University of Southern California Dissertations and Theses

Perovskite Structure Rare-Earth Transition-Metal-Oxide Catalysts

(USC Thesis Other)

Perovskite Structure Rare-Earth Transition-Metal-Oxide Catalysts

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content PEROVSKITE STRUCTURE RARE EARTH TRANSITION METAL OXIDE CATALYSTS by George Wayne Berkstresser A Dissertation Presented to the FACULTY OF THE GRADUATE SCHOOL UNIVERSITY OF SOUTHERN CALIFORNIA In Partial Fulfillment of the Requirements for the Degree DOCTOR OF PHILOSOPHY (Chemical Engineering) February 1973 INFORMATION TO USERS This dissertation was produced from a microfilm copy of the original document. While the most advanced technological means to photograph and reproduce this docum ent have been used, the quality is heavily dependent upon the quality of the original submitted. The following explanation of techniques is provided to help you understand markings or patterns which may appear on this reproduction. 1. The sign or "target" for pages apparently lacking from the docum ent photographed is "Missing Page(s)". If it was possible to obtain the missing page(s) or section, they are spliced into the film along with adjacent pages. This may have necessitated cutting th ru an image and duplicating adjacent pages to insure you com plete continuity. 2. When an image on the film is obliterated with a large round black mark, it is an indication that th e photographer suspected th at the copy may have moved during exposure and thus cause a blurred image. You will find a good image of the page in th e adjacent frame. 3. When a map, drawing or chart, etc., was part of th e material being p h o to g rap h e d the photographer followed a definite m ethod in "sectioning" the material. It is custom ary to begin photoing at the upper left hand corner of a large sheet and to continue photoing from left to right in equal sections with a small overlap. If necessary, sectioning is continued again - beginning below th e first row and continuing on until complete. 4. The majority of users indicate th a t the textual content is of greatest value, however, a somewhat higher quality reproduction could be made from "photographs" if essential to the understanding of the dissertation. Silver prints of "photographs" may be ordered at additional charge by writing the Order Department, giving the catalog number, title, author and specific pages you wish reproduced. University Microfilms 300 North Z aeb Road Ann Arbor, Michigan 48106 A Xerox Education Company BERKSTRESSER, George Wayne, 1944- PEROVSKITE STRUCTURE RARE EARTH TRANSITION METAL OXIDE CATALYSTS. University of Southern California, Ph.D., 1973 Engineering, chemical 1 University Microfilms. A X E R O X Company. Ann Arbor, Michigan THIS DISSERTATION HAS BEEN MICROFILMED EXACTLY AS RECEIVED. UNIVERSITY O F SOUTHERN CALIFORNIA THE GRADUATE SCHOOL UNIVERSITY PARK LOS ANGELES. CALIFORNIA B 0 0 0 7 This dissertation, written by George Wayne Berkstresser under the direction of h..is. Dissertation Com mittee, and approved by all its members, has been presented to and accepted by The Graduate School, in partial fulfillment of requirements of the degree of D O C T O R O F P H I L O S O P H Y Duis Date. February 1973 DISSERTATION COMMITTEE Chairman PLEASE NOTE: Some p ag e s may have i n d i s ti n c t p rin t. Filmed a s r e c e iv e d . University Microfilms, A Xerox Education Company ACKNOWLEDGMENTS I feel very fortunate to have had Drs. J. M. Whelan, C. J. Rebert, and F. A. Kroger serve as members of my dissertation committee. They have given me much free dom in the pursuit of my research so that I could develop the skills necessary for independent study, yet they have closely followed my progress and provided wise counsel when the need arose. I wish to express a special appreci ation to Dr. Whelan for his very thoughtful guidance during the research and in the preparation of the dissertation. The contribution of Dr. E. G. Partridge must also be recognized since his guidance in my early graduate work developed many skills I found so useful during the disser tation research. The importance of basic mechanical skills to the success of an experimentalist has clearly been demonstrated to me during this research. Thus, I wish to note my appre ciation to the instruction and assistance given by Mr. Jim Emerson in preparation of glassware, and to Mr. James Scott in fabrication of other apparatus. Every student must encounter the university bureaucracy during his studies, but through the guidance of Mrs. Ruth Toyama 1 feel my task has been made much easier. Mrs. Toyama's very conscientious preparation of the manuscript of the dissertation is sincerely appreci ated. Most importantly, I wish to thank my parents for their encouragement and assistance, without which I could not have performed this work. iii TABLE OF CONTENTS Page ACKNOWLEDGMENTS ......................................... ii LIST OF FIGURES ......................................... vii LIST OF TABLES .......................................... X CHAPTER I. INTRODUCTION ...................................... 1 II. PRELIMINARY EXPERIMENTS .......................... 13 Introduction ...................................... 13 Experimental ...................................... 14 Results of Kinetic Study ........................ 15 Electrical Conductivity ......................... 19 Conclusions ....................................... 23 III. EXTENDED SURVEY OF PEROVSKITE CATALYSTS......... 26 Introduction ...................................... 26 Experimental ...................................... 27 Carbon Monoxide Oxidation ....................... 27 Nitric Oxide Reduction .......................... 30 Hydrogen Treatment of Catalysts ................ 35 Water Shift Reaction Over LaCo03 ............... 36 Carbon Monoxide Decomposition Over LaCo03 ...... 38 Conclusions ....................................... 42 iv Page IV. CATALYTIC ACTIVITY OP PEROVSKITE STRUCTURE OXIDES ............................................ 44 Introduction ...................................... 44 Experimental ...................................... 45 Comparative Catalytic Activity ................. 51 Effects of Ce and Sr Additions to LaCo03 ...... 58 Effects of Ce Addition to LaCr03 ............... 71 Estimation of Specific Surface Areas of LaCo03 Series ............................................ 74 Properties of the Perovskite Oxides ........... 83 V. DISCUSSION........................................ 94 Introduction ...................................... 94 Summary of Kinetic Data ......................... 94 Catalytic Activities of Doped and Undoped Zirconia and Thoria ............................. 96 Catalytic Activities of Doped LaCo03 and NiO .. 98 Comments on Reaction Mechanism Model .......... 105 Conclusions ....................................... 114 VI. EXPERIMENTAL MATERIAL PREPARATION.............. 117 Supply of Chemicals .............................. 117 Preparation of Catalysts, Survey Experiments .. 119 Preparation of Perovskite Structure Oxide Catalysts ......................................... 120 Alternative Preparation Techniques ............ 129 Nitrate Decomposition ...................... 129 Precipitation Technique ................... 131 Freeze Drying of Nitrate Solutions ....... 134 v Page VII. EXPERIMENTAL APPARATUS DESIGNS.................. 135 Preliminary Screening Experiments .............. 135 Flow System................................. 135 Reactor Design .............................. 135 Gas Analysis System........................ 139 Experimental Procedure ..................... 140 Experimental Errors ........................ 141 Primary Kinetic Experiments ..................... 142 Flow System................................. 144 Reactant Stream Preparation ............... 147 Reactor Design .............................. 149 Reactor Furnace Design and Control ........ 158 Gas Analysis System........................ 162 Experimental Procedure ..................... 166 Electrical Conductivity Apparatus .............. 167 Apparatus Design ........................... 168 Sample Preparation ......................... 173 Measurement of Conductivity Type ............... 173 Gas Adsorption Apparatus ........................ 176 Apparatus ................................... 177 Operating Procedure ........................ 179 REFERENCES............................................... 183 APPENDIX I. DENSITIES OF SINTERED PEROVSKITE SAMPLES ................................... 187 APPENDIX II. LaQ 9Ce0>ICr03 AS A POSSIBLE THERMISTER MATERIAL*................................. 190 vi LIST OF FIGURES Figure Page 1. Survey Experiments on Catalytic Oxidation of Carbon Monoxide .............................. 16 2. Catalytic Oxidation of Carbon Monoxide, Effects of Reactor Design ...................... 20 3. Electrical Conductivity of Catalysts of the Survey Series ................................... 22 4. Catalytic Oxidation of Carbon Monoxide, LaCo03 and LaQ>gSr0 ^Co03 Perovskites ........ 29 5. Catalytic Reduction of Nitric Oxide by Hydrogen over LaCo03 Catalyst ................. 31 6. Catalytic Reduction of Nitric Oxide with Hydrogen over LaQ>gSr0 ^Co03 Catalyst ........ 32 7. Water Shift Reaction Catalyst Using LaCo03 Perovskite ....................................... 37 8. Decomposition of Carbon Monoxide over LaCo03 Catalyst ......................................... 40 9. Variables for Plug Flow Reactor Design........ 48 10. Comparison of Catalytic Activity for Carbon Monoxide Oxidation .............................. 52 11. Carbon Monoxide Oxidation over LaCo03, Effect of Water on Activity........................... 55 12. Carbon Monoxide Oxidation over Lao.9Sro.lCo°3' Effect of Water on Activity................... 57 13. Catalytic Oxidation of Carbon Monoxide, LaCo03 Series Perovskites ...................... 60 14. Comparison to Arrhenius Rate Law, CO Oxidation over the LaCo03 Series Perovskites ............ 64 15. Effect of Water Upon LaCo03 Series Catalysts for Carbon Monoxide Oxidation ................. 65 vii Figure Page 16. Reaction Rate Variation with CO and O2 Con centration, CO Oxidation over Lan QCen ,CoO, Catalyst .............................................67 17. Reaction Rate Variation with CO and 0? Con centration, CO Oxidation over LaCo03 Catalyst. 68 18. Reaction Rate Variation with CO and O2 Con centration, CO Oxidation over LaQ gSrQ ^Co03 Catalyst .......................... I.... I........ 69 19. Reaction Rate Variation with CO Concentration over LaQ#gCeQ 1Co03 Catalyst, Variation of Temperature .!................................... 70 20. Catalytic Oxidation of Carbon Monoxide over LaCr03 Series Catalysts ........................ 73 21. Carbon Monoxide Oxidation over Ce and Sr Doped LaCo03 Catalyst ................................. 76 22. Sintered La^ gSrn iCo03 Extruded Rods, 1700X Magnification ............................ 78 23. Sintered LaQ QSrn ^Co03 Extruded Rods, 8500X Magnification ............................ 79 24. Sintered LaQ gSrQ ^Co03 Die Formed Wafer, 430X Magnification .............................. 81 25. Stoichiometry of La._xSrxCo03 and La^_xThxCo03 Perovskites ....................... 84 26. Valence Distribution of Co in La, Sr CoO«» Perovskite ........................7............. 87 27. DC Electrical Conductivity of Selected Perovskite Oxides ............................... 90 28. Activation Energy for Carbon Monoxide Using Doped Oxide Catalysts .......................... 103 29. Solubility of La(N03)3 in Water ............... 130 30. Experimental Apparatus, Preliminary Survey Experiments ...................................... 136 viii Figure Page 31. Experimental Reactor, Preliminary Survey Experiments ...................................... 138 32. Precision Distribution of Fractional Conversion ....................................... 143 33. Gas Manifold System, Reactant Stream Preparation, Primary Experiments .............. 145 34. Gas Manifold System, Sampling Ports and Reactor Station, Primary Experiments ......... 146 35. Capillary Tube Flowmeter Design .............. 148 36. Laboratory Reactor, Pyrex Glass, 20 mm Diameter Chamber ................................ 151 37. Laboratory Reactor, Pyrex Glass, 10 mm Diameter Chamber ................................ 152 38. Laboratory Reactor, Quartz Glass ............. 153 39. Laboratory Reactor Furnace .................... 160 40. Reactor Furnace Power Control Circuit ........ 161 41. Gas Analysis Flow System...................... 163 42. DC Electrical Conductivity Cell ............... 169 43. Assembly Detail of Conductivity Cell ........ 170 44. DC Electrical Conductivity Measurement Circuit .......................................... 172 45. Determination of Sign of Seebeck Coefficient . 175 46. Gas Adsorption Apparatus ...................... 178 ix LIST OF TABLES Table Page I. Relative Catalytic Activity for CO Oxidation with 02 ............................... 17 II. Relative Catalytic Activity of Metallic Oxides for Oxidation of C O ............................. 24 III. Relative Activity of LaCo03 and LaQ 9Slo.lCo03 for Reaction of MO with H2 to Products ...................................... 34 IV. Catalytic Studies of CO Oxidation Using Li Doped NiO Catalyst .............................. 100 V. Element Requirements for Catalyst Compounds .. 125 VI. Catalyst Preparation Formulas .................. 126 x CHAPTER I INTRODUCTION The work described in the following pages was moti vated in part by a hypothesis originated by R. J. Brook and J. M. Whelan. They suggested that the simultaneous pres ence of significant ionic and electronic conductivities were conducive to gaseous oxidation reactions catalyzed at the surface of oxides. Further inquiry was indicated by the apparent success of this hypothesis to predict the relative catalytic activities of undoped zirconia, calcia doped zirconia, and lanthia doped thoria for the oxidation of carbon monoxide with an excess of oxygen. The cata lytic activity of these oxides were compared with those of NiO and LaCoO^ by Brook and Whelan with the assistance of W. Bradford. NiO was selected for comparison as an example of a well studied high activity CO oxidation cata lyst and LaCo03 as an oxide with a high electronic con ductivity which could be doped in directions to enhance its ionic conductivity. These preliminary experiments indicated the following order of decreasing catalytic acti vity as indicated by an increasing temperature necessary for 50% conversion of CO to CO2: (1) La Co 0 2, (2) NiO, (3) ^'o.85^*®0.15^1.925' Zr02* and (S) Zr0.94Ca0.06°1.94 * 1 2 The first phase of the present work was to verify the earlier ranking of the above oxides as heterogeneous CO oxidation catalysts by repeating the previous experi ments but with improvements in apparatus design to achieve a greater accuracy in the kinetic data. Results of these will be described in detail later. These results were generally in accord with the earlier observations and pro vided the information necessary for the first of many choices regarding the major thrust of this dissertation research. These choices will be noted for consideration as possible areas for future research. A comparison of the catalytic activity towards CO oxidation revealed ThQ g5^ao 15°! 925 to be superior to Zr0.94Ca0.06°1.94* i°ni - c conductivities of the two catalysts are about equal, yet the Tho.ss^O. 15°1.925 exhibits a significantly greater electronic conductivity. This observation may be interpreted in terms of a model of an internally short circuited fuel cell. The solid electrolyte may be used as a membrane in a galvanic cells CO(g,Pco) C02(g,PCo2) Zr02 + CaO °2^'poJ when operated at high temperature and at different oxygen pressures at the two electrodes in the range of oxygen 3 pressures where electrical conduction is exclusively ionic may be used as a fuel cell. The electrical circuit is completed through an external connection for electronic carrier flow. The half cell reactions occurring in such a cell are: Anode: Vq + 2e* + * 502(g.P(,2) =■ 0* Cathode: CO(g,Pco) + 0Q - C02(g,PCo2) + V” + 2e* Net reaction: C0(g,PC0) + * 502(g,P02) * C02(9*PC02^ In writing these electrode reactions, the symbolism of Kroger and Vink (1) is used to represent the atomic and electronic defects in the electrolyte, i.e., Vq* =• oxygen anion vacancy 0g = oxygen anion at oxygen lattice site e' 3 electron where superscript dots (*) and dashes (') represent posi tive or negative charge, respectively, of the defect rela tive to the lattice site of the host crystal. If the external circuit is removed the CO oxidation reaction may proceed only by transport of oxygen through the electrolyte membrane. Such transport requires the simultaneous diffusion of ionic and electronic defects (oxygen ion vacancies and electrons or holes). If the 4 rates of the processes at the two surfaces do not limit the rate of reaction in the galvanic cell, diffusion through the electrolyte will be the rate controlling process. This behavior is analogous to the process for oxidation of a metal as modeled using Wagner's theory of the tarnishing reaction (1). This analysis yields an expression for the net rate of reaction: rate cC <Ttet^ dp(0)/dx where, <r = electrical conductivity te = transference number, electronic specie t^ = transference number, ionic specie dp(0)/dx ■ gradient in chemical potential of oxygen x = distance across the membrane. This rate can be evaluated if the parameters r , te, t^, and dp(0)/dx are known so that the expression may be inte grated. This model is now extended to consider each elec trode reaction as occurring on the same surface of the electrolyte. The same processes present for the internally short circuited cell are considered to occur on the elec trolyte particle. For simplicity the effects of surface diffusion of adsorbed reactants is omitted from the analy sis; the adsorbed CO is thus assumed to be immobile. 5 Therefore, the reaction rate would be influenced by the magnitude of tet^, xr , and dji(0)/dx on the surface or in a shallow surface layer and within the bulk of the solid electrolyte. Although one does not normally know the value of teti or r in the surface region, those changes in the bulk composition which tend to increase the magnitude of a parameter in the bulk are thought to also increase its val ue in the region of the surface. Modification of the gra dient in chemical potential is thought to proceed through bulk composition changes which affect the fractional cov erages of reactants on the surface at a fixed pressure. Returning to the observations on Tho.SS^O. 15°1.925 and ZrQ 06°1 94 as C0 oxidation catalysts, one notes some agreement with the above model. The former catalyst has a much larger bulk electronic transference number. Assuming other parameters unchanged, the kinetic observa tions are consistent with the qualitative prediction of the above model. Changes in the bulk composition affect the proper ties of the solid near and at its surface, and thus, in turn, the process of adsorption of gases on the surface. General trends may be noted. Those changes in composition which increase a given bulk defect such as a hole or an oxygen vacancy will tend to also increase the surface density of that induced defect. Changes in these 6 densities, in turn, will affect the rate of adsorption and surface coverage of adsorbed reactants and products of heterogeneous catalytic reactions. This might be expected for several reasons: (1) by changing the surface concen tration of selected adsorption sites, and (2) by affecting the surface distribution and penetration depth of the space charge associated with adsorption of species which are either polarized or ionized. In short, changes in the bulk concentrations of both neutral and charged defects will affect the density of native surface states and those states which interact with adsorbed gaseous species. In addition, the mobilities of adsorbed gaseous species and surface defects resulting from the bulk composition are interrelated. From this viewpoint the rate of reaction on the oxide catalyst is considered in terms of the rate at which these changes at the surface proceed. These pro cesses include both equilibration of the surface with the gas phase and the interactions of the surface with the defects present in the bulk solid. One reasonable choice for this present research was to pursue in greater detail, the relative catalytic activ ities of moderately simple oxides such as doped and un doped Zr02, Th02, or Ce02 for which a fairly extensive background exists regarding bulk defect concentrations as functions of composition. Examination of the correspond- 7 ence of these changes to variations of selected kinetic parameters would be a reasonable research direction. An attractive alternative starting point would be the determination of the relative rates of simple oxida tion and reduction reactions in which a solid electrolyte was electrically biased to produce variable oxygen activi ties at the surface. The electrodes used for this cell must be carefully designed. In order that contact between the gas phase and the electrolyte is maintained, comb shaped electrodes might be used. Experimentally such studies would be relatively straightforward provided cata lytic effects of the necessary electrodes (usually metal lic) were circumvented. This might be done by using for contacts metals which are relatively inactive as cata lysts, e.g., Au rather than Pt. Catalytic effects could also be reduced by masking the comb type metallic contacts by inert oxides such as Si02. Using the technology em ployed for semiconductor devices and integrated circuits, it is possible to fabricate the electrodes with dimensions such that the potential drop between the masked finger electrodes is reduced to an acceptable level, so as to maintain sufficiently uniform surface oxygen activities. The attractiveness of this type of attack using the relatively simple oxides still has its original appeal as providing a means for separating the roles of bulk defects 8 in oxides which exhibit significant heterogeneous catalytic activity. However, this activity was set aside in favor of a more exploratory investigation of the rare earth- transition metal perovskites, of which LaCo03 is one of the selected examples. The rare earth-transition metal perovskites have only recently been studied as oxidation catalysts. The superior relative activity of LaCoO^ with respect to NiO observed in the initial experiments suggested that it, or a modification, might be useful as a practical catalyst for automobile emissions control and other air pollution abatement applications. This possibility, coupled with the relatively unknown features of these perovskites as catalysts prompted the direction of this research towards their study. Examples of the rare earth-transition metal perov skites have long been known. Interest in them developed during the 50's and early 60's as potentially useful mag netic materials. It resumed to an extent as a result of the high temperature stability and relatively high 'me trical conductivity exhibited by some of its members iuch as La0>gSr0>jCr03 and LaQ ^SrQ^^ 003. The latter h r r ; conductivity of 500 (ohm-cm) at 200°C. Because e, \hs.*.e characteristics they have been considered as mater' - ■ - for electrodes for fuel cells and magnetohydrodynamir . wer 9 genexators (2). A survey of the chemical litexature performed during the initial period of this research did not reveal any description of LaCo03 or other perovskites as catalysts. Well after our work was in progress, Libby and his co workers (3,4) reported results on hydrocarbon isomeriza tion using LaCo03 and speculated on its use as an oxida tion catalyst. Their suggestions were examined briefly by Bauerle and his group (5) and found LaCo03 to be a superior CO oxidation catalyst. This conclusion has also been stated by Voorhoeve and coworkers (6) in a very recent publication of a brief series of survey experiments. The preparation of perovskite structure catalysts has also been the subject of a patent description by Tseung (7) recently abstracted in Chemical Abstracts. The defect chemistry of the rare earth-transition metal perovskites is more complicated than for the Zr02 and Th02 oxides. The high electrical conductivity of LaCo03 is attributed to a mixed valency of the Co ions. +3 Additions of Sr, which are assumed to occupy La sites as Sr£a defects, cause an increase in the conductivity most likely by increasing the concentration of Co^Q defects (Co+4 ion the equivalent of a hole), the conductivity being dominated by holes. Substitution of a small fraction of the La+3 cations by Sr+2 in LaCrO^ likewise increases the 10 electrical conductivity. Additions of Sr into the lattice at La sites should also favor the formation of oxygen va cancies. The extent to which this occurs in the various rare earth-transition metal perovskites is most likely small in some cases, but not necessarily so in others. These points will be discussed in greater detail in sub sequent sections. The foreknowledge that one of the perovskites might be a useful catalyst combined with the possibility of a wide range of compositional modifications which might be used to satisfy the requirements of a pratical catalyst, made further study of these rare earth-transition metal perovskites very attractive for study. Some of the attri butes for a practical catalyst might include high catalyt ic activity, surface area stability, as well as the neces sary chemical and thermal stabilities in both oxidizing and reducing atmospheres. The flexibility in selecting compositional variables for optimum bulk and surface pro perties includes the use of various rare earths or their admixtures, partial substitution of the rare earth cations with those having a valency other than +3, and the selec tion of transition metal or an admixture of transition metals. Such choices determine properties such as elec trical conductivities, sintering rates, and catalytic activities by fixing the types and concentrations of 11 defects such as ionized acceptors and donors. These de fects may include anion and cation vacancies. The preliminary experiments on CO oxidation using the catalyst set of Brook and Whelan were examined with more controlled experiments. These results are presented in Chapter II. These tests confirm conclusions drawn from the studies of Brook and Whelan. The superior catalytic activity which LaCoO^ demonstrated served to confirm the desirability of exploring the properties of the perovskite catalysts. The performance of the rare earth-transition metal perovskites were more completely examined by survey experi ments which are described in Chapter III. The catalytic activities of LaCo03 and a Sr doped LaCo03 were compared for CO oxidation and NO reduction reactions. Additionally, several reactions involving CO, NO, and H2O were briefly examined. Employing much more controlled experiments (through improvements in design of reactors, reactor furnaces, gas analysis procedures, and control of catalyst preparation and reactant stream composition), the oxidation of CO with O2 is discussed in Chapter IV. These experiments were directed to evaluate effects of partial substitution of La with Ce or Sr in the LaCo03 and LaCr03 catalysts. Also, the effects of water vapor in the reactant stream upon 12 catalytic activity wexe examined. The data on catalyst activities were obtained using a reactant stream of 2.8?Cm CO and 2.89£m O2 for a series of temperatures. At about 200°C the dependence of reaction rate upon both CO and O2 concentrations were determined for the above reactant stream composition. All of the experimental data from this research are summarized and discussed in Chapter V. Consideration was given to formulation of a model to describe the catalytic activities of doped LaCo03 catalysts. The proposed model is critically analyzed and necessary experiments recommend ed to establish the validity of these assumptions. Description of the many items of apparatus and catalyst preparation technique are reserved for the last two chapters. Details on catalyst preparation are pro vided in Chapter VI along with some comments on alternative techniques which may prove to be more efficient preparative techniques. Apparatus descriptions are given in Chapter VII, where design and performance of the laboratory reac tors, reactor furnaces, and the other equipment are dis cussed. CHAPTER II PRELIMINARY EXPERIMENTS Introduction These initial survey experiments were executed to evaluate more carefully the relative catalytic activities of the oxide series examined by Brook and Whelan: Th0.85La0.15°1.925' Zr0.94Ca0.06°1.94' Zr02' (4) NiO, and (5) LaCoO^. As noted in the introduction, the thoria and zirconia based oxides were selected to examine the internally short circuited fuel cell hypothesis. The bulk concentration of defects and mobilities of oxygen ions and electronic charge are known for these compounds. At the lower temperatures the ionic conductivities (anion mobility) of Th o .8 5 La0 .1 5 ° 1 .9 2 5 and Zr0.94Ca0.06°1.94 are are essentially equal and much larger than that of Zr02* whereas the electronic conductivities for modest oxygen pressure (corresponding to experimental conditions during CO oxidation) vary greatly and decrease in the order: (1) Tho.85La0.15°1.925' ^ Zr02' and ^ Zr0.94Ca0.06°1.94* Based upon the internally short cir cuited fuel cell model as discussed in the introduction, one would expect the following ranking of catalytic acti vity for CO oxidation at low temperatures: 13 14 (1) Tho.85La0 . 1 5 ° 1 .9 2 5 ' ^ Zr02* and Zr0 .9 4 Ca0.06°1.94* The activity of NiO was observed to serve as a frame of reference for comparison of present catalysts with other oxide catalysts commonly applied to CO oxidation. Interest in the pervoskite LaCo03 stemmed from the ideas advanced in previous discussions. Experimental The preparation of the various oxides used in the phase of study are described in detail in Chapter VI. These oxide catalysts were prepared for testing as loose powders which pass through a 325 mesh sieve (noted here after as -325 mesh powder). As an aid in the interpreta tion of the kinetic data, the specific surface areas of these powders were determined from the N2 adsorption iso therms using the B.E.T. adsorption theory (8). The speci fic surface areas computed from the B.E.T. equation are as follows: Catalyst Specific Surface, m2/gm LaCoOo NiO 4.2 3.6 Th0 .8 5 La0 .1 5 ° 1 .9 2 5 Zr02 Zr0 . 94Ca0 . 0 6 ° 1 .94 (insufficient sample) 3.2 2.7 Kinetic observations were performed in a flow re actor system; its description is given in Chapter VII. 15 Catalyst charges were controlled to yield equal bulk powder volumes. The mass loadings from this procedure were about 1 gram. A reactant stream of 3.4%m CO and 3.2%m 02 in He was used in these evaluation trials. Reactant stream flow rate was about 95 cm^/min (25°C and 1 atm). Results of Kinetic Studies The results of the catalytic oxidation of CO over these oxides are presented in Figure (1). Relative acti vities of the various catalysts are judged by comparison of the temperatures necessary to achieve 20% conversion of CO to C0-. These data are shown in Table I. The 20% A t conversion was selected so that the reactor operation may be modeled by the differential reaction mode of operation. The net effect of this procedure is that the temperatures are those required for equal reaction rates, since effects of reactant concentration variations on rate of reaction are minimized. In addition thermal effects due to the heat of reaction liberated in the reactor are less severe at low fractional conversions. Inspection of the data in Figure (1) reveals a clear difference in the catalytic activities of Zr02 and ZrQ 94^0 q6°1 94 at lower temperatures. The change in activity rank at about 725°C appears to be the result of a large increase in the temperature dependence of the Percent Conversion CC to CO, Figure 1 -- Survey dxperiitonts on Catalytic Oxidation of Carbon I ion oxide Reactant Stream 3.470m CO tT 3.2fon 02 in 1Ie CM ' 1 o\: :;ato 95 cc/’ min (25 °<j, 1 atm) Full = 1 "ra: 100 80 60 o 40 20 0 0 200 800 Mear Reactor TcM^erature , C O' Table I — Relative Catalytic Activity for CO Oxidation with 02 Reactant Stream Flow Rate 3.4%ra CO in He 95 cc/min 3.2%m 02 (25°C, 1 atm) Temperature for 20% Catalyst Conversion of CO LaCo03 195°C 18 reaction rate over the ZrQ g4Cao 06°1 94 catalyst above 700°C. This behavior was not examined further during these studies. A second comparison between Tho.85LaO.15®1.925 and z*o.94Ca0.06°1.94 indicates a clear superiority in activity of the lanthia doped thoria preparation. The catalytic activity of LaCo03 is very much superior to these previous oxides. Compared to NiO, LaCo03 demonstrates a remarkable activity for the catalytic oxidation of CO with The experimental program used to produce the kinetic data had deficiencies which prohibited the evaluation of absolute reaction rate constants. Primarily, the problems resulted from the reactor design. Mass transport effects in the reactor were not adequately considered, and to a lesser extent an uncertainty existed in the amount of cat alyst present in the reactor because of the loading tech nique. The latter concern proved not to be too serious, as indicated by a comparison of data presented in Figure (1) on LaCoOj with a full and half full loading of the re actor . The 50% reduction in the amount of catalyst led to only about a 20°C shift in the position of the locus of the fractional conversion versus temperature curve. Thus, small variations in the amount of catalyst in the reactor would not change the general conclusions drawn from the data. An equivalent problem is the variation in specific 19 surface areas for the different catalysts. This does not appear to affect the ranking of relative activities be cause the reactor loading experiments covered changes about twice the differences resultant from the known speci fic surface area variations. The mass transport effects may be examined using different reactor designs. The zirconia catalyst tests were repeated using a 0.5 gram charge in a quartz reactor of improved design. As will be shown in a later section of this thesis, the improved quartz reactor design minimized mass transport limitations on overall reaction rate, and thus clarifies the inter pretation of the experimental kinetic data. The results of these comparative tests are presented in Figure (2). Changes in the fractional conversions observed between the two reactor designs are about a factor of two, equal to that anticipated because of the catalyst load differences between the two tests. From these comparative experi ments, the conclusion is drawn that the mass transport processed in the survey studies do not significantly af fect the accuracy of the data on relative catalytic acti vities. Electrical Conductivity Data Data on the electrical conductivity of the oxides studied in the catalyst survey was compiled from sources lercent Conversion CO to CO 20 Figure 2 -- Catalytic C::idacion of Carbon i.ono::ide, Jffect of Reactor Design Reactor Design Catalyst Load Open symbol, survey 1 gm. solid symbol, trimary 0.5 gm. 100 ZrO, ^ 80 Zr 0.94 0.05 1.94 60 A0 20 - 400 6 00 700 800 900 500 Reactor Temperature, *C 21 in the literature (9-15). These conductivity data are presented in Figure (3) as a function of temperature. Note the large difference in observed conductivity of NiO when tested as a single crystal and a polycrystalline specimen. Such a behavior suggests some caution when assessing differences in conductivity of different speci mens of less than an order of magnitude as necessarily being significant. Careful interpretation of experimental procedures and results is required to assess smaller dif ferences in conductivity as being valid. The electrical conductivity of the thoria-lanthia mixtures could not be found in the literature; thus, thoria-yittria mixtures were inspected and assumed to behave similarly to the thoria-lanthia mixture. As noted in Figure (3) only the doped zirconia and thoria are ionic conductors. The other materials exhibit conductivity due to hole transport. Over _2 the range of 10 to 1 atm oxygen pressure, Wimmer (16) reports the thoria mixture to have an electronic transfer ence number, te, of about Q..3 where electronic conductivity is due to hole transport. Calcia stabilized zirconia ex hibits a negligible hole conductivity (te ^ 10-3) at the above oxygen pressures and the temperatures noted in Fig ure (3). Hole conductivity of Zr02 is greater than that observed present in calcia stabilized zirconia. Electrical Conductivity, l/(ohm-cm) 22 Figure 3 — Electrical Conductivity of Catalysts of the Survey Series. (All data on polycrystalline, sanples except t There noted.) * 4 " A LaCoOo (9,10) +2 10 NiO (11) single crystal ih0.85^0.15C1.925 (14^ 10 ■4 ( 13) 6 10 NiO (12) 10 -10 10 0 200 400 600 800 120 1000 Temperature, *C 23 Conclusions A comparison of the data on catalytic activity from Figure (1) and electronic conductivity data from Fig ure (3) reveals a positive correlation between the two properties. Offered as only a qualitative rule, a more active catalyst is one which exhibits a higher electronic conductivity. Although all the oxide catalysts considered here are bulk p-type semiconductors, the presence of an n-type conductivity cannot be judged a detriment to cata lytic activity. This shall be demonstrated in some later discussions of perovskite catalysts. The survey experiments described in present discus sions have confirmed the preliminary results of Brook and Whelan on the relative catalytic activities of doped zir conia and thoria and undoped zirconia. These data are consistent with the qualitative prediction from the short circuited fuel cell model. For these oxides improvement in catalytic activity results when the magnitude of tfit^ and t terms are increased. The rare earth-transition metal perovskite, LaCoC^, proves to be a very active CO oxidation catalyst. An ap preciation of this activity is gained by comparison with NiO and other metallic oxides commonly applied to CO oxidation. Presented in Table II is a compilation by Krylov (17) of the catalytic activities of many oxides 24 Table II — Relative Catalytic Activity of Metallic Oxides for Oxidation of CO Catalyst Log^o k Catalyst k ai203 -0.04 Cu2° 1.71 Ce02 0.85 Fe2°3 1.22 CoO 2.42 NiO 2.00 c°2o3 1.58 Th02 0.43 CojO^ 2.32 Si02 -0.43 Cr 203 1.02 ZnO 1.33 CuO 2.08 ZrOn 1.04 Reaction rate data in units of g-raole CO/sec-cm2 catalyst surface determined at 150°C and expressed relative to activity of NiO. Value of k for NiO assigned a value of 2.00. 25 towards CO oxidation. These comparisons are presented us Log10 (reaction rate) at 150°C expressed relative to NiO which is assigned a value of Log^Q (reaction rate) * 2.00. These reaction rates are compared using units of g-mole C0/sec-cm2 of catalyst surface to express the rate of re action. One may quickly observe that few simple oxides are superior to NiO in catalytic activity, and these are so by less than a factor of about 3. Clearly, LaCo03 is a catalyst whose activity is competitive and probably superior to other common binary oxides. Because of the promise which LaCo03 appears to offer as an oxidation catalyst, it is very desirable to examine this perovskite for catalysis of several other simple re actions. In addition, modification of the composition by substitution of a fraction of the La with Sr is considered to test the ideas advanced in the introduction on property modification of the rare earth-transition metal perov- skites. These tasks form the core of work examined in the following section of the thesis. CHAPTER III EXTENDED SURVEY OF PEROVSKITE CATALYSTS Introduction The preceding experiments have demonstrated LaCo03 to have a high activity for the catalytic oxidation of car bon monoxide. This prompted an interest in a further sur vey of its activity and that of the Sr doped modification, La0>9Sr0#1C0O3, towards several oxidation-reduction reac tions. LaQgSrQ'1C0O3 differs from the undoped compound by having a much higher electronic conductivity at low temperatures. This is attributed to the higher hole con centration associated with compensation of the effective charge of Sr^a by Co£0 defects which represents mobile holes, and to a much lesser extent to formation of anion vacancies. The reactions investigated using LaCoC^ and LaQ gSr0 1C0O3 as catalyst were: 1) CO + ho2 - co2 2) NO + H2 = * 5N2 + H20. Additional observations were made on the reactions: 1) CO + H20 = co2 + h2 2) CO + NO + H20 » Products 3) 2C0 = C + C02 26 27 as catalyzed by LaCoO^. Experimental The preparation of LaCo03 and La0 gSrQ.1C0O3 was performed by the nitrate solution decomposition technique as described in Chapter VI. Again the catalysts were tested as -325 mesh powders in the reactor system employed in the preliminary survey experiments. A gas chromato graph became available for this work and it was used for partial or complete analyses of the gas streams. Analysis for all reactants and products except NH3, N02, and H20 could be performed quantitatively. Reaction progress was evaluated from reactant and product stream analyses using prior chromatograph calibrations to evaluate component concentrations. For the studies of NO reactions the yield of NH3 as a product was obtained by a material balance. Reactant NO and the product concentrations of N2, N20, and NO were determined by chromatographic analyses. H20 and N02 were removed from the gas analysis samples using a cold trap cooled by an acetone/dry ice thermostat. Carbon Monoxide Oxidation The two catalysts were tested using a reactant stream containing 3%m CO and 3?£m 02 in He. Approximately 1 gram of the -325 mesh powder catalyst was used in the reactor which processed 100 cc/min (25°C and 1 atm) of 28 reactant stream. Results of this study are presented as the fractional conversions of CO as functions of tempera ture in Figure (4). Also examined in this series of CO oxidations were the effects of 2%%m H20 vapor in the re actant stream upon the catalytic activities of the two catalysts. Comparison of the data for CO oxidation in dry reactant streams indicates LaCo03 to be superior to LaQ gSrQ ^CoOj. Some uncertainty in this conclusion due to the differences in surface areas of the catalysts was resolved by inspecting the sensitivity of fractional con versions to reactor loadings of La0>gSrQ ^Co03. A compar ative test was performed using reactor loadings of 1 gram and then h gram of -325 mesh powder. As seen by inspec tion of Figure (4), reduction of the charge of LaQ gSrQ ^CoO^ resulted in a 20°C increase in required temperature for achievement of 20% conversion of CO. The ratio of specific surface areas of LaCoO^ to LaQ gSr0 iCo03 is about 3*5 to 1. Therefore, La0 gSrQ ^CoO^ catalyst sample activity need be increased by a factor of 3.5 in order to judge relative catalytic activities of LaCo03 and LaQ^SrQ^Cot^ on the basis of equal specific surface areas. From the above data on per formance of LaQ 9SrQ 2C0O3, it is apparent that the frac tional conversion curve for a 1 gram reactor charge should be shifted about 20°C towards lower temperatures. Percent Conversion CO to CO Figure 4 -- Catalytic Oxidation of Carbon Ronoxide, LaCoCg and LaQ ^Sr^ ^CoO^ Ferovskites. ■eact'nt Ctrean 37.: 1 CO a - 7. n r- e Flow nto lO'O cc/r.ln (25 °C, 1 at") i.pncrcr i - ■ • - . O ?! CM Symbol O □ 60 40 20 0 0 100 200 300 400 500 Temperature, C LaCoG^, Dry Reactant Stream LaCoO^, 2%7»m ^ C in Reactant Stream ka0.9Sr0.1Co03* Dry Reactant Stream Lan c>Srn .CoO-, Dry Reactant Stream, u.* u.i j 25Q ^ Reactor Load 30 Inspecting Figure (4) again one notes that LaCo03 is supe rior by a margin of 10 to 15°C at about 20% conversion* which sustains the conclusion of its superior catalytic activity with respect to LaQ gSrQ ]Co03 for CO oxidation. The decreased catalytic activity of LaCo03 in a re actant stream containing 2*s%m H20 is significant and im portant. As seen from the data shown in Figure (4), over the range of temperatures explored, CO conversions in the reactant stream containing 2^%m H20 were less than 1/2 those in the nominally dry reactant stream. This fact was useful in planning the primary kinetic experiments, where upon a more quantitative evaluation of the effect which water has upon the catalyst activity was performed. It should be noted that practical applications of a CO oxida tion catalyst will involve exposure to H20. Nitric Oxide Reduction The evaluation studies for NO reduction were per formed using a reactant stream of 1.8 %m NO, and 6%m H2 in He passed through the reactor at a rate of 100 cc/min (25°C and 1 atm). Catalysts are the same lots used in the CO oxidation experiments. A pretreatment condition of 12 hours at 500°C in 6%m H2 was used to assure catalysts were in a reduced condition. The results for LaCo02 are pre sented in Figure (5), and the results for LaQ gSrQ j^oO^ are found in Figure (6). The relative activity of these Percent Conversion 1 1 0 to Products Figure 5 — Catalytic Reduction of Nitric Oxide with Hydrogen over LaCoC^ Catalyst. 100 NO 80 60 40 20 100 200 300 Temperature, C 400 500 600 Reactant Stream 1.8%n KO . „ 6.0%n I in He Flow Rate 100 cc/min (25 "C, 1 atm) Reactor Load 1 gm. of -325 mesh powder Percent Corrversion 1 5 0 to Products Figure 6 — Catalytic Reduction of Nitric Oxide with Hydrogen over Lao.9Sro.lCo03 Catalyst• 100 NO 60 20 100 200 300 Temperature,*C 400 500 600 Reactant Stream 1.87,m NO . . . 6 .07o m H0 in r ‘ e Flow Rate lOO cc/min (25 *C, 1 atm) Reactor Load 1 gn of -325 mesh powder 33 catalysts is judged from an inspection of product distri butions as functions of temperature. Mote that the speci fic surface area ratio of LaQ gSr0 ^CoOj to LaCoO^ is about 1 to 3.5. The MO reduction product distributions are strongly temperature dependent as are evident in Figures (5) and (6). As an aid in making a comparison of the catalysts, several statistics are tabulated in Table III. The superiority of Lao.9Sro.lCo03 *s clearly evident from these comparisons, since all judgments relating the promo tion of NO reduction reactions are in favor of this cata lyst. At higher temperatures the differences in M2 and NH3 production over the two catalysts were less evident. One must bear in mind the surface area advantage of the LaCo03 catalyst has of 3.5 to 1 over that of La0.9Sr0.1Co03- The reactions of NO with H2 are significant ones with respect to NO emissions control from internal combus tion engines. Some H2 would be present in the exhaust gases containing CO, C02, H20, at equilibrium, and smaller amounts of NO and hydrocarbons. One approach is to first reduce the NO in a reducing exhaust mixture containing CO, C02, HjO, and H2 followed by an oxidation step to complete the combustion of CO and hydrocarbons to COj. Catalysts which reduce NO to NH3 are unsatisfactory as the NH3 is then reoxidized to NO in the second stage used to remove 34 Table III — Relative Activity of LaCoO^ and LaQ gSr0 ^CoO-j for Reaction of NO with H2 to Products Required Temperature to Satisfy Condition Required Condition LaCo03 La0.9Sr0.lc°03 10% Conversion of NO 250°C 175°C Maximum N20 concentration 300 210 Removal of all N20 360 310 Maximum NH^ concentration 350 350 50% of NO to N2 340 280 90% of NO to N2 (est) 500-575 490 35 the CO. Estimates for the maximum temperatures which might be used to remove 90% of the NO are about 500°C. Attempts to study the reactions of NO with CO in He streams containing H20 were unsuccessful because of the difficulties in making the required gas analyses of the reactant and product streams. Resolution of the CO and NO chromatographic peaks was incomplete with a 6 ft length column of Molecular Sieve 5A. Good resolution of H2. CO, C02, N2, N20, and NO was achieved with a 12 ft Porapak Q column at -75°C. Unfortunately, the low temperature con trol for the chromatography columns was prohibitively ex pensive using the C02 expansion cooler of the gas chroma tograph and impractical with acetone/dry ice thermostats because of the lengthy operating periods needed to study the catalysts. Hydrogren Treatment of Catalysts The pretreatment conditioning of the catalysts for study of NO reduction experiments was exposure to 6%m H2 in He for 12 hours at 500°C. This process resulted in transformation of the catalysts into more reduced states, as evidenced by the formation of water in sufficient a- mounts to condense in the exit lines of the reactor. The bulk of the H20 formed in the first few hours of the con ditioning treatment. A qualitative estimate of the amounts 36 of water formed when LaCoO^ and LaQ gSr0 ^CoOj were treated with H2 indicates the latter compound yields less water. Based on these observations the stoichiometry of LaQ gSrQ ^CoOq is less sensitive to the oxygen partial pressure. This indicates that the model of Sr incorpora tion into the crystal at a La site with at least partial charge compensation by formation of oxygen vacancies does have some validity. Water Shift Reaction Over LaCo03 The continuing examination of this catalyst has included observations on the reaction, CO + h2o - co2 + h2 A reactant stream of 2.1%ra CO and 2)£%m H20 in He flowing at 100 cc/min (25°C and 1 atm) was used to test the proper ties of LaCo03 as a catalyst. Approximately 1 gram of -325 mesh powder was charged into the reactor. Pretreat ment conditioning of the catalyst was exposure to the re actant stream for 72 hours at 600°C. The results of this experiment are presented in Figure (7). This reaction is not promoted as well as NO reduction by H2 since 20% con version of CO is not achieved until the temperature ap proaches 350°C. The comparison of the activity of LaCoO^ with other common catalysts is not easily made from this data, but it would appear that LaCoO^ is not outstanding. Percent Conversion of CO Figure 7 — Water Shift Reaction Catalysis using LaCoO^ Perovskite. 1 0 0 CO + 80 Equilibrium CO 60 40 20 400 600 700 200 300 500 Temperature, "C Reactant Stream 2*l%m CO 2»5%m H 2O in He 1.8%m ITO Flou_kate 100 cc7nin (25 *C, 1 atm) Reactor Load 1 gm» of -325 mesh powder 800 Ul 38 An additional examination of the water shift reac tion was performed by adding NO to the reactant stream at a concentration of 1.8%m. The results of this work are shown in Figure (7). Although consumption of CO is in creased, this process appears to remain closely associated to the water shift reaction as indicated by the decreasing conversion of CO at higher temperatures. This suggests that the CO reaction with NO is not particularly fast in comparison to the other reactions. The water shift reac tion produces H2 which, in turn, reacts with NO. This con sumption of H2 then allows a greater conversion of CO via the water shift reaction. Thus, it appears that the order for decreasing reaction rates are: (1) CO - » • H20, (2) NO + H2, and (3) CO + NO, for the conditions represented in this experiment. Carbon Monoxide Decomposition Over LaCo03 During trials of preconditioning LaCo03 with CO in He, the appearance of CO2 in the product stream led to an examination for the occurrence of the catalytic CO dispro- portionation reaction, 2C0 * C + C02 Observations on the kinetics of the reaction were obtained after a conditioning treatment of the LaCo03 in H2 in He for 12 to 18 hours at 600°C. These data are shown as 39 fractional conversions of CO versus temperature in Figure (8). A proper question to raise is whether the observed consumption of CO is due to reduction of the catalyst by CO to generate C02, rather than the disproportionation re action. This was examined by testing the material balance on carbon. From experience on CO oxidation experiments, analyses of product and reactant streams yield a precision of ±1.5% on the total carbon. After a pretreatment period with 6% H2 in He, the CO decomposition test was performed using LaCo03. For the reactor conditions used to carry out the material balance check, the product stream con tained about 5% of the input CO as C02, but a total CO con version of 13% based upon CO observed in the product stream. There is no doubt that carbon is being produced and remains in the reactor. After completion of the CO decomposition reaction study, the reactor is purged with He. Product stream is then diverted to bubble through a Ba(OH)2 solution. When 02 is added to the reactant stream the Ba(OH)2 solution quickly clouds with a white precipi tate. This confirms the generation of carbon in the re actor . The kinetic observations on CO decomposition were separated into two trials. After the experiments of trial #1, the catalyst was treated with a 3%m 02 in He mixture Percent Conversion of CO Figure 8 -- Decomposition of Carbon Monoxide over LaCoO^ Catalyst* Reactant Stream 2%ra CO in He Flow Rate 100 cc/min (25'C, 1 atm) 1 gm of •325 mesh powder 100 80 60 40 20 ' 68 Hours of Operation 200 300 400 Temperature, ”C 500 100 600 700 800 41 at 500°C to burn off deposited carbon. The catalyst was then treated at 625°C with 6%n H2 for 18 hours, and obser vations on catalytic activity for CO disproportionation continued. Since solid carbon is a reaction product, an imme diate interest is raised on possible deactivation of the catalyst by coverage of the surface with carbon. This possibility was examined by analysis of data from the test performed with a 68 hour exposure of LaCo03 to a 2%ol CO in He reactant stream at 625°C. At the reactor test condi tions approximately 5% conversion of CO is observed. From the known CO flow rate, one may deduce that this involves formation of about 0.8 mg-mole of carbon. A conservative estimate of 1 ft2 surface area coverage per carbon atom, combined with the measured 4.2 m^/gm specific surface area of LaCo03 permit computation of a surface coverage by car bon. This calculation indicates sufficient carbon was produced to form about monolayers. Yet the catalytic activity after 68 hours of operation agrees well with the kinetic data obtained when the LaCo03 was reduced in H2 and exposed to the reaction conditions for less than 8 hours. The conversion versus temperature data indicate that appreciable fractions of CO are decomposed between 230 and 300°C. These temperatures are quite low in 42 comparison to those required for the heterogeneous decom position of CO with Fe203 (17). For Fe203 the maximum rake of decomposition occurred at 500°C for a pure CO re actant stream at 1 atm. Negligible reaction was noted below 400°C and above 800°C. Near 500°C, oxides of Fe, Co, and Ni caused deposition of carbon. The maximum rate occurs at about 700°C for CoO and NiO. The oxides of Mg, Ca, and Ba are low activity catalysts, while the oxides of Cu, Ag, Zn, Al, Ti, Si, V, Cr, Mo, W, U, and Mn appear inert towards catalysts of CO decomposition. C and SiC are also reported to be inert. Of the free metals only Fe, Co, and Ni are strongly active, while Mg, Al, Ce alloys, Ti, Cr, and Mn are less active. Inert metals in clude Ag, Cu, Zn, Si, Mo, W, Pd, and Pt. Conclusions These examinations of CO oxidation and NO reduction demonstrate clearly the ability to modify the catalytic activity of LaCo03 by substitution of La with Sr. The major point of interest is the observation that a change in catalyst composition which favors the NO reduction re action has a converse effect on the CO oxidation reaction. This conclusion was based only upon a few observations, but the hypothesis is advanced that substitution of +2 valence cations for La may decrease activity towards catalysis of oxidation reactions, and that substitution 43 of +4 valence cations for La may aid in catalysis of oxi dation reactions. The converse behavior is speculated for reduction reaction applications. The catalytic activity of LaCo03 for CO oxidation with 02 is decreased when the catalyst is doped with Sr. The addition of Sr is known to increase the electrical conductivity of the oxide, and is thought to yield forma tion of some oxygen ion vacancies. One would anticipate a Sr doped LaCo03 to be a more active catalyst, if the short circuited fuel cell model were applicable to this catalyst. The experimental data indicate this model is not satis factory for interpretation of the LaCo03 based catalyst system. Additional catalytic studies of LaCo03 have suggest ed the ranking of the activity for several reactions in volving CO, H20, and NO. The mechanism of NO reduction in the presence of CO and H20 appears to be via reaction with H2 generated by the water shift reaction. Finally, examination of CO decomposition over LaCo03 proved quite interesting. Over the duration of the testing, a deactivation via deposited carbon is not appar ent which raises the interest in the form and distribution of the carbon on the surface of the catalyst. The cata lytic activity of LaCo03 for this reaction appears excep tional in view of the activity possessed by many metals and their simple oxides. CHAPTER IV CATALYTIC ACTIVITY OP PEROVSKITE STRUCTURE OXIDES Introduction In the earlier sections, the catalytic activities of LaCo03 and its Sr doped modification were explored to provide relative activities for several reactions relevant to controlling emissions from internal combustion engines. The results contained in this section are those of more extensive studies of the rare earth-transition metal perovskites as CO oxidation catalysts. The goal was to provide absolute reaction rate data for selected catalysts. LaCo03 and its Sr and Ce doped modification were the perov skites most extensively studied. Data are also included for LaNi03> LaCr03 and its Ce doped modification. Objec tives of these studies were twofold; first, to provide quantitative rate data, and secondy, to indicate a trend in the properties of the perovskites which might be useful in designing optimum catalysts for various applications. The catalytic oxidation of carbon monoxide with oxygen was used to explore the catalytic properties of the perovskite oxides. It was selected because of its common usage as a test reaction, and the relative simplicity of the required gas analyses. The reactant stream test con dition of 2.8%m CO and 2.8?£m 02 was selected for use in 45 these experiments to provide a continual excess of 02 in the reactor. This condition also approximately corres ponds to the reactant concentrations of CO and 02 observed in automobile exhausts to which these catalysts might be applied for purposes of air pollution abatement. Helium was chosen as the carrier gas because of its lack of inter ference in the gas chromatographic analyses. Experimental Improvements in the experimental procedures and apparatus were introduced to enable more quantitative con clusions to be drawn from the data on reaction kinetics. The modifications in the laboratory reactor, reactor fur nace, reactant stream preparation technique, and the gas analyses procedures are reported in detail in Chapter VII. The improvements in reactor design were analyzed and shown to provide adequate preheating of the reactant gases to the reaction temperature prior to entry of gases into the cata lyst bed. Temperature gradients within the catalyst bed which originate from the heat of reaction were shown to be small under conditions of the experimental procedure. A heterogeneous catalytic reaction using porous catalyst particles involves a series of transport and re action processes for completion of the overall reaction process. Reactant gases must diffuse from the bulk gas to the surface of the catalyst particle. Then, since porous 46 catalyst particles generally have most of their surface area in the internal pores and crevices, reactants must diffuse into the interior of the particle in order to reach the available surface. Once the reactants reach the surface, the processes of adsorption surface reaction, and desoprtion occur. Products then diffuse out of the porous particle and into the bulk gas. Several analyses were performed to examine the rates of these diffusional mass transport processes in relation to the expected rates of the surface reaction. Reactant stream composition, stream flow rate, reactor temperature, and a catalyst particle diameter of 100 jm are conditions typical of the experi ments performed in this study and were considered in the analyses. The results clearly demonstrated that neither bulk diffusion nor intraparticle diffusion would be the rate limiting processes in the overall reaction process. Therefore, the results of present experiments describe overall reaction rate is governed by the surface catalytic reaction processes and not by mass transport processes. The laboratory reactor was constructed so as to conform to operational characteristics of an ideal plug flow isothermal reactor. This type of reactor is charac terized by the flow of fluid through the reactor with no difference in the longitudinal velocities between any fluid elements. Lateral mixing of the flowing fluid is 47 permitted for this model. In a plug flow reactor the composition of fluid varies with position along the flow path. A material balance for reaction components is considered within the differential volume, dV. For reactant A, the material balance is, Referring to Figure (9) for identification of terms, one may note that the material balance is formulated for the volume element, dV as: Input = Output + Disappearance by reaction (1) Input = Fa , g-moles A/sec Output = Fa + dFA , g-moles A/sec Disappearance by reaction = volume of element, dV 3 Introducing these terms into Equation (1), PA = <rA + dPA) + <-rA)dv and noting that, From these expressions, one obtains the basic design equa tion, f'Ao'^A - (-rA>dV <2> 48 Figure 9 -- Variables for Plug Flow Reactor Design K- dV ’ Ao Ao Ao dX 0 “ Af ?Af Af Distance Through Reactor v dV dX. Concentration of reactant A, gnole/cm Molar flow rate of reactant A, gmole/sec Fraction of reactant A converted 3 Volumetric flow rate, cm /sec Differential volume element in reactor Differential conversion in volume element dV 49 where, (-rA) * reaction rate, g-mole A/sec-cm2 dS - increment of catalyst surface area, cm2. A variety of experimental techniques are available for the study of heterogeneous reaction kinetics. The simplest procedure is to operate the reactor in a differ ential reaction mode. Such a technique considers the re action rate to be constant at some average value through out the reactor. Therefore, the concentration dependence of the reaction rate requires operation of the reactor at low levels of conversion in order to satisfy the assump tion of a constant reaction rate. Integration of the plug flow design equation is trivial for this case. The result is, ^“rA^ aver FAo^xA.out ~ xA.in^ S ( 4) where, = mean reaction rate, g-mole A/sec-cm2 - molar feed rate of component A, g-mole A/sec xA,out - fraction of component A converted to products present in product stream XA,in = fraction of component A converted to products present in reactant stream S = surface area of catalyst in reactor, cm2 Note that under conditions of Xa,in * t*ie accuracy 50 evaluation of the (Xft Qufc - XA ^n) term of Equation (4) is less affected by the analyses errors of the gas composi tions. Thus, all kinetic data were generated by performing the studies with a CO and 02 reactant stream. The XA>out term was evaluated by analysis for C02 and computation of the equivalent CO conversion. The results of examination of the LaCo03 series of catalysts demonstrate an approximate reaction rate depend ence upon the reacatn concentration of the form, <-**>-<‘V * where, PA = reactant A partial pressure for both the CO and 02. By differentiation of the above expression, the relationship below may be derived, A (-rA> 1 /A PA\ * 2 Therefore, for a +20% variation in relative reactant con centrations, a +10% variation in reactant rate may be ex pected. The present experiments were performed using a 100% excess of 02; thus, for an observed 20% conversion of CO to C02 reactant concentrations vary 10% for CO and 5% for 02 from the mean concentration. This would lead to a variation of the reaction rate of about +7%% about the mean value. Since the bulk of the kinetic data were 51 obtained for CO conversions less than 10%, no serious error is introduced by treating the data as being those for an ideal differential reactor. Comparative Catalytic Activity Another assessment of the catalytic activity of the perovskite compounds relative to NiO was performed to eval uate the catalytic activity of LaCo03 and LaNi03. The preparation of these compounds are described in Chapter VI. They were used as -200 mesh powders with the following specific surface areas: NiO ■ 3.6 m^/gm, LaCoC>3 = 1.2 m^/gm, and LaNiO = 7.0 m^/gm. The comparative study was done using a 3%m Co, 3%m 02, and a trace of H20 in a re actant stream of He carrier gas. The results for the CO oxidations are presented in Figure (10) as reaction rates in g-mole CO/sec-cm^ as functions of temperature. This type of data allows a proper basis for a comparison of catalytic activities. A caution on the strict interpreta tion of the data is evident from the NiO data. At about 240°C, a 50% reduction in activity was noted after opera tion over a period of 18 hours. Later work suggests this behavior was due to deactivation of the catalyst by the trace amount of water present in the reactant stream. For the reactant stream containing this amount of H20 (esti mated to be less than 0.1%m), one concludes a superiority of LaCoO^ over NiO by about a factor of 10, and a Figure 10 — Comparison of Catalytic Activity for Carbon iionoxide oxidation 10 LaNiO Initial m o 10 NiO React ant Stream. 3&a CO . tI 3%m 0o in He Reactor Load -200 mesh pov/der 200 160 220 240 Temperature, *C 180 280 300 53 superiority of LaNiC>3 over NiO by a factor of about 5. Since LaCo03 exhibited a somewhat higher catalytic activi ty, and has a greater stability at temperatures above 1000°C, further investigation of LaNi03, and its modifica tions with Ce or Sr doping, was not pursued in order to lessen the required amount of experimental work during this research. The precision of the reaction rate data noted from tests using different sample sizes of LaNi03 with replica tion of tests was about +30% below 200°C. These experi ments were performed by procedures thought appropriate to yield steady state operation for each test condition ex amined. Thus, it was difficult to explain the poor pre cision of these data solely upon a transient condition of the catalyst resultant from a continued interaction with the reactant stream. Microscopic inspection of the cata lyst particles did reveal a wide range of particle sizes. It was concluded that metering of reactor charges by weighing did not adequately meter total catalyst surface areas because of variations in the particle size distribu tions between samples. To minimize this difficulty, sub sequent comparative examinations of the LaCo(>3 and LaCr(>3 catalyst series were based upon powders with -200 to +325 mesh and -120 to +200 mesh size ranges, respectively. The preceding results on deactivation of NiO 54 emphasized the need for establishing of an initial treat ment of the catalyst to insure a reproducible starting reference state. This reference state should be one in which the surface and bulk electronic and atomic defects are in equilibrium with the solid and gas phases. Re arrangement of those atomic defects necessarily involves the least rapid diffusional mass transport; therefore, the rate of equilibration is aided by increased tempera ture. Since most of the present series of experiments were performed in a Pyrex glass reactor, a maximum of 400°C was arbitrarily established for long term operations. The selected treatment process was exposure of catalyst in the reactor to the reactant stream at 400°C for a se lected period of time. With such considerations on the significance of a pretreatment process the examination of catalyst properties is continued. The effects of water vapor on LaCo03 were assessed by experiments as follows. CO oxidation rate data on a sample immediately after an activation period of 12 hours in a reactant stream of CO and 02 with a trace of H20, and then after 14 hours in a reactant stream also containing 2%Xm H20 are given in Figure (11). Over the temperature range of 200 to 240°C, catalytic activity was reduced by a factor of 3 when water was present in large concentra tions. A similar set of observations were executed using 55 Figure 11 — Carbon iionoxide Oxidation over LaCoOg, Effect of Jater on Activity* Reactant Stream Flow P'.ate 125 cc/nin (25 *C, 1 atm) Symbal o Reactor Load 505 mg of -200 mesh powder, 1*2 m /gm P40 20 280 220 240 260 200 180 Temperature, *C Activated form, trace of 1^0 in reactant stream* Equilibrated in 2'2%n Hyp concentration in reactant stream* 56 La^ gSrQ ^CoO^* These results are summarized in Figure (12) and clearly show the effect of the trace amounts of H20 in the reactant stream. After a period of activation and equilibration for 12 hours at 200°C in the reactant stream, oatalytic activity was reduced by a factor of 1.5. Again, a deactivation of a factor of 3 was noted for the 2%% H20 experimental condition. The catalytic activity present at an equilibrated condition in the trace H20 reactant stream is reproducible. The observations for this experiment were obtained during the following sequence of treatments. First, activate catalyst at 400°C in reactant stream for 12 hours, equili brate at 223°C and observe the catalytic activity. The catalyst is again activated and equilibrated at 221°C and the activity noted. Then, without activation of catalyst equilibrate in the reactant stream at 191°C, then 200°C, and lastly at 212°C, observing the catalytic activity at each temperature. As may be observed by inspection of this data in Figure (12), there appears to be no evidence of a hysteresis. Thus, the conclusion that differences in catalytic activity are induced only by the presence of the water at each temperature appears to be well support ed. These observations indicated the desirability to more completely remove the water from the reactant stream. Figure 12 ■ - Carbon lion oxide Oxidation over Lag 9^0 1C0O3, Effect of Water on Activity. ) V. actant Stream 3. j 1 co _ i , y . 1 xn rie Elowllate 125 cc/nin (25 'C, 1 atm) Symbol O □ lionet or T,c~ / 494 r.g of -2 nosh porder , O.G 1 . 1"/ 40 o CJ 0 30 n o •H w 3 > : 20 o cj 280 260 240 220 200 180 Temperature, ’ C Activated 4Hr/400C in reactant stream, data obtained rapidly after activation. Activity after 12Kr/200 C in reactant stream, trace of H2O present. Activity after 12Kr/220 C in reactant stream with 2%?«m 1^0 present. 58 This task was accomplished by placement of a liquid nitro gen cooled cold trap in the mixed He-C0-02 reactant gas stream upstream of the reactor. The reactant gases which are so dried shall hereafter be referred to as bone dry. Effects of Ce and Sr Additions to LaCo03 An appreciation of how the catalytic activity of LaCo03 would be modified by partial substitution of +2 and +4 valence cations for the +3 valence La cation was a pri mary objective of this research. These modifications were achieved by partial substitution of Sr and Ce for La. As will be shown later, the replacement of 10% of the La by Ce made p-type LaCoOj n-type, and a corresponding 10% addi tion of Sr made the material more p-type and increased its electrical conductivity. The reactant stream containing 2.75%m CO and 2.75%m 02 in He was dried by passage through a liquid nitrogen cooled cold trap. The catalysts were in the form of -200 to +325 mesh powders prepared by crushing tablets sintered in air for 24 hours at 1250°C. A pretreatment condition of 12 hours at 400°C in the reactant stream was used to es tablish the reference condition prior to observation of the catalytic activities. Activities were determined using two separate reactor charges each for LaCo03 and LaQ 9Sr0>1CoO3, and four separate loadings for 59 LSq qC6q ^CoO^. The results on LaQ gCe0 ^CoO^ were based on observations from both the 20 mm and 10 mm diameter Pyrex reactors described in Chapter VII. The kinetic data for CO oxidation with this series of catalysts are pre sented in Figure (13). As noted earlier, the preferred form for expressing catalytic activity is in terms of a rate constant based on the unit area of the catalyst. Reproducing and deter mining amounts of catalyst with specific surface areas has been a major difficulty in this program. First, the facilities available for synthesizing the catalyst samples were inadequate for preparing more than several grams of the reactively sintered materials. Grinding facilities were more limited and separation of the ground particles was limited to screening. Using the preparation/grinding facilities available it was possible to estimate surface areas of samples ground to -200 mesh sizes using a conven tional B.E.T. adsorption apparatus which was constructed for this purpose and described in Chapter VII. A sample with a total surface area of about 20 m2 was required. As was shown, simple weighing of the -200 mesh powders proved to be an inadequate means of metering amounts of the powders with reproducible surface areas. This was due to the wide ranges of particle sizes in the -200 mesh samples. To reduce variability due to fines, so that 60 Figure 13 -- Catalytic Or.idation .of Carbon Monoxide, LaCoO^ Series of Perovskites. ■ C . c ' ;.rv\to Reactor Load 'I. / j n - . a . 1( 0 cc/:iin -200 to +325 2 . 7 5 . in Ke (25 C, l atn) ' nosh pouder tor. a La CoO LaCcO La CoO 140 220 160 200 240 180 Temperature, *C 61 comparative experiments could be made, it was necessary to select samples with mesh sizes between >200 and +325 mesh. This greatly limited the amounts of available use ful material which was initially in short supply; so much so that the B.E.T. surface area measurements of the -200 to +325 mesh powders could not be performed. The subse quent reproducibility of the kinetic data was good when based on these powder samples. This was at the price of estimating the surface areas more indirectly than desired. From analysis of certain kinetic data, inspection of catalyst particles with a scanning electron microscope, and determination of particle size distributions as de scribed at the end of this chapter, the conclusion was drawn that all catalysts in the LaCoC^ series could be treated as having a surface area of about 0.1 m2/gm. The variation from this value for each material could not be concluded from available data; thus, the present assump tion is made of an equivalence of specific surface area of each catalyst. Support for this is given at the end of this chapter. The frustration with the inability to measure sur- face areas of about 0.1 m /gm for about 1 gram samples was in fact, an incentive for the recent development of a modified B.E.T. apparatus having this capability. It was jointly conceived with Vijayakumar and Whelan and is 62 presently being fabricated. Inspection of Figure (13) reveals a consistent trend. The addition of Sr decreased the catalytic activi ty, whereas Ce doping increases the catalytic activity of LaCoO^. Relative to LaCoO^ at 200°C, the Lag gCeg ^CoOg is about twice as active a catalyst, while the LaQ gSrg ^CoO-j is only half as active. The precision of all these kinetic data is noted to be on the order of +10% of the mean. In terms of the general precision of catalytic data, and the requirements for activity resolu tion, this level of precision is acceptable. Data on the Lag gCeg^^003 compound were obtained from different re actor designs, yet the data lie within the expected pre cision tolerance; thus, channeling of gas through the catalyst bed is not thought a serious problem for these studies. An activation energy and frequency factor for the reaction was evaluated from these data using a least square regression fit of data to the rate expression form, r = rQ exp(-E/RT) where the terms are, r = reaction rate, g-mole CO/sec-gm rQ = frequency factor, g-mole CO/sec-gm E = activation energy, k-cal/g-mole 63 R = gas constant, k-cal/g-mole-K T = temperature, °K These data are replotted to test the Arrhenius rate law form in Figure (14), with the straight lines representing the least square regression fit for each set of data. The activation energy and frequency factor parameters are tabulated below: Catalyst rQ, g-mole CO/sec-gm E, k-cal/g-mole LaQ#gCe0 j^CoOj 1.1 x 103 19.1 LaCoO^ 6.3 15.2 LaQ gSrQ jCo03 8.1 x 10-2 11.8 Again, a consistency in trend is noted from the type of modification performed on the LaCo03 catalyst. The addi tion of Ce increased the activation energy and frequency factor magnitude, whereas the presence of Sr caused a re duction of these parameters. The influence of water vapor upon the catalytic activities were observed by comparing the above data with those obtained for a similar reactant stream except for the addition of 2.5%ai H20. Rate data for the wet stream are presented in Figure (15). The effect of the water is quite remarkable. Approximate deactivation factors associated with the H20 additions are summarized below: 64 Figure 14 -- Comparison to Arrhenius Rate Law, • CC Oxidation ovox* the LaCoOg Seri css ierovshites • P.prctant_ Stream 2.7.';,:'. Cl 2.75. n 0o in He Flow. Rate ? . CC cc/: ' . i n (25 ’ C, 1 at:-) ;ono Cry -o 1000/T, ‘ K"1 ; j ' _ L.pad •o -:-325 •o i.v.er 10 u 3 a -6 9 8 7 6 i o <u a O O I o u «s c4 § 2 10 -7 220 Figure 15 — Effect of U^ter upon L&CoOo Scries _______________Catalysts for Carbon Iiono::ide 0;:iclat:ion. Reactant Stream 2.757<m CO 2.75%m Oo in He 2.57-m H^O F1ct; Rate 100 cc7min (25 #C, 1 atm) / / / A / o/ / °/ / A / / Reactor Load -200 to +325 mesh potriler Symbol Catalyst o la0.9Ce< □ LdCoO^ A 0.1Co03 240 260 280 300 Temperature , *C 320 340 360 Ul 66 Catalyst Deactivation Factor at 275°C 100 LaCo03 120 5 Clearly, the Lag gSrg ^co°3 catalyst is very much less sensitive to water than the other materials, although no firm explanation is presently available. One shall be postulated in the discussion section. The dependences of reaction rates upon CO and 02 concentrations were evaluated under bone dry conditions. These data were obtained by variation of one component concentration while the other reactant was maintained at about 3%m. The reactor was operated in the differential reaction mode. Data for Lag,9ceo.lCo03 at 151°C are pre sented in Figure (16), and the results for LaCo03 and Lag gSrQ ^CoO-j at 200°C given in Figures (17) and (18), respectively. Additional work on Lag gCeg ^CoO-j rate data was acquired at higher temperatures for a better compari son with the data for the other catalysts. Rate measure ments at 200°C were not feasible because of the greater activity of the Ce doped catalyst and the necessity to operate the reactor within the small conversion range. This information is provided in Figure (19). From these data reaction rate order with respect to each reactant Figure 16 -- P.eaction Rate Variation with CO and O2 Concentration, CO Oxidation over Lag gCeQ iC o 0 3 Catalyst. Reactant Stream Flow Rate Reactor Load Temperature Variable CC and 0£ lOO^cc/min -200 to +325 151 'C Bone L'ry (25 *C, 1 atm) mesh powder 8 6 2 7 10 8 10 10 I 1 o 0> V) _r* p CO °c CO n = 0.29 o rl o «- e v $ 4J Oi 10 2 1 4 6 8 10 CO C o n c e n t r a t i o n , O^ C o n c e n t r a t i o n , % n Reaction Rate, gmole CO/sec- Figure 17 — Reaction Rate Variation v/ith CO and O2 Concentration, CO Oxidation over LaCoO^ Catalyst* Reactant Stream Flow Rate Variable CO and Oo 100 cc/min Bone Dry (25 *C, 1 atm) CO^ ACO n * * 0.54 2 4 6 8 10 1 CO Concentration, %m R.eactor Load Temperature -200 to +325 200 *C mesh poixler 2.7%m CO 10 A 1 u e a co o2 n - 0.51 o o © H « % O 4J § •H U 8 10 1 2 6 02 Concentration, %m S Figure 18 — Reaction Rate Variation with CO and O2 Concentration, CO Oxidation over LaQ^gSrQ^jCoO^ Catalyst. Reactant Stream Flow Rate Reactor Load Temperature Variable CO and 0£ 100 cc/min -200 to +325 200*C Bone Dry (25*C, 1 atm) mesh powder 2%m CO 6 rco*po? n 3 0.21 8 CO co n 3 0.47 6 4 2 •H 7 10 10 1 2 CO Concentration, %m 6 8 10 09 Concentration, %m 70 Figure 19 -- Reaction Rate Variation with CO Concentration over bag oCeQ^0003 Catalyst, Variation or j.enperature. Reactant Stream Flow Rate Reactor Load Variable CO lOO^cc/nin -200 to -i-325 2.95/Ri C0 (25'C, 1 atm) mesh powder ione Dry “rCO * PC0 V * 0 ’ 6 4 2 •H 4 6 8 10 2 1 CO Concentration, %m 71 was determined assuming a rate expression of the form, <-rA) oc (PA)n where, (—rA) = reaction rate, g-mole CO/sec-gm (PA) = reactant partial pressure, atm n = reaction order These results are summarized below: Catalyst Temperature Order for CO Order for O ta0.9°*0.1CoO3 151°C 0.29 0.50 La0.9Ce0.1CoO3 176 0.45 - LaCoO-j 200 0.54 0.51 La0.9St0.1CoO3 200 0.47 0.21 Data on Lan qCen .CoO, at 151, 165, and 176°C indicate CO reaction order approaches about 1/2 at 200°C. Effects of Ce Addition to LaCr03 Chemical and thermal stability has been noted as one of the requirements for practical emissions control catalysts. In some applications, the reaction temperatures are solely limited by catalyst stability limits. The LaCrO^ series are of interest because of the stabilities demonstrated by selected examples. This series should, in general, have lower sintering rates and lower electrical 72 conductivities than the corresponding LaCoO^ and LaNiO-j ones. Because of the latter property, a general lower catalytic activity was expected. The following experi ments were carried out to elucidate this point and to pro vide a broader base for future studies on the perovskites containing mixtures of the transition metals. The LaCrO^ series catalysts were prepared in a man ner similar to that used for LaCoC>3 preparation. Details are given in Chapter VI. It was not possible to densely sinter the LaCrO^ series because of the higher temperatures required which were beyond the range of the furnaces avail able for this work. In order to obtain a sufficient quan tity for testing, the LaCrO^ series catalysts were select ed within the range of -120 to +200 mesh size particles. The quartz reactor used to test this series of catalysts was similar in design to the Pyrex glass reactor used for the LaCoC^ series. Details of the reactor design are pre sented in Chapter VII. To examine the trend seen in the LaCo(>3 series that doping with Ce improves the CO oxidation catalytic activity, the catalyst . 9Ce0. lCr03 was comPared with the undoped LaCr03. These two compounds were tested using a 2.8%m CO and 2.8?£m 02 in He reactant stream. A bone dry reactant stream and one with 2*0(m H20 vapor were examined. These results are presented in Figure (20). Inspection Figure 20 — Catalytic Oxidation of Carbon Monoxide over LaCrOq Series Catalysts — i -------- J— i ---------- 1 ------ La~ «Ceft ,CrO LaCrO, La CrO Bone Dry 2.5%m Ho0 LaCrO Reactor Load ••120* to +200 mesh po\/der Flow Rate 100 cc/min (25*C, 1 atm) Reactant Stream 2.0/on CO 2.8%m O- VariablS llo0 320 420 280 300 340 360 380 400 Temperature, *C 74 of the data reveal that indeed, Ce doping enhances the catalytic activity towards CO oxidation. This results from an increase in both the pre-exponential factor and the activation energy for the overall reaction as is in ferred from inspection of the data of Figure (20). Such behavior is consistent with observations upon the LaCo03 series catalysts. The response of the LaCr03 series of catalysts when exposed to water vapor is interesting. Unlike the behavior observed for the LaCo03 series, these catalysts are appar ently less sensitive to deactivation by water. Indeed, for LaCr03 it appears the catalytic activity is improved in the presence of water at higher temperatures. This warrants a fuller investigation in the future. Many facets of this program suggest worthwhile areas for future study. Establishing this was an early goal of the present work. It is with some regret, that the leads developed could not be more fully pursued in the present work because of time limitations. Estimation of Specific Surface Areas for the LaCo03 Series An estimate of the specific surface area of -200 to +325 mesh powders can be made by comparison of reaction rate data for two-sieved powders of the two sizes. Note that the data from Figure (15) show LaQ 6q ic<>03 and LaQ gSrQ 1Co03 as having about equal catalytic activities 75 for a 2.7556m CO, 2.7556m 02, and 2.556m H2O reactant stream. For the same test conditions, a sample of LaQ.9Sr0.lCo03 of -200 mesh particle size was examined. A comparison of these data with Lag gCeg ^CoOj of -200 to +325 mesh size is provided in Figure (21). Observe that at 240°C the catalytic activities differ by a factor of 6.5. Assessing this difference as due only to specific surface area vari ation, the -200 to +325 mesh powder is estimated to have a specific surface area of about 0.1 m2/gm as compared to the 1.2 m^/gm for the -200 mesh powder which was measured using a B.E.T. apparatus and N2 adsorption isotherm. These catalysts have approximately a 9556 theoreti cal density for small particles, and about a 7556 bulk density for the large sintered wafers. Thus, one is un certain about assessing the total surface area from the external particle surfaces as representative of the true specific surface area of the particle. For example, a cube 75 jum on a side and 6 gm/cc density has a specific surface area of about 0.01 m2/gm, if only external surfaces are being considered. The -200 to +325 mesh particles have a mean diameter near 75 jum, and a bulk density near 6 gm/cc, yet appear to have a specific surface area of 0.1 m2/gm. This fact makes the non-permeable particle model unlikely. The converse model of a completely per meable particle is not supported since elimination of 76 Fisv.ro 21 -- Carbon Ilonoxide Oxidation over Ce and 3r Doped LaCoO^ Catalyst. ~.y\rc'c .it Strrqart "V:_to 2 • i 10 1 cc/r.iin -o 10 OU3 -200 nosh por .?dor 8 La0.9Ce0.1Co°3 -200 to +325 mesh 6 4 2 • r l 7 10 200 220 260 240 2 8 0 3 0 0 Temperature, *C 77 the small diameter, and thus, small volume particles greatly reduced the specific surface area of the powder. A possible explanation of this conflict is suggest ed from an inspection of scanning electron microscope photographs of several sintered LaQ gs*Q ^CoOj parts. Rods of this compound were extruded with a glycerol/ water lubricant, dried, and sintered by heating 48 hours at 1250°C in air while contained in a Pt tube crucible. Bulk density of these rods was determined to be about 5.5 gm/cc, which compares well with that of pressed and sin tered wafers used to prepare powdered catalysts. Inspec tion of photos shown in Figure (22) reveal the machine- formed exterior surface of the rods has the typical appear ance of a well sintered part. Grain size is approximately 3 to 5 microns. There were few voids or passages to the interior of the pellet visible at a magnification of 1700X. A view of the surface produced by a fracture shows much more roughness and a greater permeability to the in terior of the part. These factors are better seen in the higher resolution photographs of the exterior and interior surfaces presented in Figure (23). At 8500X magnifica tion, inspection of the exterior surface indicates a mean grain size of 2 to 3 microns and a low permeability. The roughness of the fracture surface is better shown and indicates penetration to a 5 to 10 micron depth is 78 Figure 22 — Sintered LaQ^SrQ ^CoOj Extruded Rods, 1700X Magnification Exterior Machine Formed Surface 10 yum i ------ 1 Interior Fracture Formed Surface 79 Figure 23 — Sintered I*ao.9Sr0.1Co°3 Extruded Rods 8500X Magnification Exterior Machine Formed Surface 2 urn I t Interior Fracture Formed Surface 80 possible through many of the pores. A single lower resolution photograph (430X) is presented in Figure (24) to exhibit the character of a die-pressed and sintered wafer of I*ao.9®rO. 1^°®3 such as used to prepare catalyst powders. Material so fabricated appears more permeable, yet bulk density is about 5.2 gm/cc. At higher resolutions it is expected that the die- pressed wafers would appear similar to the extruded rods. These visual inspections of the microstructure of the sintered particles suggest a model for the catalyst particles as ones having a permeable surface layer of 5 to 10 microns in depth and a nonporous interior. Thus, effective surface area is that which results only from the surface layer region. The -200 to +325 mesh particles have a mean diameter of about 75 juun, thus, a surface layer of 5 jim would occupy about 20% of the cube volume. As sume that the surface layer has properties equivalent to a layer of agglomerated 2 jam cubic particles having a den sity of 6 gm/cc. Such a surface would yield a specific surface area of about 0.5 m /gm. Therefore, the average specific surface area of the catalyst 75 jam particle as it is modeled above would be about 0.1 m /gm. Although this model is an approximation, it does appear to be con sistent with the kinetic data and B.E.T. method estimates of the specific surface areas for the more finely divided Figure 24 — Sintered LaQ gSrQ jCoO^ Die Formed Wafer 430X Magnification External Die Formed Surface in Horizontal Plane of Photograph Interior Fracture Formed Surface in Vertical Plane of Photograph 82 -200 mesh cut of particle sizes. This model for the specific surface area is then extended to all other catalysts by the assumption of a similar microscopic structure for the particles. As long as the particle sizes are about equal so that the frac tional surface layer volumes are about constant, all cata lyst particles should exhibit equivalent specific surface areas. An estimate of the mean particle sizes for LaQ gCe0 2C0O3 a^d LaQ gSrQ iCo°3 of "200 to +325 mesh was performed by examination of particle size distribution from a sample of about 150 particles. The characteristic dimension of a particle was taken to be the geometric mean of the two largest dimensions normal to each other, as the particle is viewed under a microscope. This size distri bution was analyzed by application of a logarithmic- probability relation as described by Hatch and Choate (18). From a plot of cumulative frequency versus particle size on a special logarithmic-probability graph paper, the geometric mean diameter and geometric standard devia tion of the distribution was evaluted. The results of this treatment are tabulated below: 83 Catalyst Mean Diameter Standard Deviation 70 jim 10 jam La0.9Sr0.1CoO3 61 10 The agreement of these dimensions to within ±10% is suf ficient to indicate that specific surface area variations between the LaCo03 catalyst series should not be suffi cient to void the previous conclusions on relative cata lytic activity. Properties of the Perovskite Oxides the oxidation of CO catalyzed by rare earth-transition metal perovskites, several properties of these oxides are examined below. Of much interest is the identification of the types of defects which might be present in these compounds when appreciable amounts of Sr+^ or Ce+4 cations are pre sent in these pervoskite catalysts. The data of Schroder (27) and Jonker and Van Santen (28) on the distribution of cobalt ion valence in LaCo03 when doped are presented in Figure (25). Schroder observed a maximum solubility of Th+4 in LaCo03 corresponding to a mole ratio of Th to La of 0.04. Presumably the Th+4 cation substitutes for a La+3 cation in the crystal lattice. The controlled valence model applied to this compound requires In order to further discuss the kinetic data for Percent of Co in Valence State Figure 25 -- Stoichiometry of Lai ^Sr^CoOg and La-^ _xT1v: Co03 ^s^ovskites SO •50 40 <r + o o r a ^ 20 0 CM + o o 20 20 0 20 40 60 80 100 +4 +2 %a Tli %a Sr 3 Percent Substitution for La Symbol Data Sotirce Schroder (27) Jonker and Van Santen (28) ------ Theoretical model for no vacancies in crystal lattice. 85 compensation of the effective charge of the defect Th£a by an adjustment of the valence of another cation. This would be thought to be the Co cation by formation of COqq +2 +3 defects, i.e., a Co on a Co lattice site. The data of Schroder are consistent with this model of the defect structure of Th doped LaCoO^* Both Schroder and Jonker and Van Santen examined the doping of LaCoO^ with Sr. Their samples were prepared from La203, SrC03, and C0CO3 mixtures fired in pure oxygen at 1150 to 1250°C. Firing at lower temperature caused an appreciable loss of oxygen although the pervoskite struc ture was preserved. At high Sr content the preparations had a tendency to lose oxygen. These data on valence distribution of cobalt re sultant from Sr doping are in agreement up to about 60% replacement of La+3 cations by Sr+^ cations, but disagree at higher doping levels of Sr. No explanation for the disagreement is offered for the 60 to 100% Sr range of data. Interpretation of Co valence distribution is con sistent with the incorporation of Sr at La sites (Sr^a defects) with charge compensation by adjustment of Co valence (Co£0 defects) for Sr^a mole fractions up to 0.4. Jonker and Van Santen published the raw experimen tal data on cobalt valence distribution upon Sr doping of LaCo03 in addition to a smooth curve to represent the 86 data. This information is presented in Figure (26) for inspection and interpretation. Upon inspection of the data in Figure (26), one notes that about half of the data below 40% Sr level lie below the theoretical prediction of cobalt valence distribution. A least square regression of this portion of data yields a functional relationship, Y - 0.7 + 0.95X where, Y = percent of Co present as Co+^ X = percent substitution of La with Sr From this fit of experimental data, one concludes that the observed Co+* content is about 10% below amounts expected V • on the basis of SrLa compensation by CoCo in this region of Sr concentration. Such behavior is in accord with the recent suggestion by Brook and Whelan that incorporation of Sr+2 into LaCo03 lattice at La+^ cation sites may lead to formation of anion vacancies. As indicated by the small deviations of cobalt distribution from the valence control model, at least at Sr concentrations less than 0.4 mole fraction, this process of incorporation accounts for only 10% of the charge compensation. At higher Sr contents, more vacancies may be formed. Due to the limited variance from the controlled cobalt valence model, it is necessary to inspect and Percent of Co as Co 87 Figure 26 -- Valence Distribution of Co in La, SrCoO, Ferovskite. X " X X Experimental data of Jo niter and Van Santen (28) 100 80 60 40 20 0 0 20 60 80 40 100 Percent Substitution of +2 + 3 Sr for La 88 comment upon experimental procedures used by Jonker and Van Santen. The cobalt valence distribution was deter mined by dissolution of the perovskite sample in a HCl/FeCl2 solution, followed by titration with permanga nate for the excess Fe+2. The gravimetric factor for Co content of the pervoskite is based upon the exact formula La^ ^Sr^oO^. Then, by comparison of observed total oxi dizing power of sample to that which would be expected from a 100% Co+3 sample, the fraction of cobalt present as Co+4 may be estimated. The experimental difficulty pre sent in this analysis lies in the fact that Co+^ is a very strong oxidizing agent and may lead to side reactions. Consider for example some reactions of C0O2 possible with FeCl2 aqueous solution: Reaction________________________ Eo C0O2 + 4H+ + 2Fe+2 = Co+2 + 2H20 + 2Fe+3 + 1.32 v Co02 + 4H+ + 2C1"1 = Co+2 + 2H20 + Cl2 0.7 Co02 + 2Fe+2 ■ Co+2 + 02 + 2Fe+3 0.1 Thus, 02 and Cl2 gas may be produced when the samples are dissolved in the reagent. Jonker and Van Santen modified the standard analysis by placement of samples in a vial containing the frozen reagent, then sealing the vial. These vials were then heated 20 hours at 120°C to complete dissolution and reaction. The possibility of an in 89 complete reaction of any 02 of Cl2 produced with the ex cess FeCl2 is thought to be slight; if not, it would lead to a low value for the fraction of Co*4 present in the sample. The electrical conductivity of the perovskite catalysts was measured using a four terminal direct cur rent apparatus described in Chapter VII. The experimental data are presented in Figure (27) as conductivity versus temperature. Jonker (10) has investigated the semicon ducting properties of LaCoC>3. He concluded that it was a semiconductor with an energy gap of about 0.3 eV width. Both the conduction and valence bands have a high density of states. Mobilities of the charge carriers is low, e.g., at room temperature the hole mobility is about 0.3 cm2/V-sec, and the electron mobility being approximately 0.03 cm2/V-sec. A weak thermal activation energy of 0.05 eV has been found for the hole mobility, whereas the elec tron mobility activation energy has a higher value of 0.13 eV. LaNiC>3 has a very high conductivity for an oxide (200 (ohm-cm)-* at 100°C in air) which decreases slightly with temperature suggesting that it is metallic, or a degenerate semiconductor. LaCoOj is also a fair conductor with a conductivity of about 1 (ohm-cm) * at 100°C in air. Doping of LaCoC>3 with Ce modifies the conductivity by less Conductivity, l/(ohm-cm) Figure 27 -- DC Electrical Conductivity of Selected Perovskite Oxides. .+3 10 Svnbol Oxide A LaiiiO ^ o . ^ o . i 000: LaCoO. La CoO La CrO LaCrC CrO 10 10 0 200 400 600 800 Temperature, *C 91 than a factor of 2 or 3, but changes the conductivity type from p to n, whereas additions of Sr increases the p-type conductivity greatly. Lag^SrQ ^CoOj has a con ductivity of about 350 (ohm-cm)-1 at 100°C in air and in creases slowly with temperature. The LaCr03 compound is less conductive than the other perovskites studied having —2 —1 a maximum conductivity of 4 x 10 (ohm-cm) at 800°C in air. Again doping with Ce does not vary conductivity greatly at high temperatures, but does yield a larger dependence of conductivity upon temperature. Sr additions again increase the conductivity yielding about 20 (ohm- cm)”* at 100°C in air for LaQ ^SrQ ^CrOj. The color of LaCr03 changed from a light green to a dark brown when the Sr was added. All other pervoskites studied had a grey metallic luster after sintering, and a dull black appear ance in the form of fine powders. The types of mobile charge carriers were identified in the above perovskites by measuring the sign of the Seebeck coefficient. The apparatus used to perform this measurement is described in detail in Chapter VII, along with the experimental procedures. Samples used were sin tered wafers having a bulk density of 15% of the theoreti cal density, containing particles with diameters on ap proximately a few microns. It is assumed that these mea surements reflect the bulk properties, rather than a 92 surface layer characteristic. The results of these deter minations are summarized below: Conduction Type n (electrons) p (holes) n n“ P- P P+ The superscript (+) or (-) notation place on the conduc tion type symbol are to indicate the concentration of carriers relative to the undoped compound. These judg ments are made from the magnitude of voltages observed when sign of the Seebeck coefficient was determined, and what one would expect to occur from the type of dopant added. The LaCo03 series exhibit properties expected from the reports in the literature when doped with a +2 or a +4 cation. Sr doping of the LaNi03 did not yield a Compound_______ La0.9Ce0.1CoO3 LaCoO-j La0.9Sr0.lCo03 La0.9Ce0.1NiO3 LaNi03 La0.9Sr0.1NiO3 La0.9Ce0.1CrO3 LaCr03 La0.9Sr0.1Cr03 93 p-type semiconductor, and Ce doping of LaCr03 did not yield an n-type semiconductor. CHAPTER V DISCUSSION Introduction The breadth in physical properties of an oxide that one may consider for correlation with catalytic activity is large. Description of the bulk defect chemistry is considered a very promising starting point for describing the catalytic activity of metallic oxides. The experimen tal work described in the preceding chapters has been directed towards study of oxides such as stabilized zir- conia, a good ionic conductor at high temperatures, to the opposite extreme, the rare earth-transition metal perov skites that have electrical properties characteristic at moderate temperature of degenerate or nearly degenerate n- type and p-type semiconductors. Consideration of the heterogeneous catalytic process follows a summary of the kinetic observations. Summary of Kinetic Data From the preliminary screening experiments a ranking of relative activities for catalytic oxidation of CO with O2 was determined for a variety of oxides. These results are recalled below: 94 95 LaCo03 > NiO > Th0 85La0>15O1>925 > Zr02 > Zro .95Ca0 .06° 1.94 A decision was made to investigate further the rare earth- transition metal pervoskites because of the high catalytic activity shown by LaCoO^. Observations on a more quantitative level demon strated the ranking of the catalytic activities for tran sition metal substitution in the rare earth perovskites to be: LaCoO^ > LaNi03 > LaCrO^. The influence upon catalytic activity of doping of these perovskites with altervalent cations was examined, and the observed ordering as represented by the LaCo03 series, La„ Ce„ CoO_ > LaCoO. > La„ Sr CoO_ 0.9 0.1 3 3 0.9 0.1 3 in the region of 200°C in a bone dry reactant stream. Another series of experiments on the reaction of MO with H2 catalyzed by these materials revealed the activity ranking to be: La0.9Sr 0.lCo03 > LaCo03 as judged from consumption of NO and the product distribu tion. 96 A limited examination of the catalyst composition variation effects in the LaCrO^ series indicated the order towards the catalytic oxidation of CO to bet La0.9Ce0.1CrO3 > LaCr03 The continued observation of a significant deacti vation by small amounts of water vapor for the perovskite catalysts led to a more precise examination of this effect. For this comparison, experiments were performed using a bone dry reactant stream and one with 2*0&m H20. The kinet ic data were extrapolated to 275°C for comparison. The re sult was observation of a deactivation factor of about 100 for LaQ gc®0 lCo03 and LaCo03» but only about a factor of 5 for LaQ gs*0 iCo03• T*ie activation energy for CO oxida tion is 27+2 K-cal/g-mole for these catalysts in the presence of 2%&n H^O. Similar studies performed on the LaCrO^ series showed quite a different behavior. La QCe_ -.CrO. was deactivated by water by a factor of U • 7 v • X J about 2, with no change in activation energy; however, the LaCrO^ showed an increase in the activation energy for CO oxidation from 11 to 20 K-cal/g-mole when water was added to the reactant stream. Catalytic Activities of Doped and Undoped Zirconia and Thoria_____________________________________________ A new hypothesis has been described previously in 97 this thesis as the "fuel cell model" to explain the hetero geneous catalytic oxidation of CO with O2 on doped and un doped zirconia and thoria. Properties of these oxides which seemed to correlate with the catalytic activities were the bulk electronic and ionic conductivities. It was hypothesized that significant mobilities for electronic charges and ions (oxygen ions in these materials) in the bulk were indicative of mobilities of these components on the surface or in a shallow layer of the bulk near the sur face. The simultaneous drift of ions and compensating charge on, and or, in the bulk provides a means of trans porting a reactant (such as oxygen) from one site on the surface to another site. If this transport is the rate limiting process controlling the overall reaction rate, such a model is worth considering for application to cata lysis. In one sense this view is an extension of Wagner's theory for tarnishing of a metal in which the rate of oxi dation is limited by transport through the oxide layer by the mechanism of simultaneous diffusion of ions and elec tronic charges. The driving force for the diffusion is the gradient in the chemical potential of the mobile ions. For surface reactions, as viewed with the fuel cell model, the chemical potential gradients are also influenced by the catalyst compositional changes which modify the surface concentrations of the reactants. 98 The extension of this model to other oxides such as NiO and LaCoO^ is viewed as somewhat improper, because the latter oxides have appreciable activities at lower temperatures where bulk ionic diffusion constants are re latively small. These oxides also have substantially higher electronic conductivities. As was described earli er, there has been little encouragement to be found in the literature for the existence of significant concentrations of vacancies or interstitials in the perovskites studied in this research. These defects are required for a high bulk ionic diffusion constant. Inspection of some of the features of the catalyzed oxidation of CO with O2 over LaCo03 and NiO are useful in establishing some of the trends of behavior of LaCoO^ relative to the established catalyst NiO. Catalytic Activities of Doped LaCo03 and NiO One of the important topics of this research was an evaluation of the effects of doping upon the catalytic activities of the perovskites. Experimentally, this was performed by measuring the catalytic acitivity of LaCoO^ in which 10% of the La cations were substituted with either Sr or Ce cations. Analogous experiments may be done using NiO when doped with either Li or Cr cations. Such experiments on NiO are reported in the literature. A tabulation of NiO and Li doped NiO catalysts for 99 CO oxidation with 02 are presented in Table IV. These were taken from the review article by Bickley and Stone (20). Awareness of the divergence of results reported by various investigators is necessary so that one may appre ciate the difficulty of arriving at a "correct conclusion" in this area of catalysis. Much discussion of these re sults has been presented in the literature to explain the lack of agreement; the consensus is that a lack of under standing of how foreign cations are incorporated into the NiO lattice for a particular preparation procedure is central to the difficulty of interpretation of the kinetic data. From these discussions, it is also apparent that the concentrations of foreign ions in the bulk and near the surface may be different. The same comment applies to all defects present in the lattice. One can then appreci ate how the divergence in the kinetic data are not unrea sonable, if the methods of preparation and conditions of use produced different equilibrium states in the differ ent investigations. Some of the common kinetic parameters for doped NiO are in qualitative accord with the limited results for the LaCo03 series of catalysts. These are noted below only to supplement the known experimental validity of present data on doped LaCoO^ catalysts. Brelanski and Deren (26) report the reaction rate dependence of CO Table IV — Catalytic Studies of CO Oxidation Using Li Doped NiO Catalyst Date of Work, Reference Doping^ of NiO with Li Catalyst ^ Preparation Range o f ^ Study Variation of Arrhenius Law Parameters with Respect to Undoped NiO, r = rQ exp(-E/RT) E H) r0 (5) r(6) Parravano (1953) #21 0.01*m 640°C 3 hours 230-380°C Increase 14-18 - Schwab & Block (1954) #22 up to 5%m 850°C 3 hours 300-400°C Decrease 15-13 Decrease Increase Keier, et.al. (1958) #23 up to 8%a 900°C 2 hours 20-350°C Increase 5-20 — Cimino, et.al. (1958) #24 1.0%a 950°C 3 hours 210-260°C Increase 13-19 — — Dry & Stone (1959) #25 up to 2.8%a 1000°C 3 hours 300-400°C Decrease 14-12 Decrease Increase (1) Percent of Li; £m = mole percent, %a = atomic percent. (2) Highest temperature and longest heating time for preparation of catalyst. (3) Range of temperature over which kinetic experiments performed. (4) Change in E with increased Li content, K-cal/g-mole. (5) Change in rQ with increased Li content. (6) Change in r with increased Li content. 101 oxidation with 02 upon the reactant pressures over p-type NiO, pure and doped with Li20. A few of their results are tabulated below: Catalyst Conductivity Type Temperature Order for CO Order for 0 NiO P 250°C 1.0 0.0 0.2%a Li in NiO P+ 250 0.25 0.75 3.4%a Li in NiO P++ 250 0.33 0.20 Dry and Stone (25) report the CO and O2 concentration de pendence of reaction rate catalyzed by a Cr doped NiO to be first order for CO and zero order for O2 at 400°C, simi lar to Brelanski and Deren's results for undoped NiO. Re call from the previous chapter the observed reaction rate dependencies for the LaCo03 series perovskites to be: Conductivity Order Order Catalyst Type Temperature for CO for 02 La0.9Ce0.1CoO3 n 176°C 0.45 0.50 LaCoO^ P 200 0.54 0.51 La Sr ,CoO_ 0.9 0.1 3 P+ 200 0.47 0.21 Comparison of behavior of the perovskites with the p-type doped NiO reveals the p-type La^ gSrQ iCo03 yields a reac tion rate dependence on O2 concentration which is similar to that resultant from strongly doped NiO. Behavior of 102 the n-type LaQ 9CeQ 1Co03 and p-type LaCoO-j is not similar to observations made on the NiO system. Commonly, investigators report a fit of data on overall reaction rate to the Arrhenius rate law expression (rate = Aexp(-E/RT)). Since the catalytic process is the result of a series of elementary reaction steps, the acti vation energy determined from the Arrhenius law is an apparent value for the overall process, but mainly deter mined by the rate limiting step. One cannot infer that this parameter necessarily represents the magnitude of an energy barrier which re actants must overcome in order to yield products. Such an interpretation would imply that an equivalence of activa tion energy means some correspondence of mechanism of re actions over different catalysts. The best use of the Arrhenius rate law for catalytic heterogeneous reactions is as a correlation function describing the temperature sensitivity of the reaction rate, and its utility as a means of assessing relative catalystic activity. Arrhenius law activation energies for catalytic CO oxidation using NiO doped with various amounts of Li or Cr have been determined by Dry and Stone (25) and Schwab and Block (22). These data are presented in Figure (28) along with results of this work on Ce and Sr doped LaCoO^. Although the scatter is marked for the Activation Energy, kcal/gmole Figure 28 — Activation Energy for Carbon Monoxide Oxidation Using Doped Oxide Catalysts* Svnbol Data Source A Schwab and Block (22) O Dry and Stone (25) □ This work 20 A a IS 16 12 10 10 5 0 5 10 Percent Foreign Cation Cation Catalyst Cation Cr NiO Li Ce+4 LaCo03 Sr*2 104 activation energies for NiO, the trend in magnitudes is similar to that found for the n-type LaQ gCeQ j^CoOj and p-type LaCo03 and LaQ gSrQ ]Co03 catalysts. A further inspection of the data on Arrhenius rate law fit of CO oxidation data using doped NiO and LaCo03 revealed the presence of the compensation effect phenome non. The phenomenon is the observation that a reaction performed under differing conditions, e.g., with a differ ent catalyst, both the activation energy, E, and pre-ex ponential factor, A, change simultaneously. E and A change in such a way that the corresponding change in reaction rate is much less than would be the case if only E or A changed alone. These data proved to conform to the rela tionship, In(A) = a + bE This correlation must be regarded as purely empirical, primarily because neither of the Arrhenius parameters determined from overall reaction rate data represent thermodynamic properties of an elementary reaction step. When the compensation effect assumes the linear form noted above, the result is that a temperature exists where the reaction rates over a series of catalysts are all equal. This temperature is defined as the isokinetic temperature. From the data on doped LaCo03, the iso- 105 kinetic temperature was determined to be shout 90°C. Be ing conscious of the risks of extrapolating the Arrhenius plots beyond the range of experimental data, awareness of the isokinetic temperature has practical importance to ex perimental study of a catalyst system. When comparison of different catalysts are made in the region of this temperature, experimental errors are much more significant because of the small differences in catalytic activity which must be measured. Comments on Reaction Mechanism Model The preceding sections of this dissertation have covered the experimental conclusions regarding the cata lytic activities of the rare earth-transition metal perov- skite catalysts. These studies have adequately defined the general characteristics of these catalysts as they are applied to simple reactions such as carbon monoxide oxida tion with oxygen, and to a lesser extent examined their application to a number of other oxidation-reduction reac tions. Included in these earlier discussions are some comparisons between the catalytic properties of perov skites and the behavior of doped NiO and other pure metal lic oxides. Detached from consideration of reaction kinetics, several aspects of the chemical and electrical properties of these oxides have also been examined. These discussions are of value in establishing a 106 perspective in viewing some of the physical properties likely to be related to the catalytic properties of the perovskites. Unfortunately, these properties do not ade quately provide the descriptions needed for the most simple understanding of heterogeneous oxidation-reduction reac tions. Understanding of the catalytic processes are taken herein to mean an experimentally supported model for the mechanisms operative in the rate limiting step for the overall reaction of interest. At the present state of the art, it is a major task to describe such a model for a heterogeneous catalysis reaction on any oxide catalyst. The following discussion will be limited to the examination of the overall reaction for the oxidation of CO to CO2 by O2. Processes which may define the rate limiting step for this reaction might include the following: (1) adsorption of one or both the reactants as neutral species; (2) dis sociation of O2 into adsorbed oxygen atoms; (3) ionization or polarization of the reactants at selected surface sites; (4) transport of the neutral, polarized, or ionized reac tants with appropriate amounts of electronic charge to preferred surface reaction sites; (5) reaction of the re actants to form an ionized, polarized, or neutral product; (6) formation of the neutral adsorbed reaction product, if the product is formed as ion; and (7) desorption of the neutral product and its diffusion into the gas phase. A 107 suitable model should account for variations in the over all reaction rate with concentrations of the reactants and products in the gas phase near the surface, as well as consider compositional changes of the catalysts. It is beyond the limits of reasonable speculation to hope to do this for more than minor changes in the composition of a given catalyst. With this in mind, the changes in catalyst compositions will be limited to unintentionally doped p- type LaCo03, a semiconductor which can be made more p-type by doping with Sr, and which can be made n-type by doping with Ce. The doped samples studied were ones in which 10% of the La ions were replaced by either Sr or Ce ions. The consequences of several "reasonable" assumptions will be followed to indicate one of several possible routes to wards forming a model. The following assumptions are made regarding the defects in doped LaCoO^. Cerium doping results in the sub stitution of Ce+3 ions for La+^ cations. A significant fraction of the Ce ions are in the +4 valence state and as such are ionized donors, Ce£a. These donors are primarily compensated by the hopping electrons, Co^, which are responsible for the n-type conductivity. The presence of Ce£<a tends to reduce the concentration of Vq ’ and increase I the concentration of cation vacancies. The SrLa ionized acceptors are generated by substitution of Sr+^ for La+^ 108 • and are primarily compensated by the hopping holes, CoCo. Additions of Sr tend to increase the concentration of V^* and decrease the concentration of cation vacancies. Since relatively little direct information presently points to significant bulk concentrations of vacancies (about 1% of • • VQ in Sr doped LaCoO^), they will be ignored in the de velopment of a model for the catalytic process. It should be noted that the surface concentrations of vacancies should influence the adsorption and subsequent ionization of reactants, but due to the low concentration of these defects, the effects are thought of little importance. Transport of a reactant as an ion by surface diffu sion or diffusion in a shallow surface layer is enhanced by an appreciable vacancy concentration. Because of the uncertainty regarding these defects in the perovskites, the fuel cell model suggested for Zr02 and Th02 will not be pursued. Another factor for this decision is that the temperature needed for CO oxidation over LaCoC^ samples are lower than those required for Th02 and ZrC>2> Ionic diffu sion currents would also be lower if they depended only on the diffusion coefficients, i.e., the chemical potential gradients of the reactants are essentially constant for the different catalyst compositions. It may be noted that the magnitudes of the chemical potential gradients for the reactants depend partly upon the surface densities of 109 preferred adsorption sites. These site densities are likely to be subject to the catalyst dopant and its con centrations. Following Stone (29), who has reviewed the earlier work on CO oxidation over relatively simple oxides, we shall assume that adsorbed CO can ionize on the surface of LaCo03 as a donor, so that the total amount of CO adsorbed on the surface is the sum of the neutral and ionized con centrations, i.e., |C0 (sg an(} jco'(sj] , respectively. Adsorbed molecular oxygen is assumed to be in equilibrium with adsorbed neutral oxygen atoms which can ionize as ac ceptors, i.e., o'(s). Furthermore, it will be assumed that the reaction rate is not limited by the adsorption or desorption rates of neutral species or the rates of ioni zation, but rather by the product of the concentrations of ionized carbon monoxide donors and ionized oxygen accep tors. This is the essence of the charge transfer model for catalysis (30,31,32) as applied to the presumed rate limiting step: CG*(s) + o'(s) = C02(s). Inference that adsorbed CO and O might behave respectively as a donor and an acceptor is based on the conductivity changes upon CO or 02 adsorption on powdered p-type Cu20 samples. Garner, Grey, and Stone (33) noted 110 an increase in the conductance of a powdered Cu20 sample upon the adsorption and presumed dissociation of 02 into O atoms. This was interpreted as being due to the adsorp tion of oxygen and the formation of ionized surface oxygen I acceptors, 0 (s). Upon adsorption of CO, the conductance decreased suggesting the presence of CO*(s). Such conclu sions were predicated on several assumptions. These in clude one that p-type Cu20 has a native surface state density which causes the energy bands to bend downward at the surface resulting in the surface layer having a lower conductivity than if the bands were flat at the surface. The latter is unlikely. The addition of ionized surface acceptors tends to reduce the downward band bending, and thus increases the p-type conductivity of the surface lay er. This would increase the particle to particle conduct ance, and therefore, the conductance observed using a powdered sample. By the same arguments, the decrease in conductance of the powder was taken as evidence for CO*(s). An additional assumption that was necessary (and quite reasonable) to reach these conclusions was that the native surface state density was small compared to the surface concentrations of ionized oxygen and carbon monoxide. It was necessary in the argument that CO and O ad sorbed on CU2O were donors and acceptors, respectively, to assume that the Fermi level at the surface could be dis Ill placed relative to the band edges by adsorption of either CO or 02> An alternate situation is postulated for doped and undoped LaCoO^; namely, the native surface state den sity is such that the Fermi level at the surface is always pinned at an energy substantially above the acceptor ioni zation energy for O(s) and below the CO(s) donor ioniza tion energy. In many of the n- and p-type degenerate and non-degenerate III-V and II-VI seroicondcutors the Fermi energy level is pinned at an energy above the top of the valence band by an amount approximately equal to 1/3 of the energy gap. The assumption of the position of the Fermi level relative to the CO(s) donor and 0(s) acceptor levels thus appears a possible occurrence. Recalling the postulate that the controlling rate for CO oxidation was proportional to the product [co' (s (sj| , it is only necessary to make relatively safe additional assumptions to derive the predicted pres sure dependence of the reaction rate. The first two re gard the equilibrium between the oxygen surface species and 02 in the gas phase, i.e., the following processes are in a state of equilibrium: (1) 02 (g) = 02(s) (2) 02 (s) - 20(s) CoCo,s + °<s> = CoCo,s + °'ts> 112 where the subscript s is used to denote a surface species. Similarly, for carbon monoxide equilibrium is postulated for the processes: (4) CO(g) = CO(s) (5) Coc0fS + CO(s) = Co£0#s + CO*(s) and finally: (6) 2CoCos ® CoCo#s + Co£0'g Pinning of the Fermi level at the surface below the CO(s) donor ionization energy makes |co* (s)j oC p^q5' since [CO* (s)] = [co^,o Also, the concentration of CoCq is fixed by the surface pinning of the Fermi level so that jo' (s)] cC |o(s)] oC Pq2 . Thus, the rates of CO oxidation might be expected to depend on the square root of oxygen and carbon monoxide pressures. The experimental rates vary as Pq p™0' values of n and m for the doped and undoped LaCoO^at about 200°C as follows: Catalyst n for O2 m for CO La. Ce .CoO. (151°C) 0.50 0.29 LaQ gCe0 1Co03 (176°C) 0.50 0.45 LaCo03 (200°C) 0.51 0.54 LaQ gSrQ ^003 (200°C) 0.21 0.47 Considering the arbitrariness with which the assumptions 113 were made, the agreement between the predicted and observed pressure dependencies are good at about 200°C. The excep tion was a weaker observed pressure dependence for O2 in the p-type doped LaQ gSrQ ^CoOj. This suggests that on this catalyst the concentration of adsorbed oxygen atoms is higher for a given oxygen pressure than it is for Ce doped and undoped LaCo03. At lower temperatures there is a suggestion that the concentration of adsorbed CO might be increased by the presence of Ce in the catalyst. This would be inferred from the value of m less than 1/2 at 151°C. The experimental observations when water was pre sent did not include data to develop reaction rate depend ence upon reactant concentrations. Thus, it is not pos sible to establish from the available data clearly if it is a surface coverage phenomenon or a change in reaction mechanism that occurs upon exposure of the perovskite catalyst to water vapor. In terms of the model postulated earlier, the gross features for H2O deactivation would be anticipated if adsorbed H20 was a donor and occupied sites normally available to CO(s) and CO*(s). Since all envi sioned applications of these catalysts are for situations where water is present in moderate to high concentrations, further exploration of how H2O affects the catalyst acti vity is essential. 114 Conclusions One of the early questions raised was the reason ableness of investigating the use of the rare earth-tran- sition metal perovskite oxides as oxidation-reduction cata lysts. It is concluded that further work is well justi fied at the present time. This is based on several points. First, examples of these perovskites do show sufficiently high activities to be seriously considered as practical catalysts for the oxidation of CO in auto emission control. This conclusion has also been very recently stated by Voorhoeve and coworkers (6) from results of survey experi ments. These authors did not note a substantial differ ence in the activities of the LaCo03 and LaCrOj catalysts, contrary to the results of this work. They also report a degradation in catalytic activity with time at low tem peratures, which would be expected from our work if their reactant mixtures contained trace amounts of water. If the conclusion is accepted that examples of these perov skites do offer a potential for practical use, the desir ability for initiation of further work is very strong. A wide range of opportunities exist because of the extensive range of compositional changes which are available for miximizing the properties desired in practical catalysts. The option of widely modifying the compositions makes these materials attractive ones to study from a 115 fundamental viewpoint in which the goal is a better under standing of the role of defect chemistry in heterogeneous catalysis. Towards this objective, a better understanding of the bulk defect chemistry is desired along with work which relates the bulk defects to surface or near surface defects. Rates required for establishing gas-surface-bulk catalyst equilibria as function of temperature, gas, and catalyst composition are desired. These studies will be more complicated to interpret than ones based on doped and undoped Zr02, or Ce02 *n whi-ch th® bulk defect chemistry is moderately well understood. It is recommend ed that studies of heterogeneous catalysis be continued for these materials. Besides the directions for studying oxide catalysts as probably best exemplified by prior work with NiO, it is suggested that attention be given to the electrical and catalytic properties of the perovskites as a function of doping which affect oxygen activity of the oxide. Lastly, it is suggested that a significant frac tion of future work on the perovskites be directed towards heterogeneous catalysis of reactions involving nitrogen monoxide, the other nitrogen oxides, and ammonia. A good NO reduction catalyst is badly needed for application to automobile emission abatement. The results described in previous sections of the thesis suggest the perovskites 116 may be useful in this application. Another area of interest is exploration of catalytic reactions of hydrocarbons using the perovskite oxides. Such studies should also include inspection of oxidative dehydrogenation of hydrocarbons. Some brief results have been reported by Libbay and coworkers (3,4) on the hydro carbon isomerization and hydrogenation reaction catalyzed by LaCo03. CHAPTER VI EXPERIMENTAL MATERIAL PREPARATION Supply of Chemicals The reactant streams employed to study the various reactions were prepared by metering required amounts of each reactant into a flowing stream of helium. High pres sure cylinders of each gas served as supply sources of the reactants. The identity and minimum purity of each gas are tabulated below: __________Gas Puritv. minimum Helium, He 99.99592m Oxygen, 02 99.6 Carbon Monoxide, CO 99.3 Carbon Dioxide, C02 99.8 Nitrous Oxide, NjO 98.0 Nitric Oxide, NO 99.0 Hydrogen, H2 99.99 The supplier for all of these gases was Air Products and Chemicals, Co. When use of nitrogen gas was required, it was obtained from the laboratory wall service, Minimum purity of the nitrogen was 99.592m. Soluble nitrate salts were the primary sources of 117 118 all elements used in preparation of the oxide catalysts. From such solutions the oxides of each element were pre pared by thermal decomposition of (1) the nitrate solu tion, (2) an oxalate precipitated from nitrate solution, or (3) the hydroxide or carbonate precipitated from the nitrate solution. A variety of manufacturers were purchase sources for these reagents. The reagent, supplier, and supplier's label of quality are tabulated below: Reagent Supplier Strontium nitrate Mall. Cerious nitrate R.0.1. Chromium nitrate J.T.B. Cobaltous nitrate J.T.B. Nickelous nitrate J.T.B. Lanthanum nitrate A.I.V. Thorium nitrate R.0.1. Lanthanum oxide A.I.V. Zirconium oxide (Zircoa A) Z.C.A. Stabilized zirconium oxide (Zircoa C) Z.C.A. Label of Quality Analytical Reagent Baker Analyzed Baker Analyzed Baker Analyzed Lot #9199 99.95% Lot #30570 119 Supplier Abbreviation Index Mall. - Mallinckrodt Chemical R.0.1. - Research Organic/Inorganic Chemical J.T.B. - J. T. Baker Chemical A.I.V. - Alfa Inorganics, Ventron Z.C.A. - Zirconium Corporation of America Preparation of Catalysts, Survey Experiments The initial survey experiments on catalytic oxida tion of carbon monoxide included inspection of ZrC^* NiO, Zr0.94Ca0.06°1.94' ^ O . S S ^ O . I S ^ ^ S ' and LaCo03* ' Ihe zirconium oxides and nickel oxide are commercial prepara tions which were annealed 24 hours in air at 1200°C in an Al2Og boat. The ThQ 351^0 925 and LaCo03 were pre pared in the laboratory by thermal decomposition of nitrate solutions. Preparation of catalysts by nitrate solution decom position is a simple preparative scheme, but it has certain disadvantages in accuracy and convenience of execution. A brief description of this technique is presented and the deficiencies are noted: 1) The amount of each nitrate salt was computed from batch size and the listed formulation for each compound including the hydration. For these preparations, the empirical formula 120 assumed were La (NO3) 3 • 6H20, Co (NO3) 2 ’ 6H20, Cr(NO3)3*9H20, and Th(NO3)4•6H20 . The accuracy of the resultant preparation is then limited by the accuracy with which the water of hydration is known. The degree of hydration may vary with storage conditions. 2) Weigh nitrates and place into a 34 mm OD X 250 mm length quartz test tube and add water to form a slush. The inclined test tube was heated with a Fisher burner until the solution had been evaporated to dryness and a majority of the nitrates decomposed. 3) Contents were removed from the test tube, ground to a fine powder in an agate mortar and pestle. The powder was then returned to the quartz test tube and heated in air at 800°C for 12 hours to complete decomposition of the nitrates. 4) Material again was ground to a fine powder, transferred to an AI2O3 boat and heated in air at 1200°C for 24 hours. The material was again ground to a -325 mesh powder and placed in storage. Preparation of Perovskite Structure Oxide Catalysts The primary objective of this research was the examination of the oxides, LaCo03, LaCrC>3, and LaNi(>3 in 121 pure form and with substitutions of 10% of the La by Ce or Sr. Before preparation of these compounds, it was desired to improve upon the accuracy and convenience of execution inherent in the nitrate decomposition technique. Selected citations on preparation of LaCoO^ (10,27,28,35- 41), LaCrOj (2,35,41-43), and LaNi03 (44-47) were found after a review of literature listed by Chemical Abstracts. After examination of the various preparative schemes de scribed, several important features for a successful pro cedure became evident: (1) use of La£^ 20^)3 as a source of La due to the superior storage stability of the oxalate; (2) use of a (NH4)2C03 precipitate of Ce, Sr, Ni, Co, and Cr as satisfactory sources of these elements; and (3) use of homogeneous small particle mixtures compressed into dense tablets prior to the final series of heatings. The reaction temperatures and times required are primarily a function of particle size and gross homogene- iety of the initial solid mixture. Experience during this work demonstrated the following reaction conditions to be adequate: LaCo03 series 48 hours in air at 1250°C LaCr03 series 48 to 72 hours in air at 1250°C LaNi03 series 96 hours in air at 800°C. These reaction times are split into 2 or 3 segments during 122 which material is ground and repressed into dense tablets. A large supply of La2 (C204)3 was prepared so that during preparation of numerous catalysts a single source of supply for La could be utilized. The preparation of the oxalate is presented briefly: 1) La2°3 powder was dissolved in a slight excess of HNO3, then neutralized to a pH of 3 to 7 with NH^OH. 2) Precipitate lanthanum oxalate by gradual addi tion of a 100% excess of ammonium oxalate. 3) Filter and wash the precipitate with H20 until pH of spent wash is about 5. 4) Dry the product in an air oven 24 hours at 240°F, then grind to a -200 mesh powder in a Pyrex mortar and pestle. The supply of La2 (0304)3 stood for 60 days exposed to the laboratory atmosphere. It was then transferred to a CaCl2 desiccator for storage. After 3 weeks' storage in the desiccator the material was analyzed for La content. The assay was performed by ignition of the oxalate sample con tained in a Coors porcelain crucible for 6 hours in air at 900°C. La203 was assumed to be the only product present after ignition. Result of the assay was 42.8 ± 0.1%w La as compared with the theoretical assay of 48%w for 123 hydrated lanthanum oxalate having the formula La2(C204)3-2H20. The remaining reactants were prepared by precipita tion of the carbonate or hydroxide from nitrate solution with ammonium carbonate solution. The quantities of each stock prepared was determined from requirements for a sup ply of 10 to 15 grams of each transition element, and 3 to 5 grams of Ce and Sr. The preparation of carbonate preci pitates of Ce, Sr, Co, and Cr were performed by the following technique: 1) Dissolve a 100% excess of in water. 2) Dissolve required amount of nitrate salt in water and add dropwise to the rapidly stirred carbonate solution. 3) Filter and wash the precipitate with water. 4) Dry the material in an air oven at 240°F for 12 hours, then grind to a -200 mesh powder in a Pyrex mortar and pestle. The preparation of the Ni stock was modified due to the formation of a soluble ammonium complex in the excess ammonium carbonate solution. Precipitation of the nickel carbonate was successful when ammonium carbonate was added dropwise to a rapidly stirred solution of nickelous nitrate at 65°C. An additional drying of the Co stock of 124 12 hours in air at 350°C resulted in partial decomposition of the material to the oxide. The stocks were stored in CaCl2 desiccator immedi ately after preparation, except for the Co stock which stood in the laboratory atmosphere for 60 days before desiccator storage. Assay of these stocks was similar in execution to that used for the analysis of hydrated La^CC^O^)^. The results of these determinations with in clusion of the gravimetric factor for elements from their oxides are presented below: Stock Ignition Product Stock Assay Oxide Assay Cr Cr2°3 55.1 + 0.1%w Cr 68.4%w Cr Co C03O4 72.1 + 0.05%w Co 73.4%w Co Ni NiO 56.0 + 0.05%w Ni 78.5%w Ni Ce Ce02 78.5 + 0.05%w Ce 81.4%w Ce Sr SrO 61.1 + 0.3%w Sr 84.57tw Sr The catalysts were prepared by combination of oxa late and carbonate stocks in proportions necessary to yield 10 grams of product oxide. Presented in Table V are the mass of each element required for preparation of the 10 gram lot. Based upon these specifications and the assay of preparative stocks the amounts of each stock was computed. The data on preparation formula are presented in Table VI. Due to the availability of an adequate 125 Table V — Element Requirements for Catalyst Compounds CeQ iLao 9M03 seri-es/ 19 gram batch Element Ce La M Cr Co Ni 0.584 0.567 0.567 5.212 5.065 5.065 2.168 grams 2.388 2.381 Formula Weight 239.8 gm/g-mole 246.8 246.5 LaM03 series, 10 gram batch Element La M Cr Co Ni 5.813 5.649 5.655 2.176 grams 2.397 2.390 Formula Weight 238.9 gm/g-mole 245.8 245.6 SrQ iLao 9M03 s®*ifis, 10 gram batch Element Cr Co Ni Sr La M 0.376 0.365 0.365 5.363 5.213 5.213 2.313 grams 2.457 2.448 Formula Weight 233.0 gm/g-mole 239.9 239.7 126 Table VI — Catalyst Preparation Formulas Ceo.lLao.9MO3 series, 10 gram lot of compound Element Ce Stock La Stock M Stock Cr 0.745 12.191 3.940 grams Co 0.721 11.841 3.311 Ni 0.723 11.838 4.254 LaM03 series, 10 gram lot of compound Element La Stock M Stock Cr 13.588 3.949 grams Co 13.193 3.325 Ni (Prepared from nitrate decomposition) SrQ iLao 9M03 series, 10 gram lot of compound Element Sr Stock La Stock M Stock Cr 0.616 12.536 4.049 grams Co (Prepared from nitrate decomposition) Ni 0.598 12.182 4.373 127 supply of Lag 9SrQ iCo03 an<* LaNi0. j from previous prepara tive work, these two materials were not prepared by the present technique. X-ray analysis data, microscopic in spection, and electrical conductivity data did not provide any reason to void these preparations. Preparation of the catalyst from above reactant stocks is capable of producing oxide mixtures accurate in composition to better than 1%, a level assumed satisfactory for purposes of this research. The work on assay of stocks by ignition to 900°C revealed no problems associated with violent decomposition of the initial charge as the material dehydrates and the oxalates and carbonates decompose. This preparative technique thus offers successful solu tions to the primary problems with the nitrate decomposi tion procedure. The synthesis steps for preparing the catalysts from these stocks is presented briefly below: 1) Combine preparative stocks and bulk mix in a Pyrex beaker with a spatula. 2) Complete mixing in a Diamonite mortar and pestle. Material is ground to a very fine powder. 3) Transfer material into a covered 50 cc Pt crucible. 4) Decompose the stocks by ignition in air for 128 12 to 18 hours at 900°C. 5) Grind and mix the product in a Diamonite mortar and pestle to a -325 mesh powder. 6) Compress powder into a 25 mm diameter by 2 mm thick tablet using no lubricant and a load of 15,000 lbf. A die fabricated from mild steel is used in formation of the tablet. 7) Transfer material to a 20 mm wide x 250 mm length quartz boat. The boat is lined with a 0.001 inch thick Pt foil that is folded over to form a cover for the boat. 8) Ignite to the appropriate temperature in air for a period of 24 hours. 9) Repeat sequence of steps (5) through (8) until completion of the required reaction time. After completion of preparative reaction, a portion of the sintered tablet was saved for fabrication of electrical conductivity specimens. The remaining material was then ground to a desired range of particle sizes. The Diamonite mortar and pestle was used for grinding, and a 2 inch diameter glass sieve set used to classify material by the particle size. The excess material may be recycled through the preparative process to increase recovery of material of the desired particle size. 129 Alternative Preparation Techniques The preparation of catalysts is of primary import ance in this research program; thus, a significant frac tion of the effort was directed towards consideration of this topic. Several aspects of accuracy of preparation, convenience of technique, and control of product purity have been touched upon in previous discussions. Several explorations on alternative techniques have been examined to approach each problem. Nitrate Decomposition The decomposition of nitrate solutions prepared from weighed mixtures of crystals is unsuitable because of variability in the state of hydration. This diffi culty is large overcc-ie by storage as a saturated solu tion. Inspection of the literature data (48-50) and observations made in this work on solubility of La(NO^)3 in water indicate that caution must be exer cised on assessment of the long term storage stabili ties. Selected solubility data are presented in Figure (29) to illustrate these uncertainties. Apparently this variability is due to the existence of several hydrates of LatNO^)^. Periodic analysis of solutions and storage in an ice/water thermostat should make the saturated nitrate solution a workable satisfactory preparation source. 130 Figure 29 — Solublity of LaCKO^)^ in Water. 62 CO « 58 *56 54 40 20 25 30 15 Temperature, C Symbol Data Source A, □ Friend (48) O Latimer and Hildebrand (49) O International Critical Tables (50) • This work 131 Additional work on decomposition of nitrate solu tions has shown that a more controlled decomposition process results from use of a large diameter shallow evaporation dish, and the more uniform heating of con tents by use of an electric heating mantle or an air oven. Observation of some etching of the Pyrex glass ware after decomposition at 350°C suggest advisability of fused silica or vycor glassware during this process. Precipitation Techniques Because of the advantage of hydroxide or carbonate precipitates in storage stability and simplification of the decomposition procedures, precipitates of all elements except lanthanum were used during this research. The disadvantage of this technique is the high losses of transition metal cations due to their soluble ammonium complexes. This precludes use of (NH4)2C03 for coprecipitation of the elements for cata lyst preparation. Search for an alternate precipitation agent resulted in examination of oxalate salts produced by precipita tion of the elements from nitrate solution with 10%w h2c2°4 solution. Observations of precipitation trials indicated only the LaCo03 compound with either Ce or Sr doping were successful. Microscopic examination of coprecipitates of La and Co after ignition in air at 132 750°C for 3 to 4 hours provides a means of qualitative evaluation of this technique, due to the difference in color and shapes of crystals from the La and Co oxalates after decomposition. Since the size and shape of crystals is retained after the ignition of the oxa lates, one may determine the origin of coprecipitated crystals by the size, shape, and color. The predomi nant crystal form present in the coprecipitate is that derived from cobaltous oxalate. These crystals were smaller and less well formed than that produced solely from a cobalt nitrate solution. Those light colored crystals associated with a lanthanum oxalate origin were poorly formed and seldom physically separated from the dark colored crystals of cobaltous oxalate origin. These observations indicate that oxalate coprecipi tation leads to more than a mechanical mixture of the individual oxalates. Achievement of a true homogeneous coprecipitate is not attained, but the procedure does promote a high degree of mixing at the microscopic level. This certainly provides an advantage over pre parations from bulk mixtures of components, even when further mechanical grinding and mixing operations are to be performed. Considerable care was exercised to avoid contamina 133 tion of these preparations. This required use of am monium salts for stock preparation, since volatile residues are produced upon ignition of this reagent. Relaxation of this imposed requirement would allow use of KOH and K2CO3 as precipitation agents. The level of contamination of products by potassium entrapped in the precipitate or not removed in the wash operation is un known but should be low. Present speculation is that the effect of this level of impurity should be insigni ficant when Group II or IV element substitution for La is above a few percent. Qualitative observations on precipitation of CrCl^ and NiCl2 from 75°C solutions with an excess of KOH, and with a KOH/K2CO3 mixture, revealed production of very fine particle size precipitates that can be satisfactorily filtered; with Whatman No. 50 filter paper the filtrates are clear and colorless. Thus, these agents appear to provide a quantitative copreci pitation and yield solids having desirable character istics for preparation of the oxide catalysts. Solubility product data for the lanthanide series rare earths (51) validate the assumption of a quantita tive precipitation from alkaline solutions. The alka line earth elements do not satisfactorily precipitate, thus, modification of the precipitation agent to a 134 KOH/KjCO^ mixture. Freeze Drying of Nitrate Solutions The techniques used for preparation of the catalysts and those additional procedures discussed previously are satisfactory for compound preparation, yet a re maining deficiency is the low specific surface area resultant from extensive sintering occurring at the reaction temperatures required for compound formation. Typical specific surface areas of the -325 mesh powders is about 1 m2/gm. Recently reported in the literature is the prepara tion of the perovskite compounds through use of a freeze drying technique (7). Apparently, the particle size and homogeny of the freeze dried mixtures is such that solid state reactions required for compound forma tion may be performed at about 500°C within a reasonable time period. The patent abstract describing this pre paration state production of material having a specific surface area of about 40 m2/gm. CHAPTER VII EXPERIMENTAL APPARATUS DESIGNS Preliminary Screening Experiments The purpose of the initial screening experiments was to identify gross differences in the catalytic activi ties of the various oxides. Apparatus used for this were relatively simple as will be evident from the following discussions. Flow System A schematic diagram of the reactant stream prepara tion manifold, laboratory reactor and furnace control, and the gas analysis manifold is presented in Figure (30). The carrier gas flow rate was controlled by a needle valve, and the reactant gases metered into the helium stream through 0.004 inch ID steel capillary tubes. Provisions were made for use of either helium or nitrogen carrier gas. The reactant stream flows through the reactor and passes into a manifold system for analysis of CO2 concentration by collection in a cold trap. Waste gases then were conducted into an exhaust hood for disposal. Reactor Design A sketch of the flow reactor used in these 135 — & n J t±j M Figure 30 -- Experimental Apparatus, Preliminary Survey Experiments. A Helium carrier gas supply B Cold trap to dry helium C Nitrogen purge gas supply D Reactant gases supply E , Capillary f lov/meters F Reactant stream flow measurement port G Reactant stream flow indicator H Reactor and electric furnace I Port for by-pass to exhaust K Port to exhaust L Measurement cold finger M Mercury manometer 136 137 experiments is presented in Figure (31). The reactor was constructed of clear quartz tubing to provide an inert container for the high temperature studies. At operating flow rates of 100 cc/min (25°C and 1 atm), the residence time in the reactor preheat section is about 2 minutes. This lengthy time provided adequate opportunity to preheat the gases to reaction tempera ture before passage over the catalyst. Temperature gradients in the reactor were minimized by placement of a stainless steel pipe about the reactor. The reactor was heated in a 1-3/4 inch ID by 24 inch length tube furnace manufactured by Marshall. Temperature control was performed by a Leeds and Northrop Speedomax H with a series 60 control unit which operated a mercury sole noid for on-off control of power to the furnace. Since no thermocouple well is present in the reactor, the reactor temperature was taken to be the mean of values observed at four equally spaced intervals along the catalyst bed section between the reactor and the pipe insert. It was observed that these temperatures rarely varied more than +3°C from the mean value. No analyses were performed on the mass transport to the surface of the catalyst bed or inside the bed to evaluate the importance of these effects on the kinetic observations. Since major differences in catalytic Figure 31 — Experimental Reactor* Preliminary Survey Experiments. 8 OD 6 ID 18 OD 15 ID 160 25 -* 570 3 Height Placement of Reactor in Furnace 304 SS Tube 350 Inlet Exit 250 Measure Furnace Temperature Dimensions in at These Locations g Millimeters 139 activities were to be explored, it was thought that use of a constant volume catalyst bed of uniform prepara tion would adequately serve the purpose of these screen ing experiments. Gas Analysis System The analysis of fractional conversion of CO to CO2 was performed by determination of the quantity of C02 produced during a measured time period. The collection of CO2 from the product stream was performed by passage of gas stream through a cold trap immersed in a liquid nitrogen thermostat. After collection period the pro duct stream flow was diverted from the collection cold trap; the manifold from the cold trap to the manometer was evacuated while the liquid nitrogen thermostat re mained about the cold trap. The C02 was then trans ferred by diffusion into a measuring cold finger by re arrangement of the liquid nitrogen thermostat. The cold finger and manometer were then isolated, the C02 permitted to warm to room temperature, and the pressure observed using the mercury manometer. From previous calibration, the volume of the entrapped gas for a given manometer reading was known. By comparison of the rate of C02 production evaluated from the above observations with the CO feed rate, the fractional conversion of CO to C02 was computed. A dual system of 140 cold traps permitted efficiency in execution of dupli cate observations for each reactor condition. Experimental Procedure Identical operating procedures were employed during the screening experiments to assist the interpretation of the kinetic data. Due to simplicity of the reactor design and the system's behavior, uniformity in the execution of experiments was quite essential. The operating schedule is presented briefly below: 1) The reactor catalyst chamber is charged with -325 mesh powder which is distributed to ap proximately half fill the chamber without com paction. 2) The reactor is assembled and positioned in the flow stream. The reactant stream is initially prepared using nitrogen carrier gas. The fur nace is set to operate at 600°C for 24 hours, during which period the catalyst is conditioned. 3) After the conditioning period, the carrier gas is changed to helium. Furnace temperature is reset to the desired level. 4) A series of product stream analyses were per formed at various reactor temperatures to ade quately define the fractional conversion versus 141 temperature performance of the catalyst. About 2 hours was required for a steady state temper ature after a 10°C change in furnace set point. 5) Upon completion of observations, the reactor is emptied and cleaned with aqua regia, rinsed, and dried. Experimental Errors The simplicity of the experimental apparatus proved not to limit the accuracy of the gas analysis and re sultant accuracy of fractional conversion computations. Flow through the capillary flow meters was measured by displacement of Hg from a gas buret. Stability of the flow meters was observed over long periods of time to be about +1% of their set value. Assurance that the product collection cold traps operate with a high efficiency was established by ob servation of the CO conversation at aboe 350°C over the LaCoO^ catalyst. The mean conversion observed was 99.4% with a precision of +1.3%. The expected conver sion was 100%, since a 99+% conversion was observed for this catalyst at about 300°C. Thus, it appears that the cold trap efficiency is greater than 99%, and the un certainty present in fractional conversion data is comparable to the relative error present in the CO flow rate. A conservative estimate on the accuracy 142 of conversion values above 10% would be +3% of the ob served conversion value. The large number of manual operations necessary for the execution of the experimental runs introduced an additional source of inaccuracy. Its effects were in spected. From the 72 duplicate sets of observations made during the program, a cumulative polygon of pre cision in fractional conversion values was constructed. This result is presented in Figure (32). As one may note, 75% of the observations occur within the expected limits of error, with 90% of the data within the +5% range of precision. Primary Kinetic Experiments The apparatus used for examination of catalytic activities of the oxide catalysts evolved from that em ployed in the preliminary experiments. Design changes were instituted to provide improvements ins (1) prepara tion of the reactant stream mixtures, (2) accept use of a gas chromatograph for stream composition analysis, and (3) the use of an improved reactor design and furnace con figuration. Completion of these modifications permitted observations to be made which more accurately represented the actual consequences of the reaction rates associated with surface processes on a given catalyst. Cumulative Fractional Frequency 143 Figure 32 — Precision Distribution of Fractional Conversion. 0.8 0.6 0.2 Precision* Fercent of Mean Conversion 144 Flow System A schematic diagram of the reactant stream prepara tion manifold is presented in Figure (33). The carrier gas flow rate was controlled by a combination of supply pressure and adjustment of the Nupro needle valve. A cold trap provides for removal of impurities from the helium stream. The floating ball type flow meter was used only for a qualitative observation of reactant stream flow rate. Through arrangement of the manifold valves, water vapor may be added to the reactant stream from the saturation bubbler. The reactant gases enter the flow system through the provided entry ports. Mix ing of the reactant stream was effected in a 15 cm length of 8 mm ID tube packed with 1 mm diameter Pyrex beads. The stream sampling ports and reactor station flow system schematic is presented in Figure (34). The gas manifold system was operated at about 2 psig, thus pro viding for sample flow to the chromatograph. A cold trap was placed immediately before the reactor to allow for removal of water from the reactant stream when bone dry test conditions were desired. The reactor station is designed to allow easy replacement of reac tors of various designs in one of several furnaces. A pressure head for product stream sample flow is Figure 33 — Gas Manifold System, Reactant Stream Preparation, Primary Experiments. J A Carrier Gas Supply B Regulator, Air Products, # 02-1000 C Valve, Choke, # 321-2M4B D Valve, Whitey, # 1GS4 E Valve, Rupro, i"2SA F Pressure Relief, Air Products, # E33-4-1121 G Connector, Cajon, # 4UT-6 H Cold Trap 1 Flowmeter, Fisher-Porter, # 36-541-04 J Uater Saturation Bubbler K Reactant Supply Ports h Reactant Mixing Chamber, Flow to Reactor M By-paS3 Flow to Exhaust Figure 34 — Gas Manifold System, Sampling Forts and Reactor Station, Primary Experiments* To From mixing chamber < t A Reactant Stream Sample Port B Cold Trap C Reactor and Furnace D Product Stream Sample Port E Manometer F Product Stream Backpressure Bubbler Sample Port Components a Adapter, Cajon, # 4UT-1-2 b Valve, Uhitey, # 1GF2-A c Adapter, Cajon, # 2UT-1-2 146 147 provided by a bubbler arrangement on the product stream. Reactant Stream Preparation Several of the necessary operating procedures and conditions need to be reported on the reactant stream preparation. After experience in the system operation, the supply pressure of helium to the manifold was set at 30 psig, and the flow rate control performed by use of the needle valve. Experience also revealed no ap parent short range effects of not employing the He stream cold trap during kinetic experiments. A tap water filled dewar flask is placed about the saturation bubbler to insulate this unit from variations in room temperature. Observation of this temperature is made after completion of an experiment. Generally, the temperature was 20 ± 1°C. The control of reactant gases flow rate was per formed by use of 304 stainless steel capillary tubes located downstream of the cylinder regulator. A schematic diagram of this device is presented in Fig ure (35). The high pressure adaptor is a length of \ inch stainless steel tube in which a brass plug was brazed and a 0.012 inch diameter hole was drilled. The low pressure adaptors are prepared from Swagelok #100-R2 brass fittings which were plugged and 0.012 Figure 35 — Capillary Tube Flotaneter Design High Pressure Adaptor Tube Low Pressure Adapter Fitting 0.004" ID 304 SS Capillary C = ® — Alignment Rod Alignment B lock Teflon Tube 0.075" CD 12/5 0-Ring Socket Joint Reactant Port Adapter Fitting 149 inch holes drilled to accept capillary tube. Depending upon the desired flow rate, the length of capillary tubing is selected. During this work a 24-inch length of 0.004 inch ID tube satisfactorily provided 1 to 5 cc/min of CO or O2 at 10 to 30 psig supply pressures. During preparation of the flow meters, it was found satisfactory to cut the capillary tube with a pair of electrical scissor snips. The capillary tube is sealed to adaptors by a few drops of Apiezon W wax melted by an electric hot air blower. Teflon tubing (AWG #24, Markel and Sons) was used as a flexible conduit from the flow meter to reactant entry port. The reactant gas flow rates were measured with a 2 cc volume soap bubble type flow meter operated at the gauge pressure maintained in the reactor flow system during the kinetic experiments. This is done to compen sate for flow rate variations due to presence of a variable downstream pressure in the capillary tube. The reactant gas flows are then corrected to 0°C and 1 atm for computation of molar flow rates. Reactor Design The reactor used in these observations was an iso thermal plug flow type. Results of previous kinetic observations and anticipated flow conditions were con sidered in the design of the present reactors to assure 150 that restrictions of heat and mass transport would not limit validity of the observations on reaction kinetics. The reactor designs are presented in Figures (36) and (37) for the LaCo03 experiments, and in Figure (38) for the LaCrO^ observations. Preheating of reactant stream to the reaction temperature was performed in the coil or annular sections of the reactor. Reactant gases then pass downward through the catalyst bed supported by a porous quartz or Pyrex disc depending on the re actor. The two Pyrex reactors were deliberately differ ent in design to check the significances of the preheat coil dimensions and diameters of the reaction chambers. Design of the quartz reactor reflected modifications made to simplify its fabrication and to be compatible with the furnace used for operations above 600°C. The performance of the preheat coil was checked by inspection of the Gratz solution to the problem of laminar flow heat transfer in a circular tube with a fully developed velocity profile and uniform fluid temperature at entrance to the heating zone. The solution to this problem is taken from Knudsen and Katz (52). This expression provides for computation of fluid temperatures as a function of radius and length in the heating zone. The solution is presented below: Figure 36 — Laboratory Reactor, Pyrex Glass, 20 nm Diaxaeter Chamber. 151 Open end of tube to load reactor Thermocouple Probe 25 Prepare from Corning Glass #36060 Funnel, Coarse Filter 20 Turn coil on 25 mm ID mandre ^ Pyrex tube L o Dimensions in millimeters Figure 37 -- Laboratory Reactor, Pyrex Glass, 10-ram Diameter Chamber* 152 . . . Open end of tube 7 “ “ Thermocouple Probe 25 100 Prepare from Coming Glass # 36060 Funnel Medium Filter turn coil on 25 mm mandrel q TH f t od Pyrex trube Dimensions in millimeters Figure 38 -- Laboratory Reactor* Quartz Glass 500 200 150 v__ 15 ID T„. 18 OD Tube 12/5 0-Ring Connector Detail of reaction chamber 20 ID / 23 OD Thermocouple Probe 3 ID Tube 5 OD luDe J? Tube 13 OD Tube Fritted Disc 15 to 40,u/>w. Pore Diameter 150 175 Dimensions in millimeters 153 where, T(r) = fluid temperature at radius r and length x Tw a wall temperature in heating zone Tm = fluid temperature at entrance to heating zone cn* Bn* ^n( £— ] = constants determined by boundary 'rw I conditions, tabulated in Knudsen and Katz Npe = Peclet number, Cpfu(2 rw ) A rw = radius of tube r = radius x = length into heating zone. Evaluation of the expression for the reactor design geometry: x - 60 cm, r = 0 cm, rw = 0.2 cm and the helium physical properties evaluated at the average reactor conditions of 100°C and 1 atm, Cp = 1.24 cal/gm-C (Rohsenow & Choi, 53) P » 1.3 x 10-4 gm/cc (ideal gas law) —4 k = 4 x 10 ca/sec-cm-C (Rohsenow & Choi, 53) ji = 2.2 x 10”^ poise (Perry, 4th ed., 54) and the flow conditions of 300 cc/min of He at the mean temperature of 100°C. Checking for laminar flow condi tions by inspection of the Reynolds number, then evaluation of Equation (1) yields a centerline tempera 155 ture equal to the wall temperature to within less than 0.01°C. A similar calculation performed for the quartz reactor design, substituting the equivalent diameter for an annulus and appropriate reactor dimensions, also results in a confirmation of the adequacy of the design. Mass transfer from the bulk gas to the surface of the catalyst particle is analyzed by comparison of mass transfer coefficient with observed reaction rate con stant. A correlation for mass transfer coefficient for forced convection around a sphere is given in Bird, Stewart, and Lightfoot (55) as, "Sh ' [da^] * 2-° + °-6 where, NSH s Sherwood number 9 = mass transfer coefficient d = sphere diameter DA B = diffusivity of A into B - fluid bulk velocity Pf = fluid density Mf = fluid viscosity evaluated for the assumed conditions of: CO at 2%m, temperature of 200°C, pressure of 1 atm, gas velocity of 1 cm/sec, and a particle diameter of 0.01 cm. The dv„ Pf V 2 Mf _Pf°A,B V 3 (2) 156 physical properties of He are evaluated for the reactor condition of 200°C and 1 atm pressure, = 9.8 x 10~5 gm/cc (ideal gas law) Cp = 1.24 cal/gm-C (Rohsenow & Choi, 53) Mf = 2.6 x 10~5 poise (Perry, 4th ed., 54) kf = 4.1 x 10"4 cal/sec-cm-C (Rohsenow & Choi, 53) D = 2 cm^/sec (Bird, Stewart, & uo'He Lightfoot, 55) Evaluation of Equation (2) yields a Sherwood number of 2.1. The mass transfer rate from the bulk gas to the particle surface is expressed by, rate = 0 (cb - cs) (3) where, cb - concentration of reactant in bulk, g-mole/cm^ c8 = concentration of reactant at particle surface. Evaluation of the mass transfer coefficient, , yields a value of 400 cm/sec. A first order reaction at a uniformly accessible catalyst surface is expressed by, rate = kabs(cs) (4) where, kabs = f^rst order rate constant, cm/aec. External diffusion will not limit overall reaction rate when the first order rate constant satisfies the condition, 0 > > ^abs (5) Observed reaction rates for CO oxidation on LaCo03 at 200°C are about 10”^® g-mole CO/sec-cm2, which leads tion of Equation (5) is easily satisfied and external diffusion from bulk gas it not a rate limiting process. The remaining aspect of mass transport is analyzed by computation cf the intraparticle diffusion effects. The effectiveness factor for a spherical catalyst particle and a first order reaction is computed by the expression taken from Satterfield (56), to an estimate of ka^s as 10~^ cm/sec. Thus, the condi ^ ~ “57 ITANH (4>s) " T 7 (6) where, and where R - particle radius, cm Sv = volume specific surface area, cm2/cm3 De^ = effective intraparticle diffusivity, cm /sec The effective diffusivity is computed by correlation presented in Satterfield, 158 where, 0 = porosity of particle T = tortuosity factor of pore structure. From bulk density measurements the porosity is estimated to be about 0.3, but no value is known for the tortuo sity factor. A value of 10 is assumed for ^ and the sensitivity towards reaction rate noted. The parameter, <£s, is then evaluated to be about 0.13 — 3 2 2 3 using values of kabs =10 cm/sec and Sv = 10 cnr/cm . Thus, evaluation of the effectiveness factor leads to a value of near unity, and the conclusion is that intra particle diffusion is not limiting the observed reaction rate. Reactor Furnace Design and Control An inconvenience observed during the preliminary work was the slow transient response of the electric tube furnace used to control reactor temperature. Since this furnace requires 6 to 8 hours to cool from 600°C to 200°C, catalyst transients were difficult to observe. In addition, the versatility of the experi mental program was limited unduly by the slow tran sients. For experiments below 500°C, the imposed service limit for Pyrex glassware reactors, a pair of tube furnaces were constructed to provide rapid transient 159 response, with a 4 to 6 inch hot zone of +3°C of mean temperature. These requirements were satisfied with the design described in Figure (39). This design fea tures 3 heating elements, a central zone 5 inches long, and two end zones of 3-1/2 inches in length. A second feature is the use of relatively poor insulation about the central 8 inches of the furnace to induce high radial heat losses to the water cooled exterior tube wall. Although intended for automatic power control to the central heaters, and manual control of the end zones, the control scheme described in Figure (40) proved to satisfy the design criteria on performance. Balance of the end zone heaters is achieved by adjust ment of the 500 ohm rheostat in parallel with the bot tom element. Final smoothing of the temperature profile is obtained by placement of a 41 mm ID x 49 mm OD x 125 mm length Inconel tube in the central section of the furnace. All control and monitor thermocouples are located in slots milled in the external surface of the Inconel tube. The furnace used in conjunction with the quartz reactor required service capability to 1000°C maximum temperature. For this service the 1-3/4 inch x 24 inch length Marshall tube furnace described in the prelimi nary catalysis work was used. The temperature control Figure 39 — Laboratory Reactor Furnace 160 1 Alumina core, 50mm ID x 65mm OD, 12/inch ext* thread 2 Aluminum silicate fiber insulation 3 Steel tube, 5 inch ID x 5 3/8 inch OD 4 Copper tube coil, 5/16 inch OD 5 Bubble alumina insulation, -10 to +20 mesh 2% 3-j °a° V J m m Selector IN lie ter Furnace Controller Power — V ([Package 115 VAC o Neut Figure 40 -- Reactor Furnace Power Control Circuit. Controller: L&N # 6261-2110 Power Package: L&N # 11906-223 MV Meter: Honeys/ell # 105X211 Thermocouple placement Furnace control - 6" depth Monitor probes - 3,5,7,9" depth 161 162 system was also identical to previous application. Gas Analysis System The composition analysis of reactant and product streams was performed with a Hewlett/Packard 7620A gas chromatograph. A schematic diagram of the sample and gas analysis system is presented in Figure (41). Since a switching valve was not available for selection of the stream to be analyzed, this task was completed by use of O-ring connector union of 1/8 inch OD copper line to the sample port valve. A needle valve on the exit of sample line was used to control stream flow to about 6 cc/min. The dual loop gas sample valve serves to meter a 2 cc sample into the chromatograph. The chroma tograph operating conditions used for CO oxidation ex periment analyses are tabulated below: Detector: Thermal conductivity bridge Temperature, 100°C Bridge current, 250 ma Gas Flow: Helium, High purity grade Flow rate, 50 cc/min through each side of thermal conductivity bridge Injection Port: Temperature, 100°C Gas Sample Valve: Temperature, 85°C Pressure, 1 atm Loop, dual 2 cc volume Column Oven: Temperature, 100°C Figure 41 — Gas Analysis Flow System. Reactant Gas Sample Valve 1 - Porapak Q 2 - Molecular Sieve 5A Product ’ ---0 - Q Sample Ports Exhayst & Flow Control Valve General Analysis for CO, C2, and C02i Siritching Valve £"F1ow Control Valve Thermal Conductivity Detector Exhaust 1) Sample flowrate of 6 cc/min 2) Close sample port valve 3) Delay before injection of sample into chromatograph 4) Observe elution of components stcg.aa Reactant Product Delay Period 30 to 40 sec 15 sec Cplurtn _E f f luen£ Molecular Sieve 5A Porapak Q Component Elution Order Oo then CO CO + 02 then C02 <n 164 Columns: 1) Porapak Q, 50-80 mesh, %" x 12' 2) Molecular Sieve 5A, 60-80 mesh, h” x 6' The analysis of stream composition is based upon assumption of a linearity between mass of a component in a sample and the resultant chromatograph peak area. Thus, gas samples must all be collected at identical temperature and pressure so that samples of equal vol ume result in equal mass of component for analysis when concentration of specie in sample loop are equal. These criteria then led a procedure for the temporary termina tion of the sample stream flow and allowed the sample loop pressure to become atmospheric. Experience has shown 30 to 45 seconds of delay are necessary for this process on a reactant stream sample, and about 15 sec onds delay is satisfactory for a product stream analy sis. The original analysis routine was intended to analyze for CO, 02, and C02 in the single sample. This is performed by storage of CO and 02 in the Molecular Sieve 5A column while the C02 elutes from the Porapak Q column. Unfortunately, a fault in the chromatograph piping was detected which could not readily be correct ed. It allowed a leakage loss of components from the Molecular Sieve 5A column section during the storage period. Although very good precision for the CO and 165 O2 analysis were obtained, the results on these analyses are about 15% below those obtained by continuous elution from the Molecular Sieve 5A column. Therefore, the plan was adopted to perform reactant stream analysis only for CO and 02 by observation of Molecular Sieve 5A ef fluent. The analysis of product stream is then executed only for C02 by observation of Porapak Q effluent. A calibration of the chromatograph peak area to component concentration was executed. The results of this examination are tabulated below: Several determinations for the CO calibration lead to an accuracy of about ±3%. Through use of these cali bration factors, the reactant conversions are computed. Since reactions were studied at dilute conditions, no significant volume change of gas stream occurs. This permits the direct computation of CO conversion by observation of CO2 concentration in product stream, since no C02 was conducted into the reactor during kinetic observations. Component Peak area count percent of component CO 5300 4800 6050 166 Experimental Procedure A series of kinetic observations were performed in a similar manner except where special effects were under examination. The general operating procedure used during kinetic studies on LaCoO^ and LaCrO-j catalysts is presented briefly below: 1) Reactor is charged with 250 to 1000 mg of cata lyst transferred from a weighing crucible. Powder is distributed into a uniform bed and reactor is sealed. 2) After positioning reactor in the manifold, the reactant stream flow began, liquid nitrogen thermostat placed on reactant stream cold trap, and furnace set to operate at 400°C for 12 hour s. 3) After conditioning period, furnace control is reset to desired temperature. Final adjust ments of reactant concentrations were made by adjusting the control of the carrier gas or component flow rates. Gas chromatographic analyses were used to judge the component con centrations. 4) A series of kinetic runs were performed to es tablish the temperature dependence of the re action rates. Stream composition analyses 167 performed as required. Approximately 1% to 2 hours were allowed for steady state operation of reactor after a 10 to 15°C change in the re actor temperature. 5) Next, a series of kinetic observations were executed to obtain the reaction rate dependence upon reactant concentrations. Variation of only one component at a time was used to explore this phase of behavior. 6) Next, a series of experiments on the kinetics were completed to observe the effects of H20 in the reactant stream at the component con centrations used in part (4). The temperature dependence of the reaction rate for these condi tions is determined. 7) Upon completion of kinetic observations the re actor is removed from the manifold, opened, catalyst recovered, and the reactor cleaned in aqua regia, rinsed, and dried. Electrical Conductivity Apparatus An apparent relationship between electrical conduc tivity and CO oxidation catalytic activity of oxides di rected the preliminary screening experiments. Thus, to aid the correlation and the interpretation of kinetic data on the perovskites a program of measurement of the DC 168 electrical conductivity was performed. Apparatus Design A four terminal technique was employed for the pre sent study. The conductivity cell is shown in Figure (42). A detailed assembly of the method for securing the sample and contacts is indicated in Figure (43). The specimen is a lmmxlmmxl2mm bar of sintered material sawed from a tablet made by firing pressed discks of powder as is described in Chapter VI. A des cription of sawing and shaping of the specimens follows the discussion of apparatus design. The sample is posi tioned between two blocks upon which 0.010 inch diameter Pt wire contacts are secured. The sample is held secure by the arrangement of slots in the support rod and cur rent lead blocks, and a compressive load from the spring. Current supply leads are secured to the top plane of the specimen at a 9 mm separation, and the voltage drop leads are secured at a 3 mm separation on the central zone of the bottom plane of the specimen. A Chrome1/Alumel thermocouple is secured about 10 mm from the specimen for observation of the temperature in the cell. As indicated in Figure (42), envelope and gas flow arrangements were adapted from several of the kinetic reactors of previous description. Electrical leads are individually insulated with short lengths Figure 42 — DC Electrical Conductivity Cell* C B A 600 C 23/15 O-Rlng Connector 12/5 O-Ring Connector A Current B Voltage C Thermocouple 6 ID 8 OD Tube 15 ID 18 CD Tube Dimensions In millimeters Detail of Samole Holder 1 - Spring 2 - Tension Shim 3 - Current Lead Block 4 - Specimen 5 - Voltage Lead Block 6 - Bearing 6 *-7 7 - Support Rod 10 Figure 43 -- Assembly Detail of Conductivity Cell. 1 - Compression Spring 5 - Voltage Lead Block 2 - Spring Shim 6 - Contact Block Bearing 3 - Current Lead Block 7 - Support Rod ^ 4 - Specimen o 171 of quartz capillary tubes and pass out of the cell through an Apiezon W wax sealed port. Specimens may be examined at various temperatures under controlled atmospheres and various temperatures. The external measurement circuit is presented in Figure (44) . All elements are of standard design and normal assembly for use in measurement of voltages of > 10 juV at source impedances < 5000 ohms. The ac curacy of this conductivity cell and measurement cir cuit was examined by testing the resistivity of a short length of Kanthal resistance wire. A 50 cm length of B&S 20 gauge Kanthal wire was connected in series with a 100 ohm precision resistor to a DC power supply. From voltage drop along the 50 cm length of wire, poten tial across the resistor, and wire diameter, a resisti- —4 vity of 1.44 x 10 ohm-cm was determined. This com- - 4 pares quite well with the value of 1.43 x 10 ohm-cm stated by the manufacturer for room temperature wire. A 10 mm length of this wire was then mounted into the conductivity cell as would a sintered sample be handled. Observations were made at room temperature and resulted — 4 in a value of 1.47 x 10 ohm-cm. Thus, an ability to measure electrical conductivity of a sample to about ±2% was demonstrated. Figure 44 -- DC Electrical Conductivity Measurement Circuit. Switch: 1 Pole, 3 Position Voltage EMF Current EilF 100 'ohm Resistor L&N #403OB Thermocouple EMF Std. Cell liey DC Power Supply HP #6207B DC Kuli Detector L&N # 9834-1 Digital Voltmeter HP #3440A Potentiometer L&N Type K-3 # 7553-5 fo 173 Sample Preparation The specimens used in these observations were pre pared from the 2mm thick sintered tablets produced dur ing preparation of the perovskite catalyst powders. These samples were sintered to a minimum of 75% of theoretical density and are polycrystalline. The speci mens are fabricated from the sintered chips by a series of sawing and surface grinding operations using a pre cision wafer cutting machine. SiC cutting blades were used in sawing (Norton Co., #37C180-ROR30, 4" diameter x 0.010" thick), and Al2(>3 grinding wheels were used to surface grind samples (Norton Co., #39C100-18VK, 6" diameter x 3/16" thick, fine grade). The surfaces are not polished after forming of the specimens, but repre sent the result of either to the above forming opera tions. During these machining operations the sample is secured to the sample carriage fixture with Apiezon W wax. The wax is removed from the specimen by solvent extraction in hot toluene for several hours, then dried in an air oven at 250°F. Measurement of Conductivity Type To complement the observations upon electrical con ductivity, the type of mobile charge carrier was also determined for the perovskite catalysts. The measurement 174 was based upon an observation of the sign of the Seebeck coefficient Using the apparatus shown in Figure (45). The Seebeck effect is the phenomenon of an electric current being produced in a closed circuit composed of two different conductors if one junction of the dissimilar materials is maintained at a temperature different from the other junction. Microscopically the Seebeck effect arises because the chemical potential of the mobile charge carriers in a conductor depends upon the temperature. The presence of a temperature gradient in a material causes a gradient in these chemical potentials unless an electric field is established. The polarity of this field speci fies the sign of the charge of this carrier. For the case of a hole conductor the cold junction shall be observed to be more positive than the hot junction, and visa versa for the case of an electron conductor. A temperature gradient is impressed across the sample by operation of the hot plate at about 250°F and circulation of cooling water at 75°F. Platinum foil con tacts and leads are used to minimize thermal gradient ef fects in the measurement circuit. Observation of an elec tric potential and its polarity which are developed across the specimen was made using a digital voltmeter having an input impedance of 10 megohms. Execution of the test is straight-forward and 175 Figure 45 -- Determination of Sign of Seebeck Coefficient* Brass block, %x%x% inch 2 - Copper cooling coil, 1/8 inch OD, 72 F water 3 - Alumina insulator, 2mra thick 4 - Specimen, 5 x 10 x 2 mi thick 5 - Electric hot plate, 250 F 6 - Pt foil 5 x 10 mm x 0.006 inch thick 7 - Pt wire lead, 0.010 inch diam. x 100 mm length 8 - Cu wire lead, 0.015 inch diam. 9 - Digital voltmeter, HP #3440A 10 megohm input impedance 0 176 apparently requires only normal care in its performance. The specimen may be an irregular shaped chip of sintered material from 1 to 3 mm thick. About 5 minutes was allowed for steady state operation before the sign and magnitude of the potential was recorded. Potential magnitudes ranges from 0.5 to 25 millivolts which was within the ap plicable range of the digital voltmeter. Gas Adsorption Apparatus The desirability of evaluating catalytic activity on a rate of reaction per unit surface area required pre paration of a gas adsorption apparatus for determination of the adsorption isotherm of N2 at about 77°K and computa tion of surface area using the B.E.T. equation (8). The basic assumption in the derivation of the B.E.T. equation is that the Langmuir equation applies to each layer ad sorbed, with the added postulate that for the first layer the heat of adsorption may have some special value, where as for all succeeding layers, it is equal to the heat of vaporization of the liquid adsorbate. A further assumption is the evaporation and condensation can occur only from or on exposed surfaces. Design of the apparatus used in this work is similar to one described by Benson and Garten (57). This particular design permits a rapid evaluation of specific surface areas down to about 1 m2/gm provided 177 sufficient sample is available to yield a total surface area of 10 to 20 m^. Apparatus A schematic diagram of the equipment is presented in Figure (46). The vacuum service was provided by a mechanical pump having a liquid nitrogen cooled cold trap. This allows pumping to about 25 millitorr for the degassing operations. Helium gas for dead volume measurement, and nitrogen gas for adsorption measure ments were obtained from pressurized cylinders and purified by passage through a liquid nitrogen cold trap. The saturation pressure of the nitrogen is determined directly during the experiment. The pressure measure ments were made with a Hg manometer read with a catho- meter to +0.1 mm. The uniqueness of this apparatus design is derived from the technique used to obtain the adsorption iso therm. The objective is to determine the amount of gas adsorbed as a function of the partial pressure of the adsorbate. Using this apparatus design, the adsorption isotherm is determined by observation of the incremental amount adsorbed after pressure in the dose volume is increased in a series of steps. The incremental amount of gas adsorbed is computed using the following equation: Figure 46 -- Gas Adsorption Apparatus A - Cathometer G - N2 Supply Cylinder B - Hg Manometer H - He Supply Cylinder - Dose Volume Bulb - Sample Bulb - M2 Thermometer Bulb - Gas Purification Trap Vacuum Gauge Mechanical Pump Trap Mechanical Pump vent 178 179 An [76O | 82.06 | 295] [PdoseVdose + PbulbVdead pads ^vdose ” vdead j] where the observed variables are defined below: An, mg-moles of gas adsorbed in dose operation Pdose* Pressure in dose volume at start of measurement, torr Vdose' vo^ume of dose chamber of apparatus, cm-* Pbulb* Pressure sample bulb at start of a dose operation, torr Vdead' v°lurae free space in sample bulb, cm^ Pads' pressure in combined dose and sample bulb at equilibrium of dose step, torr. The volume of the dose chamber of the apparatus was determined by transference of CO2 from a bulb of cali brated volume into the cold finger of the dose bulb. The transfer operation was performed by freezing of CO2 in the cold finger by use of a liquid nitrogen thermo stat. Upon expansion of gas to room temperature, the pressure in the dose chamber is measured with the Hg manometer. Computation of the volume was performed using known pressure, temperature and g-moles of CO2 from calibrated bulb and assumption of the ideal gas law. Operating Procedure A brief description of operation procedure is 180 presented for the determination of the adsorption iso therm using N2 and He gas for dead volume determination. This presentation follows below: 1) A weighed amount of sample is placed into the sample bulb. Excess free volume is reduced by placement of glass rods into the long neck of the sample bulb. 2) The sample bulb is attached to the manifold, and then evacuated to 25 millitorr. Degassing is assisted by heating sample bulb to about 350°C for 12 hours, while continuing to pump a vacuum. A small electric tube furnace (1-1/2" ID x 6" length) is used to heat the sample bulb. 3) Valve #2 (see Figure (46)) is closed to isolate the sample bulb from the manifold. A liquid nitrogen thermostat is positioned about the sample bulb and thermometer bulb. 4) N2 is admitted into the thermometer bulb and allowed to condense until bulb . i . . half filled. Appropriate valve opening and closures are performed to connect the thermometer bulb with the Hg manometer. After about 20 minutes, the pressure is observed. This is PQ, the satura tion pressure of N2 at the temperature of the thermostat. 181 5) The manifold is evacuated prior to performing gas dosing operations. Valves are arranged to admit N2 into the dose bulb until a pressure of about 40 torr is attained. Observe the pres sure in the dose with the Hg manometer; this is t > Pdose* Open valve to admit gas into the sample bulb. About 20 minutes are allowed for equili bration of the first dose of gas; following steps only require about 10 minutes for equili brium. Observe the pressure using the Hg mano meter; this is Pa(js. 6) The sample bulb is isolated from the manifold. Note that during the first dose step the value of Pfcuib is zero. After this first step the value of pijU2i)^s the pads value of the preceding dose step. The pressure in the dose volume is increased in stages of about 40 torr, and the gas dosage operations are continued. The B.E.T. equation is accurate between relative pres sures (P/P0) between 0.05 and 0.3; thus, observations are terminated at about 300 torr. Determination of the sample bulb volume is performed by procedures simi lar to first gas dosage step, except that He gas is used. At 77°K, He does not adsorb to a significant amount. 182 From the raw data, one now computes, Va, the volume of gas adsorbed (expressed at 0°C and 1 atm) as a func tion of relative pressure, P/PQ of N2. The B.E.T. equa tion may be rearranged so that a plot of the factor (P/PQ)/ Va (l-P/PQ) is a linear function of (P/PQ). From the slope and intercept values, the volume of gas required to form a monolayer is computed by the formula, Vmono = V(slope + intercept) From an assumption of the surface area coverage per molecule of gas (16.2 &2 for N2) the surface area of the sample is computed. This calculation reduces to, vmono • ^.3 7 ~ , s ~ (sample mass)' m ' REFERENCES 1. Kroger, F. A., "Chemistry of Imperfect Crystals," North-Holland Publishing Co., Amsterdam, 1964. 2. Meadowcroft, 0. B., Br. J. App. Phys., Series 2, 2, 1225 (1969). 3. Libbay, W. F., Science, 171. 499 (1971). 4. Pedersen, L. A., and Libby, W. F., Science, 176. 1355 (1972) . 5. Bauerle, G. L., Service, G. R., and Nobe, K., 161st ACS National Meeting, Los Angeles, Ca., 1971. 6. Voorhoeve, R. J. H., Remeida, J. P., Freeland, P. E., and Matthias, B. T., Science, 177. 353 (1972). 7. Tseung, A. C. C., Ger. Offen. 2,119,702, Chem. Abstr., 76, 213 (1972). 8. Brunauer, S., Emmett, P. H., and Teller, E., J. Am. Chem. Soc., 60, 309 (1938). 9. Heikes, R. R., et.al., Physica, 30,, 1600 (1964). 10. Jonker, G. H., Phillips Research Reports, 24, 1 (1969) . 11. Austin, I. G., Proc. Phys. Soc., 90,, 157 (1967). 12. Morin, F. J., Bell Syst. Tech. J., 37,* 1047 (1958). 13. Kingery, W. D., et.al., J. Araer. Ceram. Soc., 42, 393 (1959) . 14. Patterson, J. W., et.al., J. Electrochem. Soc., 114. 752 (1967). 15. Johansen, H. A., and Cleary, J. G., J. Electrochem. Soc., 111. 100 (1964). 16. Wimmer, J. M., Bedwell, L. R., and Tallen, N. M., J. Amer. Ceram. Soc., 50, 198 (1967). 183 184 17. Krylov, 0. V., "Catalysis by Nonmetals," Academic Press, New York, 1970. 18. Baukloh, W., and Henke, G., Metalliverloschoft, 19. 463 (1940. 19. Hatch, and Choate, J. Franklin Inst., 207. 369 (1927) . 20. Beckley, R. I., and Stone, F. S., "Electronic Pheno mena in Chemisorption and Catalysis on Semiconductors, Moscow, 1968," ed. Hauffe, K., Gruyter Co., Berlin, 138 (1969). 21. Parravano, G., J. Am. Chem. Soc., 75, 1448 (1953). 22. Schwab, G. M., and Block, J., Z. phys. Chem. (Frank furt) , .1, 42 (1954) . 23. Keier, N. P., Roginski, S. Z., and Sazonova, I. S., Akad. Nauk (Otdel. Fiz. Nauk), 2JL, 183 (1957). 24. Cumino, A., Molinari, E., and Romeo, G., Z. phys. Chem. (NF), 16, 101 (1958). 25. Dry, M. E., and Stone, F. S., Disc. Faraday Soc., 28. 192 (1959). 26. Bielanski, A., and Deren, J., "Electronic Phenomena in Chemisorption and Catalysis of Semiconductors, Moscow, 1968," ed. Hauffe, K., Gruyter Co., Berlin (1969) . 27. Schroeder, J., Z. Naturforsch, 17B. 346 (1962). 28. Jonker, G. H., and Van Santen, J. H., Physica, 19. 120 (1953). 29. Stone, F. S., Advances in Catalysis, ,13., 1 (1962). 30. Wolkenstein, T., Advances in Catalysis, 12., 189 (1960). 31. Garrett, C. G. B., J. Chem. Phys., 33., 966 (1960). 32. Hauffe, K., Reviews of Pure and Applied Chemistry, 18, 79 (1968) . 33. Garner, W. E., Gray, T. J., and Stone, F. S., Proc. Roy. Soc. (London), A197. 294 (1949). 185 34. Enikeev, E. H., et.al., Dolk. Akad. Nauk, S.S.S.R., 130. 807 (1960). 35. Yakel, H. L., Acta Cryst., 8, 394 (1955). 36. Wold, A., et.al., J. Am. Chem. Soc., 79. 6365 (1957). 37. Raccah, P. M., and Goodenough, J. B., Phys. Rev., 155. 932 (1967). 38. Dwight, N., and Raccah, P. M., J. Phys. Chem. Solids, 28, 549 (1967). 39. Askham, F., et.al., J. Am. Chem. Soc., 72, 3799 (1950). 40. Remeika, J. P., J. Am. Chem. Soc., 78, 4259 (1956). 41. Wold, A., and Ward, R., J. Am. Chem. Soc., .76, 1029 (1954). 42. Pornoi, K. I., and Timofeeva, N. I., Izv. Akad. Nauk. S.S.S.R., Neorgan. Materialy, 1_, 1953 (1965). 43. Keith, M. L., and Roy, R., Amer. Min., 39.* 1 (1954). 44. Wold, A., et.al., J. Am. Chem. Soc., J79* 4911 (1957). 45. Wold, A., and Arnott, R. J., J. Phys. Chem. Solids, 9., 176 (1959). 46. Goodenough, J. B., and Raccah, P. M., J. App. Phys., 36, 1031 (1965). 47. Cassedanne, J., Anais Acad. Brasil Cienc., 36, 13 (1964). 48. Friend, J. A. N., J. Chem. Soc. (London), 824. 1430 (1935). 49. Latimer, W. M., and Hildebrand, J. H., "Reference Book of Inorganic Chemistry," 3rd ed., Macmillan Co., New York, 1951. 50. , International Critical Tables, 9., 227. 51. Karenman, I. M., Zhur. Obsckei Khira, 25. 1859 (1955). 186 52. Knudsen, J. G., and Katz, D. L., "Fluid Dynamics and Heat Transfer," McGraw-Hill Book Co., New York, 1958. 53. Rohsenow, W. M., and Choi, H., "Heat, Mass, and Momentum Transfer," Prentice Hall Co., New Jersey, 1961. 54. Perry, R. H., Editor, "Chemical Engineer's Handbook," McGraw-Hill Book Co., New York, 1963. 55. Bird, R. B., Stewart, W. E., and Lightfoot, E. N., "Transport Phenomena," Wiley and Sons, New York, 1960. 56. Satterfield, C. N.,"Mass Transfer in Heterogeneous Catalysis," M.I.T. Press, Boston, 1970. 57. Benson, J. E., and Garten, R. L., J. Catalysis, 20, 416 (1971). APPENDIX I DENSITIES OF SINTERED PEROVSKITE SAMPLES Using the specimens prepared for measurement of DC electrical conductivity, the bulk density of sintered cata lyst particles were determined. These samples were pre pared utilizing techniques described in detail in Chapter VI. Starting from -200 mesh powder, wafers were dry pressed in a mild steel die at 15,000 lbf load. These wafers were then sintered in air using covered Pt foil lined boats at the following furnace conditions: Catalyst Sintering Condition La0.9Ce0.iCo03 24 HR at 1250°C LaCoO-j 24 HR at 1250°C ^O.S^O.l00^ 24 HR at 1250°C La0.9C*0.1Nl°3 48 HR at 950°C LaNiO^ 48 HR at 950°C La0.9Sl0.1NlO3 48 HR at 950°C La0.9Ce0.lCr03 48 HR at 1250°C LaCr03 24 HR at 1325°C Lan QSrr t .CrO. 24 HR at 1250°C From the conductivity specimen dimensions and mass, the following bulk densities were computed: 187 188 Observed Density Theoretical Density 5.69 gm/cc 6.04 7.2 gm/cc 5.47 3.93 7.2 4.13 3.61 6.75 4.04 The theoretical density of LaCo03 is reported by Askham (39) to be 7.2 gm/cc. From the reported rombohedral pseudocell dimensions the density of LaNiO^ and LaCrO^ were estimated by ratio of cell volumes and formula weights. Cell volume was approximated by the cube of the cell length since the angle is near 90 degrees. These data are summarized below: Compound Formula wt. Length Angle Reference LaCo03 245.9 gm/g-mole 3.83 A 90°40' 36,40 LaNi03 245.7 3.83 90°43' 45 LaCrO, 239.0 3.88 90°15' 36 The density of the sintered particles is well below the theoretical density, but this is not thought to repre sent the density of particles at the microscopic scale. To examine this a specific gravity determination via the Catalyst La0.9Ce0.1CoO3 LaCo03 La0.9Sr0.lCo03 LaNiOo La0.9Ce0.1CrO3 LaCr03 La0.9Sr0.1CrO3 189 ASTM method D 153 - 54 (reapproved 1970) was performed on -325 mesh LaNiO^ powder. Duplicate determinations result in a mean value of 6.68 gm/cc. Similar measurements were not made on the other compounds. Since LaNiC^ was pre pared at the lowest temperature, densification via sinter ing is expected to be least effective on this preparation. From qualitative observations on strength of LaCoO^ series sintered parts, it is believed that particle density in excess of 95% of theoretical are obtained. Judgment of the LaCrO-j series lead to similar assessment, although sintered part strength is intermediate to LaNiO^ and LaCoO-j. APPENDIX II La. „Cen ,CrO_ AS A POSSIBLE THERMISTOR MATERIAL 0.9 0.1 3 Another point of interest noted from the conductiv ity data is the temperature dependence of Lag gCeg ^CrO^. The fractional change of resistivity with temperature is expressed by the function, dp/dT = A E/R p t2 where, p = resistivity, ohra-cm T = temperature, K AE — activation energy, cal/g-mole R = gas constant, cal/g-mole-K From the conductivity data the terra (dp/dT)/p was evalu ated to be 0.2%/°K and 0.1%/°K at 1000°K and 1500°K, res pectively. This sensitivity to temperature compares to that observed for EMF changes of a Pt/13% Rh-Pt thermo couple junction which are also 0.2%/°K and 0.1%/°K at 1000°K and 1500°K, respectively. This behavior of the La0.9CeO lCr°3 compound suggests usage as a thermistor sensor for high temperature applications. Note that the sensitivity to temperature is larger at lower tempera tures. Application of this material appears promising; 190 191 yet more work on possible fabrication problems need to be explored.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

An Experimental Investigation Of Heat Conduction Through Liquid-Liquid Phase Boundaries

PDF

Radiative Transfer Through Plane-Parallel Clouds Of Small Particles

PDF

A Study Of The Behavior Of Thixotropic Fluids

PDF

The Generation Of Predictor Equations By Recursive Linear Regression

PDF

Investigations On The Flow Behavior Of Disperse Systems

PDF

The separable programming problem and its application to the optimal formulation of elastomeric compounds

PDF

Thermal Conductivity Of Liquids

PDF

Mechanistic Study Of Air Pollution

PDF

The Separation Of Multicomponent Mixtures In Thermally-Coupled Distillation Systems

PDF

Vapor-Liquid Equilibria For The Benzene - N-Octane System Near The Regionof The Critical-Locus

PDF

Removal And Separation Of Particles (In Solid Organic Chemicals And Metals) By Crystallization

PDF

Phase Behavior In A Multicomponent Heteroazeotropic System

PDF

Absorption And Scattering Of Thermal Radiation By A Cloud Of Small Particles

PDF

Prediction Of Enthalpy Of Saturated Paraffin Hydrocarbon Mixtures

PDF

Dental Enamel Electrochemistry

PDF

New Photoluminescence Line Series Spectra Attributed To Decay Of Multiexciton Complexes Bound To Lithium, Boron And Phosphorus Centers In Silicon

PDF

High Temperature Hall Effect Measurements Of Cadmium Rich Cadmium-Sulfide

PDF

Self Diffusion And The Defect Structure Of Cadmium-Sulfide

PDF

Thermal Conductivity Of Binary Liquid Mixtures

PDF

Nucleation And Solubility Of Monosodium-Urate In Relation To Gouty Arthritis

Asset Metadata

Creator Berkstresser, George Wayne (author)

Core Title Perovskite Structure Rare-Earth Transition-Metal-Oxide Catalysts

Degree Doctor of Philosophy

Degree Program Chemical Engineering

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag engineering, chemical,OAI-PMH Harvest

Language English

Contributor Digitized by ProQuest (provenance)

Advisor Whelan, John M. (committee chair), Kroger, Ferdinand A. (committee member), Rebert, Charles J. (committee member)

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c18-785951

Unique identifier UC11364295

Identifier 7309304.pdf (filename),usctheses-c18-785951 (legacy record id)

Legacy Identifier 7309304

Dmrecord 785951

Document Type Dissertation

Rights Berkstresser, George Wayne

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the au...

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus, Los Angeles, California 90089, USA