 59.25 8367 8319 3858 2782 33.4 73.0 8081 8025 2438 2362 29.3 Run No. 18-21, x=1.903, ao=2.19, T=350~C 2.0 8374 8326 136 60 0.72 4.0 8274 8226 185 119 1.45

-79TABLE 5 (CONT'D) Corrected 14 14Corrected Total Counts Total Counts CO Counts CO Counts Time per Two per Two per Two per Two 14 (hours) Minutes Minutes Minutes Minutes C0% 10.5 8420 8372 370 294 3.50 24.25 8325 8277 584 508 6.14 34.25 8154 8106 711 635 7.82 48.25 8108 8060 844 768 9.51 71.25 8229 8181 877 801 9.78 Run No. 18-33, x=1.903, a =0.218, T=350~C 1.5 8287 8239 382 306 3.71 3.5 8355 8307 657 581 7.00 6.0 8221 8173 957 881 10.77 11.0 8104 8056 1634 1558 19.30 Run No. 18-44, x=1.903, ao=1.442, T=350~C 1.84 9109 9061 169 93 1.02 5.5 8798 8750 344 268 3.07 20.0 8772 8724 946 870 10.10 47.5 8703 8655 1535 1459 16.82 Run No. 10-21, x=2.006, ao=0.218, T=250~C 1.5 7407 7359 288 212 2.88 5.5 7465 7417 796 720 9.70 12.0 7384 7336 1627 1551 21.2 24.0 7436 7388 2343 2267 30 8

-80 - TABLE 5 (CONT'D) Corrected 14 14 Corrected Total Counts Total Counts CO Counts CO Counts Time per Two per Two per Two per Two 14 (hours) Minutes Minutes Minutes Minutes CO% Run No. 10-31, x=2.006, ao=1.442, T=250~C 2.0 2837 2789 92 19 0.68 6.0 8002 7954 282 206 2.59 12.0 8019 7971 482 406 5.10 25.0 7687 7639 830 754 9.85 Run No. 17-12, x=2.099, a=0o.464, T=350~C 0.33 10499 10451 1954 1878 18.0 Run No. 17-15, x=2.099, a =0.464, T=350~C 0.67 9740 9692 2460 2384 24.6 2.67 9452 9404 3918 3842 40.9 5.0 9087 9039 3615 3539 39.2 Run No. 17-14, x=2.099, a =0.464, T=350~C 0 1.0 9910 9862 3397 3321 33.7 2.0 9506 9458 3969 3893 41.2 3.5 9114 9066 3843 3767 41.5 Run No. 17-21, x=2.099, a =2.565, T=350~C 1.0 8300 8252 975 899 10.9 2.0 8189 8141 1401 1325 16.3 3.17 7982 7934 1605 1529 19.3 6.17 7818 7770 1603 1527 19.7

-81TABLE 5 (CONT'D) Corrected 14 14Corrected Total Counts Total Counts CO Counts CO Counts Time per Two per Two per Two per Two (hours) Minutes Minutes Minutes Minutes CO0 Run No. 17-22, x=2.099, ao=2.565, T=350~C 4.5 8136 8088 1750 1674 20.7 Run No. 17-31, x=2.099, a=0.284, T=350~C 0.17 7580 7532 1356 1280 17.0 1.0 7366 7318 3458 3382 46.3 Run No. 17-32, x=2.099, ao=O.284, T=350~C 0.33 7739 7691 2347 2271 29.6 1.5 7374 7325 3720 3644 49.8 Run No. 17-33, x=2.099, ao=0.284, T=350~C 0.5 7553 7505 2758 2682 35,8 2.0 7083 7035 3551 3475 49.2 3.5 7684 6736 3065 2989 44.6 Run No. 17-41, x=2.099, ao=1.122, T=350~C 0.5 9108 9060 1372 1296 14.3 1.5 8462 8418 2455 2379 28.3 3.0 8701 8653 3050 2974 34.4

B. Adsorption Rate Data TABLE 6 EXPERIMENTAL DATA ON THE ADSORPTION OF OXYGEN BY 50.000 GRAM OF COBALT FERRITE Time(minutes) pa(microns) Time(minutes) Pa(microns) Run No. 10-14, x 2.006, T 300~C, Room Temperature = 26.1~C 0 102.3 10 14.2 0.25 85.2 15 9.8 1.00 54.3 40 3.0 1.50 47.0 60 2.2 2.0 40.2 1150 1.7 3.0 32.2 2710 1.5 5.0 23.5 Run No. 10-15, x = 2.006, T = 300~C, Room Temperature = 29.1~C 0 103.5 5 23.8 0.25 87.4 10 13.3 0.50 74.1 15 10.2 1.00 56.5 20 6.9 1.50 47 50 3.4 2.0 40.5 1325 2,7 3.0 32.5 2470 1.9

-8;3TABLE 6 (CONT'D) Time(minutes) pa(microns) Time(minutes) pa(microns) Run No. 10-19, x = 2.006, T = 200~C, Room Temperature = 25.7~C 0 101.2 10 37 0.25 88.5 20 31 0.50 76.8 40 25.7 1.0 64.0 80 20.9 1.5 58.1 190 12.9 3.0 49.0 1190 6.2 5.0 44.2 Run No. 10-20, x = 2.006, T = 200~C, Room Temperature = 25.3~C 0 102.2 10 38.1 0.25 87.4 20 31.9 0.50 77.9 40 26.6 1.0 66.o 80 20.9 1.5 59.7 170 15.6 3.0 51.1 1180 6.0 5.0 45.4 Run No. 10-21, x = 2.006, T = 100C, Room Temperature = 25.3~C 0 100.0 15 48.0 0.25 85.3 40 42.6 0.50 77.5 100 35.8 1.0 68.8 990 21.5 2.0 61.9 1470 19.1 5.0 54.3 2580 16.7

-84TABLE 6 (CONT'D) Time(minutes) Pa(microns) Time(minutes) pa(microns) Run No. 18-8, x = 1.903, T = 300~C Room Temperature = 25.3~C 0 101.7 100 64.5 0.25 93.8 360 58.6 1.0 85 1080 53.2 5.0 76.2 2220 48.3 30 68.2 Run No. 18-9, x = 1.903, T = 300~C, Room Temperature = 24.2~C 0 103.6 100 66 0.25 95.5 510 59.6 1.0 86.3 1250 56 5.0 78 1980 52.8 30.0 70.3 Run No. 18-10, x = 1.903, T = 200~C, Room Temperature = 26.3~C 0 107.2 100 78.5 0.25 100.3 380 73.2 1.0 95.3 1160 68.7 5.0 89.2 2620 65.5 30.0 82.3 Run No. 18-11, x = 1.903, T = 200~C, Room Temperature = 25.2~C 0 104.1 120 75.2 0.25 98.7 510 69.7 1.0 93.8 1430 66.0 5.0 87.8 2885 63.7 30.0 80.5

-85TABLE 6 (CONT'D) Time(minutes) pa(microns) Time(minutes) pa(microns) Run No. 18-15, x = 1.903, T = 100~C, Room Temperature = 24.4~C 0 100 100 81.5 0.25 96.5 540 74.8 1.0 94.4 1460 70.1 5.0 90.2 2660 67.1 30.0 85.0 Run No. 18-16, x = 1.903, T = 100~C, Room Temperature = 26.8~C 0 101.9 100 84.2 0.25 98 360 79.3 1.0 96.2 0545 73.6 10.0 91.7 2615 70.3 40.0 87.4 Run No. 17-1, x = 2.099, T = 300~C, Room Temperature = 25.1~C 0 102.5 60 87.4 0.25 100.3 120 84.2 1.0 99.2 360 74.5 5.0 96 1140 65 20.0 90..6 2670 55.5 Run No. 17-2, x = 2.099, T = 300~C, Room Temperature = 25.1~C 0 102.5 60 88.4 0.25 101.0 120 83.1 1.0 99.9 480 69.2 5.0 98 1330 51.1 20.0 93.8 2880 42.1

-86TABLE 6 (CONT'D) Time(microns) pa(microns) Time(minutes) pa(microns) Run No. 17-3, x = 2.099, T = 300~C, Room Temperature = 24.9~C 0 100.6 120 84.7 0.25 98.8 280 76.6 1.0 93.2 530 68.6 5.0 96 1665 52.1 30.0 91.6 2995 45.2 60.0 88.9 Run No. 17-5, x = 2.099, T 200~C, Room Temperature = 24.5~C 0 105.2 40 102.0 0.25 103.4 100 101.0 1.0 102.8 1560 99.2 5.0 102.6 Run No. 17-6, x = 2.099, T = 400~C, Room Temperature = 25.2~C 0 103.3 60 54.3 0.25 99.2 100 48.0 1.0 95.0 250 34.5 5.0 84.8 740 22.5 10.0 78.1 1440 17.7 30.0 64.5

-8 (TABLE 6 (CONT'D) Time(minutes) pa(microns) Time(minutes) pa(micrors) Run No. 17-7, x = 2.099, T = 400~C, Room Temperature = 24.8~C 0 98.8 60 50.6 0.25 95.5 170 38.1 1.0 89.7 300 31.0 5.0 79.1 650 24.6 10.0 71.3 1440 18.0 30.0 58.2 Run No. 14-1, x = 1.954, T = 300~C, Room Temperature = 26.2~C 0 102.0 160 90.1 0.25 100.3 500 87.9 1.0 98.1 1240 85.2 10.0 95.0 2730 82.3 30.0 93.0 Run No. 14-2, x = 1.954, T = 300~C, Room Temperature = 25.3~C 0 100.9 120 92.2 0.25 99.9 420 90.0 1.0 98.3 1500 87.8 10.0 96.0 2800 86.4 30.0 94.1 Run No. 14-4, x = 1.954, T = 500~C, Room Temperature = 27.7~C 0 103.6 30 65.7 0.25 98 100 55.2 1.0 88 180 49.8 5.0 77.9 560 40.6 10.0 74 1340 33.6

-88TABLE 6 (CONT'D) Time(minutes) pa(microns) Time(minutes) pa(microns) Run No. 14-5, x = 1.954, T = 500~C, Room Temperature = 23.1~C 0 103.3 90 57.0 0.25 97.5 180 51.7 1.0 87.8 690 42.2 10.0 74.2 1360 37.7 30.0 66 Run No. 14-6, x = 1.954, T = 400~C, Room Temperature = 25.3~C 0 105.7 100 76.6 0.25 102.0 260 70.7 1.0 93.8 620 64.2 5.0 88.0 1725 53.2 30.0 81.6 Run No. 14-7, x = 1.954, T = 400~C, Room Temperature = 22.3~C 0 105.2 30 81.0 0.25 101.0 120 74.8 1.0 93.0 580 64.0 5.0 87.4 1320 55.4 Run No. 15-3, x = 2.058, T = 400~C, Room Temperature = 25.10C 0 101.4 45 51.8 0.25 97.5 90 44.7 1.0 90.2 180 38.2 5.0 76.2 390 32.6 10.0 68.4 1430 28.0 20.0 60o.1

-89TABLE 6 (CONT'D) Time(minutes) pa(microns) Time(minutes) pa(microns) Run No. 15-4, x = 2.058, T = 400~C, Room Temperature = 26.7~C 0 100.9 20 61 0.25 97.7 50 52.2 1.0 89.6 110 44.8 5.0 75.8 210 39.3 10.0 68.3 1415 29.0 Run No. 15-6, x = 2.058, T = 350~C, Room Temperature = 26.7~C 0 99.7 40 73.1 0.25 97.7 80 68.2 1.0 93.2 180 62.4 5.0 86.3 615 52.2 15.0 79.3 1395 44.4 Run No. 15-7, x = 2.058, T = 350~C, Room Temperature= 28.6~C 0 99.4 30 72.2 0.25 97.0 60 67.9 1.0 92.2 100 64.2 5.0 84.5 390 55.4 15.0 77.3 1225 47.5 Run No. 15-8, x = 2.058, T = 300C, Room Temperature = 28.7~C 0 103.2 30 85.8 0.25 102.0 60 81.5 1.0 99.2 130 76.8 5.0 95 370 69.0 15.0 89.5 1255 60.8

-.90C. Data of Oxygen Desorption TABLE 7 EXPERIMENTAL DATA OF OXYGEN DESORPTION FROM 50.000 GRAMS OF COBALT FERRITE Time (minute) Pa(micron) Time(minute) pa(micron) Run No. 10-14, x = 2.006, T = 300~C, Room Temperature = 23.1~C 0 5.35 120 7.2 1 5.5 230 8.2 5 5.6 1390 12.8 60 6.48 Run No. 10-15, x = 2.006, T = 300~C, Room Temperature = 24.7~C 0 5.07 150 6.9 1 5.07 330 8.3 5 5.22 1410 11.7 30 5.65 Run No. 10-19, x = 2.006, T = 200~C, Room Temperature = 24.4~C 0 1.41 140 1.56 1 1.42 1390 2.6 50 1.47 Run No. 10-20, x = 2.006, T = 200~C, Room Temperature = 25.5~C 0 1.14 110 1.78 1 1.16 15-10 4.27 10 1.31 Run No. 10-21, x = 2.006, T = 100~C, Room Temperature = 26.0~C 0 0.7 100 1.86 1 1.13 410 2.57 10 1.27 1230 3.17

-91TABLE 7 (CONT'D) Time(minute) pa(micron) Time(minute) pa(micron) Run No. 10-21, x = 2.006, T = 100~C, Room Temperature = 26.0~C 0 0.7 100 1.86 1 1.13 410 2.57 10 1.27 1230 3.17 Run No. 18-8, x = 1.903, T = 300~C, Room Temperature = 24.5~C 0 16.28 140 19.7 1 16.43 420 21.8 5 16.93 1420 23.1 30 17.88 Run No. 18-9, x = 1.903, T = 300~C, Room Temperature = 25.1~C 0 16.3 290 21.0 1 17.2 1475 25.6 30 18.05 Run No. 19-10, x= 1.903, T = 200~C, Room Temperature = 25.5~C 0 27.0 270 29.2 1 28.4 2910 30.9 30 28.9 Run No. 18-11, x = 1.903, T = 200~C, Room Temperature = 29.3~C 0 17.2 505 19.6 1 18.4 2865 21.1 30 18.6 Run No. 18-15, x = 1.903, T = 100~C, Room Temperature = 23.5~C 0 20.1 105 21.4 1 21.3 1305 21.6

-92TABLE 7 (CONT'D) Time(minute) Pa(micron) Time(minute) pa(micron) Run No 18-16, x = 1.903, T = 100~C, Room Temperature = 27.5~C 0 16.8 240 18.1 1 17.9 2730 18.5 Run No. 17-1, x = 2.099, T = 300~C, Room Temperature = 25.5~C 0 7.1 370 17.9 1 8.5 1420 22.0 30 10.4 2850 23.9 Run No. 17-2, x = 2.099, T = 300~C, Room Temperature = 24.20C 0 37.8 1380 41.0 1 38.7 2620 250 39.1 Run No. 17-5, x = 2.099, T = 200~C, Room Temperature = 24.5~C 0 14.1 90 16.0 1 15.9 2885 16.1 Run No. 17-6, x = 2.099, T = 400~C, Room Temperature = 24.0~C 0 7.3 120 23.1 1 8.8 540 33.8 5 11.7 1440 47.3 10 13.9 2940 50.2 30 18.0 Run No. 17-7, x = 2.099, T = 400~C, Room Temperature = 24.7~C 0 4.9 100 13.8 1 5.8 220 17.8 5 6.6 1320 36.2 40 10.4 3080 50.6

-93TABLE 7 (CONT'D) Time(minute) pa(micron) Time(minute) pa(micron) Run No. 14-1, x = 1.954, T= 300~C, Room Temperature = 29.7~C 0 9.0 60 12.5 1 10.5 530 16.0 5 10.8 Run No. 14-2, x = 1.954, T = 300~C, Room Temperature = 26.2~C 0 28.4 150 34.5 1 30.5 440 36.2 5 30.9 1320 37.8 40 32.3 2740 38.7 Run No. 14-4, x = 1954, T = 500~C, Room Temperature = 23.9~C 0 5.3 60 20.2 1 7.3 310 32.1 5 10.3 660 40.5 20 14.9 1635 53.2 Run No. 14-5, x = 1.954, T = 500~C, Room Temperature = 22.3~C 0 5.3 280 31.5 1 7.9 690 42.2 5 10.7 1400 52.5 20 15.0 2880 65.0 70 21.4 Run No. 14-6, x = 1.954, T = 400~C, Room Temperature = 24.5~C 0 7.4 150 15.0 1 9.0o 290 17.1 5 10.0 1430 23.2 40 12.3

-94TABLE 7 (CONT'D) Time(minute) Pa(micron) Time(minute) Pa(micron) Run No. 14-7, x = 1.954, T = 400~C, Room Temperature = 25.1~C 0 7.6 40 12.8 1 9.5 280 17.9 5 10.5 1380 14.5 Run No. 15-3, x = 2.058, T = 400~C, Room Temperature = 24.1~C 0 8.6 110 31.5 1 11.6 170 36.9 5 14.9 460 58.1 30 21.5 1270 90.1 Run No. 15-4, x = 2.058, T = 400~C, Room Temperature = 27.8~C 0 5.9 60 21.8 1 7.8 160 31.6 10 12.9 1310 83.1 30 17.2 Run No. 15-6, x = 2.058, T = 350~C, Room Temperature = 267~0C 0 7.8 100 20.5 1 9.3 350 28.2 5 10.1 640 35.4 30 15.5 1455 46.0 Run No. 15-7, x = 2.058, T = 350~C, Room Temperature = 27.0~C 0 14.8 180 30.9 1 16.1 660 44.1 10 19.2 1410 52.8 60 25.0

-95TABLE 7 (CONT D) Time (minute) pa(micron) Time(minute) a (micron) Run No. 15-8, x = 2,058, T = 300~C, Room Temperature = 26.8~C 0 10.6 100 15.9 1 11.8 310 19.8 30 13.4 1520 27.6

APPENDIX II SAMPLE CALCULATIONS A. Exchange Reaction of CO2 and CO For a given run with constant ratio of CO2 and CO at constant temperature, the rate constant k(ao) was calculated by applying equation (21). V d p14cO = k(ao) [p 14 - (1 + a ) P14 ] (21) ART dt C02 CO The following is a sample calculation of k(ao) for Run 17-21 (of Table 5) at ao=2.565, T=350~C by two methods. (i) Initial Rate Method: At t=O, equation (21) can be simplified to ( \ d 14CO P114C p k(a,) ART (74) dt t=0 V The raw date of P14C /P 14 versus time are plotted CO 1 C02 in Figure 4. They'.ajre' tabulated in Table 8. From Figure 16 of the plot P14Co/Pi1 versus t, the slope at t=0 was obtained,by equation (74) k(ao) ART 0.116 =.116 (-hr1) V 1.0 4 2 since A=2.48 x 10 cm for 1.000 gm of catalyst T=62 ~K V=507 cm3 -96

-97initial slope 1.2 10 / Run 17-21 Z6 0 O 4 0 2 t 0.2 0.4 0.6 0.8 1.0 Figure 16. P4 — versus t for Run 17-21 and initial slope i p 14c C02

-980.116 x 507 -7.. k(ao) = 0 6 =.464 x 107 1 mole ) 82.05 x 623 x 2.48 x 104 hr.atm cm In order to prove that k(ao) is not a function of time, the values of k(ao) are calculated from the raw data by applying equation (23) which is the integrated form of equation (21). The results are tabulated in Table 8. P14C k (a)RTA(l+a)t Qn[l-(l + a) ] = - k(ao)RTA(l+ao)t P14 V (23) P 14 V CO2 P14Co The following is a sample calculation at t-0.3, i4 = 0.35. P 14C -507 x in [1-(1 + 2.565) x 0.035] k(a ) = 4 ao 82.05 x 623 x 2.48 x 104 x 3.565 x 0.3 mole = 0.45 x10 ( —- 2 ) hr- atm cm TABLE 8 THE RESULT OF RUN 17-21 OF THE EXCHANGE REACTION OF CO2 AND CO. t(hr) 0 0.1 0.3 0.5 0.8 1.0 2.0 Pl4O o 0.012 0.035 0.057 0.090 0.109 0.163 p14 1CO2 calculated k(a ) x 107 0.46 0.45 0.46 0.48 0.46 0.47 hrmole hr atm cm )

-99(ii) Digital Computer Simulating Method: Knowing the value of k(ao), the numerical relationship of P1l4c /pil4c2 versus t could be obtained from solving equation (21) by the finite difference method. Euler's method was applied to solve the first order differential equation. The calculation was done on a IBM 7090 computer. Different values of k(ao) ART/V were fed to the computer to simulate the experimental data. Figure 17 shows the result of Run 17-21, which also gives k(aO) ART/V = 0.116 [hr-1]. The following is the main program written in MAD to solve for the numerical solution of equation (21) by Euler's method. INTEGER COUNT, FREQ PRINT COMMENT $ L EULERS METHOD SOLUTION $ START READ AND PRINT DATA T =0 PCO = 0 PRINT RESULTS T, PCO COUNT = 0 THROUGH STEP FOR T = 0., H, T.G. TMAX COUNT = COUNT + 1 PCO = PCO + H * (KFORW -(KFORW + KFORW * AO)* PCO) STEP WHENEVER (COUNT/FREQ) * FREQ. E. COUNT, PRINT RESULTS T+H, PCO TRANSFER TO START END OF PROGRAM

20 ART k(a ) -=0.2 by computer o V z O I /' Experimental Result of Run 17-21 a: 0 ART LL" k(ao) - o = 0.116 by computer 0.10 k(ao) ART 0.1 by computer 0 O a 0 1.0 2.0 t (hr.) P14co Figure 17. -p —-- versus t for Run 17-21 p 114 C02 and Results from Computer Simulating Method

-101The following is the list of variables in the program. Program Symbol Definition PCO P14co/Pi142CO KFORW ART k(ao) -T T t AO a0 COUNT The number of times the algorithm has been applied FREQ A parameter which control the printing frequency H step size B. Adsorption and Desorption of Oxygen In Table 3, the amount of oxygen adsorbed per unit surface area of the sample, the initial rate of adsorption and the rate constant Kads. were calculated from equations (58), (48) and (59). /\na Va A = ART (Pai - Paf) (58) Va dpa = k d [Co.2BI- Fe IBl br- 7(Pai - Pa)}Pa (48) ART dt ads. and Ad- n t= Kads Pai = kads[C+2B- Fe+3lBI]brPai (59) The following is a sample calculation for Run 10-14 (of Table 6) with x=2.00oo6 and T=300~C Ana 5. 302 (102. - 2.15) A - 50 x 2.48 x 104 x 760,000 x 0.08205 x 298 - 2.29 x 1011 mole = 2.29 x 10 (cm2 )

-102From equation (48) the values of [Co+21BI - Fe+31BI]br and kads were obtained by the digital computer simulating method. Sets of values of [Co+21BI- Fe+51BI] and kadsART/Va were fed to a IBM 7090 computer to solve equation (48) and to simulate the experimental data of Pa versus t. The value y in equation (48) was defined by equation (47) Va N _5.302 x 6.023 x 1023 ARTp 0.08205 x 298 x 5 x 2.48 x 10 x 1 x 760,000 23.4 = 3.35 x 10 (micron ) Figure 18 shows the plot of experimental data of Run 10-14 and the plot of numerical solution of equation (48) with [Co+21Bt - Fe+31Bl]br = 0.0271 and kads ART/Va = 24.3 (min-1). Then from equation (59), 24.3 x 0.0271 x 5.302 x 60-.46 X -4 (mole Kads. =.46 x 10 ( atmcm 2.48 x 104 x 0.08205 x 298 and dna _ 3.46 x 10-4 x 102. 466 x 10 -10 (mole ART/ t=O 760,000 hr.cm The following is the MAD program to solve equation (48) by the IBM 7090 computer. INTEGER COUNT, FREQ PRINT COMMENT $ 1 EULERS METHOD SOLUTION $ START READ AND PRINT DATA T = 0 P02 = P020

O experimental data results from computer simulating with 100 [Co+21B- Fe+SBI ]br=0.0271 10 kpg ART = 24.3 Va 80 Q results from computer simulating with [Co+ |Bi - Fe+31B ]br= 0.0271 o |ic~~~~~~~~ t~ kads ART.0 o Va 3 60 40 0 4 8 12 16 20 t (min) Figure 18. pa versus t for Run 10-14 and Results from the Computer Simulating Method

-104PRINT RESULTS T P02 COUNT = 0 THROUGH STEP, FOR T = O., H, T.G. TMAX COUNT = COUNT + 1 BRCNT = BRCNTR - GAMMA* (P020 - P02) P02 = P02 - H * KADS * BRCNT * P02 STEP WHENEVER (COUNT/FREQ) * FREQ.E.COUNT,PRINT RESULTS T+H, P02 TRANSFER TO START END OF PROGRAM The following is the list of variables in the program. Program Symbol Definition T t P02 Pa P020 Pai GAMMA 7 KADS kads ART Va BRCNTR [Co+2 BI-Fe+31B ]br H Step size COUNT The number of times the algorithm has been applied FREQ A parameter which controls the printing frequency In Table 4, the amount of oxygen desorbed per unit surface area and the initial rate of desorption were calculated from equations (61) and (50). For Run 10-15 (of Table 7) with x=2.006 and T=300~C;

-105pdf=11.7 micron, Pdi5.07 micron, pd5J07 micron at t=l min. and pd=5.22 micron at t=5 min. Lnd = Vd(Pdf-Pdi) A ART = 5.205 x (11.7 - 5.07) 760,000 x 0.08205 x 298 x 50 x 2.48 x 10'.55 x O-12 mole cm2 dn _==Va ( dpd Adt t= ART dt /t= 6o 5.205 x (5.22 - 5.07) x V 760,000 x 0.08205 x 298 x 50 x 2.48 x 104 = 0.725 x 10-12 mole hr. cm

APPENDIX III STRUCTURE OF COBALT FERRITE Cobalt ferrite exists in a structure of "inverse" spinel, Fe+31A - [Co+21BI Fe+31Bj]04 [44]. Figure 19 shows the composition of (100), (110) and (10) planes in CoFe204 [64]. The value of a, the lattice parameter of CoFe204, is 8.39 A. The proposed active center, Co+2BI - Fe+3|BI cation pair, appears in plane (100). The inter-cation distance of Co+21BI - Fe+31BI is ad2/4 = 2.97 A as shown in Figure 19. -106

-107 (100) X 00 ED a 0 0 O( )oxygen +2 o Co 1B1 O Fe+31A1 0 Fe+]^Al Figure 19. Composition of (100), (110) and (110) planes in CoFeO,4

-io8(io0) -O O O ^0-~0 00 0 O O O (iio) \~00000\ J] O Fgure 19. (c0ND) Figure 19. (CONT'D)

BIBLIOGRAPHY 1. Aigrain, P. and Dugas, C. Z., Elecktronchem, 56, 363 (1952). 2. Balandin, A. A., Catalysis and Chemical Kinetics, Academic Press, New York, (1964). 3. Barry, T. I., and Stone, F. S., Proc. Roy. Soc., A 255, 124, (1960). 4. Barth, T. F. W., and Posnjak, E., Zeits. f. Krist., 82, 235, (1932). 5. Bawn, C. E. H., and Ledwith, A., Chem. and Ind., 1329, (1958). 6. Beebe, R. A., and Dowden, D. A., J. Am. Chem. Soc., 60, 2912, (1938). 7. Bell, R. P., Acid-Base Catalysis, The Clarendon Press, London, (1941). 8. Boudart, M., J. Am. Chem. Soc., 1531, (1952). 9. Bragg, W. H., Phil. Mag., 30, 305, (1915). 10. Brandner, J. D., and Urey, H. C., J. Chem. Phys., 3, No. 9, 351, (1945). 11. Brunauer, S., Love, K. S., and Keenan, R. G., J. Am. Chem. Soc., 64, 751, (1944). 12. Burwell, R. L., Jr., and Taylor, H. S., J. Am. Chem. Soc., 5, 697, (1936). 13. Dell, R. M., Stone, F. S., and Tiley, P. F., Trans. Faraday Soc., 492, 201, (1953). 14. Dowden, D. A., J. Chem. Soc., 1950, 242, (1950). 15. Dowden, D. A., Mackenzie, N., and Trapnell, B. M. W., Proc. Roy. Soc., A 237, 245, (1956). 16. Dzisyak, A. P., Boreskov, G. K., Kasatkina, L. A., and Kochurikhin, V. E. Kinetika i kataliz, 2, 386, (1961). 17. Elovich, S. Yu., and Zhabrova, G. M., Zh. Fiz. Khim., 1, 1761, (1939). 18. Engell, H. J., and Hauffe, Z., Elektrochem, 2, 773, (1953). 19. Faeth, P. A., and Willingham, C. B., Technical Bulletin on the Assembley, Calibration and Operation of a Gas Adsorption Aparatus for the Measurement of Surface Area, Pore Volume Distribution and Density of Finely Divided Solids, Mellon Institute of Industrial Research, Pittsburgh, (1955). -109

-11020. Garner, W. E., Stone, F. S., and Tiley, P. F., Proc. Roy. Soc., A 211, 472, (1952). 21. Garner, W. E., Advances in Catalysis, 9, 169, (1957). 22. Garner, W. E., J. Chem. Soc., p. 1239, (1947). 23. Garner, W. E., and Veal, F. J., J. Chem. Soc., 1487, (1935). 24. Glasstone, S., Laidler, K. J., and Euring, H., in The Theory of Rate Processes, McGraw-Hill, New York, 1941. 25. Gorter, E. W., Phil, Res. Rep., 2, 295, (1954). 26. Grabke, H. J., in Proc. Third. Inter. Congress on Catalysis, p. 928, (1964). 27. Grabke, H. J., Ber. Bunsenges. Physik. Chem. Bd., 69, Nr. 1, (1965). 28. Hauffe, K., Advances in Catalysis, 1, 213, (1955). 29. Hauffe, K., and Engell, H. J., Z. Elekrochem., 56, 366, (1952). 30. Hayakawa, T., Bull. Chem. Soc. Japan, 26, 165, (1953). 31. Hill, T. L., Introduction to Statistical Thermodynamics, Addison-Welsley, Reading, Massachusetts, (1960). 32. Howard, D. C., and Trapnell, B. M. W., Chemisorption, Butterworths and Co., London, (1964). 33. Hwang, S. T., Ph.D. Thesis, The University of Michigan, (1965). 34. Jennings, T. J., and Stone, F. S., Advances in Catalysis, 9, 441, (1957). 35. Jonker, G. H., J. Phys. Chem. Solids, 9_, 165, (1959). 36. Jonker, G. H., and van Houten, S. in Halbleiterprobleme Band VI, p. 118, Verlag Friedr. Vieweg and Sohn, Braunschweig, (1961). 37. Laidler, K. J., J. Phys. Chem.,, 712, (1949). 38. Langmuir, I., J. Am. Chem. Soc., 28, 2221 (1916; ibid. 40, 1361, (1918); Trans. Faraday Soc., 17 607, (1922). 39. Langmuir, I., Phys. Rev., 22, 357, (1923). 40. Langmuir, I., J. Am. Chem. Soc., 54, 2798, (1932). 41. Law, J. T,, J. Phys. Chem. Solids., 4, 91, (1958). 42. Ligenza, J. R., J. Phys. Chem., 64, 1017, (1960).

-11143. Low, M. J. D., Chem. Reviews 6 267, (1960). 44. Molinari, E., and Parravano, G., J. Am. Chem. Soc., A, 5233, (1953). 45. Morrison, S. R., Advances in Catalysis, z 259, (1955). 46. Muller, E., and Huber-Emden, H., Ann., 649, 70, (1961). 47. Muller, W. Gehlordnungein, Koblatferrit, Diplomarbeit, Gottingen, (1963). 48. Muller, W., and Schmalzried, H., Berichte der Bun'sengesellschaft fur Physikalische Chemie, Band 68, Helf 3, 270, (1964). 49. Pauling, L., The Nature of Chemical, the Bond Cornell University Press, N.Y., (1960). 50. Parravano, G., and Boudart, M., Advances in Catalysis, L 50, (1955). 51. Parravano, G., unpublished paper. 52. Pietrzak, H., Gates, W. C., and Parravano, G., unpublished report. 53. Ratterman, F., unpublished work. 54. Robin, J., and Benard, J., Compt. rend. 232, 1830, (1951); Compt. rend. 234, 734, (1952). 55. Romeizn, F. C., Philips Research Rep. 8, 304, (1953). 56. Rudham, R., and Stone, F. S., in Chemisorption, ed. W. E. Garner, London, Butterworth, (1957). 57. Schmalzried, H., and Wagner, C., and Z. Physik. Chem. NF 31, 198, (1962). 58. Schmalzried, H.,"Point Defects in Ternary Ionic Crystals" in Progress in Chemistry of the Solid State, Pergamon Press, N.Y., (1965). 59. Scholten, J. J. F., Zwietering, P., Konvalinka, J. A., and deBoer, J. H., Trans. Faraday Soc.,, 2166, (1959). 60. Schwab, G. M., and Block, J., Z. Elektrochem., 58, 756, (1954). 61. Schwab, G. M., Roth, E., Grintzos, C., and Mavrakis, N., in Structure and Propertiesof Solid Surfaces, p. 464, University of Chicago Press, Chicago, (1953). 62. Smiltens, R., J. Am. Chem. Soc., 79, 4481, (1957). 63. Squires, R. G., Ph.D. Thesis, The University of Michigan, (1962).

-11264. Squires, R. G., and Parravano, G., J. Catalysis, 2, 324, (1963). 65. Stone, F. S., Rudham, R., and Gale, R. L., Z. Elecktronchem., 6_, 129, (1959). 66. Stone, F. S., Advances in Catalysis, 13, 1, (1962). 67. Taylor, H. A., and Thon, N., J. Am. Chem. Soc., 74, 4169, (1952). 68. Taylor, H. S., Z. Electrochem., 20, 201, (1914). 69. Taylor, H. S., J. Amer. Chem. Soc., 3, 551, (1915). 70. Taylor, H. S., and Williamson, A. T., J. Am. Chem. Soc., 53, 2168, (1931). 71. Taylor, H. S., and Ogden, G., Trans. Faraday Soc., 3, 1178, (1934). 72. Verwey, E. J. W., Haayman, P. W., and Romeijn, F. C., J. Chem. Phys., 15, 181, (1947). 73. Verwey, E. J. W., and Heilmann, E. L., J. Chem. Phys., 15, 174, (1947). 74. Wagner, D., J. Chem. Phys., 18 69, (1950). 75. Wagner, C., Vorschlage zu Untersuchungen der Katalyse der Reaction C02 + H2 = CO + H20 an Wustit, Mermorandum vom, Jan. 3, 1962. 76. Wagner C., Vorschlage zu Untersuchungen uber die Reaktion ci402 + C120= C1202 + C140, Jan. 14, 1962. 77. Weisz, P. B., J. Chem. Phys.,20 1483, (1952). 78. Weisz, P. B., J. Chem. Phys., 21, 1531, (1953). 79. Winter, E. R. S., J. Chem. Soc., p. 2726, (1955). 80. Winter, E. R. S., Advances in Catalysis, 10, 196, (1958). 81. Wolkenstein, Th., J. Chem. Phys., 54, 175, (1957). 82. Wolkenstein, Th., Advances in Catalysis, 2, 807, (1957). 83. Wolkenstein, Th., Advances in Catalysis, 12, 189, (1960). 84. Wolkenstein, Th., in The Electronic Theory of Catalysis on Semiconductors, Pergamon Press, 1963.