er HETEROGENEOUS REACTOR DESIGNBUTTERWORTHS SERIES IN CHEMICAL ENGINEERING SERIES EDITOR E, BRUCE NAUMAN Rensselaer Polytechnic Institute ADVISORY EDITORS ANDREAS ACRIVOS Stanford University JAMES E, BAILEY California Institute of Technology HOWARD BRENNER Massachusetts asttute of Technology MANFRED MORARI California Institute of Technology ROBERT K. PRUD'HOMME Princeton University Enlargement and Compaction of Particulate Solids Nayland Stanley-Wood Fundamentals of Fluidized Beds John G. Yates Heterogeneous Reactor Design Hong H. Lee Liquid and Liquid Mixtures 4.S. Rowlinson and F. L. Swinton Phase Equilibria in Chemical Engineering Stanley M. Walas Solid Liquid Separation Ladislav Svaroveky HETEROGENEOUS REACTOR DESIGN _ Hong H. Lee Associate Professor of Chemical Engineering University of Florida BUTTERWORTH PUBLISHERS Boston * London Sydney * Wellington + Durban + TorontoConyigt © 1985 by Raterwoh Poise, [No pat ofthis publication may be reproduced, stored in a retcivlsytem, or tranemite in ay form ot by any means, electronic, mechanical, photocopying. sording, or otherwise, without the pre write, Dermison of the publiher ioary of Congress Cataloging in Pablation Data Les, Hone #. Heteropencous reactor dvgn, elude inde. 1, Heterogeneous catalysis, 2. Cotalyst. 2. Chemical reactor 1. Tie. Trise.cxsLaae 1904 660295 saITIT ISBN 0-405.95073-4 Buterworth Publishes 30 Montvale Avenue ‘Stoneham, MA 02180 wosreseaay Printed in the United States of America TO MY MOTHERCONTENTS. Preface 42 PART 1 CATALYSTS, REACTIONS, AND KINETICS CATALYSTS AND CHARACTERIZATION Introduction Catalysts and Reactions Catalyst Preparation A Description of Catalyst Impregnation Characterization of Catalysts Characterization of Support Notation Problems. References KINETICS OF CATALYTIC REACTIONS Surface Reactions ‘Mechanisms and Kinetic Models Interpretation of L-H Kinetics Notation Problems References EXPERIMENTAL REACTORS AND INTRINSIC KINETICS Experimental Reactors and Transport Criteria Determination of Intrinsic Kinetics ‘Complex Reactions: Synthesis of Kinetic Structure Notation Problems References PART I CATALYST PELLETS AND PARTICLES REACTION/HEAT AND MASS TRANSPORT Introduction Fluxes 10 8 18 35 a 48 2 35 35 2 63 68 69 70 1 16 at 88 90 9s 95 98vii Contents a3 “4 Pei 46 a 48 49 410 os 52 33 34 35 56 57 58 o1 62 63 os Internal Effectiveness Factor and Pellet Shape Internal and External Transport Processes ‘Why Is Diffusion Important? Generalized Internal Effectiveness Factor Reactor Point Effectiveness Nonuniform Activity Distributions Multiple Reactions ‘Transport Criteria and Pellet Design Notation Problems References CHEMICAL DEACTIVATION Introduction Nature of Deactivation Deactivation Kinetics Noncatalytic Gas-Solid Reactions Ditfusion-Deactivation Reactor Point Eifectiveness and Global Rates Catalyst Regeneration Deactivation by Coking Notation Problems References PHYSICAL DEACTIVATION AND SINTERING Introduction Effect of Physical Deactivation of Catalyst ‘Surface Diffusion Mechanisms and Kinetics of Sintering Relative Importance and Sintering Behavior Reactor Point Effectiveness Notation Problems References MULTIPHASE REACTIONS Introduction Mass Transfer Coefficients ‘Transport Effects and Global Rates Global Rate for Trickle-Beds: Nonisothermal Pellets Notation Problems References, 101 106 us 7 120 126 130 138 143 45 19 Ist 151 152 135 158 164 168 180 182 190 193 199 201 201 208 au 21 22 226 229 233 235 235 236 243 247 249 249 1 92 93 os 9-6 oo 98 99 10 1041 102 10-3 10-4 10-5 un nt 2 13 res ins 16. Contents SELECTIVITY AND STABILITY Selectivity Affected by Transport Processes Selectivity Affected by Deactivation ‘The Stability Problem for a Single Pellet AA Single Wire Problem and Slope Conditions ‘The Case of Unit Lewis Number and Extensions Notation Problems References PART III REACTOR DESIGN AND ANALYSIS FIXED-BED REACTORS Introduction Conservation Equations Simplifications Other Considerations Numerical Methods Examples and Patterns Reactor/Heat Exchanger Coupling: Autothermie Reactors Variations of Fixed-Bed Reactors Design of a Countercurrent Reactor/Heat Exchanger with Parametric Sensitivity Notation Problems References DESIGN AND ANALYSIS OF FIXED-BEDS: APPROACH OF REACTOR POINT EFFECTIVENESS Introduction Adequacy of Approximations Equivalence of Fixed-Bed Models Design and Analysis of Fixed-Bed Reactors. Multiple Reactions Notation Problems References FLUIDIZED-BED REACTORS Introduction Fluidization ‘Two-Phase Models Further Considerations ‘A Multiphase, Variable Bubble Property Mode! Nonbubbling Beds at High Velocities 21 251 258 261 263 206 214 27 280 285 285 286 290 293 296 312 314 323 325 331 331 332 333 337 383 390 393 401 403 407 413 415 422x Contents 3 Bl B2 41 42 43 a4 Notation Problems References MULTIPHASE REACTORS Introduction Slurry Reactors Trickle-Bed Reactors Notation Problems References OPTIMIZATION Introduction Equilibrium Constraints Catalyst Deactivation Notation Problems References PART IV TRANSPORT PROPERTIES TRANSPORT PROPERTIES AND EXPERIMENTAL METHODS Pellet Transport Properties ‘Mass and Heat Transfer Coefficients Reactor Transport Properties Determination of Transport Properties and Coeflicients Notation Problems References Subject Index Author Index 404 26 29 431 BI 433 437 445 487 443 49 449 450 458 474 a7 49 483 483 491 493 499 508 506 307 509 513 PREFACE EE ‘This book is on the design of reactors for heterogeneous, solid-catalyzed reactions. AAs such, the catalyst phase and bulk-fluid phase are treated separately as opposed 10 homogencous treatment, and all subjects are discussed with reactor design as the focal point. This text is intended primarily for instruction at the graduate level, but itis also intended for practicing engineers who have had a preliminary course in reactor design. Parts of the material can be used for an undergraduate course when the emphasis is on heterogeneous reactor design, ‘The material in this text is presented in the order of the logical steps that ‘one would follow in proceeding from catalyst selection to reactor design, Accord. ingly, this book is divided into four parts. Part I (Chapters 1 through 3) provides fan understanding of catalysis, support and catalyst preparation, and surface reac. tions. It also provides basic characterization techniques useful in choosing a catalyst system and in providing basic information on the chosen system. Part II (Chapters 4 through 8) deals with the catalyst pellets and particles, in which the reaction takes place, using global rates a the focal point. It provides a quantitative capability for obtaining global rates in terms of the intrinsic kinetics and those parameters Pertinent to diffusion and catalyst deactivation. Part III (Chapters 9 through 13) treats the design and analysis of various types of reactors. It provides specific ‘means of relating the reaction taking place in the catalyst pellets to the overall conversion in the reactor. Part IV (Chapter 14) contains a detailed treatment of transport properties, ‘The general tone and the common thread in all subjects treated are such that students can recognize reactor design as the focal point. The approach allows the students to acquire a rather self-sufficient capability for handling the design of heterogencous reactors. The mechanics of carrying out the reactor design, whether they involve numerical integration of the reactor conservation equations oF construction of a design diagram, are detailed so that students ean put into practice what they have leamed. In presenting the material, emphasis is placed on fundamentals and generality. ‘The origin of and the rationale behind the basic relationships used are explained in terms of the fundamentals of physics and chemistry. These fundamentals ore used to describe quantitatively the macroscopic behavior of the physical and chemi. al rate processes underiying the reactor design. Arbitrary kinetics are used throught ut the text so that the results can be applied to any given kinctic scheme, ‘The {eatment of catalyst deactivation by both physical (sintering) and chemical racana is rather complete, as is the application of the results to the design and analysis, of a reactor undergoing catalyst deactivation. An approach based on “reactorxii Presace point effectiveness” is introduced in this book and is utilized to simplify the design, ‘and analysis of heterogeneous reactors. Reactor point effectiveness, which is the ratio of global rate to intrinsic rate for fresh catalyst expressed in terms of the bulk uid, makes it possible to transform seemingly untenable complex design problems into manageable ones. Since the global rates can be expressed explicitly for arbitrary kinetics of the main reaction and catalyst deactivation, almost all hhcterogeneous reactor problems can be treated as a homogeneous reactor problem whether the reaction is affected by diffusion, catalyst deactivation, or both. This leads to a design practice in which catalyst deactivation is taken into account during the design phase, as opposed to the usual practice of designing a reactor based on the activity of fresh catalyst and then operating the reactor so as tO account for the declining activity with time. ‘The material in this text has been used at the University of Florida for the last five years at the graduate level, and for two years at the undergraduate level. AAt the graduate level, for which this book is primarily intended, it has been found. to be much more effective to emphasize concepts and reasonings and to maintain the continuity of contents and tone rather than to dwell on number-plugging exam- ples. Typically, two lectures a week are followed by a recitation in whieh problems relevant to the material presented in the lectures are solved. Accordingly, examples do not appear in the text as such, but rather as a part of the text whenever such iilustrations are necessary to help maintain the thread of concepts and reasonings. In a typical 15 week semester, Part I is covered in four weeks with major emphasis ‘on Chapter 1. Chapters 2 and 3 are usually combined in presenting the material, Part II is covered in five weeks with major emphasis on Chapters 4 through 6 Chapter 7 is usually covered in conjunction with Chapter 12. Part III is covered in five weeks, Chapter 13 is not covered. Part IV is covered in one week in the form of a summary with emphasis on experimental methods. Not all material can be covered in one semester and therefore some sections are usually left for Students to read, At the undergraduate level, emphasis is placed on the use of results and number-plugeing examples so as to ensure that students can utilize the results given. Work was mainly confined to simple kinetics and plug-flow reac- tors and only parts of chapters were covered. Senior students who took the course felt very comfortable with the design of heterogeneous reactors and were very enthusiastic, as indicated by many wito voluntarily solved the problems on their programmable calculators. By the time graduate students take this course, they ‘usually have had courses in kinetics and transport phenomena. The undergraduate students will have had a course in homogeneous kinetics and reactor design. Problems at the end of each chapter are listed in the order that the material is presented. Suggestions for solving problems are given when it is appropriate. Notation is given for each chapter to avoid confusion due to changes that occur in the notation. Each chapter starts with an introductory section in which the subject matter to be covered is outlined and each chapter ends with a summary section. The chapter on detailed treatment of transport properties is reserved for the end of the book so that better continuity for reactor design can be maintained. ‘The author is indebted to J.B. Butt for his direct and indirect involvement, Preract xii in this endeavor. He is also indebted to J.M. Smith, whose encouragement set fhe endeavor into motion and who showed continuing interest in its progress ‘The sugzestions and criticism of EB. Nauman and R.L. Kable are gratefully acknowledged. Past and present students of the author also contributed in a signi cant way. They are: LM, Akell, DJ. Miller, J.C. Hong, LA, Toor, KI. Klingeran, and E.C. Stassinos, To Jeanne Ojeda, Linda Padgett, and in particular Meliseg Michaels, who did most of the typing, I express my thanks. Finally, I wish to ‘express my special gratitude to my wife, Sukie, for help and encouragement, Hong H. LeePART | eS Catalysts, Reactions, and KineticsCHAPTER 1 Catalysts and Characterization 1-1 INTRODUCTION A catalyst is a substance that accelerates (or sometimes decelerates) the rate of approach to chemical equilibrium. In so doing, itis neither consumed nor is its effectiveness reduced unless it is deactivated in the course of reaction. Only heteroge- ‘neous solid catalysts are considered in this book. Catalysts are usually metals or ‘metal compounds. The catalyst surface exposed to fluid reactants is responsible for the catalytic effect. It is natural then that the catalyst be made to have 2 high exposed surface area per unit weight. On the other hand, the reactor that contains the catalyst should be as small as practically possible. Therefore, the catalyst is usually spread on a substance of high surface area. Such a catalyst is called a supported catalyst ‘Ultimately, one would like to tailor a supported catalyst for the reaction under consideration ina reproducible manner. Unfortunately, however, this capabil- ity does not yet exist and even the characterization techniques necessary for this capability are far from satisfactory. However, there are some basic tools that are rather well established for the characterization of catalysts and supports. These will be treated in some detail, ‘This chapter gives a description of catalysts, reactions, and catalysis, catego- rized in terms ofthe type of catalyst. Some detailed procedures for catalyst prepara tion are described qualitatively to provide some insight into the variables that ‘can affect the performance of the prepared catalysts. An approach for quantifying the preparation procedures is presented for impregnation to provide a basis for further studies in this important and yet unexplored area of catalyst preparation ‘Treated in detail are those well-established basic tools for the characterization of catalysts in terms of eatalyst dispersion and specific activity. Catalytic activity and correlations for it follow as a natural consequence of the discussion. Total surface area, pore volume, and pore size distribution essential for the characteriza- tion of the support are then treated in detail. The premise on which engineers eal with reactions and reactors is invariably “given a catalyst.” This chapter provides a basic level of understanding required in moving toward eliminating this restriction,4 HETEROGENEOUS REACTOR DesiGn 1-2. CATALYSTS AND REACTIONS. Solid catalysts can be grouped into three types: supported metal catalysts usually used for reforming reactions, acid and zeolite catalysts for cracking reactions ‘and metal compounds (oxides and sulfides) for partial oxidation reactions. A list of typical functions and examples of catalysts is given in Table 1 {cis well understood that a catalyst cannot change the ultimate equilibrium dictated by thermodynamics but accelerates the rate of approach to equilibrium Why such an acceleration takes place can be explained with the aid ofthe energy in energy associated with the different steps in a simple exothermic reaction, The Gotted line is for the homogeneous reaction with its activation energy of Enon, while the solid line is for the heterogeneous catalytic reaction. The activarey ‘netay for the limiting step of heterogeneous reaction is usually lower than that ‘ate ofa catalyzed reaction should be faster than the corresponding homogencous reaction, each of the steps in the catalyzed reaction requiring a smaller activation energy than Brom. The diagram also helps understand why no significant increase in rate is achieved by any catalyst when high temperatures are required, as in fome endothermic reactions, in order for a substantial amount of the product to be present at equilibrium, ‘The mechanisms of catalysis for each group of the catalysts listed in Table 1-1 ate dificult if not impossible, to generalize, However, a few statements ean be made on each group of catalysts. Almost all the metal catalysts are trensition metals having d orbitals. These orbitals are not completely filled due to band ‘overlap or electron overflow from the d to the s band (Gates, Katzes, and Schur 1979), While itis not clear how this bulk atom behavior is related to the surface atom behavior, itis clearly plausible that a singly occupied surface orbital can interact with a molecular orbital of an approaching gas species, resulting in a bonding atthe metal surface. This chemisorption can be regarded as the formation of a surface compound. In order for a catalyst to be effective, the stability of the GomPound should not be too high, since this would lead to complete coverage of (2.very litle surface coverage by the compound. When bonding level is adequate for a given reaction, the reaction can proceed at a satisfactory tate ‘Metal compounds (mostly metal oxides) are usually used for partial oxida- tons, The catalyst are typically transition metal oxides in which oxygen is readily {ansferred to and from the structure. These compounds are usually nonstoichioner sie. Other types of oxides are those in which the active species is chemisorbed ‘oxygen, as molecules or atoms. The behavior of most oxidation catalysts Satterfield 1980) can be interpreted within the framework of a redox mechanism. This postu- Solos, E.G. (Heterogeneous Katayee, Verlag ‘homed theoctaly thatthe esta at et be Tat te ' kes than the activation energy ofthe ling sop hemi, Wernheim. Germany, (1971) the homogeneous ate even when Ex CATALYSTS AND CHARACTERIZATION. 5 Table 1.1. Classification of Sol Catalysts (after G.C. Band 1974) Group Action Example Meas tydropeton Sehyaropenaion Fe, NA PALPL AR hyrogeolyoe hysrncting) Jt Compounds “emnomdcing —osttan NiO, 290, MnO, cides and sus ehyérogenation Crs, BO," Mes fis destitution” WS, Inslator oxides detindation ADs, SOs, Meo Acid andwolte” cracking 510," Alo, (amorphous or erysaline) ‘Aditon ot hyropen ares single bond to cause piting ito two molecules, bs Cae + Hy 20H, lates that the oxide catalyst is first reduced by the gas molecule being oxidized and then the reduced catalyst is in turn oxidized by oxygen from the gas phase The form of the oxygen in the catalyst can be either chemisorbed or lattice oxygen In the case of lattice oxygen, the active species is presumably the O*- ion ‘One important fact about the acid catalysts is that neither silica nor slumina ‘nor @ mechanical mixture of the two dry oxides is an active cracking catalyst, but a cogelled mixture of silica and alumina containing mainly silica is highly active. The incorporation of alumina into silica, even at very low concent Energy ——+ Reaction path > re Ancictay agra orcas ecto (Aer CN Sater, Herpes sabato sh pron af Meow i one Compe ee ea6 HereroceNeous Reactor DESIGN ante ¥ Figure 1.2 Postuleted structures of silica-alumina causing Bronsted and Lewis acids. (After GN, Satterfield, Heterogeneous Catalysis, © 1980; with permission of McGraw Hill Book. Company, New York) ‘ous in the formation of sure ss that catia cracking exeton tis beeved Shit ata ome ef tse tes conte rose anor Lov ste Tes of aciiy (ater 1980) ay te ronan n eae os teey Sad lrg by Pasig If an aluminum ion, whieh ie ison oc ace ne oho a sion on hich untae nse lace comping es Teahera, the net neptive charge mst be sabied y's saxty ee a suchas a proton This canbe prodeed bythe weston of ae eons hydro group onthe alominum atom. The rasiong seston ee ae sluminun and te sicon are both tetrsadraly cordate, Be ae iets i hese, water driven of and Briel aed sts ae een Levis ad Stes a shown in Fight 12: Some mel atoms te sey eae bated and some fourcordated. The alomisum aims esrb aetna Ise with ydrocuton ofr inated cttentom nas erat eo for the two kinds of sites: i aa i i rena ow OO = en ng St ee a i i and \ \ ' § ee sg i | i Similar arguments can be advanced to explain the acidity of various other mixed ‘oxides containing metal atoms of differing valence, such as SiO,-Mg0, Si0s-Z:0, CATALYSTS AND CHARACTERIZATION 7 Figure 13. The cuborctahedral unit of faujasite zeolites, structure of reali and X and Y zeolites (From let to sght). (Barrer 1968) and Al,O,-MgO. Mechanistic explanations of catalytic eracking are usally based ‘on the formation of carbonium ion (Voge 1958), which these acidic sites initiate. “The aluminosilicate catalysts considered above are amorphous. In recent ears, rytallinealuminosilicates (Bond 1974) have been developed with the general formula M,(AI0:)(SiO,),-2H,0- these are known as zeolite. When M isa mono Positive cation (e.g. sodium or ammonium), v equals x; for divalent cations, v is 5/2, and so on. There are great many naturally occurring minerals of this kin, but the mordente and faujasite ean now be synthesized, The basi unit ofa foujasite is the regular cubo-octahedron o sodalite unit (Fig 1.3) consisting of 24 tetrahedra of cther SiO{- oF AIO; Two general types of structore can then aris as shown in Figure 13, depending onthe way in which the sodalite units are joined together. When they are joined through their square faces, A zeolite resulis in which »/x is unity; when joined through their hexagonal faces, the X zeolite (y/x, 1.25) or the Y zeolite (y/x. 1.5 to 3) results. These last two structures have quite large capes joined by smaller openings and therefore avery high internal surface area im the form of pores of fixed geometry. The effective pore diameter is determined by the kind ofeaton that balances the negative charge on the structure. Cations cccupy one of three distinct positions, and their size thus controls the sizeof the openings. With A olite, the elective pore diameter is 0.3 nm when the cation is potassium, 04 for sium, and 0.5 for calcium. Zeolites are usualy frst made in the sodium form: the sodium ions can then be exchanged quite simply for ther cations. ‘The wef catalytic properties of zeolites hinge on three factors: the regular crystalline structure and tniform pore size that allows only molecules below certain size to react, the presence of stronaly acidic hydroxyl groups that can initiate carbonivm-ion reactions and the presence of very large electrostatic elds in the neighborhood of the cations that can thus induce reactivity in reactant molecules, Catalytic activity therefore depends heavily on the nature of the ation, Which also seems able to affect the acidity of the hydroxy! groups. These zeolites axe typically used for skeletal rearrangements and cracking of hydrocarbons. Shape- Selective catalysis can also be realized using zeolites as reviewed by Chen and Weisz (1960. Some heterogeneous catalysts of industrial importance are summarized in “Table 12 (Satterfield 1980) along withthe types of reactors commonly used with them8 HETEROGENEOUS REACTOR DISICN ‘Table 1.2 Some Heterogencous Catalysts of Industrial Importance (after CAN, Satterfield Heterogeneous Catal. © 1980; with permission of McGraw-Hill Book ‘Company, New York) Reaction Dehydrogenation Calle butane) + butenes and Cal (bute ‘iene) ‘Butenes — CaM (butadiene) CaHhiCaHls— CH,CH=CH, (ethyl benzene — styrene) (CH, oF other hydrocarbons + HO — CO +H: (steam reforming) (CHy)CHOH — CH,COCH, + Hy (Gsopropanol — acetone + hydrogen) (CHsCH(OH)C.Hy — CHCOCAHs + Hs Hydrogenation OF edibe fats and oils Various hydrogenations of fine organic chemicals Gitte 3H Cathe Ne 31 2NH Calla ~ Cotte (elective hydrogenation of ‘CoH impurity in CH, from thermal- cracking plant) usta $021 40.50, aH, +20,2N0 +3840 NH + CH + air HEN (Andrussow pro- cass) Cll oF 12-CeHl(CHs)2 + 0; > CuHA(CO),0 (naphthalene or o-nylene + tir“ phthalic anhydride) OF Calls + Os — CHO, (butene zene + ait — maleic anhydride) ee HE EE Cataiyst and Reactor Type (continuous operation unless otheraise noted) (Crs0y-AlOs (Bxed bed, etic) Fe.Os promoted with C:03 and K:COs, of CaNi(PO.)e (xed bed, continuous, in presence of steam) FeOs promoted with Cri0s and K,CO, (fixed bed) Supported Ni (xed bed) 200 [Raney Ni, or Nion a support (surry reactor, batch or continuous) Pd on C or other support (slurry reactor, batch or continuous) (ether supported mictals may also be used) 1Ni or noble metal on support (fined bed) Fe promoted with AliOs, K,0, C30, and MgO (adiabatic fixed beds) 'P6 on Al,Os or sulfide Ni on support (adia- bate fixed bed) 'V.Os plus K:S0, on SiO» (adiabatic, xed beds) 90% Pt-10% Rh wire gauze, oxidizing con- ditions 90% PI-l0%% Rh wire gauze, under net re- duction conditons Supported V0 (multitube fxed bed) Supported V0, (multitube fred bed) cate + $0.01. ninety CH.OH + 0, — CHO + He and/or HO iH + 0) — CHs=CHCHO (aerolein) ‘and/ot Cs = CHCOOH (acrylic aid) GN 20, chemcHeN + 0 (Complete oxidation of CO. hydrocarbons, in pollution control, as of auto exhaust cit. +to,+ enscoon— HCOOCH=CHs ny axa) cate LO c+ Acid-eotalyzed reactions Catalytic oracking Hydrocracking Isomerization Catalytic reforming Polymerization Hydration, e., propylene to isopropyl aleo- hol Reactions of synthesis gas CO+2H,—CHLOH 0+ th chs + 140 (thant ©. ypartin .cerToseh ‘yay oer Ovedocnation Ca 2801440, = cata + 0) Hrsodeaoton 80, + 21,8 —+ 35 + 2H,0 (Claus process) HO + CO— CO, + Hy (water-gas shift) CATALYSTS AND CHARACTERIZATION 9 Supported Ag ‘Ag or FeO, MoO, (CuO or multimetalic oxide compositions Complex metal molybdates or multimetalic ‘oxide compositions (fui bed) Pr or Pt Pd, pellet or monolith support Pd on acid-resistant support (vapor phase, sultitube fixed bed) Promoted ferite spinels Zeolite (molecular sieve) in SiOz » AL Os ma ‘wie (fd bed) Metal (e4,, Pd) on zeolite (adiabatic fixed beds) Metal on acidified AlsOs (xed bed), zeolites Pt, Pt Re, or Pt Iron acidified ALOs(ad- ‘batic, xed or moving bed) HPO, on clay (fixed bed) Mineral acid or acidtype ion-exchange resin (xed bed) ZnO promoted with Cris, oF Cu! - Zn0 promoted with Cri0y or ALO: (adiabatic, Fixed beds or multitue fixed bed) Supported Ni (fixed bed) Fe with promoters (Rud bed) (CuCh/ALO, with KCI promoter (Co Mo/AlOs oF Ni - —Mo/AL0, (adi: atic, fixed beds) ‘AlOs (xed beds) FeO, promoted with C05 (adiabatic fixed bed) fora second, lower temperature stage, Cu * ZnO supported on AliOs or Sid,10 HereRocencous Reactor Desicn 1-3 CATALYST PREPARATION Mot catalysts are either finely divided meal inthe frm ofeytalites supported on a case sch as alumina or sles, or metal onde eter a ce ot nspported, Meal slide catalyts ae usally prepared fest the ods asd then teated with hydrogen slide or another sullen compound is ones inne the slide. For the preperation of supported catalyst parle ofthe sosponieg tater are rough into contact wth metal sluions in the orm of sub che ses ota. The resulting mintre eared, washed, dre, and er fos int desired shape, The ctl then activated to conver into is soos form by heating to cause calnation or decomponon, flowed by setae 4 ictal call i desired The preparation method ig tamed the enconon method or the “impregnation method” depending onthe manner whee oe Darts of support are brought nto contact wth te metal salen, Inthe precipitation method, he mixing of two or more salon O sispen- sions of material causes precipitation ofthe catalyst For instances seed nickel eal eould be prepared from nickel nate anda suspension of even by pesipitation wth ammonium hydroxide. Binder, ements, te orcs ee thay beaded at this or ater stage, The al sine and hope te ceed catlyt ae determined by the forming rove, which may aso afte nee oe ad presi dstibution Large ores canbe trodoced ito ental by oan, ‘ig ne organic powders into the mixture, which can te burned out aes, ‘The impregnation method isthe easiest vay of making catalyst; While fne partis ofthe supporting mater! ean be used in abostony eepeanns the Wal indo practice He to tue the support inte fal nee ‘hus simian the ering and forming stop, Two methods of omsacu we Used: dry impregnation (impregnation to incifent wena) and neice De impregnation i more commonly wed indy since the composition os bach, of slion uid for immersion wl hang as addional sport re mprepret, ‘utingin nonuniform activity fom one pelt to anther In tery erie zethod the supports contacted a by Sprang, With & sluice of pees oncentation, corresponding in anty tthe oul known pore vakonare gle ls, Oxide sppors suchas alumina andere ely we by agucousvolnine, 4s re mont activated carbons sine they havea let of cheated soe them: Capilry frees ensure that guid wil be sucked ap into te cae ne Structure Ifthe support isnot ely wetted as ina carbon thats Nghly seh oF without chemisorbed oxygen, an organi sohent may be une ot erent may be impregatd onde vertu, “The pretation method generally provides a mre uniform ctl distribu kon within the pelt than the impregnation method. Some sontcel orc sane oe and its csibaton canbe exerci. However the Raa artus ay be tied if oo or more metal compounds are prea! thar presptaten aifens ae ‘The impregnation metho speed forthe preparation of male seal catahon since i's economically desirable to spread out the metal inas acy ech a form as posible. The noble metals anually presenta eoceattaton of | mide CATALYSTS AND CHARACTERIZATION TT or less of the total. In contrast, the precipitation method is desirable when a high percentage of metal is to be incorporated, up to 20 to 40 wt, onto the support lsed for base metals such as copper and nickel. It may be very difficult to obt such high loadings by impregnation or even by multiple impregnations. While the usual dry impregnation method leads to a supported catalyst with higher catalyst loading toward the surface, use of the immersion method can lead to an essentially uniform deposit of adsorbed catalyst if sufficient time is allowed for the diffusion of reagent species into the interior of the pellet. The drying process can affect the catalyst distribution within the support. ‘The crystallite size ofa supported metal catalyst may also be altered if a considerable portion ofthe soluble metal is occluded rather than adsorbed. Initially, evaporation ‘occurs at the outer surface, but the liquid evaporated from small pores is replaced by liquid drawn from large pores by capillarity, possibly causing a nonuniform distribution of catalyst. In the precipitation method, the dried catalyst particles are formed into granules, spheres, tablets, and extrudates. In either method, the dried material is calcined to activate the catalyst In the calcination process, extraneous material such as binders as well as volatile and unstable anions and cations that have been previously introduced but are not desired in the final catalyst are eliminated. Before calcination, crystallites are usually in the form of metal salts. If a metalic catalyst is the ultimate goal, ‘conversion to the oxide form is frequently sought prior to reduction to the metal. If a mixed oxide catalyst is desired, a substantially elevated firing temperature ‘may be required to cause mixing by diffusion of individual species to form a desired ‘compound or crystalline phase. In any event, the catalyst should be heated to a temperature atleast as high as will be encountered in the plant reactor to preserve the integrity of the catalyst under reaction conditions. Otherwise, bound water or carbon dioxide in the catalyst may cause decompositions, leading to breakup and dusting of pellets, which in turn lead to excessive pressure drop and premature reactor shutdown. The choice of temperature, on the other hand, should not be so high that the catalyst forms stoichiometric compounds or 2 solid solution with the support. The final form of supported metal catalysts is arrived at by reduction of the oxide at an elevated temperature by contact with flowing hydrogen or hydrogen diluted with nitrogen, the latter for safety reasons. Anderson (1975) has shown from thermodynamic calculations that the reduction of the metal oxide to the ‘metal is highly favored in the region of $70 to 770K for all group VIII elements, whereas the oxide form is highly favored for chromium, vanadium, tantalum, tta- nium, and manganese. Sometimes reduction is carried out in situ in the plant reactor to avoid hazards associated with handling a pyrophoric material ‘There are other ways of preparing catalysts. The iron catalyst used for ammo- nia synthesis is made by fusion of naturally occurring magnetite, FesO,, with ‘small amounts of potassium carbonate, alumina, and other ingredients. The best known example of a metal catalyst prepared by leaching is Raney nickel, developed bby Raney in 1925. The catalyst is prepared from a nickel-aluminum alloy by leaching ‘out much of the aluminum with caustic solution to leave behind @ porous nickel12. HeTeRocentous REACTOR DESIGN catalyst. Typically @ 50:50 nickel-aluminum alloy is reacted with @ 20% solution of sodium hydroxide. To achieve maximum activity and structural stability some ‘aluminum must be left behind, and some hydrated alumina is also formed and retained. This may act as a physical promoter. Raney nickel is pyrophoric, so it ‘must be handled carefully; it is usually stored under water or an organic solvent. The leaching method can also be applied to other metals alloyed with aluminum to make, eg, Raney cobalt, Raney iron, etc, and the method may also be used to produce metal catalysts in unusual forms. tis obvious from the foregoing discussion that many factors are involved in the preparation ofa catalyst that undoubtedly affect its final performance. While some efforts are being made to quantify these various effets, catalyst preparation is still an art, and detailed arcane procedures need to be followed to achieve repro- dducibility and desired properties. The most overriding consideration in preparing ‘catalyst should be the chemical composition of the surface. The final form of the catalyst should also be stable to heat and to fluctuations in process conditions, and mechanically stable to attrition ‘The desired catalyst particle size is determined by the process in which it is to be used. For fluidized-bed or slurry reactors, particles usually range from about 20 t0 300 pum in size. In fuidized beds, the lower limit is set by the difficulty of preventing excessive carryover of fines through cyclone separators inthe reactors; the upper limit is set by the poorer fluidization characteristics of larger particles, In slurry reactors, powders that are too coarse may be difficult to suspend; those too fine ae dificult to remove by filtration. The catalysts used in fixed beds generally range from about 1.5 to 10 mm in diameter and have about the same length. Small pellets (1 ~ 2 mm) may cause excessive pressure drop through the bed; larger ones cause difusion limitations ‘The supports used for the catalyst are usually alumina, siliea, or activated cearbon, Because stability is needed at high temperatures, the form of alumina is either 7 or m-alumina, These are usually the first choice for supports. However, active aluminas can dissolve or become soft and mushy under acidic conditions— conditions under which silica is stable. The relative nonreactivity of silica upon calcination with other catalyst ingredients may also be a significant factor in the choice of supports. Carbon supports are used primarily for noble metals and for reactions in which the strong adsorption of earbon for organic molecules may be ‘an asset. An unusual support form known as a monolith or honeycomb is used in some automobile catalytic converters, where a very low pressure dop is required fo minimize power loss from the engine. This isa single block of material containing. within it an array of parallel, uniform, straight, nonconnecting channels. Promoters are often used for supported catalysts. These can be grouped into physical and chemical promoters. A physical promoter is an inert substance that inhibits the sintering of crystallites of the catalyst by being present in the form ‘of very fine particles. An example is the incorporation of a small amount of alumina in the conventional iron catalyst used for the synthesis of ammonia, To be effective, the physical promoter must generally be of considerably smaller particle size than that of the active, catalytic species, it must be well dispersed, and it must not CATALYSTS AND CHARACTERIZATION 13, react or form a solid solution with the active catalyst. A chemical promoter causes 4 chemical effect~it changes the chemical composition of the catalyst ‘More details on catalyst preparation can be found in the book by Satterfield (1980), from which the bulk of this section is derived, and the books by Linsen (1970), Anderson (1975), and Moss (1976). The ultimate goal for catalyst preparation isthe capability to tailor a catalyst for any given application. This goal, however, is dificult to attain due to the complexities involved in catalyst preparation. Nevertheless, some progress has been ‘made, which is considered in the following section. 1-4 A DESCRIPTION OF CATALYST IMPREGNATION ‘The two most important properties of prepared catalysts are the chemical state of the catalyst and its distribution within the support. Consistency is essential ven ifit does not give complete uniformity since otherwise the behavior of reactions and reactors becomes unpredictable. This is precisely the reason why arcane proce- dures are followed in preparing catalysts. The preparation step that can change the chemical state of a catalyst is calcination. While all steps involved in the preparation affect the catalyst distribution, impregnation and drying steps are ‘mainly responsible for the final distribution, ‘The effect of impregnation conditions, and to a certain extent that of drying conditions, on catalyst distribution can be ascertained through an idealized model. To begin, uniform distribution before drying can always be assured by allowing sufficient contact time for the solute equilibration between the impregnating solution ‘and the support surface. While this can easily be accomplished in a laboratory, such an equilibration is not always possible for industrial supported catalysts for which the dry impregnation method is typically used. Since the origin of nonuni- formity is the distribution resulting from the impregnation step, attention here is {focused on impregnation using the single pore model of Vincent and Merrill (1974). ‘The impregnation process can be considered to consist of two steps: the initial fling of pores when the impregnating solution is brought into contact with particles of the supporting material and equilibration that follows when the pores are filled The filing step can be adequately described by a plug-low model since the ratio of pore length to diameter is much larger than unity. For a cylindrical pore of radius R and length L (Figure 1.4), a solute balance can be written as: aay where C= solute concentration of solution in the pore, mol/volume locity of penetrating solution 49= concentration of solute adsorbed on support surface, mol/surface area14 HeTerocencous Reactor DEsicn ee ey -——_ . ———_4 Figure 1.4 Single pore model. (Vincent and Merrill 1974; reprinted with permission from ural of Catalysis. Copyright by Academic Press.) ‘The penetration velocity v is not constant and the correlation used by the above authors is: aay where-y = surface tension B= viscosity ‘The physical properties of water may be assumed for the impregnating solution, ‘Typical nonuniform catalyst distributions (Chen and Anderson 1973) shown in Figure 1.5 suggest that there definitely exists a resistance to solute removal from ‘the solution since otherwise the profiles should shaw stepwise distributions if enough solutes are present. Solute removal from the impregnating solution onto the pore wall can be modeled as the sequential events of mass transfer at the liquid-solid interface followed by adsorption onto the pore wall. in many cases, the adsorption/ desorption step is the controlling step. The solute balance for the solid phase in such cases is: saeco fe where ky and ke are the adsorption and desorption rate constants and q is the solute concentration on the pore wall in mol/surface area. At equilibrium, the coverage @ is given by the Langmuir isotherm: ont -te a 1+ RE aay Where q, is the saturation concentration in mol/surface area, and k = ke/ke. ‘The balance equations are rendered dimensionless to give: CATALYSTS AND CHARACTERIZATION 15 Figure 15 Typical profiles of catalyst loading: dry impregnation method. (Chen and Ander- son 1973; reprinted with permission from Industrial and Engineering Chemistry, Product ‘Research and Development. Copyright by American Chemicel Society) ap el ay gta tt 1s) w_K Spa -0)-07Ke 16 Fe DOO) OK) ce) whete y= C/o t= thy, x= 2/L, 9 = 24/RCo KDR, Ke em tc Bhp ‘concentration of bulk solution 1, = time required to fll the pore ‘The concentration as well as the coverage is zero until the front of the penetrating solution reaches a point of interest. Therefore: (tay (ate) ‘The method of characteristics (Forsythe and Wason 1960) can be used in a straight- forward manner to solve this system of equations (see Problem 1.10). Typical profiles of coverage at the time pores are just filled (r = 1) are shown in Figure 1.6 (Vincent and Merrill 1974). The equilibrium parameter K,, determines the16 HeTeocentous REACTOR DESIGN 10 os Kiros = kino 040s 08 a7 as 05 48 x Figure 16 Profiles of coverage 0 as determined by the single pore model at r= I. (Vincent nd Marl 1974; rept with permission from ural of Cts, Conyght by Ace final steady state level of coverage, and K/n determines the rate of approach to the sendy stat as evident om Eg, 16 It Sen ftom Figure 16 tha be seedy sate coverage i racod a the pore mouth when Kn & taioals ene ey =-8) A lower value of7 extends the plateau shown the figure farthest pore eter Ones the por i fle, the main mechonsm for solute movement is that of ition. Thrfor, the slut equilibration between stagnant slot aed in pore wal wil be described bys ae ra oa (1.8) 2S ya 9-o7ny as whe Pest number Pe en by rent «13 CATALYSTS AND CHARACTERIZATION 17 Here, the diffusivity D is that of solute in solution. Initial and boundary conditions are the profiles of y and @ at + = 1 obtained from the solutions of Eqs. 1.5 and. 1.6. Depending on the impregnation method, one of the boundary conditions takes the following form (Or) = 1 for the immersion method autay ay, aelenw 0 for the dry impregnation method tt) ‘An assumption made on the immersion method has been that the amount of solution used is large enough for the bulk solution concentration to remain constant through- ‘out impregnation. The symmetry around the pore center provides the other bound- ary condition: y al. (1.2 ‘An estimate on the time required for complete equilibration can be obtained by solving: ayy, Be ar Poe ay since this corresponds to an equilibration process in which no solute is removed at the pore wall. For the purpose of obtaining the equilibration time, one may envision a pore filled with water being brought into contact with an impregnating solution. This system reaches its steady state (Churchill 1944) exponentially with the factor of exp(~7?Per). When the argument of the exponential is larger than 5, deviation from the steady state is less than 1%. Therefore, the equilibration time may be set as Fay = S/aPe In order to obtain the final catalyst distribution before calcination, the drying process has to be described with the profiles at the end of impregnation as initial conditions. This drying process is much more complicated than the impregnation process and no satisfactory results are available. For one thing, the single pore ‘model is inadequate. A model pore cannot be treated independently of other pores since the liquid in larger pores is drawn into smaller pores as evaporation proceeds. [As the evaporation front moves down pore, concentrations reach saturation, and diffusion of solute may occur even after complete equilibration. It has been ‘observed (Komiyama et al. 1980), however, that relatively fast drying docs not alter significantly the distribution resulting from the equilibration, On the other hhand, relatively slow drying results in segregation of the catalyst toward the external surface of the particle.18 Heterocentous REACTOR Desicn The parameters defining the catalyst concentration profiles can be obtained from equilibration runs. Let the concentration of solute remaining after complete equilibration and drying be gu. This quantity, which can be measured, is a comsin tion of adsorbed and occluded solutes, ie pimaottg gy Therefor a table of Cy versus ga can be obtained fom equilibration curs in hich the bulk concentration ofthe impregnating alton i re Te acne For the equation runs, C= Cy andthe parameters Fad go: Ree) can be determined ftom Eq, L4 wt inte falloning ores teat @ Nak) Co" gy om where gi determined from the able of C versus dn. The coverage on the extemal ‘surface can be obtained by solving Eq. 1.6 with the condition that y is unit there (see Problem 1.1): ‘i i 1 gy a Tri {of ‘This result is valid for both pore-filling and eq R is usually obtained by setting the ratio of the pore volume to BET surfres ‘area (see Section 1-6) equal 0 R/2. ‘The single pore model is equivalent to the model for a slab-like particle. In fact the conservation equation for the liquid phase (Eq. 1.5) can be weiten ia a general form applicable to slab-like, cylindrical, and spherical particles (Lee 1984. The catalyst distribution at the end of the pore-flling period (Z), which is the ‘approximate distribution that results after fast drying, is given by: sovetaat sig li-eo[-A(ie e-em 1-5 CHARACTERIZATION OF CATALYSTS 1 Js undoubtedly true that there is no unique way of characterizing a catalyst. ‘There are, however, some basic pieces of information required to determine how ‘fective a catalyst is for a given reaction and how effectively the catalyst is used CATALYSTS AND CHARACTERIZATION 19 ‘The latter requires knowing the exposed surface area of the catalyst and the former requires Knowing the rate of reaction per unit exposed surface area, Since only the exposed catalyst surface ig involved in reactions and the metals and metal compounds used as catalysts are usually expensive, the effective utilization of & ‘catalyst requires a good dispersion ofthe eatalyst on a support. For a given catalyst, the exposed surface area is a measure of the number of active sites responsible for the reaction. Therefore, to a first-order approximation, this information can be used to determine the effectiveness ofa particular method of catalyst preparation ‘and ultimately to choose the most effective preparation method. The ratio of the number of surface metal atoms to the total number of metal atoms present is termed dispersion (percentage exposed). While this term is usually used for metal catalysts, the same definition can be used for catalysts formed of metal compounds. ‘The dispersion of supported platinum catalyst, for instance, can readily exceed 0.5, meaning that more than half the platinum atoms dispersed on the support are surface metal atoms. In contrast, a platinum particle of 1 jm in size yields a dispersion of only about 0.001. Specific activity is defined as the number of molecules, ‘that react per unit exposed catalyst area per unit time. For metal catalysts, this is frequently referred to as the turnover number, or turnover frequency when the basis is on the number of sites rather than the area. The specific activity may be used in place of the turnover number in view of the fact that the number of sites is an clusive quantity to determine, particularly for the catalysts consisting ‘of metal compounds. The specific activity of a catalyst is a basic measure of true catalytic activity: this activity can be used as the basis for choosing a particular catalyst. Here again, knowledge of the exposed surface area is required for the determination of the specific activity ‘The catalytic activity inferred from the specific activity is often correlated to certain quantities to gain some understanding of heterogeneous catalytic reac- tions. As the exposed surface area is required for characterizing the specific activity, ‘methods of determining the surface area will be treated first. This will be followed. by correlations developed for characterizing the catalytic activity. persion and Specific Activity ‘Chemisorption is the usual method for the measurement of exposed surface area ‘of a metal catalyst. For a catalyst formed of metal compounds, however, the chemi- sorption method fails, usually yielding a fraction of the total catalyst surface area; ‘only rudimentary methods such as specific poisoning have to be tailored for specific application. In the case of acid catalysts, a measure of the number of acid sites rather than the surface area is determined via titration techniques. A plot of the specific activity versus the dispersion is often used as a way of characterizing catalytic reactions, as shown in Figure 1.7. The reactions repre- sented by the solid line are termed structure-insenstive (facile) reactions. The specific activity of these reactions is independent of the surface area and is thus structure-insensitive, The reactions following the behavior of the dotted lines are20 HeTtRoceNcous REACTOR Désicn o Dispersion Figure 1.7 Speci activity versus dispersion, (After JJ. Carberry, Chemical and Catalyie Reaction Engineering, © 1916; with permission of McGraw-Hill Book Company, New York) {termed structure-sensitve (demanding) reactions. The specific activity does depend fn the surface area, or more precisely on the size and shape of the crystallites of the catalyst, and therefore is structure-sensitive. A few comments are in order ‘egarding the igure. For one, the rate of most reactions will increase with increasing dispersion, although the possible exceptions are shown in curves (1) through (3) since the actual rate isthe specific activity multiplied by surface area. For instance the rate of structure-insensitive reactions increases linearly with dispersion. For another, the betiavior of the specific activity at and close to a dispersion of unity should be viewed with caution. When the dispersion approaches unity, all the ‘metal is present in the form of atoms that are extremely reactive. These metal ‘toms readily react with gaseous molecules to form, for instance, platinum oxide in the presence of oxygen. The catalytic activity observed in such eases is that of platinum oxide rather than platinum itself. Studies have shown that simple hydro- §enation reactions on various metals are usually structure-insensitive but that reas. tions in which C—C bonds are broken, such as hydrogenolysis and skeletal ‘isomerization, are structure-sensitive. A typical example of a structure-semsitive reaction involving the hydrogenolysis of ethane is given in Table 1.3. ‘A gas can adsorb on a solid surface ether physically (physisorption) or chemi cally (chemisorption), the latter involving actual bonding at the surface between the gas molecule and the metal as explained in Section 1.2. While the magnitude ‘of the heat of adsorption (defined below) is often used to differentiate physisorption from chemisorption, the most useful way of distinguishing them is the way in which gas molecules cover the surface: only a monolayer of adsorbed gas molecules CATALYSTS AND CHARACTERIZATION 21 Table 1.3._Ethane Hydrogenolyss over Rhodium (Yates and Sinfelt 1967) Specific Activity (milimotes ethane con- Coytalite Sze State of Dispersion @ verted hr m! of catalyst) Bulk Rh 2560 079 5% Rh on SiO (sintered) 27 oat Intermediate 1-10% Rh on SiO; Raat 8~16 Very high: 0.1~ 0.3% Rh on SiO; <2 4a (Reprised wth permision fom Journal of Catalysis. Copyright by Academic Pres) ig involved in chemisorption, whereas multilayers are observed in physisorption. Differences between physisorption and chemisorption are summarized in Table 14 ‘The first attempt to quantify adsorption was made by Langmuir based on the assumption of a uniform, energetically homogeneous solid surface. The postula- tions are that the rate of gxs adsorption is proportional to the frequency of gas ‘molecules striking vacant surface sites and that the rate of desorption is proportional to the fraction of surface sites covered by gas molecules. This leads to the expres rats = kap(l— 8) (ts) Table 14 Physisorption versus Chemisorption (Carberry 1976)" Physsorption Chemtorpion Adsorbent All solide Sone tis All ess below critical point Some chemi re Adsortate mer ca Temperature range Low temperature Generally high tem perature ofthe order of| Heat of sdortion Low, ~ AH ligueiction (< High. ofthe ore 15 keal/mel) chemical reaction © 15 teame Rate and activation Very rapid, low enesey Nonactivate, low erey eres tivated, high enerey Coverage Mottager Monolayers ess Reversity Richy reversible Often ireverble Ue Surac ae, pote sce Active surface area, catalysis kinetics "FL Emmet, Johns Hopkins University (1959)22, HETEROGENEOUS REACTOR DESIGN Yes = ha ay Where ka, ka adsorption and desorption rate constants, respectively 0 = fraction of surface sites covered by gas molecules, coverage (P= gas partial pressure K = ka/ka, adsorption equilibrium constant Atequilibrium, the two rates are equal and therefore the coverage @ can be expressed Kp. iim (1.20) ‘The temperature dependence ofthe adsorption equilibrium constant K is expressed K 0 exp(Q/RyT) aan where Ko is a preexponential factor, and Q is termed the heat of adsorption. ‘According to the energy diagram shown in Figure 1.8, the heat of adsorption is equal to: O=k~E, (1.22) Where Band Ea are activation energies for desorption and adsorption, respectively For gases used for typical chemisorption experiments on transition metals suck 4 Ha On CO, and COs, the activation energy for adsorption is small and the i ~ surtace 3 i Tavorbod ste Adsorption path Figure 18. Energy diagram for gas adsorption CATALYSTS AND CHARACTERIZATION 23 activation energy for desorption is often taken as the heat of adsorption. Since adsorption is an exothermic process, the heat of adsorption is positive, Therefore, desorption is favored as the temperature is increased, as evident from Eq, 1.2L Real solid surfaces are not energetically homogeneous as foreseen by Langmuir. He treated such heterogeneous surfaces as a composite of a number of homogeneous surfaces at different energy levels: 3 _Kp o- Se 4.23) where K's are equilibrium constants for different surfaces, and m, is the fraction of the total surface occupied by the i surface. Various models for real surfaces hhave been proposed since then for the adsorption isotherms and these are summa- rized in Table 1.5. The BET equation for physisorption will be treated in detail in the section to follow. The difference between the Langmuir model and the other real surface models lies in the dependence of the heat of adsorption on ‘coverage. It has been known for real surfaces that the heat of adsorption decreases with increasing coverage. Assuming a linear dependence of the heat of adsorption ‘on coverage leads to the Tempkin and BLK isotherms. The Freundlich isotherm results when the heat of adsorption is assumed to decrease exponentially with coverage; the heat of adsorption is constant for the Langmuir model ‘The fact that chemisorption leads to monolayer coverage is utilized in measur- ing the exposed surface area of metal catalysts, as was first demonstrated by Spenadel ‘and Boudart (1960) for a platinum catalyst, although the origin ofthe chemisorption Table 1 Adsorption Isotherms (Thomas and Thomas 1967) Nome therm Equations Arpleaions Chemisorption and Langone ‘em Physisorption Freundlich Chemizorption (at races) hye ome emisrsion (al Tempkin canon Bi -Love-Keer ac ae Rea sures com 4 exptQn/) = 0.20 Bronauer Emmet Teller Pent (BET) fy 4. ate constants With permison from Inoduction tthe Principles of Heterogeneous Catalysis Copyright by ‘Academic Pres Tae (London Ltd)24 HETEROGENEOUS REACTOR DESIGN ‘are summarized in Table 1.6 along with the relative strength of chemisorption, In order to arrive at the exposed metal surface area, the volume of the chemi. sorbed gas is measured, from which the number of adsorbed molecules can be calculated. Given the chemisorption stoichiometry (the number of surface atoms covered for each molecule of gas adsorbed) and the surface area occupied per metal atom, the number of molecules adsorbed can be converted to the exposed surface area of metal. For hydrogen, the stoichiometric number is almost always 2 since the hydrogen molecule usually dissociates upon adsorption and each hydro gen atom is adsorbed on one metal atom. For carbon monoxide, the number is cither 1 or 2, depending on whether it adsorbs in a linear form in which it covers ‘one metal atom, or in a bridged form covering two metal atoms. The number of Surface atoms per unit area of metal varies slightly withthe erystallographie plane, but for all metals it is about 10° atoms per square meter, e.g, 1.5 to 1.6 * 10% for Fe, Co, and Ni; 1.25 to 1.33 X 10! for Pt, Pd, Ir, and Rh; and 1.15 x 10! for Ag (Anderson, 1975). The gas most appropriate for chemisorption is hydrogen, but oxygen and carbon monoxide are also useful. The major problems associated With chemisorption measurements are the preparation of uncontaminated metal Surfaces, determination of the presence or absence of adsorption on the support, and determination of the proper stoichiometric number. For instance, the stoiehio metric number may be affected by the crystallite size, which may change during the preparation stage of cleaning the metal surface. Clean surfaces can be prepared by reduction with hydrogen at temperatures exceeding about 400°C for several hours, possibly preceded by an oxidation cycle, and then outgassing under high vacuum for several additional hours. The optimum temperature and pressure will vary with the system and must be established experimentally. Studies must also be made with the support by itself to establish its possible contribution to the cxerall adsorption. High-area supports contribute excessively to hydrogen adsorp. tion at 77 to 90K. Preferred conditions for dispersed platinum are about 273 to 300K and about 0.01 to 0.03 kPa, and for dispersed nickel, about 273 to 300K Table 1.6 Chemisorption of Gases on Metals (after G.C. Bond 1974) ee mre Gane. Bond 1978) Group Meals Os > Cie > Gilly > CO > Hy > > Ny pan ee Lae Len oe ee A TM Fe + + See ae Be Nico + + + Peau EH EH Br RRP, Phir + + + Pane one ME By Mn, Gs + + + eH Eee c AL AL oF + Sn DMITRY Nat = CoE E Mg Ag. Za + - a - ea “+ strong chemisorption = unobservable = weak chemisorption CATALYSTS AND CHARACTERIZAHON 23 ‘Table 1.7 Comparison of Methods (Freel 1972) Electron Microscopy emisorgtion cae crs ‘erage Cotte Sie (am) re Surfoe/Pt Bulk on 098 ls 036 2 032 o2 ra 035 ais as aia Reprinted with permission from Journal of Catalysis. Copyright by Academic res. and 20 kPa (Satterfield 1980). A recent review by Farrauto (1974) concludes thet, with the exceptions of Pt and Ni using Hz and of Pd using CO, chemisorption results should be used with caution for the determination of the surface area. It should be recognized, however, that a reliable determination of the relative surface area of a metal catalyst can always be made given proper care, even if this measure- ‘ment does not determine the absolute surface area. A chemisorption measurement should be complemented by the X-ray line broadening method and electron micros- copy. An example of a comparison of two of the three methods is given in Table 1.7 in terms of dispersion for a supported platinum catalyst with 2 wt% loading. ‘The X-ray line broadening method utilizes the Scherrer equation: ke Feos 6 (24) where B= net breadth of an X-ray diffraction line r= crystallite size d= wavelength of X-ray applied @=Bragg angle H ‘=a constant, usually equal to unity As the equation indicates, the breadth of an X-ray diffraction line is inversely Proportional to the crystallite size and therefore the line gets increasingly sharper as the crystallite size increases. The line becomes too diffuse to be useful for sizes smaller than approximately 5 nm, whereas it becomes too sharp for sizes larger than 50 nm. The useful range is 5-50 nm. The method of electron microscopy is treated in some detail in Chapter 6. ‘A description (Farrauto 1974) of apparatus and procedures used for chemi sorption is given below. ‘Static Vacuum System. Usually made of glass and equipped with appropriate evacua- tion devices, such as a roughing and difusion pump, the static vacuum system is ‘the most commonly used apparatus for measuring gas adsorptions. The general princi- ple involves measuring the amount of gas remaining in the manifold system after contact with a sample, By knowing the amount of gas initially present and subtracting from it the amount remaining after equilibration with a sample, the extent of adsorption26 HETEROGENEOUS REACTOR DESIGN can be determined. Most often a pressure device, such as a mercury manometer or commercial mechanical device, is used to follow pressure changes caused by gar. sample interactions. Pressure changes in a constant volume system would then be Proportional, through the gas laws, to adsorption (or desorption) of a given quantity of gas It is assumed that the manifold and sample container volumes have been accurately determined. Prior to gas adsorption, it is common practice to pretreat or condition the ‘atalyst surface, Frequently high temperatures, about $00°C. are employed. Therefore, 4a sample furnace isan essential part of the apparatus. After pretteatment, evacuation at Pretreattent temperature, and cooling to adsorption temperature, itis necessary {determine the dead space volume (the volume inthe sample tube which the adsorbate ‘would occupy provided no adsorption occurred). Helium is most offen sed for this Purpose. After evacuation, the adsorbate is added to the manifold and its presure ‘noted, Subsequently, itis expanded into the sample chamber, and adsorption, Iran}, commences. The pressure is monitored ntl no further variation with le fe now ‘The pressure over the sample can then be increased via a gas burette and readings ‘gsin taken until equibrium is established. When there is no longer gas uptake ty {he sample with increasing pressure, the desirable potion ofthe isotherm is completa, and the total volume adsorbed, expressed at S.T-P. per gram, can be determined ‘Anderson (1968), Continuous Flow Apparatus. A stream of gas composed of adsorbate and inert farrier is passed through a previously outgassed catalyst, usually purged with om inert gas at a high temperature, and the concentration of adsorbate monitored fre, quently via a thermal conductivity cell. Assuming the latter is used for detection, {ondltions are adjusted so that no as will adsorb (accomplished by selecting a suitable {temperature or by allowing for sample bypass) and the bridge balanced. When conde tions are such that gas will adsorb, the detection system responds to 2 change in {he composition ofthe gas mixture, and a response occurs proportional to gas adsorbed, One may also adjust the conditions to monitor the desorption of adsorbate from the catalyst. Further details can be found in the article by Eberly (196, ‘The major advantage of the method is rapid determinations. However, ditfue Sonal limitations and slow signicant chemisorption proceses may not be readily detected. The adsorption portion of the measurement may also include some physical adsorption, However, ths can be minimized by proper temperature and partial pressure diestments. The apparatus is suited for physical adsorption (that i BET) as well 4s chemisorption, Methods for determining the exposed surface area of oxide catalysts are ‘much more rudimentary than those for metal catalysts and acidic catalysts. Specific poisoning is suggested by Kndzinger (1976) for measuring the surface area, Gandhi and Shelef (1973) suggested the use of the Freundlich isotherm for the determination ofthe surface area of copper oxide using NO as the adsorbate. Recently, » method applicable to any catalyst, including metal compounds, has been developed (Miller and Lee 1984) for the surface area measurement based on selective physisorption, ‘The quantity typically used in place of the exposed surface area for ccidic CATALYSTS AND CHARACTERIZATION 27 catalysts is the number of acid sites, or equivalently, the acid amount in mol/g. For the determination of the acid amount (Satterfield 1980), a sample of the solid acid as a powder is suspended in an inert nonaqueous liquid, e.g., benzene, and is titrated with a base, utilizing an indicator. The titrating base must be a stronger base than the indicator, and n-butylamine, pK ~ +10, is often used for this purpose. AS the base is added it adsorbs on acid sites, the strongest ones frst, and ultimately it displaces indicator molecules from the solid. When the indicator has been substan- tially replaced, the color changes. At this equivalence point [B]/[BH"] ~ 1, where [B] and [BH] are the concentrations of the neutral base and its conjugated acid If the pKa (= log Ke) value of the indicator is, say +3.3, then the amount of ‘base added is equivalent to the amount of acid sites having the Hammet acidity function He $ +33, where: (B) BH: Ho= pKa + log 25) and where Ko is the equilibrium constant of dissociation of the acid. By amine titration with indicators having different pK, values, the amount of acid sites having strengths exceeding the various corresponding values of pK (~ Ha) may be determined. This method gives the sum of Bronsted and Lewis acid sites. The titration method is not suitable for molecular-sieve zeolite catalysts in which the pore size is so small that the indicator molecules cannot penetrate into the interior. Some other disadvantages of the method are that water must be rigor ously excluded, since water may react with the surface and alter the acidic character of the solid the time required for equilibrium may be long, amounting sometimes to days; and measurements are made under conditions much different than those ‘occurring during a reaction. Another method isto measure the amount of a gaseous base, such as ammonia, Pyridine, or quinoline, adsorbed at elevated temperatures under a specified set of Conditions. This has the advantage of allowing the study of a catalyst under condi- tions more nearly similar to those of reaction. Ammonia has been used extensively. Catalysts may be compared in terms of the amount of ammonia adsorbed as a function of the temperature over @ range such as 150 to 500°C. A minimum temperature of about 150°C is necessary to eliminate physical adsorption. By infra- ted spectra it is possible to distinguish between Bronsted and Lewis acid sites. More details on characterizing the acidity of solid catalysts are described in reviews by Formi (1974), and by Bensei and Winquist (1978). Catalytic Activity and Correlations ‘The catalytic activity, expressed in terms of the specife activity, is often correlated to the heat of adsorption for metal catalysts. The heat of adsorption is a measure Of the stability of the bonding between the adsorbed gas and the surface metal atom, AS pointed out in Section 1.2, the stability of this bonding should not be28 HETEROGENEOUS REACTOR DésicN ‘ Figure 19° Effect of heat of ad- i sorption on catalytic activity: the Adsorption _,'__ Adsorption Caaeniatateaatl : too weak ‘too strong ino ta Catalytic activity Parameter measuring strength of adsorption too high or too low for the catalyst to be effective. This effect is known as the Sabatier’ principle, or as the voleano curve, shown in Figure 1.9. A specific example (Bond 1974) for the hydrogenation of ethylene is shown in Figure 1.10. Hi the rates of hydrogenation relative to rhodium are given in terms of the initial heat of adsorption of hydrogen (AH). Because of the correlations that exist between the initial heat of adsorption and the periodic group number, or between the initial Figure 1.10 An example of the ‘volcano curve for the hydrogena- tion of ethylene. Open points, eva porated metal films: fled points, siliea-supported metals: circles, first row transition metals: squares, second row transition metals: trian. les, third row transition metals (After G.C. Bond 1974) loaqq (relative rate) 100 150-200 1 BH, (kd mot") CATALYSTS AND CHARACTERIZATION 29 Figure 1.11 Variation of \ the initial heat of adsorption a of oxygen on evaporated \ films: circles are forthe fist transition series, squares the second transition series, and ‘riangles the third transition a series. (After G.C. Bond 1974) 250 3 E150 dy i N = ° 4 100h oN \ \ av 50} Wa MIA MM, WM, Periodic group number heat of adsorption and the heat of formation of the highest metal oxide per atom ‘of metal, the abscissa can also be the periodic group number or the heat of formation, ‘Some typical examples are given in Figures 1.11 (Bond 1974) and 1.12 (Tanaka and Tamaru 1963). It is more difficult to find useful correlations between catalytic activity and chemical or physical propertis for metal oxides than for metals. One of the problems is the difficulty of obtaining the exposed surface area and thus the specific activity. ‘The chemisorption of gases is more complex on oxides than on metals. To start with, the adsorbed molecule may be attached either to a cation or to an oxide ion. Further, it has long been known that the adsorption of reducing gases such 4 hydrogen and carbon monoxide is substantially irreversible in the sense that, when desorption is attempted, only water and carbon dioxide, respectively, can be obtained. The process has clearly led to reduction of the surface, and it may bbe guessed that the adsorbed species resided or were associated at least in part with oxide ions. It is also difficult to find a suitable chemical or physical property tocorrelate with the catalytic activity. Many of the transition metals exhibit variable valence and in such cases several oxides are possible, each with its distinctive structure, stability, and catalytic ability. Some of the cations at the surface may be in a higher oxidation state than those in the bulk. It is, however, possible to30 Hereroceneous REACTOR Desicn 300 200 Hg, keal/mot 100 o 100 200 300 — AH, keal/g atom of metal Figure 1.12 Correlations between the heat of adsorption and the heat of formation of the highest rhetal oxide per atom of metal, (Tanaka and Tamara 1963; reprinted with permission from Journal of Catalysis. Copyright by Academic Press.) proceed a little. If the breaking of the bond joining the relevant oxygen species to the surface isthe rate-determining step in oxidation reactions, for which oxide catalysts are usually used, an inverse correlation should be found between the catalytic rate and the oxygen-catalyst bond strength. This quantity has been obtained either from the temperature dependence of the oxygen dissociation pressure or fom the het of formation ofthe oxide per oxygen atom. While reasonable correla- tions have been found for the oxidation of hydrogen and of many hydrocarbons, they are still only approximate and based on too few examples. For acidic catalysts, the catalytic activity expressed in terms of rate constant or percent conversion is usually correlated to the acid amount. Typical examples are given in Figures 1.13 (Johnson 1955) and 1.14 (Ward and Hansford 1969) The Bronsted acid groups shown in Figure 1.14 were obtained from infrared peak intensity. ‘As evident in the preceding discussions on the metal and metal oxide catalysts, the heat of adsorption plays a key role in characterizing a catalyst. The heat of adsorption can also be used to gain an understanding of the nature of catalytic reactions. Three different experimental methods can be used for the determination of heats of adsorption: adsorption experiments (isotherms), the calorimetric method, and temperature-programmed desorption (TPD). In the first method, the adsorption isotherms obtained at different temperatures are used. The values of d In P/dT at constant » (volume of gas adsorbed) are calculated as a function of » through CATALYSTS ANO CHARACTERIZATION 31 04 02 03 of 05 06 ‘Acid amount, mmotig Figure 1.13 Propylene polymerization activity versus acid amount fora series of SiOy-AlLOs catalysts Johnson 1955): Catalyst A BC DE F G Al:0;(wt%) 012 032 104 205 356 103 25.1 numerical differentiation of the curve obtained by cross plotting. These values are used in the Clausius-Clapeyron equation: @inPY __Q- (oF). -ar to obtain the (differential) heat of adsorption. The value of Q thus obtained is called the isosteric heat of adsorption. The calorimetric method, in which the heat liberated upon the adsorption of gas on a cooled surface is actually measured, tives the average value of the heat of adsorption over the range of surface coverage studied. Differential heats of adsorption can also be determined calorimetrically by admitting small quantities of vapor at a time or by differentiating integral data. Because heats of adsorption frequently decrease with increasing surface cover- age, and not always in the same way, itis convenient to use the heat extrapolated to zero surface coverage, the so-called initial heat of adsorption. As indicated ‘earlier, this initial heat of adsorption is usually used for correlations. ‘Temperature-programmed desorption (TPD) (Cvetanovic and Amenomiya 1972) yields not only the heat of adsorption but also information on groups of different active sites, which are useful in characterizing reactions. While the Rash filament method in which an adsorbate is desorbed from a rapidly heated filament in an ultra-high vacuum environment has been used for a long time, the usefulness cof the TPD method lies in the fact that the adsorbed gas i desorbed in a programmed32, HETEROGENEOUS REACTOR Desicn Catalyst activity (% conversion) ° L 1 L 04 08 12 is 20 Concentration of Gransted seid groups (arbitrary units) Figure 1.14 Activity for o-aylene isomerization of SiOs-AlO; catalysts of various AliOs contents: dark circles are for samples heated to 500°C and open circles are for those heated to 425°C. (Ward and Hansford 1969; reprinted with permission from Journal of Catalysis. Copyright by Academic Press.) ‘manner from a conventional catalyst under conditions approaching more closely those prevailing during reaction. The bed of catalyst particles in the sample cell is quite shallow in this method, and in essence quite similar to the arrangement Of the differential reactor typically used for kinetic experiments for heterogeneous reactions. The particles are presorbed with a suitable gas that chemisorbs on the solid adsorbent and are then subjected to thermal desorption. The carrier (typically helium), which elutes the desorbed gas, is passed through a conductivity cell for ‘concentration measurements. In the absence of any appreciable diffusional effects, the mass balance for the shallow bed (differential bed) is that for a continuously-stirred tank reactor: 4 a, (Oma) = F (Cin = a“ Raa OTIC (1.26) CATALYSTS AND CHARACTERIZATION 33 where 6 = fractional coverage of surface V, = volume of solid particles F carrier gas flow rate centration of desorbed gas in the exit carrier stream mount of chemisorbed gas when 6 is unity per unit volume of solid particles Gin = zero since no adsorbate is fed to the bed at the start of thermal desorption. oe ‘The heating program is such that the temperature increases linearly with time: T=T.+Bt 27 where 8 is a constant. Combining Eqs. 1.26 and 1.27 yields: Vin 40 __ Yim dT 40 Fd F der Ot (1.28) ‘The net rate of desorption is equal to the rate at which the desorbed gas is carried away by the carrier: FC= Velvnke@ — kaC (1 ~ 8)] (129) where ka and ky are the rate constants for desorption and adsorption, respectively. If the surface is energetically homogeneous, the desorption rate constant is given by: a= ke expl—Ea/RoT) (13) The heating rate f can be set at a value high enough to neglect readsorption, ‘Then, Eqs. 1.28 and 1.29 can be combined to give: ote ge Mee rym sn Voruko c= F O exp(—Ea/RgT) (1.32) where Eq, 1.30 has been used. Data are obtained in the form of C versus time and T versus time, which yields C versus 7. At an extremum on this curve, C/AT = 0, oF from Eq, 1.32: ac a (68-4480) eu-eunry