Catalysis and automotive pollution control IV: proceedings of the Fourth International Symposium (CAPoC4), Brussels, Belgium, April 9-11, 1997

Studies in Surface Science and Catalysis 116 CATALYSIS AND AUTOMOTIVE POLLUTION CONTROL IV This Page Intentionally Le...

Author: N. Kruse | A. Frennet | J.-M. Bastin

21 downloads 1001 Views 33MB Size Report

This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!
Report copyright / DMCA form

DOWNLOAD PDF

Studies in Surface Science and Catalysis 116 CATALYSIS AND AUTOMOTIVE POLLUTION CONTROL IV

This Page Intentionally Left Blank

Studies in Surface Science and Catalysis A d v i s o r y Editors: B. Delmon and J.T. Yates

Vol. 116

CATALYSIS AND AUTOMOTIVE POLLUTION CONTROL IV Proceedings ofthe Fourth International Symposium (CAPoC4), Brussels, Belgium, April 9-11, 1997

Editors N. Kruse, A. Frennet and J.-M. Bastin

Chimie Physique des Surfaces - Cata/yse H&t~rogbne, Universit~ Libre de Bruxe//es, Brussels, Belgium

1998 ELSEVIER A m s t e r d a m - - L a u s a n n e - - - N e w Y o r k - - Oxford - - S h a n n o n --- S i n g a p o r e - - Tokyo

ELSEVIER SCIENCE B.V. Sara Burgerhartstraat 25 ~.O. Box 211, 1000 AE Amsterdam, The Netherlands

Library of Congress Cataloging in Publication Data. A catalog record from the Library of Congress has been applied for.

ISBN 0-444-82795-1 © 1998 Elsevier Science B.V. All rights reserved. No part of this publication may be reproduced, stored in a retrieval system or transmitted in any form or by any means, electronic, mechanical, photocopying, recording or otherwise, without the prior written permission of the publisher, Elsevier Science B.V., Copyright & Permissions Department, RO. Box 521, 1000 AM Amsterdam, The Netherlands. Special regulations for readers in the U.S.A.- This publication has been registered with the Copyright Clearance Center Inc. (CCC), 222 Rosewood Drive, Danvers, MA 01923. Information can be obtained from the CCC about conditions under which photocopies of parts of this publication may be made in the U.S.A. All other copyright questions, including photocopying outside of the U.S.A., should be referred to the copyright owner, Elsevier Science B.V., unless otherwise specified. No responsibility is assumed by the publisher for any injury and/or damage to persons or property as a matter of products liability, negligence or otherwise, or from any use or operation of any methods, products, instructions or ideas contained in the material herein. Transferred to digital printing 2005 Printed and bound by Antony Rowe Ltd, Eastbourne

CONTENTS PRELIMINARIES Foreword ...........................................................................................................................xi Introductory Remarks and Outlook ................................................................................ xiii Acknowledgments ............................................................................................................ xv Financial Support ............................................................................................................ xvi Organizing committee ................................................................................................... xvii Scientific advisory board .............................................................................................. xviii

GENERAL LECTURES

Global trends in motor vehicle pollution control. a 1997 update M.P. Walsh .........................................................................................................................3

Contribution offossil fuels and air pollutants emissions in Belgium since 1980. The role of traffic W. Hecq .............................................................................................................................. 5

Auto Emissions after 2000. The Challenge for the Catalyst Industry R.A. Searles ......................................................................................................................23

Diesel engine development routes towards very low emissions P.L. Herzog ....................................................................................................................... 35

THREE WAY CATALYSTS

Novel Pd-based three-way catalysts R. van Yperen, D. Lindner, L. Muflmann, E. S. Lox, T. Kreuzer .................................... 51

Comparative behaviour of standard Pt/Rh and of newly developed Pd-only and Pd/Rh three-way catalysts under dynamic operation of hybrid vehicles S. Tagliaferri, R.A. K6ppel, A. Baiker ............................................................................. 61

Comparative three-way behaviour of Pt, Pd and Rh single and combined phases in a full gas mixture with oscillating feedstream

J.R. Gonz~ilez-Velasco, T.A. Botas, R. Ferret, M.A. Guti6rrez-Ortiz .............................. 73

Effect of alkaline addition on hydrocarbon oxidation activities of palladium three-way catalyst H. Shinjoh, N. Isomura, H. Sobukawa, M. Sugiura ......................................................... 83

Ethanol oxidation on three-way automotive catalysts. Influence of Pt-Rh interaction A. Marques da Silva, G. Corro, P. Marecot, J. Barbier .................................................... 93

Reduction of NO by CO on manganese promoted palladium catalysts

J.F. Trillat, J. Massardier, B. Moraweck, H. Praliaud, A. Renouprez ............................ 103

vi

Light-offperformance over cobalt oxide- and ceria-promoted platinum and palladium catalysts M. Skoglundh, A. T6rncrona, P. Thorm~.hlen, E. Fridell, A. Drewsen, E. Jobson ........ 113 CATALYST AGEING- POISONING

Influence of catalyst deactivation on automotive emissions using different cold-start concepts

T. Kirchner, A. Donnerstag, A. K6nig, G. Eigenberger ................................................. 125

Measurement of the ceria surface area of a three-way commercial catalyst after laboratory and engine bench aging E. Rogemond, N. Essayem, R. Fr6ty, V. Perrichon, M. Primet, S. Salasc, M. Chevrier, C. Gauthier, F. Mathis .................................................................................................... 137

The effect of the ageing procedure upon the activity of a three way catalyst working under transient conditions

R. Roh6, V. Pitchon, G. Maire ....................................................................................... 147

Causes of deactivation and an effort to regenerate a commercial spent three-way catalyst T.N. Angelidis, M.M. Koutlemani, S.A. Sklavounos, Ch.B. Lioutas, A. Voulgaropoulos, V.G. Papadakis, H. Emons ............................................................ 155

Pb poisoning on Pd-only TWC catalysts S. Sung, R.M. Smaling, N.L. Brungard .......................................................................... 165

Effect of ageing on the redox behavior of Ce in three-way catalysts

S. Irusta, A. Boix, J. Vassallo, E. Mir6, J. Petunchi ....................................................... 175

The CeO2-Zr02 system. redox properties and structural relationships

G. Vlaic, R. Di Monte, P. Fornasiero, E. Fonda, J. Kaspar, M. Graziani ...................... 185

DENOx NOBLE CATALYSTS

Kinetics of the reduction of NO by C3H6 and C3H8 over Pt based catalysts under lean-burn conditions

R. Burch, T.C. Watling ...................................................................................................199

N20 and NO2formation during NO reduction on precious metal catalysts P. Bourges, S. Lunati, G. Mabilon .................................................................................213

Mechanistic investigation on the selective reduction of NO with propene in the presence of oxygen over supported platinum S. Eckhoff, D. Hesse, J.A.A. van den Tillaart, J. Leyrer, E.S. Lox ............................... 223

Platinum-titania-sepiolite monolithic catalysts for the reduction of nitric oxide with propene in lean-burn conditions P. Avila, J. Blanco, C. Knapp, M. Yates ........................................................................233

vii

DeNOx mechanism on platinum based catalysts V. Pitchon, A. Fritz, G. Maire ........................................................................................243

Electrochemical promotion in emission control catalysis: the role of Na for the Ptcatalysed reduction of NO by propene I.V. Yentekakis, A. Palermo, M.S. Tikhov, N.C. Filkin, R.M. Lambert ....................... 255

Promoting effect of zinc in DeNOx reaction over Pt/A1203 A. Bensaddik, N. Mouaddib, M. Krawczyk, V. Pitchon, F. Garin, G. Maire ................ 265

Catalytic properties of Palladium exchanged ZSM-5 catalysts in the reduction of nitrogen monoxide by methane in the presence of oxygen: nature of the active sites P. G61in, A. Goguet, C. Descorme, C. L6cuyer, M. Primet ........................................... 275

Influence of the platinum-support interaction on the direct reduction of NOx unde lean conditions F. Acke, B. Westerberg, L. Eriksson, S. Johansson, M. Skoglundh, E. Fridell, G. Smedler ............................................................................285

DENOx BASE CATALYSTS

A comparative study of the activity of different zeolitic materials in NOx reduction from simulated diesel exhausts M. Guyon, V. Le Chanu, P. Gilot, H. Kessler, G. Prado ............................................... 297

The effect of Al and Cu content on the performance of Cu/ZSM-5 catalysts at the exhaust of high efficiency spark ignition engines P. Ciambelli, P. Corbo, M. Gambino, F. Migliardini ..................................................... 307

Kinetic study of the selective catalytic reduction of nitric oxides with hydrocarbon in diesel exhausts B. Westerberg, B. Andersson, C. Ktinkel, I. Odenbrand ................................................ 317

Steady state and transient activity patterns of Cu/ZSM-5 catalysts for the selective reduction of nitrogen oxides J. Connerton, R. W. Joyner ............................................................................................327

Selective reduction of nitrogen oxide with hydrocarbons and hydrothermal ageing of Cu/ZSM-5 catalysts P. Denton, Z. Chajar, N. Bainier-Davias, M. Chevrier, C. Gauthier, H. Praliaud, M. Primet ........................................................................................................................335

Transient kinetic study of NO decomposition on Cu-ZSM-5 catalysts Z. Schay, I. Kiricsi, L. Guczi .........................................................................................347

Stability of cerium exchanged zeolite catalysts for the selective catalytic reduction of NOx in simulated diesel exhaust gas W.E.J. van Kooten, H.P.A. Calis, C.M. van den Bleek ................................................. 357

viii

Study on copper and iron containing ZSM-5 zeolite catalysts: ESR spectra and initial transformation of NO J. Varga, J. Halfisz, D. Horvfith, D. M6hn, J.B. Nagy, G. Sch6bel, I. Kiricsi ................ 367 KINETICS- MECHANISMS

The use of isotope transient kinetics within commercial catalyst development J.C. Frost, D.S. Lafyatis, R.R. Rajaram, A.P.Walker .................................................... 379

Kinetic study of the ethene oxidation by oxygen in the presence of carbon dioxide and steam over Pt /Rh /Ce02 /y-Al203 R.H. Nibbelke, R.J.M. Kreijveld, J.H.B.J. Hoebink, G.B. Marin .................................. 389

Three-way catalytic converter modelling. numerical determination of kinetic data C. Dubien, D. Schweich ................................................................................................. 399

NO + CO -+1/2 N: +CO: differentiated from 2NO + CO -+ N20 + CO: over rhodia/ceria catalysts using 15N180 and ISC160 reactants or time-resolution of products J. Cunningham, N.J. Hickey, F. Farrell, M. Bowker, C. Weeks .................................... 409

Investigation on the role of rhodium on the kinetics of the oxidation of CO by NO over Pt-Rh catalysts P. Granger, J.J. Lecomte, C. Dathy, L. Leclercq, G. Mabilon, M. Prigent, G. Leclercq .............................................................................. 419 MODEL SYSTEMS ] STUDIES

CO oxydation on Pd (11 O) M. Bowker, I.Z. Jones, R.A. Bennett, S. Poulston ......................................................... 431

In- situ ESR of Rh/y-Al20s and Rh/ZSM-5 S.G. Lakeev, A.V. Kucherov, M. Shelef ........................................................................ 441 MISCELLANEOUS

Substrate contributions to automotive catalytic converter performance: the role of channel shape on catalyst efficiency J. Paul Day ...................................................................................................................... 453

Evaluation and characterization of catalysts for alternative-fuelled vehicles. A study of the influence of catalyst composition on activity and by-product formation L.J. Pettersson, A.M. Wahlberg, S.G. J~irfis ................................................................... 465

SHS catalysts for purification of exhaust gases from internal combustion engines E.H. Grigoryan, I.P. Borovinskaya, A. G. Merzhanov .................................................. 477

ix

Catalytic decomposition of high-concentration nitrous oxide N20 H.C. Zeng, M. Qian, X.Y. Pang ..................................................................................... 485

Structure an activity of Cu/Cr/Sn02 environmental control catalysts P.G. Harrison, W. Azelee, A.T. Mubarak, C. Bailey, W. Daniell, N.C. Lloyd ............. 495

Preparation and study of thermally stable washcoat aluminas for automotive catalysts Z.R. Ismagilov, R.A. Shkrabina, N.A. Koryabkina, D.A. Arendarskii, N.V. Shikina...507

Ensuring substrate retention. Part 2 4 J.Kisenyi, K. Soe, P. Leason, C. Tooby, D. Pritchett, G. Morgan, M. Zillikens ........... 513 STORAGE: NOx AND OXYGEN

A catalytic NOx management system for lean burn engines J. Feeley, M. Deeba, R.J. Farrauto ................................................................................. 529

Investigations of NOx storage catalysts E. Fridell, M. Skoglundh, S. Johansson, B. Westerberg, A. T6rncrona, G. Smedler .... 537

Oxygen storage capacity of three-way catalysts : a global test for catalyst deactivation R. Taha, D. Duprez, N. Mouaddib-Moral, C. Gauthier .................................................. 549

NO Reduction by CO over Pd /CeO:-Zr02-Al:Os Catalysts R. Di Monte, P. Fornasiero, J. Kaspar, A. Ferrero, G. Gubitosa, M. Graziani .............. 559

Comparative sulfur storage on Pt catalysts: effect of the support (CeOz Zr02 and CeO2-Zr02) P. Bazin, O. Saur, J.C. Lavalley, A.M. Le Govic, G. Blanchard ................................... 571

Oxygen storage capacity in perovskite-related oxides. the role of over-stoichiometric oxygen in three-way catalysis N. Guilhaume, M. Primet ............................................................................................... 581

Influence of ceria dispersion on the catalytic performance of Cu/(CeO2)/Al203 catalysts for the CO oxidation reaction A. Matinez-Arias, J. Soria, R. Catalufia, J.C. Conesa, V. Cort6s Corberfin ................... 591

Some surface chemical features of Pt catalysts supported on Al20s and Ce02 /Al20s G. Magnacca, G. Cerrato, C. Morterra ........................................................................... 601

Fundamental Properties of new cerium- based mixed oxide as TWC component S. Bernal, G. Blanco, M.A Cauqui, P. Cochardo, M. Pintado, J.M. Rodriguez-Izquierdo, H. Vidal ............................................................................... 611

DIESEL

Improved soot oxidation by fuel additives and molten salt catalysts S.J. Jelles, J.P.A. Neefi, B.A.A.L. van Setten,, M. Makkee, J.A. Moulijn ................... 621

Investigation of copper-cerium oxide catalysts in the combustion of diesel soot D. Courcot, E. Abi-Aad, S. Capelle, A. Abouka'fs ......................................................... 625

Catalytic ceramic filter for diesel soot removal: preliminary investigations P. Ciambelli, V. Palma, P. Russo, S. Vaccaro ................................................................ 635

Catalytic oxidation of model soot by chlorine based catalysts G. Mul, J.P.A. Neefl, M. Makkee, F. Kapteijn, J. A. Moulijn ....................................... 645

Copper catalysis for particulate removal from diesel exhaust gas. Copper fuel additives in combination with copper coatings J.P.A. Neefl, S.J. Jelles, M. Makkee, J. A. Moulijn ....................................................... 655

Supported liquid phase catalysts : a new approach for catalytic oxidation in diesel exhaust particulate emission control S.J. Jelles, B.A.A.L. van Setten, M. Makkee, J. A. Moulijn .......................................... 667 AUTHOR INDEX . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

675

LIST OF PARTICIPANTS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 681

xi

FOREWORD In spite of the energy crises and the recession, an explosive growth of the world vehicle population from 50 to 700 million vehicles in 50 years was observed, as analyzed by Michel Walsh in his introductory lecture. On the other hand, in spite of the poor yield of internal combustion engines associated to the inevitable production of some gazeous pollutants, economical reasons essentially have made their use to continue and probably for still an important number of years. The resulting increase of gazeous pollutants in our atmosphere coming from exhaust gas of automobile vehicles has enhanced the problem of the elimination of these pollutants produced by internal combustion engines. This is why there has been continuing interest in the organization of meetings on the depollution problem. Catalysis was considered and has been proven to be the best solution to lower the content of exhaust ges in pollutants The use of catalytic processes started to be studied in the United states already in the early seventies. That research was mostly conducted by the two giant american auto industries: General Motors and Ford. During many years no need for international scientific exchange on the problem was considered. In the eighties, the european countries started to show some interest to that problem. It is only in june 1984 that the EC Commission proposed standarts of permissible pollutants in the exhaust gas from motor vehicles to be introduced in Europe ; these standarts were approved by the Ministers of the Environment one year later. Very quickly, a number of Academic research laboratories started working on the subject, and namely on the development of new catalysts. We thought that a need for exchange of results and of ideas had appeared and I have initiated the organization of international meetings on this topics at the University of Brussels under the title "Catalysis and Automotive Pollution Control" associated with the acronym CAPoCFour meetings have been organized in Brussels in 1986 (CAPoC1), in 1990 (CAPoC2), in 1994 (CAPoC3) and in 1997 (CAPoC4). The proceedings were published as an issue of the series " Studies in Surface Science and Catalysis" published by Elsevier, respectively as vol. 30 for CAPoC1, vol. 71 for CAPoC2 and vol. 96 for CAPoC3. The present volume contains the proceedings of the last of these meetings, CAPoC4, that took place at the University of Brussels on april 9-11, 1997. I have been the organizer and the chairman of the three first of these meetings and the Honorary President of this fourth one. I would like to take the opportunity to thank some persons for their special contribution in the organization of these meetings and the publishing of their proceedings. First of all, my colleague Andre Crucq who joined me to start these meetings and who took in charge the heavy and important role of secretary of CAPoC 1 and CAPoC2. This function has succesfully been taken over by Jean-Marie Bastin for CAPoC3 and CAPoC4.

xii It is also a pleasure to stress the continuing interest and the enthousiasm of some colleagues who were active members of the organizing committee of all the four meetings namely: Andr6 Pentenero, Michel Prigent, Ginette and Lucien Leclercq, Walter Hecq and Georges Poncelet. Finally, I am glad that after my retirement in 96 from my position heading the catalysis group at the ULB, my successor, Prof. N. Kruse, accepted to continue the organization of this series of meetings. I also thank him to have accepted to be the chairman of this fourth meeting. From the startpoint, these meetings had an important succes, in spite of the otherwise very restricted topics. These last years, the problem of pollution by the emission from the engines of automobile vehicles has been examined in one of the sessions of several more general meetings devoted to transportation problems or to fuels production. The continuing succes of the CAPoC's meetings that comes out the following table puts well in evidence the still large importance of the topic. CONGRES CAPoC1 CAPoC2 CAPoC3 CAPoC4

EUR

USA

127 197 223 212

20 23 18 10

PARTICIPANTS Other Total Industry 30 177 106 40 260 160 38 279 133 11 232 97

Acad. Labs. 71 100 119 135

PAPERS Submitted Accepted

38 66 131 88

28 42 79 68

It appears that most of the participants come from Europe. The total number of participants is rather constant from CAPoC2 to CAPoC4. On the other hand the number of participants from industry is progressively decreasing whereas that from academic laboratories increases.

A. FRENNET Honorary President CAPoC4

xiii

INTRODUCTORY

REMARKS

AND OUTLOOK

The fourth congress on catalysis and automotive pollution control (CAPoC4) was held in Brussels from 9-11 April 1997. Following the habit of its predecessors, this congress started with a number of keynote lectures both surveying the field on the whole and covering aspects ranging from vehicle/catalyst technology to legislative regulations. M. Walsh, in his presentation, elaborated on the continuing growth of global vehicle population with the highest rates being found in developing countries. Accepting that pollution knows frontiers, a clear need has been demonstrated for a worldwide move to pollution control. W. Hecq reviewed the EU emission regulations from 1970 up to now and examined the impact that they had on the emissions of the main pollutants from road vehicles. Based on measurements of pollutant concentrations in Belgium and, more specifically, in urban areas like Brussels, it became clear that certain improvements on a per-car-basis are destroyed by a general growth of the car fleet, especially diesel cars. Given the EU proposals for 2000 and 2005 emission standards of gasoline/diesel fuelled vehicles, R. A. Searles reviewed the state-of-the-art aftertreatment technology for the control of emissions. He also emphasized that in order to meet the 2005 standards, further technological improvements are necessary in catalyst performance, trapping and adsorption along with an optimization of engine managements and control systems. Based on the fact of increasing proportions of diesel fuelled engines and respective problems in achieving legislative standards, P. L. Herzog reviewed the main parameters influencing the emissions of NOx and particulates. Even in consideration of remarkable improvements in engine and combustion technology as well as in electronic control, P. L. Herzog sees the development of highly effective exhaust gas aftertreatment systems playing a key role in future development routes. There is no doubt to me that the four keynote lectures enjoyed great esteem and gave the prelude to a number of most interesting communications on various subjects in the field. The large number of accepted papers (68) made it necessary to shift some of them into a poster session. As a rule, poster and oral contributions were equally assessed and no discrimination was made in the proceedings. As its predecessors, CAPoC4 proved to be a most suitable platform for discussing technological improvements and developments along with future perspectives and challenges. In the light of new results and further legislative regulations, the following topics were intensely discussed: • low light-off behaviour based on improved catalysts and substrate formulations • efficient adsorber systems for storage of hydrocarbon emissions • electrically heated catalyst systems ahead the main catalyst or, alternatively, close coupled catalysts (at the manifold of the engine)

xiv

lean DeNOx catalysts allowing for decomposition of NO× in the oxygen-rich exhaust of direct injection gasoline engines and high speed injection diesel engines or, alternatively, NOx trapping/reduction in a hybrid approach collection and destruction of dry particulates or soot. During the conference a poll was made on the structure of the congress. Although the tenor was to keep the general format (3 day meeting, every 2-3 year's), opinion was expressed to introduce short oral communications of 10 minutes duration (plus 5 minutes for discussion). The organizing committee will take care of this point and make respective arrangements for CAPoC5. Stimulating suggestions were made on future topics. Accordingly, all participants seemed to agree that the search for new catalyst materials is of high priority in view of tighter legislative regulations. More attention should also be given to questions related to catalyst or trap deactivation due to the presence of compounds containing sulphur. The need for more research on the recovery of noble metals and the development of sensors was likewise recognised. Last but not least, participants from industry requested the production of more engine data, performance of real test cycles and development of integrated systems. There is no doubt that clean vehicle technology is a vital part of improving air quality. Challenges remain and call for technological answers. The job is not done! Catalytic air pollution control is still an area providing a considerable incentive for innovative work. It would be a pleasure for the organizers if the outcome of this research would be part of CAPoC5 subjects.

N. Kruse Chairman CAPoC4

XV

ACKNOWLEDGEMENTS

The organizers thank the Rector of the Free University of Brussels, Mr. J. L. Vanherweghem, for his interest in the meeting and the words of welcome that he addressed to the participants of CAPoC4. We are indebted to the members of both committees for their important work. The success of a congress like CAPoC4 which covered so many fields round about catalyst technology and related issues, depends on the knowledge and advise of experts. It was our privilege that a number of the most distinguished experts accepted our invitation to participate in the scientific organization of the congress and/or the selection of submitted papers. We like to thank the four keynote lecturers, W. Hecq, P. L. Herzog, R. A. Searles, M. Walsh, for their excellent presentations in the introductory session. Special thanks are due to all coworkers, members and friends of the Chair of Inorganic Chemistry at our University. Their helpfulness and motivation have largely contributed to run the congress as smooth as possible and let CAPoC4 become a most successful event. Of course, it is difficult to render prominent the particular credits of a single person of "the team". Nevertheless, we would like to address our gratitude to Mrs. Parmentier- Depuydt for taking care of whatever you was approaching. Last but not least, the organizers recognize that CAPoC4 has succeeded in attracting and gathering experts from all over the world. A number of high quality contributions were made initiating most vivid discussions either in the lecture-hall or during poster sessions. Thanks to all participants for having contributed to a most successful CAPoC4 meeting.

The organizers, J-M. Bastin A.Frennet N. Kruse

xvi

F I N A N C I A L SUPPORT

The following companies have provided financial support to this Congress. The Organizers express their gratitude to these companies for their generosity.

AlliedSignal Inc. Automobile Emissions Control by Catalyst (AECC) Degussa A G Engelhard Co Ford Motor Co Johnson Matthey Ltd Rh6ne - Poulenc Terres Rares et Gallium Shell

xvii

ORGANIZING COMMITTEE Executive Chairman:

KRUSE N. Universit6 Libre de Bruxelles, B. Honorary President :

FRENNET A. Universit6 Libre de Bruxelles, B. Secretary :

BAST1N J-M. Universit6 Libre de Bruxelles, B. Members :

BELOT G. PSA Peugeot Citro6n, F. CUCCHI C. ACEA, B. HECQ W. Universit6 Libre de Bruxelles, B. JANNES G. Institut Meurice, B. LECLERCQ L. Universit6 de Lille 1, F. LEMAIRE J. Rhone Poulenc, F. MAIRE G. Universit6 de Strasbourg, F. MONTIERTH M. Coming Keramik, D. NIEUWENHUYSB. Rijksuniversiteit Leiden, N1. PENTENERO A. Universit6 de Nancy, F. PONCELET G. Universit6 Catholique de Louvain, B. PRIGENT M. Institut Frangais du P6trole, F. SEARLES D. AECC, B. WEBSTER D. Johnson Matthey LTD, GB.

xviii

SCIENTIFIC ADVISORY BOARD :

All members of the organizing committee, and BAIKER A. Swiss Federal Institute of Technology, CH. BURCH R. University of Reading, GB CAMPINNE M. Ecole Royale Militaire, B. FARRAUTO R.J. Engelhard Corporation, USA IWAMOTO M. Hokkaido University, J. KONIG A. Volkswagen AG, D. LEDUC B. Universit6 Libre de Bruxelles, B. LOX E. Degussa AG, D. PALMER F.H.C.E.C., B. ROBOTA H.J. AlliedSignal, USA SHELEF M. Ford Motor Co., USA SCHWEICH D. CNRS - CPE, F. VAN DEN BRINK P.J. Shell, N1.

General Lectures


CATALYSISAND AUTOMOTIVEPOLLUTIONCONTROLIV Studies in Surface Science and Catalysis,Vol. 116 N. Kruse, A. Frennet and J.-M Bastin (Eds.) 91998Elsevier Science B.V. All rights reserved.

Global trends in motor vehicle pollution control : a 1997 update M.P. Walsh

3105 N, Dinwiddie street, Arlington, Virginia 22207 USA

ABSTRACT Four trends continue to drive the global market for motor vehicle pollution control equipment 9 9 The continued growth in the world's population 9 The rising affluence of many rapidly industrializing developing countries, increasing the affordability of motor vehicles 9 The increasing number of health studies showing adverse effects at lower and lower levels of pollution 9 The response of governments by adopting more and tighter emissions standards for new vehicles or other incentives to stimulate the introduction of pollution controls on vehicles. As we approach the 21 ~t century, the global vehicle population exceeds 700 millionalmost 500 million light duty vehicles, about 150 million commercial trucks and buses and another 100 million motorcycles. Each year, the vehicle population is growing by about 12 million automobiles, 3.7 million commercial vehicles and 2.5 million motorcycles per year. While the growth rate has slowed in the highly industrialized countries, population growth and increased urbanization and industrialization are accelerating the use of motor vehicles elsewhere. One result is that air pollution is an increasingly common phenomena necessitating aggressive motor vehicle pollution control efforts. The purpose of this report is to survey what is presently known about transportation related air pollution problems, to summarize the adverse impacts which result, to review actions underway or planned to address these problems, and to estimate future trends. Based on these trends, this study will assess the large and growing vehicle pollution control market, expeeially with regard to exhaust after treatment systems.


CATALYSIS AND AUTOMOTIVE POLLUTION CONTROL IV Studies in Surface Science and Catalysis, Vol. 116 N. Kruse, A. Frennet and J.-M Bastin (Eds.) 91998 Elsevier Science B.V. All rights reserved.

Contribution of fossil fuels and air pollutants emissions in Belgium since 1980 The role of traffic* W. Hecq CEESE-ULB, Avenue Jeanne, 44 1050 Brussels, Belgium For my first presentation made here in 1991 (HECQ, W., 1991), I began by recalling the role of the Rome Club Report, United Nations Conference in Stockholm, oil shocks which had revealed the dangers threatening western economies, i.e.: pollution, natural resources depletion and a strong economic dependence on imported oil. Six years later, Earth Summit + 5 Session will start in just a few weeks but the above mentioned essential problems identified six years ago stay the same. Pollution and resources depletion belong to today's concerns and dependency towards fuel imports lies on numerous uncertainties. Only wording has changed : it is a case of "sustainability" and more specifically, within the frame of this congress, an issue of "sustainable mobility". However, if the issues have remained the same throughout the years, many things have been achieved in favour of the environment. Parallel knowledge of environmental systems has improved and new issues came into sight. I shall begin this presentation by reminding you of some of the major air pollution issues in which traffic plays a major role. The second part of this presentation will give an overview of the EU legislations enforced until now and which obviously concern traffic. The third part will show and discuss results obtained in terms of emissions from vehicles in Belgium. The fourth part will consider air pollution as it is encountered in a European city in comparison with Brussels. The fifth and last part will conclude with a few suggestions. 1. M A J O R A I R P O L L U T I O N A S P E C T S I N V O L V I N G T R A F F I C Essentially, these aspects are energy related and if we refer to a geographical scale classification, three forms of pollution are a cause of concern : 9 local and urban levels; 9 local up to continental level; 9 global level.

* Walter J. HECQ

Centre for Economic and Social Studies on the EnvironmentUniversit6 Libre de Bruxelles

Local and urban levels

At local and urban levels, it is essentially a question of vicinity pollution. Several air pollutants are concerned : sulphur dioxide, SO2 (primary pollutant); nitrogen oxides NOx (primary or secondary pollutants); particulate matter PM (primary and secondary pollutants); carbon monoxide, CO, (primary pollutant); (volatile) organic compounds, HC (or VOCs) (primary and secondary pollutants), and photochemical oxidants, 03, PAN (secondary pollutants). Organic compounds account for a wide range of hydrocarbons and are found in solid, liquid, and gas forms. Effects of these pollutants are investigated today all over the world. Some research findings can be recalled. Firstly, for some of these pollutants, the WHO guidelines, when they exist, are exceeded episodically and more specifically in the urban areas. Secondly, above certain thresholds, significant health effects are observed. In fact, pollutants such as PM10, PM2.5, NO2, 03, can deeply penetrate the respiratory tracks with the consequence of a rise in medical consultations, of hospital admission, bronchitis amongst children younger than 14.... In fact, more and more people living in cities are breathing more CO, NO2 than what is good for them. Other effects such as the deterioration of buildings and monuments which increases either the repairing rate of material damaged with acid gases or the cleaning of facades soiled with sooty particles. Lastly, damage associated with this pollution is a cause of social costs which are born by the community and which are not taken into account in the benefit provided by a car trip. These costs, described as "external", are not included in the transportation prices, a factor which causes imbalances and a non-optimal distribution of resources. Their extent is undisputed but, in many cases, has still to be quantified. A number of initiatives are looking into this. As shown in studies carried out in my Centre (Bres, 1995; HECQ, W. and ALPI, I. 1995), this environmental damage, including on bronchial diseases, costs hundreds of $ million per year and this for the Brussels-Capital region only. An amount in which traffic plays a significant role. As indicated in other studies (Infras, 1995; Ecoplan, 1996; UNIPEDE, 1996), these costs are typical for European cities and concern especially public health and materials. They do not include other impacts like olfactory discomfort from VOCs, long-term effects like risk of mutagen and carcinogen diseases. Local up to continental level

At local up to continental level, pollution takes the form of acid deposition and photoxidant phenomena. On a large scale over Europe, critical loads are exceeded. Degradation of crops, forests, materials, terrestrial ecosystems ..... is produced here too. Damage from these forms of pollution is also extensively studied nowadays and assessed in physical (RENTZ, O., 1993) or monetary terms (ExternE, 1996). Global level

At a global level, we also have to take into account the global warming and ozone layer depletion. Here too, damage estimates as sea level rise, climate change .... are assessed. For the greenhouse effect, estimates range between 120 and 250 ECU/toe (Holland, M., 1996). Last, but not least, we still have to take into account resources scarcity :fossil fuels for which oil reserves, the cheapest ones and those of best quality, are concentrated in politically unstable areas of the planet and do not remain superseding today for the great majority of the vehicle fleet.

2. R E V I E W O F EU V E H I C L E E M I S S I O N R E G U L A T I O N S Obviously, decision makers did not remain unconcerned, especially considering traffic emissions. The first regulation came into force by September 1971, a few months before the first earth summit : the Stockholm conference. It was the starting point of a long sequence of amendment steps towards more and more stringent emission limit values, also associated with numerous technical standards (monitoring procedures, test cycle profiles, vehicle fleet typology,

...). Emission regulations for Europe were first introduced in order to assume a uniformity of technical prerequisite amongst car producers. This initiative belongs to the United Nations Economic Commission for Europe (UN-ECE). The European Commission found this to be a good opportunity to adopt a first vehicle standard with the Directive of 20 March 1970.

European motor emission standards for light vehicles As far as light vehicles are concerned, table 1 gives an idea on the evolution of standards for light duty vehicles.

Directive 70/220/EEC Adopted on 20 March 1970, it is the first directive concerning the reconciliation of Member State legislations relative to measures to be taken against air pollution by exhaust gases from vehicles with starting engines. It defines the relative prescriptions for the conformity of vehicles and fixes standards for 9 classes of vehicles, from less than 750 k~ to more than 2,150 kg. Only CO and HC are regulated. The standards are expressed in g/test.

Directive 74/290/EEC Adopted on 28 May 1974, it is the first amendment of the regulation of the EEC. It also only concerns CO and HC emissions. It lowers CO and HC emissions in respect to the base level.

Directive 7 7 / 1 0 2 / E E C Adopted on 30 November 1976, it fixes, for the first time, a limit value for NOx emissions, which, just like CO and HC, have a great influence on our health and on the environment. To simplify matters, nitrogen oxides are expressed in NO2 equivalent.

Directive 78/665/EEC Adopted on 14 July 1978. It is the third amendment of the first directive and its emission standards for CO, HC, NOx are more severe.

* 3 main procedures for the approval : Type I test cycle (HC and CO), before the test period, the vehicle is soaked for 6 hours at a temperature of between 20 and 30 C~ Type II test, CO determination test at low speed after fourth cycle type I; Type III test, crankcase emission procedure on chassis dynanometer.

Table 1 European 9 motor emissions standards ,for vehicles

VEHICLE TYPE

DIRECTIVE

DESCRIPTION

Light duty vehicles

70/220/EEC

Light duty vehicles

74i290/EEc (first amendment) 77/102/EEC (amending 70/220/EEC) 78/665/EEC (third amendment) 83/351/EEC (fourth amendment)

Base directive setting emission limits for c o and HC More stringent emission limits for CO and HC Introducing limits for NOx

Light duty vehicles Light duty vehicles Light duty vehicles

Light duty vehicles

88/76/EEC (fifth amendment)

Cars with engine capacity of less than 1.4 litres

89/458/EEC (amending 70/220/EEC)

Light duty vehicles

91/44i/EEC (amending 70/220/EEC)

Light duty vehicles

94/12/EC (amending 70/220/EEC)

More stringent emission limits for CO~ HC, and NOx Introducing new methods 0f HC and NOx measurements Emission limits for diesel engines More stringent emissions for co, HC and NOx Introducing particulate emission limits for diesel. Three vehicles types in function of cubic capacity Tightening iimit values for gaseous emissions set by 70/220/EEC (as amended) Consolidating Directive applying Stage 1 limits (tightening the limits imposed by Directive 70/220/EEC and its amending Directives) AND introducing requirements for evaporative emissions and durability of emissionrelated components Applying more stringent Stage 2 limits for hydrocarbons, carbon monoxide and nitrogen oxides, with separate limits for petrol and diesel cars and limits for particulates 9 from 1.1.96 for new models 9 from 1.1.97 for vehicles entering into service

Directive 8 3 / 3 5 1 / E E C Adopted on 16 June 1983. Up until then, directives concerning CO, HC and NOx emissions were only valid for gasoline fuelled vehicles. However, given the extent of the development of diesel vehicles, the EEC decided to submit them to the same standards as those of gasoline fuelled cars. So, up until now, emissions of CO, HC and NOx, from diesel vehicles are regulated. On the other hand, in the directives that follow, emissions of NOx are no longer regulated as such, but in combination with unburned hydrocarbons. This manner to regulate these two pollutants gives, to the manufacturers, the choice to reduce either NOx or HC. Directive 8 8 / 7 6 / E E C Directive 88/76/EEC, also called "Agreement of Luxembourg", fixes standards that are even more severe for gasoline and diesel fuelled vehicles of up to 3.5 t. This directive distinguishes two types of standards : type approval standard and conformity of production standard. Application dates for these standards vary according to three engine capacity categories : 9 vehicles with an engine capacity of below 1400 cm3; 9 vehicles with an engine capacity of between 1400 and 2000 cm3; 9 vehicles with an engine capacity higher than 2000 cm 3. What's more, these standards are as much applicable for gasoline fuelled cars as they are for diesel engined cars with a certain modulation and they take into account particulate emission. However, concerning emissions from vehicles with an engine capacity lower than 1400 cm 3, the decision was only taken on 18 July 1989 and brought into practice in Directive 89/458/EEC. Directive 9 1 / 4 4 1 / E E C This directive called the "Consolidated Emission Directive" was adopted on 26 June 1991. It replaces Directives 88/76/EEC and 89/458/EEC. Vehicle emissions are no longer measured with the same ECE 15 test cycle. In fact, the new test cycle combines the existing urban test cycle (ECE 15) with a test cycle (new ECE 83) simulating driving conditions outside urban areas (EUDC). It concerns a reinforcement to the extent that NOx emissions increase rapidly at high speed. These limit values with the new test cycle, make it very difficult for a gasoline fuelled car to satisfy the directive without requiring three way catalysts. Moreover, the directive anticipates a supplementary test in order to guarantee the durability of anti-pollution systems. Vehicles will have to take the test after 80,000 km and will have to comply to the same standards as those applicable to new cars. At last, limits for vehicles evaporative emissions are also given. Directive 9 4 / 1 2 / E E C In December 1993, more stringent limits from 1996 are programmed (stage 2). They are adopted in Directive 94/12/EC. With these new standards, it must be noted that production conformity must comply with the type approval limit. To summarise, thanks to this sequence of more and more stringent regulations, emission for new vehicles could be reduced by more than 95% between 1970 and now (Figure 1).

10

Figure 1: Evolution of gasoline car emission standards in E.U. Emission standards for diesel engines and other vehicles At the beginning, European regulations concerning diesel engine emissions were only effective for three 'classic' pollutants for light duty vehicles (Directive 70/220/EEC modified by Directive 83/351/EEC) and on black smoke emissions (Directive 72/306/EEC). This black smoke represents a potential danger for health. It is thus better to limit total emissions of particulates from these engines. The new limit values for particulates were reformulated according to three categories of vehicles: light duty, light commercial and heavy duty. The limit values reformulated for particulates for light duty vehicles are defined in Directive 88/76/EEC and those that follow (table 1) for the other vehicles. As far as light commercial vehicles are concerned, Directive 88/436/EEC, modifying Directive 70/220/EEC is published and specifically concerns emissions for diesel vehicles except small engines. This one is extended to the three other gaseous pollutants by Directive 93/59/EEC and Directive 91/441/EEC (for M. ~

c 70

r,.) 0

o~

C r,.) NI

- -,

0,992

Figure

t

........ ! . . . . .

0,996

8:

[

1

Lambda

"

2:65

-

1,004

1,008

.... l l

0,992

. . . . . .

0,996

"

' " ' , , , ,

, i

1

1,004

Lambda

1,0(18

Dynamic engine sweep test results for catalyst types VIII and IX. (9Pd/1Rh 60 g/ft3; 100 h, 1123 K fuel cut aged; SV=60000 Nl/l/h, Ti~et = 673 K; 1Hz •

=

-=945

55

t. ~J

.-;;-

= O 40

ca

35

~,, x 9 .-. ,-.. '" 7"

0,992

Figure 9:

0,996

.. ....

"'"

!

=_~," :: : -

......

....

}

1 1,004 Lambda

J..

Z

1,008

45

...

0,992

i

0,996

.

i

i

1 l..ambda

1,004

1,008

Dynamic engine sweep test results for catalyst types IX and X. (9Pd/1Rh 60 g/ft3; 50 h, 1123 K fuel cut aged; SV=50000 N1/l/h, T ~ t = 673 K; 1Hz +IA/F)

.~9o

9o ' ~ 1 7 6 1 7 6

o= .~ so

type IX

0

~,x...."

~

......

x

""" i'-.

. 2r ~ 8 0

,~,,x

I. >

= 65 o

70

".

\ ~ Q

0,992

0,996

1

Lambda

1,004

1,008

o,~

0,996

1

1,(]04

1,008

L~bda

Figure 10: Dynamic engine sweep test results for catalyst types IX and X. (9Pd/1Rh 60 g/ft 3 96 h, 1123 K fuel cut aged; SV=60000 Nl/l/h, T~a.t= 673 K; 1Hz •

59 3.2. Pd/Rh catalysts

The Pd-only technology developed with the new preparation method C was also tested for Pd/Rh three-way catalyst technologies. Figure 8 shows the engine test results of a standard Pd/Rh three-way catalyst technology (type VIII) compared with the same washcoat technology, however, with applying the new preparation technology for the Pd containing washcoat layer (type IX). The Rh containing washcoat layer was for both catalysts the same and was prepared by the utilization of preparation method B. The results clearly show the benefits of the novel preparation technique for applying the precious metal onto the washcoat on the overall performance of the Pd/Rh three-way catalyst. For further improvement of the activity of the Pd/Rh catalysts it was also tried to utilize the novel preparation method D for applying Rh onto one or more washcoat components. Figure 9 shows the engine test results of a Pd/Rh three-way catalyst technology, where for all three catalysts the preparation method C for the Pd containing washcoat layer was utilized. For the Rh containing washcoat layer two preparation techniques were used, i.e. method B (type IX) and method D. The method D was used to prepare catalysts where Rh was applied onto an AIzO 3 component with two different preparation parameters (type X and Xb). The results show that method D leads to comparable if not improved performance of the Pd/Rh three-way catalyst compared with the commonly known method B. As already discussed the new preparation methods are more flexible and include more parameters that can be controlled to prepare tailor-made catalysts. This can also be seen in Figure 9, where with the catalyst encoded type Xb it is shown that by changing the preparation parameters it is possible to prepare catalysts with an improved CO activity. Another important advantage of the novel preparation techniques is that they cannot only be used for the application of precious metal onto one or more washcoat components but also for the manufacturing of novel OSCs. The new preparation techniques allow the use of a smaller amount of OSC without sacrifices in catalyst performance, on the contrary. Figure 10 shows the engine test results of a Pd/Rh catalyst technology with commonly used OSCs compared with a catalyst technology where in the Rh containing washcoat layer a novel OSC is used, which was prepared by one of the new preparation technique (method D). Although the catalyst prepared with the new technology contains five times less OSC, the performance of the catalyst is comparable with the catalyst with a standard Rh washcoat layer. The CO oxidation activity is somewhat lower but the performance for the NOx reduction has improved by applying the new technology. The addition of stabilizers and promoters to the newly developed OSC should result in a further improvement in catalyst performance and will be investigated in future development programs.

60 4. CONCLUSION The development of novel preparation techniques for exhaust catalysts has resulted in a general improvement of the catalyst performance. For Pd-only catalyst technologies a preparation method was developed to apply the precious metal homogeneously and selectivitely into more than one washcoat component. These new catalysts show a significant improvement in performance compared with the standard Pd-only catalyst technologies. The same technology used to prepare the Pd containing washcoat layer in Pd/Rh three-way catalysts also results in an enhanced activity for these types of catalysts technologies. One of the novel preparation technologies was also used to apply Rh on one or two washcoat components, which led to an improved performance of the Rh containing washcoat layer.

ACKNOWLEGMENT. The authors wish to thank colleagues and coworkers for the valuable discussions and for the high quality experimental work. REFERENCES

1. B.H. Engler, E.S. Lox et al., ~tRecent Trends in the Application of Three Metal Emission Control Catalysts )>, SAE Paper 940928 (1994) 2. B.H. Engler, E.S. Lox, D. Lindner, A. Schafer-Sindlinger and K. Ostgathe, tt Development of Improved Pd-Only and Pd/Rh Three-way catalysts ~) in tt Catalysis and Automotive Pollution Contro III, A. Crucq, Ed. Elsevier (1994) 3. J.C. Summers, J.F. Skowron, W.B. Williamson and K.I. Mitehel, ~tFuel Sulfur Effects on Automotive Catalyst Performance )~, SAE Paper 920558 (1992) 4. J.Hepbum, K. Patel, M. Meneghel and H.S. Gandhi, tt Development on Pd-only Threeway Catalysts, SAE Paper 941058 (1994) 5. D.J. Ball, tt A Warm-up and Underfloor Converter Parametric Study ~) SAE Paper 930386 (1993) 6. B.H. Engler, E.S. Lox, D. Lindner and K. Ostgathe, ~ Advances in Three-way Catalyst Design to Meet more Stringent Emission Limits )~, ISATA Conference, Aachen, Germany, October 31-November 4, Automation Limited, England (19.94) 7. A. Punke, U. Dahle, S.J. Tauster, H.N. Rabinowitz and T. Yamada, (t Trimetallic Threeway Catalysts )), SAE Paper 950255 (1995) 8. R.J. Bfisley, G.R. Chandler, H.R. Jones, P.J. Anderson and .J. Shady, tt The Use of Palladium in Advanced catalysts ~>SAE Paper 950259 (1995) 9. S. Matsura, A. Akimasa, K. Arimura adn H. Shinjoh, ~tDevelopment of Three-way catalyst with Using Only Pd as Activator )) SAE Paper 950257 (1995) 10. D. Lindner, E.S, Lox, R. van Yperen, K. Ostgathe, T. Kreuzer, ~tReduction of Exhaust Gas Emissions by Using Pd based Three-way Catalysts )) SAE Paper 960802 (1996) 11. B.H. Engler, E. Koberstein and P. Schubert, ~t Automotive Exhaust Gas Catalysts: Surface Structure and Activity )) App. Cat. 48 (1989) 71-92

CATALYSIS AND AUTOMOTIVEPOLLUTIONCONTROLI v Studies in Surface Science and Catalysis,Vol. 116 N. Kruse, A. Frennet and J.-M Bastin (Eds.) 91998 Elsevier Science B.V. All rights reserved.

61

Comparative Behaviour of Standard Pt/Rh and of Newly Developed Pd-only and Pd/Rh Three-Way Catalysts under Dynamic Operation of Hybrid Vehicles S. Tagliaferri, R.A. K6ppel and A. Baiker Department of Chemical Engineering and Industrial Chemistry, Swiss Federal Institute of Technology (ETH), CH-8092 Z0rich, Switzerland

ABSTRACT The suitability of newly developed non-promoted and ceria-promoted palladium and palladium-rhodium three-way catalysts for the exhaust gas control of a hybrid drive system has been tested with periodic changes of the feed stoichiometry (~-r and pulsed-flow operation. The performance of the catalysts under dynamic operating conditions has been compared to the behaviour of analogue samples based on standard platinum-rhodium-technology. Combined use of mass spectrometry and time-resolved FTIR spectroscopy allowed simultaneous monitoring of the exhaust components. The air plug preceding the exhaust pulse in the intermittent operation of the combustion engine had a crucial impact on the performance of the catalysts. The air pulse transfers the catalyst into a defined state, which corresponds to a fully oxidized surface. Controlled application of appropriate ~-r allows to compensate the negative effect of the air plug and to achieve sufficient conversion of NOx as well as CO and HC to harmless compounds. For application in the hybrid vehicle, ceria-promoted palladium catalysts proved to be superior to the standard platinum-rhodium technology.

1. INTRODUCTION In view of new regulations for low-emission vehicles, research on hybrid vehicle technology has been intensified recently in an attempt to optimize overall vehicle performance, fuel efficiency and emissions. Generally, hybrid drive systems make use of the synergetie combination of a combustion engine, which guarantees a high range of performance, with an electric motor, which allows local emission free driving over a limited distance. Hybride concepts can basically be divided into serial and parallel configurations [ 1] with the latter showing higher overall efficiency but being more demanding with respect to emission control. As part of an interdisciplinary study we are working on emission control catalysis for an extended parallel hybrid concept [ 1-5]. The main elements of the experimental car developed at the Swiss Federal Institute of Technology are the combustion engine, operating in a fixed cycle mode, a flywheel as a short term energy storage device, an electrical machine and a continuous variable transmission. One of the features of this hybrid vehicle is the so called intermittent

62 mode, where driving energy is taken from the flywheel, which is recharged by operating an internal combustion engine for 3 s in intervals of about 17 s. On engine start-up and shut-off, the cylinders are filled with air, which consequently passes through the catalyst at the beginning of engine operation. The intermittent operation leads to a pulsed-flow operation of the TWC (Fig. lb).

a) X-Cycling l

b) Intermittent Mode

I I I i / I

I

I

I

-;k+

I .... x9time

flow rat

"l

~,: const, or cycled

air plug

......

,, ,, 17s

~ time 3s

Figure 1. Time dependent ~,-value and flow pattern during: (a) ~,-cycling between ~+ and ~,- with constant or variable length of the half cycles and constant gas flow; (b) Intermittent mode, 17 s without exhaust gas flow (engine shut off), 0.2 s pure air (air plug at engine start-up), 3 s exhaust gas flow (engine operation). Catalytic converters in automobiles are periodically forced about the stoichiometric air-fuel ratio at a frequency of about 1 Hz and a small amplitude [6]. In a recent review, Silveston [7] concluded that this periodic forcing suppresses rather than enhances conversions under normal operating conditions in the 400-600~ temperature range. Other authors reported that under cycling conditions the catalytic activities of three-way automotive catalysts can be superior compared to static conditions, depending on temperature, cycling period and feedstream conditions [8-10]. Cycling at temperatures below the light-offtemperature was found to increase conversions of NOx, CO and hydrocarbons, whereas the effect of )~-cycling was negative at higher temperatures [ 11 ]. Base metal oxides such as ceria are added to catalyst formulations in order to buffer excursions into the lean or rich region [12]. Ceria was reported to stabilize precious metal dispersion and to be involved in the storage and release of oxygen as well as in the promotion of the water gas shift reaction and the steam reforming reaction [ 13-15]. Recently, economic factors as well as the favorable low temperature performance and hightemperature resistance of Pd have lead to an increased interest in palladium as main noble metal component for three-way catalysts. Several research groups have presented a new generation of palladium catalysts with and without addition of Rh [ 16-20]. The objective of our study was to gain information about the behaviour of palladium based three-way catalysts under dynamic operation, especially under pulsed-flow operation as occurs in the intermittent mode of a combustion engine used in a hybrid vehicle.

2. EXPERIMENTAL 2.1. Catalysts The catalysts tested were supplied by Degussa AG and consisted of a ceramic honeycomb carrier with 400 cells/in 2. The washcoat loading was 110 g 1-1 with the composition (wt %) as denoted in Table 1. The catalysts had a length of 15 cm and a diameter of 2.5 cm. To reduce the

63 void volume of the catalyst to 12.73 cm 3, the outermost channels were sealed with an inert ceramic paste. Before catalytic tests, the catalysts were conditioned for 5 h at 600~ in a simulated exhaust with Z, = 1.

2.2. Apparatus Experiments were carried out in a fully computer controlled apparatus, which has been described in more detail elsewhere [3]. A synthetic exhaust gas mixture containing CO and HE at a ratio of 3:l, C3H6(500 ppm), C3H8 (500 ppm), NO (2000 ppm), O2, CO2 (12 %), H20 (10 %) and N2 (balance) was used for laboratory tests. The gas flow rate was 10.625 I(NTP) min "l, giving a gas hourly space velocity of 50'000 h "1 with regard to the total catalyst volume. The Evalue of the gas mixture, which represents the ratio between the available oxygen and the oxygen needed for full conversion of the components to CO2,H20 and N2: 2,=

2c~176176176176 2cco + ci_i2+ lOcc~m + 9Ccm, + 2Cco, + Ctl,O

(1)

was altered by adjusting the CO/H2 and the 02 flows via fast switching valves. The gas analysis system consisted of an FT-IR spectrometer (Bruker IFS-66) with a heatable gas cell (100 cm"3volume) and a quadrupole mass spectrometer (Balzers GAM 400). NO, NO2, N20, NH3, CH4, C3H6, C3H8, CO, CO2, and H20 were analysed by FT-IR spectroscopy and O2 and HE by mass spectrometry. The analytical system permitted the quantitative analysis with a resolution of up to 15 measurements per second. Table 1 Composition and denotation of tested catalysts. Catalyst denotation

Washcoat composition / wt % Pd

Pt

Rh

A1203

Pd

1

-

-

99

Pd-Ce

1

-

-

87

Pd-Rh

1

-

0.2

98.8

Pd-Rh-Ce

1

-

0.2

86.8

Pt-Rh

-

1

0.2

98.8

Pt-Rh-Ce

-

1

0.2

86.8

CeO2

12 12

12

2.3. Experimental procedure L-cycling (Fig. 1a) and pulsed-flow (Fig. lb) experiments were carried out to study the dynamic behaviour of the catalysts. Forced L-cycling with different amplitudes and frequencies was achieved by periodically changing the stoichiometry of the feed composition. To simulate the intermittent operation of the combustion engine in the hybrid vehicle, pulsed-

64 flow experiments (Fig. 1b) were carried out at 400~ and 1.7 bar. The exhaust gas was pulsed with a flow rate of 10.625 I(NTP) min1 through the reactor for 3 s, followed by a period of 17 s, with no gas flowing through the converter. The X-value was either kept constant or cycled symmetrically or asymmetrically during 3 s of the pulse. In most experiments an air plug with a flow rate of 3.187 I(NTP) min 1 and a duration of 0.2 s preceded the exhaust pulse, simulating air which is transferred into the cylinders.

3. RESULTS

3.1. Experiments with X-cycling Time resolved cycling experiments were carried out at 310~ using an amplitude of X = 1 • 0.05. In Figure 2 the changes of concentration of the exhaust gas components CO, C3H8, NO,

0.201

. . . . . . .

o,4 f

~

i ," \ ,

0.101 "-l

~o=-I,

,'

...-,,

,

Z0.15

~

-

/

~

/

/

-

1-o.6

I-0.4

:

~ooo I

'~

O'0.20 ~

0.8 0.6

8

/r%..'"'-..

0.4

o . ~ : - ~ - = ; !

o.o

~ 0"20/

. . . . . . .

o, 1

t/

O.lO-I I

|./ 0 " 0

--

'~ 5

~ \

~

[ 0

,

,,

/'=t

.

000 l ' ~ ~ ' t ' ~ 1 2 3 Time/s

1 1.0

,,,

' ....[ 2 4

/I-o6 I

/

"

'

"

'

"

'

1-0"4

.-"=-I O0 5

t

1 1.0

[o.,

~

~

'

'"

I:tIRK~ [ 0.8

o.,ol A

.... ! o.o

_~ .

0.201 0.15t

_ ;,'A:, I^..

\;

~o.oo r ' - ~

A

;

1 1.0

'

~ '

oo

. . . .

R~

o.,o

i 1.0

to8

Ao.4

o.oo~ ~ ~ . - - - - !

1o.2oi 0.15

.

.

-=

I /-~

O.lO-I / "i / "* 0 "

0

\ \

.

5

.

.

.

~

I

'

,'

.-,

~

0.00 r i ' ' , - - - ~ " ~ , ' ~ ~ ' - - - I 1 2 3 Tirre/s

o.oI~ .

R ~

i,.o 0.8

a

,-l-ao

, /t ',." .... -.1 I"0"4 ,I 0.2 4

5

00

Figure 2. Change of the concentrations of the most significant exhaust gas components with time at 310~ during X-cycling with X = 1 + 0.05 and v - 0.3 Hz for catalysts Pd, Pd-Ce, Pd-Rh, PdRh-Ce, Pt-Rh and Pt-Rh-Ce. The arrow indicates one rich half-cycle. Symbols: (.... ) CO, ( - - - ) C3H8, ( ~ ) NO, ( .... ) NH3, ( . - . ) N20, ( .... ) NO2 NH3, N20 and NO2 with time are shown for a frequency of 0.3 Hz. Propane had been chosen as low reactivity HC component to compare the performance of the catalysts, because 100 %

65 conversion was reached with propene. The black arrow indicates one rich half-cycle of 1.67 s duration. Qualitatively similar cyclic concentration-time profiles were observed for all catalysts, with the concentration maxima of both CO and propane appearing at the end of the rich halfcycle. For C3H8 a second maximum, coinciding with the NO peak at the end of the lean halfcycle, was observed. This effect was most pronounced for catalysts Pd-Rh-Ce and Pt-Rh-Ce, which also showed highest propane conversion. For all catalysts except Pt-Rh, substantial amounts of NH3 were produced. Ammonia concentrations reached their maximum in the middle of the lean half-cycle when NO started to appear. Substantial amounts of nitrous oxide were formed with catalyst Pd, whereas only small amounts were observed for the other catalysts. NO2 was not produced in significant amounts. Interestingly, CO formation was reduced for the Pd containing catalysts upon addition of ceria, whereas the opposite effect was observed for the Pt containing catalyst. When the frequency of L-cycling was increased to 1 Hz, concentrations of the exhaust gas components generally decreased markedly. The concentration-time profiles for catalysts Pd-RhCe and Pt-Rh-Ce are depicted in Figure 3 as an example. Concentrations were very low and no

0.04

0.08 '

"1"

o

'

'

'

' Pd---Rh-Ce '

i

'

'

'

'

'--Rh-C~ .... Pt'

,~176

z

o;

"** .

0.03-

**

*-***..**...*'**

.....

**..***

.~

0.06

Z & 4

~)

II II il

0.02

9 Ir"~l

" :

o o

." :

8

:

0.01

: : I

l

"~ II

o it

t I

9

I

I

: " t

" I

"L

o a * i a

o i "

o

m t

0.04 ~I

*1

I o * o

o: ,~ m , m

II ,

II

,

t ~

" ~,

, , I

'

8

@mmD

l

"

0.021

**" . . . . ~ 1 7 6 1 7 6 1 7 6 1 7 6 1. 7 6

I

0.00 2

3

Time/s

4

5 1

2

3 Time/s

4

5

Figure 3. Change of the concentrations of the most significant exhaust gas components with time at 310~ during ~,-cycling with ~, = 1 + 0.05 and v = 1 Hz for catalysts Pd-Rh-Ce and Pt-RhCe. The arrow indicates one rich half-cycle. Symbols: ( .... ) CO, (----) C3H8, ( ~ ) NO, ( .... ) NH3, ( . - . ) N20, (.... ) NO2 cyclic effects were discernible for the fully promoted platinum catalyst, whereas oscillations with intervals of 1 s were still observed for the corresponding palladium based catalyst.

66

3.2. Experiments with pulsed flow Simulation of the pulsed-flow operation of the catalytic converters was performed at 400~ Although the thermally insulated catalyst is located close to the exhaust ports in the hybrid vehicle, the exhaust temperature is expected to be uncommonly low for TWC applications due to the intermittent operation of the combustion engine. The performance of the various catalysts was examined for the nine different intermittent operating modes A to I, listed in Table 2. For modes A and B a stoichiometric exhaust with constant Z-value was used with (B) and without (A) a preceding air plug, whereas in mode C the gas mixture was kept rich to compensate the effect of the air plug. For operating modes D to I the Z-value was cycled during the pulse with a frequency of 1 Hz and an amplitude of + 0.05. Experiments were performed in pairs where cycling started either with a rich or lean exhaust gas, and was either symmetric or asymmetric (Table 2). Table 2 Examined intermittent operating modes Operating mode

a)

Air plug

Z-value, v

First half-cycle

Rich half-cycle a)

A

no

1.0

-

-

B

yes

1.0

-

-

C

yes

0.985

-

-

D

no

1 + 0.05, 1 Hz

rich

50 %

E

no

1 + 0.05, 1 Hz

lean

50 %

F

yes

1 + 0.05, 1 Hz

rich

50 %

G

yes

1 + 0.05, 1 Hz

lean

50 %

H

yes

1 • 0.05, 1 Hz

rich

60 %

I

yes

1 • 0.05, 1 Hz

lean

60 %

Portion of the rich half cycle, i.e. 60 % means 0.6 s rich and 0.4 s lean at 1 Hz.

The performance of catalyst Pd-Rh-Ce for the intermittent operating modes A, B, and H is illustrated in Figure 4, which shows the time-dependence of the gas concentrations during an exhaust pulse. With mode A, CO broke through after 0.5 s, but was eliminated in the second half of the pulse. During the period of 17 s without exhaust gas flow, the catalyst approached chemical equilibrium and CO concentration was zero at the beginning of the exhaust pulse. About 500 ppm NH3, corresponding to ca. 25 % of the NO inlet concentration in the exhaust, were continuously produced during the pulse, with the concentration being slightly higher in the second half of the pulse. NOn, C3H8, and N20 were not detected in significant concentrations. A preceding air pulse (mode B) had a dramatic influence on conversions. CO and NH3 were quantitatively eliminated from the exhaust, whereas NO broke through immediately after the

67

0.20

!

'

9

I

'

A

0.15 0.10 #

0.05

~ ....

0.00 o..9`

0.20

" .o_

0.15

t,..,, t'-

q) r

*" o (9

oo

,,=~

I ,t1-o ~ . . . . . . . . .

.....

- --- = ' ~ - - -

I

~ 1 7 6 1 7 6 '~ q ' ,=

,~176

'

'~..,.,=

r'-"~-/~

'

0.10 0.05 0.00 0.20

--- ~~ ~

'

.....

- .....

,~i~----'-'---r,,~:--~,-~-_-- .....

!

'

I

'

0.15 0.10 0.05

0.00

__!

0

1

__

,

2

Time during exhaust gas pulse / s

3

Figure 4. Concentrations of exhaust components for intermittent operating modes A, B and H with catalyst PdRh-Ce at 400~ Influence of air plug and its compensation. A: exhaust pulse with L = 1; B" air plug followed by exhaust pulse with L = 1; H: air plug followed by asymmetric L-cycling with L = 1+0.05, 1 Hz, periods 0.6 s rich/0.4 s lean. Symbols: (.... ) CO, (----) C3H8, ( ~ ) NO, ( .... ) NH3, (--.) N20, (.... ) NO2

beginning of the pulse, reaching full inlet concentration after 1 s. During the remaining time of the pulse NO concentration was reduced almost completely, showing a second smaller maximum at the end of the pulse. Best performance of cataiygt Pd-Rh-Ce was observed for the operating mode H. In this case, asymmetric cycling, starting with a rich exhaust of 0.6 s, compensated the negative impact of the air pulse on conversion of NO to N2 almost completely, without affecting conversion of the other exhaust components. To compare the effect of the intermittent operating modes A to I on the behaviour of the different catalysts, the concentrations of the exhaust components were integrated over the period of the pulse and divided by 3 s. As the inlet concentration of CO was not constant for the experiments with cycled feeds, concentrations instead of conversions are given. Figure 5 depicts the average C3Hs-conversions, N2-yields and CO-concentrations during an exhaust pulse of 3 s at 400~ Depending on the operating mode applied, strongly different performance of the catalysts was observed, e.g. C3Hs-conversion showed a maximum of 80 % for catalyst Pd-Rh with mode D, but only 26 % with mode E. Similarly, N2-yield ranged between 0 % (modes A and B) and 8 2 % (mode H) for catalyst Pd-Ce. The results indicate, that by choosing appropriate operating modes high C3Hs-conversions and N2-yields as well as low CO-concentrations can be achieved with all catalysts.

68 For the palladium based catalysts the presence of ceria had a strongly positive effect on C3Hsconversion as well as CO-concentration, independent of the operating mode applied. Moreover, N2-yields increased upon addition of ceria to the catalyst formulation, except for the fully promoted palladium catalyst and mode F. Adding rhodium to the palladium based catalyst had generally a negative impact on CO concentrations as well as on N2-yields for modes E, G and I. Interestingly, C3Hs-conversion was also significantly lower for these operating modes starting with a lean exhaust composition, which should facilitate propane conversion. As regards the platinum based catalysts, addition of ceria had a less pronounced influence on CO-concentration, C3Hs-conversion and N2-yield. Similarly as with the Pd-catalysts operating modes E, G, and I negatively influenced N2-yield and C3Hs-conversion, whereas otherwise slightly positive effects were observed. These results are also supported by the data listed in Table 3, which show the operating mode resulting in best performance with regard to highest N2-yield, highest C3H8 conversion and lowest CO concentration for the different catalysts. Evidently, ceria had a positive effect on C3H8 conversion and CO concentration as well as on N2-yield for the palladium containing catalysts. 100 60

~

40 ~ 20 0

:

100,

~

:

0.4 0.2

:

:

:

0

:

:

:

:

:

r 1.o

, 0.0

O,

~o40 i oo

-1

0

100

60 ~ 40 20 0

Pd

``9` 0 0 : : : : : , Z~ ~1001~. O' >~' 40608~02_0 I~ ~ ~ ~ ~I1"0 o G00"802.0~(."64.)e.. 0

...

=

loo- i

=

=

=

:

0 0

1.o

0.6 !~_ 60 0.4 ,,~

-~. 6040~> 20-

0.6 0.4 0.2 0.0

. 0.0

0.4 0.2 7:~

Z 100

Ii.o0.8

60 40 20

0.6

-~ 40 "~, 20 ~"

lo8o!iol

1.0

--

’

,

i

’

:

:

I ~~l~I :

:

:

:

0.2

:

:

:

0.0

15

C)

i '~

,

. 0.6 0.4 0.2 1.0

0

_~

0.6

40

0.4

20

0.20 :

,mlmmmmmt

.

.

.

0.0

.

~ Pd

cO

0.6 0.4 0.2 0.0

', . . . . .

Pd-CePd-Rh Pd-Rh-CePt-Rh Pt-Rh-Ce

r---] C3H8

~

N2

~

CO

0.0

Pd-CePd-Rh Pd-Rh-CePt-Rh Pt-Rh-Ce

Figure 5. Average C3Hs-conversions, N2-yields and CO-concentrations during an exhaust pulse of 3 s at 400~ for intermittent operating modes A to I and catalysts Pd, Pd-Ce, Pd-Rh, Pd-RhCe, Pt-Rh and Pt-Rh-Ce.

69 Table 3 Intermittent operating modes affording best performance. Selection has been based on following priority of performance characteristics at the catalyst outlet: highest N2-yield- highest C3H8 conversion- lowest CO concentration. Catalyst

Operating Mode

Nz Yield / %

C3H8Conv./%

CO Conc./%

Pd

D

60

67

0.48

Pd-Ce

H

82

94

0.00

Pd-Rh

H

76

80

0.53

Pd-Rh-Ce

H

83

96

0.00

Pt-Rh

C

83

70

0.05

Pt-Rh-Ce

H

72

84

0.02

Note the overall good performance of the rhodium free catalyst Pd-Ce, which showed almost the same characteristics as the fully promoted palladium catalyst Pd-Rh-Ce. Ceria addition also increased C3H8 conversion of the platinum catalyst, but resulted in a lower N2-yield. Generally, the palladium catalysts Pd-Ce and Pd-Rh-Ce showed similar or even superior catalytic performance compared to Pt-Rh-Ce.

4. DISCUSSION The potential of ceria for the storage of oxidizing and reducing components has a marked influence on the dynamic behaviour of the catalysts. For the non-promoted palladium catalyst (Pd), highest and broadest concentration peaks of CO and NO were observed, which can be explained by the missing storing capacity of ceria. Upon addition of ceria to the palladium catalyst (Pd-Ce), CO and NO peaks became significantly smaller and more narrow. The concomitant increase in NH3 formation can be attributed to a promoting effect of ceria on steam reforming and water gas shift reaction, which results in an increased formation of hydrogen. A similar effect is observed by comparing the performance of catalyst Pd-Rh with Pd-Rh-Ce, whereas addition of ceria to platinum only increased ammonia formation without decreasing CO and NO concentrations. Promotion of the activity of precious metal catalysts for the water gas shift and the steam reforming reaction by ceria has been occasionally reported [15]. For pulsed-flow operation experiments, the air plug at the beginning of the exhaust pulse substantially influenced the performance of the catalysts for steady stoichiometric exhaust compositions. NOx conversion to N2 strongly decreased. However, by adapting appropriate ~,cycling during engine operation, the negative effect of the air plug can be compensated. Moreover, the preceding air pulse transfers the catalysts into a defined state. Knowledge of this state, which corresponds to a fully oxidized surface, can be beneficial to improve the ~-control algorithm used. As expected, NE-yields were usually lower for operating modes G and I, starting with a lean exhaust after the air pulse. Best catalytic performance was observed for asymmetric cycling, starting with a rich exhaust of 0.6 s.

70 cycling pattems has so far not completely been exploited for optimizing exhaust catalysis.

5. CONCLUSION The suitability of newly developed palladium- and palladium-rhodium catalysts and of standard platinum-rhodium catalysts for the after treatment of the exhaust of a hybrid drive system, resulting in pulsed flow operation of the catalytic converter, has been compared. It was demonstrated, that the air pulse preceding the exhaust pulse, strongly influences the catalytic performance. The apparently negative impact of the air pulse on catalytic behaviour was found to be beneficial by virtue of transferring the catalyst into a well defined state, which can be accounted for in the closed-loop ~-control. Applying a rich exhaust during engine operation increases N2 yield, but partly lowers CO conversion. The use of an asymmetric cycling pattern with longer rich half cycles results in CO and HC conversions as well as N2 yields higher than without an air pulse. From the data presented it becomes evident that the ceria promoted palladium catalysts Pd-Ce and Pd-Rh-Ce are able to outperform conventional Pt-Rh-Ce catalyst in hybrid vehicle application. ACKNOWLEDGEMENTS

Financial support of this work by the Schweizerisches Bundesamt fiir Umwelt, Wald und Landschafi is gratefully acknowledged. The authors wish to thank Degussa AG for providing the catalyst samples. REFERENCES

1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13.

L. Kiing, A. Vezzini and K. Reichert, Symposium Proceedings 1lth International Electric Vehicle Symposium, 1992. P. Dietrich, H. HOrler and M.K. Eberle, Proc. Conference on Electric, Hybrid and Alternative Fuel Vehicles, Aachen, Germany 13th-17th Sept. 1993, p. 193. L. Padeste and A. Baiker, Ind. Eng. Chem. Res., 33 (1994) 1113. S. Tagliaferri, L. Padeste and A. Baiker, Stud. Surf. Sci. Catal., 96 (1995) 897. L. Padeste, S. Tagliaferri and A. Baiker, Chem. Eng. Technol., 19 (1996) 89. R.P. Canale, C.R. Carlson, S.R. Winegarden and D.L. Miles, SAE Technical Paper Series, No. 780205 (1978). P.L. Silveston, Catal. Today, 25 (1995) 175. H. Shinjoh, H. Muraki and Y. Fujitani, Appl. Catal., 49 (1989) 195. H. Muraki, K. Yokota and Y. Fujitani, Appl. Catal., 48 (1989) 93. E. Jobson, M. Laurell, E. H6gberg, H. Bernler, S. Lundgren, G. Wirmark and G. Smedler, SAE Technical Paper Series, No. 930937, (1993). B.K. Cho, Ind. Eng. Chem. Res., 27 (1988) 30. H.S. Gandhi, A.G. Piken, M. Shelef and R.G. Delosh, SAE Technical Paper Series, No. 760201 (1976). J.G. Nunan, H.J. Robota, M.J. Cohn and S.A. Bradley, J. Catal., 133 (1992) 309.

71 14. R.M. Heck and R.J. Ferrauto, Catalytic Air Pollution Control, Van Nostrand Reinhold, New York, 1995. 15. J. Cuif, G. Blanchard, O. Touret, M. Marczi and E. Qu6m6r6, SAE Technical Paper Series, No. 961906, (1996). 16. J. C. Summers, W. B. Williamson and J. A. Scaparo, SAE Technical Paper Series, No. 900495 (1990). 17. B.H. Engler, D. Lindner, E.S. Lox, A. Sch~ifer-Sindlinger and K. Ostgathe, Stud. Surf. Sci. Catal., 96 (1995) 441. 18. S. Matsuura, A. Hirai, K. Arimura and H. Shinjoh, Sci. Technol. Catal., 92 (1995) 445. 19. J. Dettling, Z. Hu, K. Lui, R. Smaling, Z. Wan and A. Punke, Stud. Surf. Sci. Catal., 96 (1995) 461. 20. D. Lindner, E.S. Lox, R. Van Yperen, K. Ostgathe and T. Kreuzer, SAE Technical Paper Series, No. 960802, 1996.



73

Comparative three-way behaviour of Pt, Pd and Rh single and combined phases in a full gas mixture with oscillating feedstream J. R. Gonzhlez-Velasco, J. A. Botas, R. Ferret and M. A. Guti~rrez-Ortiz Department of Chemical Engineering, Faculty of Sciences, Universidad del Pals Vasco, P.O. Box 644, E-48080 Bilbao, Spain

ABSTRACT A series of palladium, platinum and rhodium (single and combined) catalysts supported on cerium-doped 7-alumina has been prepared. The monometallic catalysts were prepared by adsorption from the metal solution, and the multimetallic catalysts by joint adsorption as well as by physical mixture of those monometallic which allowed to obtain similar final metal composition. The three-way behaviour of the prepared catalysts has been tested with full synthetic gas mixtures composed of N2, CO2, CO, C3H6, NO, 02 and H20 under reducingoxidising cycled and stationary feedstream compositions.

1. INTRODUCTION Three-way catalysts (TWC) which perform, at the same time, oxidation of carbon monoxide (CO) and hydrocarbons (HC) and reduction of nitrogen oxides (NOx), seems to be, up to now, a satisfactory and efficient solution. Fine work of these catalysts requires a composition of the exhaust gaseous stream corresponding to the stoichiometric air-to-fuel ratio, i.e. A/F=14.63 for a fuel with a H/C ratio of 1.89, which should be precisely controlled. It has been proved that monometallic Pt catalysts present high activity operating about stoichiometry, even more than some Pt-Rh formulations [1]. The reason for the addition of Rh becomes apparent when studying selectivity of the reaction under reducing conditions [1,2]. In previous work we have developed a rhodium-free catalyst with adequate activity on the simultaneous control of NO, HC and CO [3]. An increasing interest to promote the use of Pd in these catalysts, due to its low cost and large availability, appeared in the last years, which marks a tendency to substitute Pt by Pd in the conventional Pt-Rh compositions, or even

74 to develop new Pd-only formulations [4,5], maintaining the activity and durability of the catalyst. Alloying between precious metals in three-way catalysts has been proposed to lead to both negative [6] and positive [7] effects on performance The impact of Pt-Rh and Pd-Rh alloying on performance as well as the cumulative effect of both metals on overall activity is being intensively investigated to provide valuable bases for designing new formulations with enhanced characteristics [8-10]. In this work, we compare the TWC behaviour of Pd, Pt, Rh, Pd-Pt, Pd-Rh, Pt-Rh, and Pd-Pt-Rh in a simulated stationary and/or cycled environment near that existing in automobile catalytic converters, trying to discover the relative merits of each metal on the overall performance of the catalyst. Comparison of activity obtained with multimetallic catalysts prepared by co-adsorption and those obtained with physical mixtures of monometallic catalysts will contribute significantly to our understanding of the impact of the interactions between metals on performance and could provide a valuable basis for designing new formulations with enhanced characteristics.

2. EXPERIMENTAL 2.1. M a t e r i a l s

The starting alumina was SAS-1/16 supplied by La Roche. After grinding to adequate particle size and calcination in air at 700~ for 4 hours, its properties resulted in: catalyst size, 0.5-1.0 mm; surface area BET, 200 m 2 g-l; pore volume, 1.0 cm 3 gl; average pore radius, 5.3 nm; mode pore radius, 6.1 nm; isoelectric point, 7.6. The cerium oxide was incorporated by the conventional incipient wetness method from an Ce(NO3)3.nH20 aqueous solution, at 40~ and 30 mmHg. Promoter-modified alumina samples were dried at 120~ for 2 hours and calcined in air at 700~ for 4 hours to decompose the nitrate to oxide. The active phases --Pd, Pt, and R h ~ were incorporated by adsorption from aqueous solution using their corresponding salts--PdC12, H2PtC16-nH20, and RhCl~-nH20-- using 40 cm3 of solution per gram of ceria-modified alumina. The multimetallic catalysts were prepared by joint adsorption of the corresponding metallic s a l t s - - P d - P t , Pd-Rh, Pt-Rh, and Pd-Pt-Rh-- and by physical mixture of the monometallic catalysts--Pd+Pt (50/50 wt.-%), Pd+Rh (50/50 wt.%), Pt+Rh (50/50 wt.-%), Pd+Pt+Rh (33.3/33.3/33.3 wt.-.%)--. The nominal composition of the prepared catalysts was 0.5 wt.-% Pd, 0.1 wt.-% Pt, and 0.02 wt.-% Rh as the most usual in catalytic converters. After drying in nitrogen for 1 hour at 120~ final activation of the precursors was made by calcination at 550~ in a nitrogen atmosphere for 4 hours and subsequent treatment in a H2/N2=90/10 stream for 2 additional hours. The final catalysts resulted in the compositions shown in Table 1.

75 Table 1 Composition of the prepared catalysts, wt.-% Component CeO2 Pd

Pt

Rh

Pd Pt Rh

7.29 8.83 8.84

0.47 -----

0.079 ---

--0.021

Pd-Pt Pd-Rh Pt-Rh Pd-Pt-ah

8.43 8.81 9.16 8.87

0.45 0.47 --0.50

0.088 --0.081 0.087

--0.017 0.021 0.017

Pd+Pt Pd+Rh Pt+Rh Pd+Pt+Rh

(5.94+9.02)/2 (5.94+8.54)/2 (9.02+8.54)/2 (5.94+8.85+9.74)/3

0.84/2 0.84/2 --0.84/3

0.20/2 --0.20/2 0.31/3

--0.041/2 0.041/2 0.047/3

Monometallic catalysts (adsorption)

Multimetallic catalysts (co-adsorption)

Multimetallic catalysts (physical mixtures of monometallic catalysts)

2.2. A c t i v i t y t e s t s Catalytic activity data were obtained by using a conventional fixed-bed reactor at atmospheric pressure. A stainless steel tube with an inner diameter of 12 mm was chosen as the reactor tube. Catalyst (3.5 cm 8, ca. 1.8 g) was placed on ceramic wall at the lower part of the reactor. The upper part of the catalyst bed was packed with 10 cm 3 of inactive ceramic spheres (2 mm O.D.) to preheat the gas feed. The furnace temperature was controlled with a maximum variation of 2~ by an automatic temperature controller. The gas exiting the reactor was led to a condenser to remove water vapour. The remaining components were continuously analysed by non dispersive infrared (CO and CO2), flame ionisation (HC), magnetic susceptibility (O2), and chemiluminiscence (NOx). The redox characteristics of the model gas mixtures can be identified by the air-to-fuel ratio, A/F 14.63 A/F = 1+ 0.02545{[CO1 + [H 2 ] + 3 n [ C n H 2 n ] + ( 3 n + 1)[CnH2n+2 ] _ 2[02] _ [NO]} (1) To investigate the TWC behaviour of the prepared samples in an environment which resembled the exhaust A/F fluctuations in a closed-loop emission control system we used a similar apparatus to that developed previously by Schlatter et al. [11]. Two fast-acting solenoid valves allowed one to cycle between the two following feedstreams prepared in two independent gas blending systems: Reducing feedstream (A/F=14.13). It was composed of 10% C02, 1.60% CO, 900 ppm NO, 900 ppm Call6, 0.465% 02, 10.0% H20, and a balance of N2.

75

Oxidising feedstream (A/F=15.17). It consisted of 10% CO2, 0.40% CO, 900 ppm NO, 900 ppm C~H6, 1.26% 02, 10.0% H20, and a balance of N2. The prepared catalysts were tested cycling both feedstreams, with a frequency of 1 Hz, an amplitude of +0.5 A/F, and a space velocity of 125,000 h -1 (STP). The temperature was increased from 100 to 600~ at a rate of 3~ min "1, and the conversion data were continuously measured. Thus, the light-off temperature which is necessary to achieve 50% conversion, Tso, and the stationary conversion at the normal running temperature of 500~ Xsoo, were determined from the obtained activity data. Once the conversion-temperature profiles obtained, the experiment was continued at 500~ but shifting the cycled feedstream to some stationary feedstreams with the following composition: 10% CO2, 1.00% CO, 900 ppm NO, 900 ppm C3H6, 0.448% to 1.510% 02, 10.0% H20, and a balance of N2. These different oxygen percentages in feedtream allow us to experiment with A/F=14.33, 14.53, 14.63, 14.73, 14.93, and 15.13. From these experiments one can determine the stoichiometric window, defined as the interval of A/F inside which the conversion is equal or above 70% for all three contaminants. 3. R E S U L T S AND DISCUSSION

3.1. Activity under cycled feedstream composition Figure 1 shows the obtained CO-conversion profiles for all the tested catalysts. From this figure and similar ones for CaH6-conversion and NOconversion profiles (Figures 2 and 3, respectively), the T~o and Xsoo were determined resulting in the values shown in Table 2. Table 2 Tso and Xsoo obtained in cycled conditions for the prepared catalysts CO NO C3H6 Catalyst Tso (~ X~oo (%) Tso (~ Xsoo (%) Tso (~ Pd 322 100 325 70 313 Pt 159 100 305 90 300 Rh 245 99 253 71 268 Pd-Pt 307 100 310 76 303 Pd-Rh 312 100 316 76 307 Pt-Rh 173 100 247 91 257 Pd-Pt-Rh 304 100 307 89 301 Pd+Pt 271 100 281 90 277 Pd+Rh 267 100 267 90 273 Pt+Rh 166 100 247 90 260 Pd+Pt+Rh 249 100 257 89 264

Xsoo (%) 100 100 100 100 100 100 100 100 100 100 100

77 100 The analysis of reaction data becomes complex due to the large 80 number of reactions involved in the o< system [12]. Nevertheless, the high 60 conversions at 500~ shown in Table ~ 4o 3 confirm a very good three-way behaviour at this temperature for all 0 u 2O the tested catalysts, which are able to achieve total oxidation of CO and 0 600 300 400 500 100 200 C8H6, and high activity for NO 100 reduction, especially with the platinum-containing formulations. 80 As previously reported for platinum catalysts [13], in the 60 Pd-Pt profiles corresponding to Pt, Pt-Rh Pd-Rh and Pt+Pd, Pt+Rh, and Pt+Pd+Rh ".. Pt-Rh 40 Pd-Pt-Rh catalysts two regions can be clearly 20 observed: (i) the direct oxidation (CO+Y209 ~CO2) at low temperature 0 200 300 400 500 600 100 (T200~ when adsorbed HC and CO begin reaction with oxygen and/or NO. On the contrary, paraffins are weakly adsorbed on platinum surfaces [6,18]; in fact, if propene is substituted by methane in feedstream, the self-poisoning effect disappears from the CO-conversion profile. 0

-':

::

TM

I

: ~'

= : = .....

o

0

.....

78 The effect of the nature of hydrocarbon processed in feedstream was analysed in previous work [13]. The monometallic Pd-catalyst needs higher temperature (T~o=322~ than those needed by monometallic Pt and Rh catalysts (T~o=159/250 and 245~ respectively) to become active in CO removal. The co-adsorption of Pt and/or Rh with Pd does not improve the behaviour of the monometallic catalyst. However, co-adsorption of Rh with Pt enhances the CO-conversion in the second region (after direct oxidation) resulting in a positive synergic effect, i.e. the performance of the Pt-Rh catalyst can be described by a superposition of the performance features of the Pt-only and the Rh-only catalysts, and also very similar to the performance of the Pt+Rh catalyst. This suggests that the oxidation of CO at low temperatures (125-200~ occurs on platinum, while at higher temperatures this oxidation occurs mainly on the rhodium sites, being both functions accesible to the reactant. With the Pd-containing catalysts, the total oxidation of CO is reached in only one step, as observed with monometallic Pd and Rh catalysts, once the removal of propene has begun with oxygen (C3H6+4 89 and NO (C3H6+9NO-~4 89 The co-adsorption of Pt and/or Rh with Pd does not improve the behaviour of the monometallic Pd-catalyst, suggesting that with Pd-catalysts prepared by coadsorption the elimination of CO occurs mainly on Pd, which could be even covering part of the other metal surface. These results indicate that Pd may alloy with rhodium and segregate to the particle surface inhibiting the rhodium function as has already been postulated as a cause of deactivation in previous Pd-Rh catalysts [7-9]. The total CO-conversion is obtained once the olefin has been completely converted, as can be seen by joint observation of Figures 1 and 3. The CO-conversion reached during the first oxidation step with the multimetallic Pd+Pt, Pt+Rh, and Pd+Pt+Rh being lower than that obtained with the monometallic Pt catalyst is due to the fact that, although there is no difference in the platinum loading, each active phase is distributed in one part of the catalyst bed, which implies less metal dispersion and metal surface area to be involved in the reaction. The oxidation of propene by either 02 and NO (Figure 2) is well achieved with all the prepared catalysts, which begin to be active around 250~ followed by a sharp rise in activity to total conversion at temperature of 3000C and above. The Rh-only catalyst presents lower light-off temperature than the Pt-only catalyst which is similar to that of the Pd-only catalyst. The Pt-Rh prepared by co-adsorption gives the best light-off performance, the presence of Rh enhancing the effectiveness of Pt for HC performance. On the contrary, all the Pdcontaining catalysts prepared by co-adsorption present similar behaviour to that corresponding to the Pd-only catalyst. The performance of catalysts prepared by physical mixture can be described by a simple linear ponderation of the performance features of their corresponding single-metal catalysts, resulting in light-off curves practically coincident.

79

100 -

100 80 -

8O

iff o

60

>4) o'~ o 3: "

40

~

60 -

'~Pd' -...e--et

"~ ~ ~

" 40 -

_

~

.

~

20 0 100

200

'~176 -

300

400

500

2o 0 100

600

1~176

80

pt i Rh I

200

300

400

500

600

80

60 = :

40

Pd-Rh Pt-Rh

~

60

c

40

-" Pd-Pt -~' Pd-Rh ~Pt-Rh = Pd-Pt-Rh

0 100

200

300

400

500

0 100 r 100

600

100

200

300

400

500

600

80

_o :9 4) > r O O

A

60

I

40

-" Pd+Pt ". Pd+Rh v" Pt+Rh ~- ~Pd+Pt+Rh

4) ~= 0 o

/

40

~I

I -" :Pd+Pt

~

I~ P d + R h

2O

10o

200

300

400

Temperature,

500

600

~

Figure 2. Temperature-programmed C3H6conversion profiles obtained with all prepared catalysts

100

200

300

400

Temperature,

500

600

~

Figure 3. Temperature-programmed NOconversion profiles obtained with all prepared catalysts

The steam-reforming does not occur in our operational conditions, which also was experimentally proved by performing the propene steam-reforming reaction (removing the rest of components in feedstream) and noting that this reaction occurs at temperatures above 400~ [12]. Comparison of Figures 1 and 2 makes clear how the CO-conversion is restablished with the beginning of the propene oxidation due to a decrease in the inhibition caused by self-poisoning.

80 Concerning the light-off temperatures for C3H~ removal, although big differences cannot be observed, the best behaviour corresponds to the Pt-Rh catalyst, followed by the multimetallic catalysts prepared by physical mixture, whose CaH6-conversion profiles are practically coincident. In relation to the NO-conversion (Figure 3) notable differences were found when running with the prepared catalysts. The elimination of NO can occur through reaction with CO and with Call6 as was already mentioned above. The behaviour of each metal can be compared with results obtained from the monometaHic formulations: platinum is the most active metal at the running temperature (X~oo=90%) but needs higher temperature than rhodium to become active (T5o=305 for Pt vs. 253 for Rh), the latter allowing a conversion at 500~ of 70%. The palladium shows lower values of both T~o and Xsoo,presenting the most unfavourable NO-conversion profile in Figure 3 at all temperatures. Again the sinergic Pt-Rh interaction can be observed as this catalyst presents the best behaviour, making use of the advantages of both metals, with T5o=247~ and X5oo=91%. The rest of formulations prepared by co-adsorption present a behaviour closer to the monometallic Pd catalyst. This could be explained considering that the higher amount of palladium in the bimetallic formulations, Pd/Pt=0.45/O.088=5.1 and PdfRh=0.47/O.O17=27.6, could be responsible for covering some of the platinum or rhodium sites restricting accessibility of reactants. Finally, the physical mixture of monometallic catalysts presented good NO removal capacity, with X~oo=90%, and intermediate light-off temperatures, even when rhodium is not present in the formulation.

3.2. Activity under stationary feedstream composition Table 3 shows the limits and amplitude of the stoichiometric windows obtained for all the studied catalysts. The upper limit of the stoichiometric windows is marked in all cases by the capacity of the catalyst to reduce NO above 70% conversion. On the other hand, the lower limit is marked by the high oxidation capacity of the catalyst to oxidize both HC and CO. All the prepared catalysts obtained practically total CO-conversion under net oxidising and slightly reducing conditions, conversion decreasing with the reducing character of the feedstream. Under reducing conditions the formulations with platinum resulted more active for CO removal, followed by palladium and then by rhodium. The removal of C3H6 was total with all catalysts containing Pd and under all tested conditions, oxidising and reducing. The Pt and Rh-containing catalysts allowed high conversions under oxidising conditions, decreasing under reducing conditions till 60%. The NO-conversion under reducing conditions resulted close to 100% with catalysts containing Pd and/or Rh, in spite of the reported low capacity of palladium for the NO reduction reaction attributed to some self-poisoning by hydrocarbons [5,16]. This effect is minimised in the prepared catalyst due to

81

Table 3 Stoichiometric windows for the prepared catalysts Catalyst CO (lower) HC (lower) NO (upper) (upper=15.13) (upper=15.13) (lower=14.13) Pd 14.36 14.13 14.55 Pt 14.13 14.25 14.60 Rh 14.44 14.13 14.60 Pd-Pt 14.13 14.13 14.57 Pd-Rh 14.13 14.13 14.57 Pt-Rh 14.26 14.13 14.68 Pd-Pt-Rh 14.13 14.13 14.56 Pd+Pt 14.13 14.13 14.59 Pd+Rh 14.13 14.13 14.57 Pt+Rh 14.24 14.13 14.62 Pd+Pt+Rh 14.13 14.13 14.57

Overall L0wer.Upper 14.36- 14.55 14.25- 14.60 14.44- 14.60 14.13- 14.56 14.13- 14.56 14.26- 14.68 14.13- 14.57 14.13 - 14.59 14.13- 14.57 14.24- 14.62 14.13- 14.57

Amplitude 0.19 0.35 0.16 0.43 0.43 0.42 0.44 0.46 0.44 0.38 0.44

their high activity for C3H6-oxidation. Under net oxidising conditions low conversions were obtained with all prepared formulations. The monometallic formulations resulted in much shorter amplitude of the stoichiometric window than multimetallic formulations. None multimetallic formulation presented notable differences in the amplitude, except for some displacement to the lean conditions for Pt-containing formulations and to the rich conditions for Pd-containing formulations.

4. CONCLUSIONS All the prepared catalysts oxidised completely both C3H6 and CO at 500~ under cycled oxidising-reducing conditions, presenting differences only in the reduction of NO. The Pd-Rh catalysts have resulted in different characteristics in comparison with Pt-Rh catalysts. This characteristic of Pd-Rh catalyst is similar to that of Pd-only catalyst. Pt, Pt-Rh and Pd-Pt-Rh prepared by co-adsorption converted 90% of NO, whereas Pd, Rh, Pd-Pt and Pd-Rh converted around 7075%. All the physical mixtures of monometallic catalysts reached 90% NOconversion. The co-adsorbed Pt-Rh catalyst presented the lowest light-off temperatures for all three contaminants. For the studied TWCs the light-off performance of Pt-Rh is dominated to a large extent by the Rh function, whereas in the case of Pd-Rh systems alloying has appreciable more negative effects on performance and suppressed the Rh function for NO reduction.

82 ACKNOWLEDGEMENTS

The authors acknowledge the financial support by the Basque Government, the Spanish Education and Science Ministry (PI93-44 and AMB93-574 projects) and the University of Basque Country (EB076/94). J.A.B also acknowledges to the Basque Government by the grant to work in the present research. REFERENCES

1. Entrena, J., PhD Thesis, Universidad del Pals Vasco/EHU, Bilbao 1994. 2. C. Howitt, V. Pitchon, F. Garin and G. Marie, in "Catalysis and Automotive Pollution Control III", A., Frennet and J.-M. Bastin (editors), p. 149-161, Elsevier, Amsterdam 1995. 3. J.R. Gonzfilez Velasco, J. Entrena, J.A. Gonzfilez Marcos, J.I. Guti6rrez Ortiz and M.A. Guti6rrez Ortiz, Appl. Catal B, 3 (1994) 191. 4. H. Praliaud, A. Lemaire, J. Massadier, M. Prigent and G. Mabilon, in " l l t h International Congress on Catalysis - 40th Anniversary", J.W. Hightower, W.N. Delgass, E. Iglesia and A.T. Bell (editors), p. 345-354, Elsevier, Amsterdam 1996. 5. S. Matsuura, A. Hirai, K. Arimura and H. Shinjoh, SAE Technical Paper 950257 (1995). 6. J.T. Kummer, J. Phys. Chem., 90 (1986) 4747. 7. S.H. Oh and J.E. Carpenter, J. Catal., 98 (1986) 178. 8. H. Muraki, H. Sobukawa, M. Kimura and A. Isogai, SAE Thechnical Paper 900610 (1990). 9. J.G. Nunan, W.B. Williamson, H.J. Robota and M.G. Henk, SAE Technical Paper 950258 (1995). 10. R.J. Brisley, G.R. Chandler, H.R. Jones, P.J. Anderson and P.J. Shady, SAE Technical Paper 950259 (1995). 11. J.C. Schlatter, R.M. Sinkevitch and P.J. Mitchell, Ind. Eng. Chem. Prod. Res. Dev., 19 (1980) 288. 12. K.C. Taylor, "Automobile Catalytic Converters", Springer-Verlag, Berlin 1984. 13. J.R. Gonzfilez Velasco, J.A. Botas, J.A. Gonzfilez Marcos and M.A. Guti6rrez Ortiz, Apl. Catal. B, 12 (1997) 61. 14. B.J. Whittington, C.J. Jiang and D.L. Trimm, Catal. Today, 26 (1995) 41. 15. M. Mundschau and B. Rausenberger, Plat. Met. Rev., 35 (1991) 188. 16. T. Engel and G. Ertl, Adv. Catal., 28 (1979) 1. 17. Gonzfilez Velasco and cols., unpublished results. 18. S.H. Oh, P.J. Mitchell and R.M. Siewert, J. Catal., 132 (1991) 287. 19. Muraki, H. Shinjoh and Y. Fujitani, Appl. Catal., 22 (1986) 325.

CATALYSIS AND AUTOMOTIVEPOLLUTIONCONTROLI'V Studies in Surface Science and Catalysis, Vol. 116 N. Kruse, A. Frennet and J.-M Bastin (Eds.) 91998 Elsevier Science B.V. All rights reserved.

83

Effect of Alkaline Addition on Hydrocarbon Oxidation Activities of Palladium Three-way Catalyst H. Shinjoh, N. Isomura, H. Sobukawa, and M. Sugiura TOYOTA CENTRAL R&D LABS., INC. Nagakute-cho, Aichi-gun, Aichi-ken, 480-11, Japan ABSTRACT The effect of alkaline addition on hydrocarbon oxidation activities of Pd catalyst supported on ,/-alumina was investigated using simulated automotive exhaust gases. The hydrocarbon oxidation activity of the Pd catalyst with alkaline earth metal such as Mg, Ca, Sr, and Ba is higher than that with nothing added. On the other hand, the activity of the Pd catalyst with alkali metal such as K and Cs is lower than that with nothing added. From the results of the partial reaction orders in C3H6 oxidation, TPR, and XPS, it was concluded that the alkaline addition to the Pd catalyst increased electron density of Pd on the catalyst and weakened the adsorption strength of Pd with hydrocarbons. The addition of alkaline earth metal suppressed the hydrocarbon chemisorption on the Pd catalyst and therefore allowed the catalytic reaction to proceed smoothly. On the other hand, the addition of alkali metal caused such a strong oxygen adsorption on Pd that rejected the hydrocarbon adsorption and suppressed the reaction. 1. INTRODUCTION Automotive three-way catalysts consist of noble metals, supports with a large specific surface and some additives. Many kinds of additives, such as cerium oxide, nickel, and other compounds, are used to supplement the activity of noble metals and to improve durability of automotive three-way catalysts. Cerium oxide is generally added to store oxygen under oxidizing conditions and to release the stored oxygen under reducing conditions(1-8). Nickel compound is often added as a scavenger of hydrogen sulfide and its addition also improves catalytic activity(9-12). Recently, barium(Ba) compound has been added to some kind of three-way catalysts, for example, palladium(Pd) only three-way catalyst and NOx storage reduction three-way catalyst(13-15). In spite of the fact that automotive three-way catalysts containing such alkaline compounds are already in practical use, the effect of alkaline addition on the catalytic activities of the catalysts are not yet clear. The authors have been studying the effect of alkaline addition on the catalytic activity of automotive three-way catalysts. We have found that the addition of Ba to Pd or platinum(Pt) three-way catalysts is effective for improvement of catalytic activity under reducing conditions, and that the suppression of hydrocarbon(HC) chemisorption on the catalysts by the addition of Ba allowed the catalytic reaction to proceed smoothly (16,17). This paper systematically reports the effect of alkaline addition, that is, alkali metals and alkaline earth metals, on hydrocarbon oxidation activity of Pd three-way catalyst.

84 2. EXPERIMENTAL

2.1. Catalysts The Pd catalyst was prepared impregnating y-alumina powder (a BET area of 200m2/g) with aqueous solution of palladium nitrate. The powder was dried overnight at 110~ in air, followed by calcination at 600~ for 5h in air, pressed, crushed, and sieved into 0.5 to 1.0 mm particles. The Pd/X catalysts (X=Li, Na, K, Cs, Mg, Ca, Sr, and Ba) were prepared impregnating the Pd catalyst powder with aqueous solution of each nitrates. The powder was dried, calcined, followed by pressing, crushing, and then sieved into particles in the same procedures mentioned above. Pd and alkaline loading amounts were 0.024 and 0.17 mole to one molar y-alumina, respectively.

2.2. Catalytic activity measurements The laboratory reaction system used was a conventional flow system with a tubular fixedbed reactor as described elsewhere(18). The characteristic feature of this system is its ability to simulate various air to fuel ratios (A/F) of automotive exhaust gases using eight mass flow controllers. In this study, catalytic activity on the catalysts in simulated automotive exhaust gases was measured as a function of E, which is a normalized value of A/F by a stoichiometric -l

one in the simulated exhaust gas, at 300~ and 420,000 h space velocity. The compositions of the simulated exhaust gases for each ~, are shown in Table 1. Catalytic activity was expressed as percent conversions of NOx(NO+NO2), CO, and HC. C3H6 oxidation activity was also measured to decide the kinetic parameters on the catalysts using the same laboratory reaction system. The compositions of C3H6 and O 2 were changed from 0.017 to 0.133 vol% and from 0.1 to 1.3 vol%, respectively, and space velocity was the same as that mentioned above.

2.3. Temperature programmed reduction(TPR) measurement The TPR measurement was performed using a flow system with a fixed-bed tubular reactor as described elsewhere(19). The 5 vol% H2/Ar was used in this measurement. Space -1

velocity was 7,000 h and heating rate was a linear rate of 50~ from -30 up to 300~ The effluent from the reactor was analyzed by both a thermal conductivity detector and a quadrupole mass spectrometer.

2.4. X-ray photoelectric spectroscopy (XPS) measurement The XPS measurement was performed using PHI-5500MC with MgK~ radiation as -9 incident beam. The base pressure of the instrument was 1x 10 Tort. Before conducting the XPS analysis, a catalyst was heated in pretreatment chamber connected to XPS chamber under 10 Tort of 10vol% C3H6/N2 at 400~ and then the pretreatment chamber was evacuated to 10 Tort and the catalyst transferred into the spectrometer without exposure to air. The electronbinding energy scale was calibrated by assigning 74.2 eV to A1 2p peak position.

85 Table 1 Compositions of simulated exhaust gases (N2 balance). #:[HJCO]=I/3

~. H2/CO#

C3H6

NO

02

C02 H20

(%) 0.960

2.00

0.062

0.12

0.41

10.0

3

0.967

1.73

0.060

0.12

0.41

10.0

3

0.974

1.49

0.058

0.12

0.43

10.0

3

0.980

1.33

0.057

0.12

0.46

10.0

3

0.985

1.20

0.056

0.12

0.49

10.0

3

0.990

1.00

0.055

0.12

0.54

10.0

3

0.992

1.05

0.055

0.12

0.56

10.0

3

0.994

1.01

0.054

0.12

0.57

10.0

3

0.996

0.99

0.054

0.12

0.60

10.0

3

0.998

0.96

0.054

0.12

0.62

10.0

3

0.999

0.95

0.053

0.12

0.63

10.0

3

1.000

0.93

0.053

0.12

0.65

10.0

3

1.001

0.92

0.053

0.12

0.66

10.0

3

1.002

0.91

0.053

0.12

0.67

10.0

3

1.004

0.88

0.053

0.12

0.70

10.0

3

1.006

0.85

0.052

0.12

0.72

10.0

3

1.008

0.84

0.052

0.12

0.75

10.0

3

1.010

0.83

0.052

0.12

0.78

10.0

3

1.015

0.77

0.051

0.12

0.85

10.0

3

1.020

0.73

0.050

0.12

0.92

10.0

3

1.027

0.69

0.050

0.12

1.03

10.0

3

1.034

0.65

0.049

0.12

1.13

10.0

3

1.040

0.60

0.049

0.12

1.21

10.0

3

3. R E S U L T S A N D D I S C U S S I O N

The conversions of HC, CO, and NOx on the Pd and Pd/Sr catalysts plotted as a function of ~. in simulated exhaust gases at 300~ are shown in Figs.1 and 2, respectively. The catalytic activity on the Pd/Sr catalyst was superior to that on the Pd catalyst, in particular, under reducing conditions defined as ~. Sr, Ca > Mg > none > Li > Na > K > Cs The alkaline earth metal addition to the Pd catalyst improved the hydrocarbon oxidation activity. Similar phenomena have been observed on Pd/Ba and Pd/La catalysts, and it is concluded that the suppression of hydrocarbon chemisorption on Pd by the addition of Ba or La allows the catalytic reaction to proceed smoothly under reducing conditions(16,20). On the other hand, the alkali metal addition, especially K or Cs, to the Pd catalyst deteriorated the hydrocarbon oxidation activity. In C3H6-O 2 reaction system, the hydrocarbon oxidation activity on the Pd catalysts with alkaline compound was measured to get further information. The rate of carbon dioxide formation V(CO2) in the reaction of C3H 6 with 02 is given by the following equation (1). (1)

V(CO2)=kxP(C3H6)mx P(o2)nxexp(-AE/RT)

where P(C3H6) and P(02) are the partial pressures of C3H6 and 02, m and n are the partial reaction orders of CBH6 and O 2, respectively. The values of m and n were determined from a conventional log-log relationship between V(CO2), obtained

i00

l'"

NOx

80cC3 O3 C_ [13 > CO C_3

60-

40-

/ I

20 --

I,

0.96

I

'

'

' i

/ /

H C/" ./

/

/

J

'1

/

/

CO

!

~

1.

0.98

!

_

1

i.O0

I

......

!.

.

.

1.02

.

1

, !

1.04

A Fig.1 Conversion efficiencies as a function of X in simulated exhaust gases at 300~ catalyst.

for Pd

87

100

c

0

/'

NOx

80

60

/

~

a

/

-r-t

r._. > o

rj

/ H C

40

/

/

/

/

/

/

/CO

20

0

_l

. . . . .

t

0.96

.....

I

I

t

o.g8

_

t

I

1.00

!

t.02

.,. f

,

1.04

A

Fig.2 Conversion efficiencies as a function of X in simulated exhaust gases at 300~ catalyst.

I00

....

,

'

80 C o .r,,4

60

r_

:> c o

40

I

!

.

.

.

.

.

.

.

.

for Pd/Sr

w

none

.Z/,//"

(-) 0 ....

V " ~ql'---~-

~..~,_.~.~i.~

200

~F---Y--Y

I /f/

:/ l

40

.,- /

../ "i- :,

300

i

/

- "-

,

,

Pdloo

,

400

Temperature (~

40

3a)

~ r

-~"~o,,,,Q

.d ;X

20

~

>>-~--13----E3----E3---C3---E~-C}---t~--

200

300

400

500

Temperature (~ Figure 3. NO conversion in absence or in presence of hydrocarbon (C3H 6 or C3H8) 3a) on Pd, 3b) on Pd 65Mn35 and 3c) on Pd30 Mn70 samples. Figure 4. Hydrocarbon oxidation on Pd and PdMn samples: 4a) C3H6 and 4b) C3H8.

500

109 the oxidation of both CO and C3H6, taking in account the composition of the gas mixture. Such an inhibition by excess of oxygen is not observed on this Pd30Mn70/SiO 2 sample since the two oxidations of both CO and C3H6 nearly occur at the same temperature, - 250~ The C3H8 oxidation is not enhanced in presence of PdMn samples. Therefore, the NO conversion is not expected to be better on the bimetallic samples than on Pd~00 as observed.

4. IR STUDIES OF ADSORBED PROBE MOLECULES: MIXTURE

CO, NO OR CO-NO

In order to tentatively clear up the specific catalytic behaviour of catalysts containing Mn, adsorption and coadsorption of CO and NO have been carried out at 300 and 573 K. The adsorbed species were characterized by IR spectrometry.The IR experiments were carried out on a Brucker IFS 110 FTIR spectrometry with a 4 cm -1 resolution in the 1000- 4000 cm -1 region.The samples were compressed up to 2.105 kPa in order to obtain a thin disc of about 15 mg and a 15 mm diameter. Before any IR study, all the samples were reduced in the IR cell and evacuated at the reduction temperature then cooled under vacuum. The IR spectra of the reduced samples were firstly recorded, then the >molecules (CO,NO or the CONO stoichiometric mixture) were introduced under about 1.3-2 kPa and the IR spectra were recorded either under gaseous atmosphere or after evacuation. After smoothing and substraction of a linear background, the IR spectra of gaseous and adsorbed molecules are given by the spectra between these spectra and the initial spectra of reduced samples. On the Pd reduced samples, the CO adsorption carried out at room temperature shows the three infrared bands of adsorbed CO: linearly bonded CO at 2050 cm~, bridged bonded CO near 1950 cml and multibonded CO near 1850 cm"~. When increasing amounts of Mn are added to Pd~00, the CO species adsorbed on top (vCO at 2050 cml ) decrease and disappear quasi completely at higher Mn contents ( Mn at.% >_ 35 %). Moreover, the intensities of the two other infra-red bands at 1950 and 1850 crn~ decrease (Fig.5). These IR bands are attributed to CO adsorbed either on Pd, Mn (11) or Pd-Mn dual sites. Such a disappearance of the 2050 cm~ IR band on the bimetallic Pd-Mn samples at high Mn content is rather surprising since, in general, the relative intensity of the linear CO IR band increases when a second metal is added to Pd due to the surface Pd dilution by the second metal (12). In order to explain this unusual behaviour, it has been assumed that the Mn segregation occurs on the low coordination surface atoms which agrees with the theoretical approach of this surface segregation phenomenon (13). However, no quantitative conclusions on the amounts of bridged and multibonded CO species (vCO at 1950 and 1850 cm"~) can be drawn. Indeed, the extinction coefficients can be changed with the nature of the adsorption sites. On Pd-Mn samples, in addition to the usual IR bands between 1850- 2050 cm~ assigned to CO molecular adsorption, CO reacts with the ~ mobile )) oxygen atoms of MnOx to form carbonated species characterised by the IR bands between 1750 a 1400 cm~. For the adsorption of NO on Pdl00, three infra-red bands are recorded at 1750, 1650 and 1550 cm~ assigned respectively to linearly, bent and bridged bonded NO (14). The sequential adsorptions : CO, NO then CO, show that each gas (NO or CO) displaces the other, which means that NO or CO are adsorbed on the same sites. However, the CO IR bands are

110

........ Pd

Pd 1oo - -" PdgoMn 1o

-----Pd65Mn35

r162 r

tO

40

o "r" i

3

,. . . .

,__, 80 9-.

s

/ /" I' /

E 200

.....................................

15

20 0

0

5

10

15

20 25 time [s]

30

35

40

Figure 12: Cold-start experiment with CHC-concept and aged catalyst simplify the comparison between the different concepts it is assumed that the electrically heated pre-catalyst doesn't show any aging tendencies. Using only a fresh main catalyst the cold-start emissions of hydrocarbons are about three times higher compared to the future legal requirements (ULEV). With decreasing catalyst activity the cold-start behavior of the main catalyst gets worse which results in increasing HC-emissions. The BHC-concept reduces the cold-start emissions considerably but is most strongly influenced by catalyst deactivation. Due to the short front part of the catalyst which is heated to high temperature levels (Figure 4) the length of an inert front area of an aged catalyst has a strong influence on the HC-emissions. Future legal requirements can be fulfilled with the EHC-concept for fresh catalysts. As the main catalyst is only heated by the exhaust gas again a considerable increase of the coldstart emissions results from catalytic aging. Hence, the ULEV-limit can not be kept for lower activities than 40 %. The CHC-concept shows the lowest HC-emissions in comparison to the other cold-start concepts. This is due to the fast and direct heating of the catalytic surface where the hydrogen combustion only takes place at active parts of the catalyst. The selective heating of the active areas of the monolith by the CHC-concept can also be demonstrated experimentally. An automotive catalyst was aged artificially in a way that the first 3 cm were completely deactivated. With this aged catalyst the cold-start experiment from Figure 7 was repeated. The heating by the hydrogen combustion results in a sharp temperature rise at the rear active part of the monolith (left diagram of Figure 12). Because of the inert entrance of the catalyst the temperature profiles are shifted downstream but are comparable in hight to the experiment with fresh catalyst. For this reason high conversion rates are reached after 10-15 s (right diagram of Figure 12). The obtained maximum HC-conversion of 95 % is probably due to inaccuracies of the gas analyser because at the measured temperature level of 500 ~ to 600 ~ total combustion of the pollutants is usually obtained. 5. CONCLUSIONS In detailed simulation studies the dynamic behavior of automotive catalytic converter systems during start-up is described for the EHC-, BHC- and the innovative CHC-concept, where light-

135 off of the monolith is induced by the catalytic combustion of hydrogen. In the simulation studies all concepts show a good performance for the reduction of the cold-start emissions as long as fully active catalyst is considered. Thereby, the necessary power input varies from 1.5 kW (EHC) tp approx. 13 kW (BHC). The results change drastically if a catalyst deactivation profile is assumed as observed in real application after a certain running period. Due to the selective heating the CHC-concept gives the best conversion behavior during start-up for an aged catalyst, whereas the B HC- and EHC-concept failed to reach future legal requirements for deactivated catalysts. The simulation results are confirmed in a number of specific cold-start experiments. NOTATION aext ao ax

Cp D Gz AhR

L

m 2

[~-~] m 2 [~] m 2 ,

[-~-] kJ [ kT_~.K] [m~-~.s] [ mk2---~].s kJ [Y~7] [m]

Mj

[ kgj 1

R

r kmol ]

t

[sl

wj Z

[m]

Ol

[ kW

e ~.

[-] [ kw m---~] [-]

tkmolj j

v o

L m2t.s J

[~1 m-r~.K]

[oc]

external surface to volume area ratio of monolith geometrical surface to volume area ratio of monolith catalytic surface to volume area ratio of monolith specific heat capacity dispersion coefficient specific mass flow heat of reaction length of monolith molar weight of component j specific heat flux reaction rate time weight fraction of component j spatial coordinate heat transfer coefficient mass transfer coefficient void fraction thermal conductivity stoichiometric coefficient density temperature

Indices amb g i

ambient gas reaction step i inlet component j solid

136 REFERENCES

1. W. Held, A. Donnerstag, E. Otto, P. Ktiper, B. Pfalzgraf ,and A. Wirth : The System Development of Electrically Heated Catalyst (EHC) for the LEV and EU-III Legislation. SAE Technical Paper Series Nr. 951072, 1995. 2. A. Donnerstag, A. Degen, W. Held and K. Korbel: Erftfllen der ULEV-Norm durch elektrisch beheizten Katalysator. VDI Fortschrittsberichte Reihe 12: Verkehrstechnik/Fahrzeugtechnik, (239), 1995.16. Intemationales Wiener Motorensymposium. 3. P. Oser, E. Mtiller, G.R. H/artel and A.O. Schtirfeld Novel 9 Emission Technologies with Emphasis on Catalyst Cold Start Improvements - Status Report on VW-Pierburg BumerICatalyst Systems. SAE Technical Paper Series Nr. 940474, 1994. 4. K. Kollmann, J. Abthoff and W. Zahn: Concepts for Ultra Low Emission Vehicles. SAE Technical Paper Series Nr. 940469, 1994. 5. B.H. Engler, D. Lindner, E.S. Lox, K. Ostgathe, A. Sch/ifer-Sindlinger and W. Mtiller : Reduction of Exhaust Gas Emissions by Using Hydrocarbon Adsorber Systems. SAE Technical Paper Series Nr. 930738, 1993. 6. M.D. Patil, W. Hertl, J.L. Williams and J.N. Nagel : In-Line Hydrocarbon Adsorber System for ULEV.SAE Technical Paper Series Nr. 960348, 1996. 7. T. Kirchner and G. Eigenberger: Optimization of the Cold-Start Behaviour of Automotive Catalysts Using an Electrically Heated Pre-catalyst. Chem. Eng. Sci., 51(10): 2409-2418, 1996. 8. S.E. Voltz, C.R. Morgan, D. Liederman and S.M. Jacob: Kinetic Study of Carbon Monoxide and Propylene Oxidation on Platinum Catalysts. Ind. Eng. Chem. Prod. Res. Develop., 12(4): 294 301, 1973. 9. Y. Kanada, M. Hayasi, M. Akakaki, S. Tsuchikawa and A. Isomura" Hydrogen Added After-Bumer System. SAE Technical Paper Series Nr. 960346, 1996. 10. T.Kirchner and G. Eigenberger: On the Dynamic Behaviour of Automotive Catalysts. Catalysis Today, 1997. (in press).


137

Measurement of the ceria surface area of a three-way commercial catalyst after laboratory and engine bench aging. E. Rogemond a, N. Essayem b, R. Fr6ty a, V. Perrichon a*, M. Primet a, S. Salasc a, M. Chevrier c, C. Gauthier c and F. Mathis c.

aLaboratoire d'Application de la Chimie ~ l'Environnement (LACE), UMR 5634, CNRS/Universit6 Claude Bernard Lyon I, 43 boulevard du 11 Novembre 1918, 69622 Villeurbanne Cedex, France. Fax : 334 78 94 19 95. e-mail : pe rrich@cp mol. univ -lyon 1.fr bInstitut de Recherches sur la Catalyse, 2 avenue Einstein, 69626 Villeurbanne Cedex, France. cR6gie Nationale des Usines RENAULT, Direction de ring6nierie des Mat6riaux, 8-10 avenue Emile Zola, 92109 Boulogne-Billancourt Cedex, Centre de Lardy, 1 all6e Cornuel, 91510 Lardy, Direction de la Recherche, 9-11 avenue, du 18 Juin 1940, 92500 Rueil Malmaison, France.

ABSTRACT The ceria surface area of a commercial Pt-Rh three-way catalyst was determined after laboratory hydrothermal aging at 1173-1373 K and after 200 h on engine bench. It was measured by X-ray diffraction (XRD) line broadening analysis and by a method based on the exploitation of the hydrogen temperature programmed reduction (TPR) profiles. In this case, the hydrogen uptakes below about 900 K include the ceria surface reduction and t h a t of the oxidized noble metals. They are analyzed and discussed, assuming two possiblities for the metals oxidation state. Compared to the fresh catalyst, the TPR profiles are deeply modified by the aging treatments. The ceria seems to sinter more t h a n alumina, particularly between 1173 K and 1273 K. After aging at 1273-1373 K, the calculated ceria surface area is only 15-10 m2g -1 washcoat, which represents 20% of the BET area, instead of 40% initially. A stabilization t r e a t m e n t at 823 K under reactants leads to an additional ceria sintering, even for the more aged system. Finally, the m e a s u r e m e n t s on the engine bench aged catalysts seem to indicate a better resistance of ceria to sintering in working conditions. The presence of a pollutant layer, containing phosphorus, zinc and calcium, did not modify the accessible ceria surface area measured by TPR.

138 1. I N T R O D U C T I O N Determining the surface area of ceria in three-way catalysts is an important problem for the characterization of these systems. Indeed, cerium oxide is a key component which enhances the global activity, particularly through the oxygen storage capacity (OSC), essential characteristic for regulating the oxidative power of the catalyst in a real catalytic converter [1-3]. Moreover, it has been proposed on model systems that the ceria support may become itself the active phase [4-7]. Accordingly, there is a great practical interest to find methods which are easily usable and allow a correct estimation of this parameter for fresh as well as for aged catalysts. In preceding papers, we have described a methodology to measure the exposed surface area in model metal/ceria-alumina catalysts [8-10]. It is based on the use of temperature programmed reduction (TPR) with hydrogen. It was shown that the reduction peaks for T lower than about 900-950 K, could be attributed to the reduction of both the oxidized precious metals and the superficial ceria layer. It has been established that 3.9 ~tmol H2 are necessary to reduce 1 m 2 of CeO2 [9]. The objective of this study is to extend this methodology to the case of a commercial three-way catalyst deposited on a ceramic monolith and to follow the evolution of the ceria surface area at different stages of its life-time, including long time engine-bench testing. In this case, it is important to examine the impact of poisons on TPR curves.

2. F ~ P E R I M E N T A L All characterizations were carried out on the same commercial catalyst (COM1). The active phase was constituted of 0.152 wt.% Pt and 0.031 wt.% Rh, the corresponding Pt/Rh weight ratio being 5/1 with a 32 g ft ~ loading. The ceramic was constituted of cordierite which contained iron as impurity. The hydrothermal laboratory aging (LA) consisted in treating the fresh catalyst during 5h at 1173, 1273 or 1373 K under a 6 1.h1 nitrogen flow containing 10% H20 introduced with an automated syringe. These four catalysts were stabilized (St), i.e. treated lh at 823 K under stoichiometric reactants synthetic mixture (CO, C3H6, C~Hs, NO, 02, CO2, H20). Moreover, the catalyst was studied after aging on engine bench (EB) during 200 h, with a air/fuel equivalence ratio oscillating around stoichiometry (~ ~ 1). Table 1 gives the results of the chemical analysis performed on the initial (fresh) monolith system. The figures remained nearly the same (within + 10%) after laboratory or engine-bench aging. We used X-ray diffraction technique to determine the CeO2 crystaUite sizes from the line broadening and the hydrogen TPR to calculate the ceria surface area according to a method developed on model catalysts [9,10]. In the case of TPR, the samples were ground before characterization. 250 mg were necessary for each run. They were treated lh at 673 K under air, and then lh30 at 773 K under argon flow before the TPR run. The heating ramp was 20 K min -1 up to

139 about 1073 K, temperature which was kept constant during 45 min. All the values will be given per gram of actual waschcoat (WC) Some XPS and SEM-EDX analysis were also realized on the engine bench aged catalysts in order to measure the surface composition of the catalytic washcoat which was modified by the poisons layer deposit. Table 1 Chemical analysis of catalyst COM 1 (wt. % basis). Washcoat 30.75

Pt

Rh

Ba

Ce

C1

Fe

La

S

0.152

0.031

0.66

6.3

0.145

0.45

0.3

0.15

2. R E S U L T S 2.1. Ceria m e a n p a r t i c l e size from X ray d i f f r a c t i o n Each catalyst was studied by XRD. To improve the intensity of the diffraction lines, the washcoat was scraped off the cordierite before analysis. The XRD spectra show the presence of alumina, ceria and some residual cordierite. The precious metals are never detected, as metal or oxide. For the fresh catalysts, the lines are broad meaning poorly crystallized phases. The resolution is greatly improved after aging. No other phase like cerium aluminate or other transformation product were evidenced on the XRD spectra. The calculation of the particle size of ceria was done on the broadening of the line at 20 = 56.37 ~ which is the best resolved and does not interfere with alumina. The results are given in Table 2.

Table 2 Ceria particle size determined from XRD diagram (line 20 = 56.37~ Catalyst

Fresh LAl173 LA1273 LA1373

particle size of CeO2 (nm)

S CeO2-XRD(m~g~)"

before stabilization

after stabilization

before stabilization

after stabilization

8.5 11.8 17.8 22.2

10.3 13.9 18.1 24.4

25 18 12 10

21 15 12 9

* S = [ 6,000 / (7.15 * Diameter in nm) ] ceria 9 percentage in the washcoat

140 In the fresh state, the ceria is rather well dispersed with a mean size of around 8 nm. During the aging up to 1373 K, this size increases from 8.5 to 22.2 nm, which evidences an important sintering of the ceria particles. This evolution is identical for the stabilized catalysts, with however slightly higher sizes. The corresponding surface areas per gram of washcoat are also given in Table 2. They were calculated by assuming spherical particles and a theoretical density of 7.15 g.cm -~for ceria. 2.2. S e l e c t i v e m e a s u r e m e n t o f the c e r i a s u r f a c e a r e a by TPR

Catalysts before stabilization

Figure 1 shows the TPR profiles of the catalyst as received and after laboratory aging at 1173, 1273 and 1373 K. The main features correspond to those observed on model catalysts [ 10]. In the fresh state, there is a well-resolved peak at 570 K, ascribable to the reduction of the oxidized precious metals and the ceria surface. After aging, the curve becomes flattened with much lower H2 uptakes. When the aging temperature is increased from 1173 to 1373 K, the intensity of the first peak is reduced to nearly zero, whilst two other curve inflexions or small peaks become more distinct at around 800 and 970-1000 K.

A 570 K

--*--Fresh

~LAl173

" m " LA1273

-*"LA1373

v

O

4)

e ..r

300

400

,500

600

700

800

900

I000

1100

Tern perature (K) Figure 1. H2 TPR of the catalysts before stabilization. Heating rate 20 9 K min ~

Although the separation between surface and bulk reduction is not always straightforward, the hydrogen consumption quantities were determined for temperatures lower than 900-950 K. They are given in Table 3. From them, it is possible to calculate the ceria surface areas of each catalyst, provided that some hypothesis are done on the mean oxidation state of the precious metal before starting the TPR experiment. In this study, the calculations were made with two different hypothesis" hypothesis 1) the metals are under the Rh 3§ and Pt 2§ states, hypothesis which was found valid for the fresh systems [9] and hypothesis 2)

141 rhodium is in a 3+ state whereas platinum is in a metallic state with a very small O/Pt ratio and set to 0 in the present study [10]. This second hypothesis has been supported by a separate TPR measurement performed on a Pt/A12Oa catalyst hydrothermally aged at 1323 K. Compared to the alumina support aged in the same conditions, no additional hydrogen consumption was detected during the TPR. The experimental hydrogen uptakes for the low temperature peaks and the calculated ceria surface areas are given in Table 3 with the BET areas. Table 3 Ce~a surface area s, per gram of washc0at, measured by the TPR method. Catalyst

SBET

He exp. --~900 K }imp1 g.1

Sc~o~ (hypo 1)a m~l

Sc~o9 (hypo 2)b m2g.1

Reduction extent r

(%)

.....

Fresh LAl173 LA1273 LA1373

167 101 77 54

309 163 75 55

69 30 7.5 2.4

75.5 38 15 10

128 121 44 41

Fresh St. LA1173 St. LA1273 St. LA1373 St.

139 96 72 41

251 124 74 49

54 20 7.2 1

61 28 15 8.5

126 74 48 29

ahypothesis 1 "O/Pt = 1 and OfRh = 1.5 bhypothesis 2 O/Pt 9 = 0 and OfRh = 1.5 csee text The difference between hypothesis 1 and 2 is about 6-8 m2g"1 and corresponds to the hydrogen quantity needed to reduce Pt 2+ into Pt ~ For the fresh catalyst, hypothesis 1 is the most appropriate and leads to a ceria surface area of 69 m2g~. For the aged catalysts, the relative uncertainty between the values of SCeO2 obtained with the two hypothesis is acceptable for LAl173, but not after aging at 1273 and 1373 K. As said above, hypothesis 2 seems the most reasonable, and the calculated values of the ceria surface areas are 15 and 10 m 2 ~ 1 after aging at 1273 and 1373 K. They are effectively in better agreement with the XRD results, than with hypothesis 1. To have more informations on the evolution of the support after the aging treatments, it is possible to follow an other parameter which is the reduction percentage of the catalyst at the end of the TPR. It corresponds to the ratio between the experimental hydrogen consumption during the whole TPR, including the 45 mill step at 1073 K, and the maximum theoretical H2 consumption necessary for the reductions (CeO2 ") CegO~; PtO ") Pt~ Rh20~ -') Rh o ). For fresh and LAl173, this percentage is higher than 100% (128 and 121%) and close to that of the washcoat without precious metals (131%). It

142

decreases to 44 and 41%, for LA1273 and LA1373 respectively. We can deduce that "i) in the initial solid and after mild aging (1173 K), some reducible species other than precious metals and ceria are present in the system, and ii) the main modification of the catalyst during the aging treatments occurs between 1173 and 1273 K. In this respect, LA1173 can be considered as a weakly aged catalyst. To explain the reduction extent higher than 100%, several hypothesis were considered but were not verified. The reduction of the iron oxide present in the cordierite was not observed during a separate TPR. The presence of barium sulfate was also evidenced by XPS. In the TPR conditions, BaSO4 begins to be reduced at about 1000 K. However, the hypothesis of its reduction in the catalysts was not kept, since the reduction percentage of a platinum catalyst supported on a ceria-alumina modified with BaSOa was a little lower than that performed in absence of barium sulfate. The assumption of the hydrogenation of some carbonates species, as surface lanthanum carbonates, was also rejected, since, as evidenced by mass-spectrometry, there is no relationship between the excess reduction percentage and the formation in the gas phase at very low concentration of CH4, or even CO and CO9. The question is still under study.

Catalysts after stabilization Figure 2 exhibits the TPR profiles of the previous catalysts, after 1 h at 823 K under the reactants, and Table 3 presents the results. For the fresh catalyst after stabilization, the initial low temperature peak is split into two peaks of lower intensity. For the aged catalysts, the stabilization leads also to profiles with a lower intensity during the whole TPR. Accordingly, the hydrogen uptakes are lower than those of the initial systems. If one supposes that the metals after reoxidation at 673 K have the same mean oxidation state before and after stabilization, which means no change in their size and their state during the stabilization, the calculated ceria surface areas are lower after stabilization (Table 3). Thus, the stabilization results in an additional ceria surface loss. , =

-e-Fresh

.

St,

.

.

.

.

.

.

- - ~ , - L A l 1 7 3 St,

..e.- LA1273 St,

--e-- LA1373 St,

~

515 K

01

e

:

- - - - -

.

.

.

.

.

.

"o

2: 300

I 400

I 500

I 600

I 700

Temperature

: 800

: 900

1000 1100

{K)

Figure 2. H2 TPR of the catalysts after stabilization. Heating rate 20 K min -1. (hydrogen uptake scale about two times higher than in Fig. 1)

143 2.3. Study of the catalyst after 200 h a g i n g on an engine-bench. One of the difficulties to study this catalyst is the possible influence of the poisons deposited on the active phases during the test and originated from the gasoline or motor oil components such as Si, Ca, P, Zn, S ... [11,12]. In particular, the TPR study may become totally erroneous if additional reducible compounds are present. To take into account this influence and to evidence an eventual aging gradient along the axis of the monolith, three samples were selected after the test, at the inlet, in the middle and at the outlet of the monolith. The analysis and distribution of the poisons were done by SEM coupled with an EDX analysis. A macroporous layer of pollutants was evidenced on the surface. The analysis was done on the elements of the support (AI, Ba and Ce) as well as the poisons usually found after such a treatment, i.e. P, Ca and Zn [11]. Sulfur was searched for but was not detected. In the front side of the converter, the poisons were the only elements detected, with almost 50% Zn, 40% P and 10% Ca. It means that the poison layer is thicker than that analysed by EDX, i.e. about l~m. The Zn concentration decreased quickly in a few millimeters axially and then was not detected ( 3 CO2 (~) + 3.5 H20 (g) FAST

NO(g)

. , . . . ~ NO*

L

- 7 ~ .....

~ ~ - - - - - " ~ + O* - O*~.....~~ NO2(g)

...........................

C3H8(g)

f "

§ *. . ~ NO2*

-

--.\...

~> CxHy \~

...........................................

> N2(g), N20(g), CO2(g), H20(g)

...........

A!203 support

Scheme 2. Proposed mechanism for the C3Hs-NO-O2 reaction. Reactions above the dotted line occur on the Pt surface, while reactions below occur on the A1203 support.

The fact that the order of C3H8 oxidation in C3Hs is greater than unity is consistent with dissociative chemisorption of C3Hs involving the breaking of a C-H bond being the rate determining step, as is generally accepted [3]. Increasing the C3H8 concentration increases the rate at which adsorbed oxygen (the dominant species) is removed from the Pt surface, resulting in an increase in the number of vacant sites at which C3H8 can adsorb. This in turn results in an increase in the rate of C3Hs oxidation in addition to that due to the gas phase concentration of C3Hs resulting in the order in C3Hs being greater than unity. Adsorption of NO and NO2 on the Pt surface results in blockage of reaction sites and hence inhibition of C3Hs oxidation. Oxidation of NO to NO2 is observed over a wide temperature range (Fig. 7) and (unlike the C3H6-NO-O2 reaction) occurs in the presence of the reductant. This is consistent with a high coverage of oxygen on the Pt surface. The conversion of NO to NO2 was independent of contact time (Fig. 8) suggesting that the rate of NO oxidation to NO2 is so fast that a pseudo equilibrium was established between NO and NO2 even at the shortest contact time used. For this reason the formation of NO2 is expressed as a conversion rather than as a TOF. However, NO and NO2 are not in thermodynamic equilibrium since the conversion of NO to NO2 (about 30%) is much less than that predicted for equilibrium under these conditions (79%). In addition, the fall in conversion to NO2 with increasing C3H8 concentration (Fig. 9) is not predicted by thermodynamics. The oxidation of NO to NO2 and of dissociation of NO2 to NO was modelled using eqns 16 and 17.

206 100-, 0~80 -

o~60 e-o

=60 0

"F940

~40

(9 > e--

0

,k

,

,

&

~

0o20

o 20 150

250

350

Temperature/ ~

450

0.0

550

Fig. 7. The effect of varying temperature on the C3Hs-NO-O2 reaction. Lines are fit to kinetic model. (0 C3Hs, 9 NO, = NOx, O NO to N2, I"1 NO to N20, A NO to NO2) 4

80

3

60~

o F ~,~ 0

i ,,-:-: 1000

~:;

,:,~ o

2000

Carte Concentration / ppm

3000

Fig. 9. The effect of varying C3Hs concentration on the C3Hs-NO-O2 reaction at 310~ Lines are fit to kinetic model. Key as for Fig. 7. Feed: 1000 ppm NO and 5% 02. 4

80

3

60~..,

2

4o i

o 1

20 8

0

0 0

2

4

6

0 2 Concentration / %

8

10

Fig. 11. The effect of varying 02 concentration on the C3Hs-NO-O2 reaction at 310~ Lines are fit to kinetic model. Key as for Fig. 7. Feed: 1000 ppm NO and 1000 ppm C3Hs.

0.5 1.0 1.5 10~ w/f (g ~ n crn~)

2.0

Fig. 8. The effect of varying the reciprocal space velocity, w/f, on the C3Hs-NO-O2 reaction at 310~ Key as for Fig. 7. Feed: 1000 ppm C3Hs, 1000 ppm C3Hs and 5% 02. 4

80

%3

o

60..

0

500

1000

1500

NO Concentration / ppm

o

2000

Fig. 10. The effect of varying NO concentration on the CaHs-NO-O2 reaction at 310~ Lines are fit to kinetic model. Key as for Fig. 7. Feed: 1000 ppm C3Hs and 5% O2.

207 Attempts were made to fit the data for the rate of NOx reduction to kinetic models based on the various reaction mechanisms proposed in the literature, viz.: (i) the oxidation of NO to NO2 which then reacts with the hydrocarbon [9, 10, 11], (ii) the formation of an oxidised hydrocarbon intermediate [12,13]; (iii) reduction of the metal surface followed by NO dissociation on the reduced surface [6], possibly with the NO dissociation being assisted by other adsorbed species [1]; (iv) the formation of an isocyanate surface species as an intermediate [14]. Combinations of these mechanisms have also been suggested, such as the oxidation of NO to NO2 which then reacts with an oxidised hydrocarbon intermediate [ 15,16]. However, none of these models satisfactorily fitted the data [2]. The only correlation that was found with the rate of deNOx was with the NO2 coverage, i.e. rt~o,, oc 0t~02. This can be interpreted in terms of a mechanism in which the rate determining step is the spill-over of adsorbed NO2 onto the A1203 support. This NO2 then reacts with C3Hs-dedved species deposited on the support, possibly located close to or at the metal-support interface, to give N2 and N20. This mechanism suggests that the nature of the support should be important, and indeed, while Pt/Al203 shows deNOx activity with a CaHs-NO-O2 feed, little [8, 17] or no [ 18] deNOx activity is observed with Pt/SiO2 with the same feed. Recently, Hamada and coworkers have reported that physical mixtures of A1203 and Pt/SiO2 are active for deNOx with a C3Hs-NO-O2 feed [18]. This is consistent with the mechanism suggested above, although in this case the reaction presumably occurs by gas phase transfer of NO2, produced by oxidation of NO on the Pt surface, to the surface of the A1203, where it reacts with Calls derived species. Since Calls oxidation (by O2) seems to occur on the metal, while the deNOx reaction occurs on the support and/or at the metal-support interface the kinetics of these two reactions are very different (see above). The final expression used to fit the rate of deNOx was:

knox 0No2

rNO~= 1 + B e c~RT

(7)

where B and C are constants. The dominator of this expression allows for the fall in concentration of C3Hs-derived species on the support with increasing temperature (presumably) as a result of reaction with 02. The concentration of these species appears to be independent of reactant concentration. Without this term the model did not predict the maximum in the NOx conversion.

4. CONCLUSIONS The kinetics of NO reduction by C3H6 and by C3Hs over Pt/Al203 under lean-burn conditions have been investigated and kinetic models which satisfactorily fit the data have been developed. The state of the Pt surface depends on the relative activities of the reductant and 02. Thus, C3H6 is more reactive than O2 and hence the Pt surface is predominantly covered with C3H6-derived species, while Cans is less reactive and the Pt surface is mainly covered with atomic oxygen. This difference determines which reaction pathway for NO reduction is possible. Thus, NO reduction via NO dissociation on the Pt surface (as appears to occur with C3H6) is favoured by the negligible oxygen coverage obtained in the presence of C3I-I6, but is presumably inhibited by the high oxygen coverage in the presence of C3H8. Conversely, NO reduction via reaction of spilt-over NO2 with carbonaceous species on the A1203 (as seems to

208 occur with C3Hs) requires the high oxygen coverage obtained with C3Hs to facilitate the oxidation of NO to NO2, but cannot occur with C3H6 since NO2 formation is only possible in the absence of C3H6. The mechanism of NOx reduction needs to be known when considering how the catalyst could be improved. If NO reduction occurs on the A1203 support then the deNOx activity may be enhanced by modifying the support to better facilitate the reaction between NO2 and hydrocarbon-derived species, perhaps by adding a basic component to trap NO2 or to aid hydrocarbon activation on the support. Lowering the temperature of hydrocarbon activation by the support so that it coincides with the maximum in NO2 production may also be beneficial. Alternatively, if NOx reduction occurs via NO dissociation on the Pt, then deNOx activity can be improved by modifying the Pt to increase the rate of NO dissociation relative to 02 adsorption. This does not appear to be possible by changing the Pt dispersion or precursor (see above), but may be possible by adding a suitable promoter.

ACKNOWLEDGEMENT We are grateful to the EPSRC for financial support for this work through grant GR/KO1452. APPENDIX 1: RATE EQUATIONS FOR THE C3H6-NO-O2 REACTION

In this appendix, expressions for the rates of reaction of the C3H6-NO-O2 reaction are derived from the mechanism discussed in section 3.1 and the method of determining the parameters outlined. Applying the stationary state approximation to 0~ (using eqns 4-6) gives, 0 = dON = kNKNO CNO0V2 - kmo KNO CNO0V ON- 2kN20N2 dt

(8)

This is a quadratic in 0r~/0v the solution of which is: .

+

ON 0V

.

.

.

+

.

.

(9)

4kN2

The number of surface sites is assumed to be constant, i.e., 1 = 0v +0NO + ON + 0c

(10)

Note that in the presence of C3H6 the coverage of oxygen is believed to be negligible (see above) and hence does not appear in eqn 10. Substituting in eqns 2 and 3 gives, 0v =

A co2

1+ KNocr~o + 0 ~ / 0 v

for cc3n6 > 0

(11)

The rate of the deNOx reaction is the rate at which NO is converted into N2 and N20 (eqns 5 and 6), i.e., rNox = 2kN2 0N2 + 2kN2oONONo

(12)

209 Substituting in eqns 3 and 11 gives: rNox =

2 AE(0N / 0 v ) Co2 2 (kN( 0 N /0 v ) + kmoK No CNo) (1 + KNo CNo + ON / Ov) 2

forcc3m>0

(13)

The term in ONhas been left as 0N/0~ rather than substituting in eqn 9 for simplicity. Since adsorbed oxygen does not desorb and 9 0 atoms are required to totally combust one C3H6 molecule, the rate of C3H6 oxidation is given by (using eqns 1 and 4): rc3u6 = (2ko~ Co20v + kNKNo CNo0v2)/9

(14)

Substituting in eqn 11 gives: 1

2 2 ko~ A Co2

+

.kN . . .KN . . . .~ A 2 CN~ CO2

The oxidation of NO to NO2 and the dissociation NO2 to NO in the absence of Call6 can be represented as: NO(g) + O* --'>NO2*

kNoE,fCNO00

(16)

NO2* --> NO(g) + O*

kNo2,b 0NO2

(17)

Under these conditions the conservation of sites is: 1 = 0v + 0NO + 0NO2 + 00

(18)

Thus, the net rate of NO oxidation to NO2 is given by (from eqns l, 3, 16-18): 2 kox k NO2,fCO2CNO FNO2 kNo2,f CNo (1 + KNo CNO + KNO2 CNO2) + 2kox CO2 =

+ kNo2,b KNO2 CNO2

for CC3H6= 0 (19)

The effect of temperature on the rate constants and adsorption coefficients was assumed to be given by the Arrhenius equation and Van't Hoff isochore respectively, i.e., In (k-k~l)= - ~E'(~22 - ~-]]

(20)

lnCK-~-~)-

(21)

AH'~(-~2 ~ ) R --~-

where k~ & k2 and KI & K2 are rate constants and adsorption coefficients at temperatures T, & T2, T~ and T2 are thermodynamic temperatures, R is the molar gas constant, E, is the activation energy for the appropriate reaction step and AH~ is the standard enthalpy of adsorption of the appropriate molecule. The reactor was assumed to exhibit plug flow and transport limitations were assumed to be negligible. For experiments at constant temperature, the reactor was assumed to behave differentially, while in experiments in which the temperature was varied the effect of nondifferential conditions was included by allowing for concentration gradients along the catalyst

210 bed as described elsewhere [2]. The parameters given in Table 1 were determined using the method described previously [2]. Table 1 Parameters . . . obtained . . fr0m . . fittin . 8 the CaH6-NO-OE reaction model to experimental data. Parameter

Value at 240~

Ea/kJ tool"1(a)

8.26

x

104 ppm 1 s]

2 kox

1.46

x

10"1 0~ "l

kN2oKNo

8.92 x 10"4 ppm 1 s-I

14to)

kN2

7.33 x 10"1

14tO

KNO = KNO2to)

5.53 x 10-3 ppm "1

A (d)

7.28 • 10"2 %-1

0

kNo2,f

5.26 x 105 ppm q S"1

0

kNo2,b

1.88 x 10"l s"l

0

kNKNo

S"l

AH~

,.

kJ mol q(b)

14(c) 107

-96.7

(a)" Activation energy for the appropriate reaction step; (b) Standard enthalpy of adsorption of the appropriate species; (c)" Set to be equal to reduce the number of parameters; (d): Set equal to kox.

LIST OF SYMBOLS A B C CNO

[%-1] [-] [J mol "1] [ppm] [ppm]

co2

[%]

kN kN2 kN20 KNO KNO2 kNo2,f kNo2.b kNOx kox R

[S"l] IS "l] [S"l] [ppm "1] [ppm "1] [ppm "I s"l] [S"l] [S"l] [%-1 s-l] [J mol q K "1]

rc3.6

[s"]

rNo2 rNox t T

[s"1] [S"] [s] [K]

0i

[-]

CC3H6

Constant used in expression for 0c, defined by eqn 2. Constant in eqn 7, for rate of NOx reduction with C3H8. Constant in eqn 7, for rate of NOx reduction with Calls. Gas phase concentration of C3H6. Gas phase concentration of NO. Gas phase concentration of 02. Rate constant for NO dissociation, defined by eqn 4. Rate constant for N2 formation, defined by eqn 5. Rate constant for N20 formation, defined by eqn 6. Adsorption coefficient for NO adsorption, defined by eqn 3. Adsorption coefficient for NO2 adsorption. Rate constant for oxidation of NO to NO2, defined by eqn 16. Rate constant for dissociation of NO2 to NO, defined by eqn 17. Rate constant for NOx reduction by C3Hs, defined by eqn 7. Rate constant for oxygen adsorption, defined by eqn 1. Molar gas constant. TOF for C3H6 combustion. Net TOF for oxidation of NO to NO2. TOF for NOx reduction. Time. Thermodynamic temperature. Fractional coverage of species i on the Pt surface.

211

0~ 0v

[-] [-]

Fractional coverage of carbonaceous species on the Pt surface. Fractional coverage of vacant sites on the Pt surface.

REFERENCES

1. 2. 3. 4. 5. 6. 7.

8. 9. 10. 11. 12. 13. 14. 15. 16.

17. 18.

R. Burch and T.C. Watling, Catal. Lett., 37 (1996) 51. R. Burch and T.C. Watling, J. Catal., 169 (1997) 45. R. Burch and M.J. Hayes, J. Mol. Catal. A, 100 (1995) 13. R.A. Van Santen and J.W. Niemantsverdriet, Chemical Kinetics and Catalysis, Plenum Publishing Co., New York, 1995. T. Engel and G. Ertl, Adv. Catal., 28 (1979) 1. R. Butch, P.J. MiUington and A.P. Walker, Appl. Catal. B: Env., 4 (1994) 65. G.P. Ansell, S.E. Golunski, J.W. Hayes, A.P. Walker, R. Butch and P.J. Millington, Stud. Surf. Sci. Catal. 96 (Catalyst and Automotive Pollution Control III), A. Frennet and J-M. Bastin (eds.), Elsevier, Amsterdam, 1995, p. 577. P.J. Millington, PhD Thesis, University of Reading, UK, 1995, oh. 5. S. Naito and M. Tanimoto, Chem. Lett., (1993) 1935. T. Tanaka, T. Okuhara and M. Misono, Appl. Catal. B: Env., 4 (1994) L 1. A. Obuchi, A. Ogata, H. Takahashi, J. Oi, G.R. Bamwenda and K. Mizuno, Catal. Today, 29 (1996) 103. A. Obuchi, A. Ohi, M. Nakamura, A. Ogata, K. Mizuno and H. Ohuchi, Appl. Catal. B: Env., 2 (1993) 71. M. Sasaki, H. Hamada, Y. Kintaichi and T. Ito, Catal. Lett., 15 (1992) 297. G.R. Bamwenda, A. Obuchi, A. Ogata and K. Mizuno, Chem. Lett., (1994) 2109. G. Zhang, T. Yamaguchi, H. Kawakami and T. Suzuki, Appl. Catal. B: Env., 1 (1992) L15. B.H. Engler, J. Leyrer, E.S. Fox and K. Ostgathe, Stud. Surf. Sci. Catal. 96 (Catalyst and Automotive Pollution Control III), A. Frennet and J-M. Bastin (eds.), Elsevier, Amsterdam, 1995, p. 529. R. Burch, and T.C. Watling, Catal. Lett., 43 (1997) 19. M. Inaba, Y. Kintaichi and H. Hamada, Catal. Lett., 36 (1996) 223.


CATALYSIS AND AUTOMOTIVE POLLUTIONCONTROLIV Studies in Surface Science and Catalysis,Vol. 116 N. Kruse, A. Frennetand J.-MBastin (Eds.) 91998 Elsevier Science B.V. All rights reserved.

213

N 2 0 and NO2 formation during N O reduction on precious metal catalysts P. Bourges, S. Lunati and G. Mabilon Institut Frangais du P6trole, 92506 Rueil-Malmaison C6dex, France ABSTRACT The comparison between the NO x reduction activity of Pt, Pd, Ir, Ru, Rh on alumina and Cu on ZSM-5 shows that only platinum and copper present high activity. Copper practically does not give any N20, but platinum is very selective for N20 formation. Moreover, platinum is the only active catalyst where NO 2 formation is important. But NO reduction doesn't produce simultaneously N 2, N20 and NO 2. If NO 2 is an intermediate in NO reduction mechanism, it could participate only as adsorbed species. On platinum catalysts, the kinetics of HC oxidation, NO reduction and NO oxidation are strongly dependant on the hydrocarbon nature. Those mechanisms occur at lower temperature with long chain alkanes than with olefins and these alkanes lead to a higher N20 selectivity than unsaturated molecules. 1. INTRODUCTION Diesel and lean gasoline engines are very attractive because of their low fuel consumption. But their development could be limited by the difficulty to comply with future NO x emission regulations: new depollution techniques have to be developed. The catalytic reduction of NO x by hydrocarbons has been studied over a large number of transition metal catalysts. Cu-ZSM-5 allows high conversion rates and good selectivities to N 2 but only above 350~ This is too high for diesel application where exhaust gas temperature varies generally between 100 and 450~ Precious metal catalysts show a higher potential for low temperature reduction. Platinum is especially attractive because it starts NO reduction at 150 - 200~ A large number of reaction mechanisms have already been proposed. They include NO decomposition prior to hydrocarbon oxidation by adsorbed O species [1], NO insertion in hydrocarbon to form N-containing organic species that are further decomposed by oxygen [26], NO oxidation to NO 2 that could be reduced by hydrocarbons to N 2 but also N20 or NO [7-9]. As precious metals can catalyze NO oxidation, we decided to investigate the influence of NO 2 formation in the global NO reduction process and to correlate N20 formation to the characteristics of the catalysts and of the reacting medium.

214 2. EXPERIMENTAL Catalysts were prepared by impregnation of alumina coated cordierite monoliths. Cylindrical catalyst samples (O = 30 mm, L = 76 mm) were placed in a down-flow reactor. The reaction mixture contained 3 to 9 components among NO, 0 2, NO 2, HC, CO, CO 2, H20, N2, SO2" in most cases NO concentration was 600 ppm with 3000 ppmC HC with a space velocity of 50 000 h-1. On-line analysis was performed by chemiluminescence for NO and NO 2, IR for N20 and CO, FID for HC. Catalysts were activated during 2 hours at 600~ under reaction conditions. Tests were performed under temperature ramp at 5~ between 150 and 500~ 3. RESULTS AND DISCUSSION 3.1. NO oxidation by 0 2

3.1.1. Influence of the catalyst nature Oxidation of NO by 02 in the presence of water, CO 2 and SO2 strongly depends on the nature of the catalyst (Fig. 1). With rhodium or iridium and more particularly with palladium or copper NO oxidation is very limited below 400~ The only catalyst which is very active even at low temperature is platinum.

-- Cu

,~ 60 t"q

50

---o-- Ru

o

-,--,

40

~Rh

O

30

--o- Pd I

Z

9

.-, 9

I-.i

~ 2o

~Ir

I

o

~ 10 9 2; 0 100

200

300

400

500

Temperature (~ Figure 1. Comparison of copper and noble metal catalysts for NO oxidation by 0 2 (GHSV=50 000h ~, 600ppm NO, 5% 0 2, 10% CO 2, 7% H20, 20vpm SO2). 3.1.2. NO oxidation over Pt/A120 3

3.1.2.1. Influence of the space velocity Below 300~ NO oxidation over platinum is very sensitive to the space velocity. Above that temperature the conversion is limited by the thermodynamic equilibrium (Fig. 2).

213 lO0

r

0 Z

80

=

60

~'

40

oo 0 2:

20

s

- - a - - GHSV 10 000

~

~

~r

__.J"~

F

--.--GHSV 50000

n I=t"

I ~

.,,.,,,"

rrrr ii

I

0 200

11

300

400

500

Temperature (~ Figure 2. Comparison of NO oxidation thermodynamics limit and NO oxidation rate over Pt/AI20 3 at different space velocity (GHSV = 10 000 or 50 000 h l ; 600ppm NO, 20vpm SO 2, 5% 0 2, 10% CO 2, 7% H20 ).

3.1.2.2. In the presence of reductants On Pt/A120 3, when CO is introduced at low concentration in a gas feed containing NO and 0 2, NO is not reduced and NO 2 is detected in the gas phase at slightly higher temperatures than in the absence of CO (Fig. 3).

1oo

~ i - ( f [

[

9 9[ [

9[ [ [ [ [ [ [

9 9 9[

9 9[ [

9[ - [ - l ' ~ [ - [ - I ( - [ - [ - [ - l " [ l l / ~

]

=

80

NO (no CO) ---o--CO

~,,,,4 9

60

_-7

0

9

40

N

2o

0 Z

0

r..)

No ( c o )........

!

100

.

i

!

i

!

a

i

150

200

250

300

350

400

450

500

Temperature (~ Figure 3. NO and CO oxidation rates in NO 2 and CO 2 over Pt/A120 3. Tests without CO or with 500 ppm CO (GHSV = 50000 h l ; 0 or 500 ppm CO, 600ppm NO, 5% 0 2, 10% CO2, 7% H20, 20vpm SO2). The introduction of hydrocarbons in the feed has a larger effect on NO oxidation and reduction. In the presence of n-decane (Fig. 4), NO is not oxidised before 230~ while in the presence of ethylene (Fig. 4), NO 2 is detected only above 290~ As in the presence of CO, NO oxidation is delayed until most of the reductant is eliminated. But hydrocarbons are oxidised at higher temperatures than CO and the nature of the hydrocarbon strongly affects its own oxidation: n-decane is oxidised around 220~ while ethylene is oxidised around 280~

216

r.,/3

Q

lOO

-,~-_,_,_,'_,,','_,'-mrrtr~ .. i~()

80 60

[

"~

20

9 2:

0

i ~--~'-,

~ ~,40

100

200

300

NO --o- HC 1 ~ NO --,-- HE

400

(no HC) (decane) (decane) (ethylene) (ethylene)

500

600

Temperature (~ Figure 4. NO and HC oxidation rates in NO 2 and CO 2 over Pt/A120 3. Tests without HC or with 6000 ppmC of decane or ethylene (GHSV = 50000 h 1 ; 0 or 6000 ppmC HC 500 ppm CO, 600ppm NO, 20vpm SO2, 5% 0 2, 10% CO 2, 7% H20 ). As in the absence of reductant, NO2 concentration goes through a maximum when temperature increases, but this maximum is clearly below that observed without reductant. This can be explained if we consider that the thermodynamic equilibrium for NO oxidation should more appropriately be expressed with the reactor outlet temperature than to the reactor inlet temperature: under adiabatic conditions the temperature increase is about 120~ for the combustion of 6000 ppmC hydrocarbon. This temperature shift is well suited to explain the NO oxidation curve in the presence of ethylene. It is less adapted in the presence of n-decane probably because decane oxidation is diffusion limited and reaches total conversion only at high temperature: NO oxidation is not at equilibrium. The temperature shift for NO oxidation in the presence of reductants could have several origins. NO 2 could not be formed on the catalyst surface until the reductant concentration in the adsorbed state is very low or NO 2 could be formed but rapidly reduced by hydrocarbon or carbon monoxide. NO 2 reduction could occur on the catalyst surface between adsorbed species or in gas phase by homogeneous reaction with the reductant. NO 2 reduction in gas phase has been studied in the absence of catalyst in a small test device equipped with on-line mass spectrometer (Fig. 5). The reduction occurs at low temperature but at very low space velocity. The maximum conversion rate is about 50%. At higher temperature ethylene is oxidised by 0 2 which accounts for the decrease of NO 2 reduction rate. As NO 2 reduction rate is 50 % at maximum at GHSV of 6 000 h "l, it must be very lower at 50 000 h "1 and could not significantly contribute to the elimination of NO 2 in the gas phase. Although NO 2 can be formed at low temperature on platinum, it is not detected in the presence of a reductant. The NO oxidation sites can be blocked when their covering by reductant is high or NO 2 adsorbed species are reduced in presence of carbon monoxide and hydrocarbons before their desorption. Therefore the absence of NO 2 in the gas phase does not necessarily means that it is not present as adsorbed species.

217 600 o

3500

500

.v~,4

~

400 El

CO 2

..................... i

NO 2

300

3000

E~

25oo

&

2000

O

15oo

"~

1000

o

500

r..)9

9. Z

200

0 Z

lOO

/

/'

~

NO

/

- r - - w ,,~

0

100

i

I

200

i

300

400

u

500

0

600

Temperature (~ Figure 5. Formation of NO and CO 2 during ethylene oxidation by NO 2 and 0 2 in absence of catalyst (GHSV = 6 000 h l , 600ppm NO 2, 3000ppmC ethylene, 1% 02). 3.2. NO reduction by hydrocarbons 3.2.1. Influence of the nature of the metal The comparison between the NO x reduction activity of Pt, Pd, Ir, Ru, Rh on alumina and Cu on ZSM-5 shows that only platinum and copper present high activity (Fig. 6). Platinum allows NO x reduction by n-decane in the temperature range 180 to 300~ with a maximum conversion of 70% near 220~ Copper allows 70% NO x conversion above 380~ In the presence of water, the percentage of reduction of NO x stays inferior to 15% on palladium, iridium, ruthenium and rhodium. These results are in accordance with those proposed by Obuchi et al [ 10] who tested the activity of precious metal catalysts with Diesel exhaust. lO0 ,--.,

1

80

=

60

~

40.,

---o-- Ru

!

--w- Rh --o-Pd I

9 Z

20 u

100

150

200

250

v

300

r

350

400

450

500

Temperature (~ Figure 6. NO reduction by n-decane on different metals (GHSV = 50 000 h "1 ; 6000 ppmC n-decane, 600ppm NO, 500 ppm CO, 20vpm SO2, 5% 02, 10% CO2, 7% H20 ).

218 The comparison (Fig. 1 and 6) between NO oxidation by oxygen in the absence of hydrocarbon and NO reduction in the presence of C 10H22 and 0 2 shows that the oxidation and the reduction of NO are comparable and occur in the same temperature range with platinum or ruthenium catalysts, the conversion rates being very low with ruthenium. With rhodium or iridium and more particularly with palladium or copper NO oxidation is very limited at the temperature where NO reduction occurs. The selectivity to N 2 0 is sensitive to the nature of the catalyst (Table 1). Copper practically does not give any N20, but platinum is very selective for N20 formation in the presence of decane [3,11 ]. Table 1. N 2 selectivity and NO x maximum conversion for NO reduction by decane on precious metal over alumina and Cu/ZSM-5 catalysts.

copper ruthenium rhodium palladium iridium platinum

N20 selectivity at NO x maximum conversion in % 3 2 20 26 3 75

NO x maximum conversion in % 76 6 23 17 22 64

Platinum is therefore the only active catalyst where NO 2 formation is important and could compete or favour NO reduction. Moreover platinum is the more selective for the N20 formation. Therefore, we decided to study especially the formation of NO 2 and N20 on platinum catalysts during NO reduction. 3.3.2. NO reduction on alumina supported platinum catalysts

3.2.2.1. Influence of temperature

In the presence of n-decane in the feed, NO is reduced to N 2 and N20 at low temperature. The reduction rate goes through a maximum at 220~ and then decreases slowly to become nil near 320~ Above 230~ NO oxidation and NO reduction occur simultaneously. For example, at 260~ 31% of NO is reduced (9% in N 2 and 21% in N20 ) and 14 % of NO is oxidise in NO 2 (Fig. 7). In order to discriminate if oxidation and reduction of NO are simultaneous or consecutive we studied the influence of the contact time. This was achieved by cutting the monolith at different lengths: 19.38 to 76 mm. The dynamics of the system was kept constant by adding an inactive catalyst to maintain the same reactor length. Results obtained at 220~ and 260~ are presented on figure 8. They are expressed as a function of the relative length of active catalyst.

219 50 -"-N2

40-

/'~

----N20

30~o t,..,i

o 9 2; 0

20

/i

10-

|I

I

_,,~ ~-

0 100

~ ,,r,-',.,,=,,,,,,,.",,,'"'"-"

,~/ -~,__.

200

,

300

Ham

400

500

Temperature (~ Figure 7. NO conversion in N 2, N20 and NO 2 in presence of n-decane on platinum catalyst (GHSV = 50 000 h l ; 6000 ppmC n-decane, 600ppm NO, 500 ppm CO, 5% 0 2, 10% CO 2, 7% H20, 20vpm SO2).

& o

~9 (1.) e,~

o:

700

7000

600

-6000

7OO 600

C

500

- 5o00

~

400

-4ooo

8 ~

300

r 3000

o1,,.~ 9

_.

L~ 100 ~ o o

50

lOOO ,

N20

0

N2

lOO

Relative lengths (%)

;

--t:~ NO2 ---o- HC

6000

& = 500

2000~_._NO

2oo

7000 r,.)

"~ o

r,.)

..~

c~

5000 . .

400

4000

300

3000

200

2000 r,j

lOO

1000

=

(1) O o

,~o 0

50

100

Relative lengths (%)

Figure 8. N2, N20 and NO 2 formation as a function of the relative length of active catalyst at 220 and 260~ (GHSV - 50 000 h -I ; 6000 ppmC n-decane, 600ppm NO, 500 ppm CO, 5% 02, 10% CO2, 7% H20, 20vpm SO2). When the temperature is such that HC oxidation is about 90%, N 2, N20 and NO 2 are observed at the outlet of the catalyst. But this results from an integral effect: indeed NO reduction to N 2 and N20 occurs at the monolith inlet while NO oxidation to NO 2 occurs at the monolith outlet. On platinum catalyst, NO reduction and NO oxidation do not occur in the same conditions. NO 2 is observed at the outlet of catalyst only when hydrocarbon concentration in the feed becomes low. But NO is no more reduced in these conditions. In the first part of this work we concluded that NO 2 reduction by hydrocarbon in the gas phase is negligible in our reaction conditions. Therefore, NO 2 is not desorbed in gas phase

220 during NO reduction. If NO 2 is an intermediate in NO reduction mechanism, it could participate only as adsorbed species.

3.2.2.2 Selectivity to N20 and N 2 On platinum catalysts, NO x reduction produces N 2 and N20. We studied the influence of the nature of the hydrocarbon on N 2 0 selectivity and N 2 yield in case of alkanes, alkenes and aromatics of various chain length (% N20 selectivity = 100 - % N 2 selectivity). The N 2 0 selectivity seems to be strongly dependant on the hydrocarbon nature [12]. On platinum catalyst, it varies from 34 to 80 % as a function of the hydrocarbon type (Table 2). Burch et al. [ 12] indicated that N 2 selectivity is 100% (0% of N20 ) in the absence of water if NO reduction is carried out with toluene. But in the presence of a complete mixture including water, CO 2 and SO2, similar N 2 0 selectivities are obtained with ethylene and toluene (34%). Table 2. Selectivity in N 2 0 and yield in N 2 as a function of the hydrocarbon nature. NOx maximum conversion % n-octane n-decane decaline dodecane ethylene propylene toluene xy,!ene

50 59 36 30 23 25 39 26

N20 selectivity N 2 yield at at NO x the NO x maximum maximum conversion conversion % % 80 73 75 65 37 40 38 65

10 16 9 10 15 15 24 9

Temperature at NO x maximum conversion ~ 220 217 230 260 305 311 295 288

HC halfconversion temperature ~ 221 218 222 290 302 320 281 285

N 2 0 selectivity is higher with alkanes than with unsaturated molecules except xylene. N 2 0 selectivity is no much dependent on the temperature for maximum NO x conversion. NO x reduction occurs at the beginning of HC oxidation. HC oxidation characteristics on platinum catalyst depends strongly on their adsorption strength [13,14]. This explains why the temperatures of HC oxidation and NO reduction vary similarly as a function of hydrocarbon nature. N 2 0 selectivity stays constant as a function of temperature during NO reduction by one hydrocarbon type. It seems to depend strongly on HC oxidation mechanism. Long chain alkanes lead to a higher N20 selectivity than unsaturated molecules even if NO reduction occur in the same temperature range. The choice of hydrocarbon nature determines N 2 and N20 selectivities. But the temperature range of NO conversion and the yield in nitrogen are function of the chain length (Table 2). On platinum catalysts, the best nitrogen yield are obtained with long chain alkanes at low reduction temperature and with unsaturated hydrocarbons like toluene and ethylene at higher reduction temperature.

221 at low reduction temperature and with unsaturated hydrocarbons like toluene and ethylene at higher reduction temperature. 3.2. 2. 3. Kinetics of NO reduction The influence of the reactant concentrations on the reaction rates has been studied at low conversions on platinum catalysts in order to determine the partial reaction orders and apparent activation energies both for NO reduction and HC oxidation. A negative order is obtained for NO, a positive order for 0 2 and either a negative order for HC if it is an olefin or a positive if it is an alkane [ 15] (Table 3). Table 3. Activation energies and partial reaction orders for NO reduction by ethylene or decane.

Ea (Kcal/mol) Partial order HC NO 02

ethylene HC oxidation NO reduction 34 26 - 0,9 - 0,9

- 0,7 - 0,3

n-decane HC oxidation NO reduction 43 37 0,8 - 1,0 1,8

1,0 - 1,0 1,6

Hydrocarbon oxidation is more actived than NO reduction. This explains that whatever the reductant, NO reduction rate increases more slowly with temperature than HC oxidation rate. NO shows an inhibiting effect on HC oxidation and NO reduction whatever the nature of the reductant. Ethylene is more strongly adsorbed than n-decane and shows inhibiting effects both on HC oxidation and NO reduction. Large ethylene concentrations cause a shift of NO reduction to high temperatures. Decane is more smoothly adsorbed and promotes NO reduction: the temperature for NO reduction decreases when decane concentration increases. Strongly adsorbed hydrocarbons induce an inhibiting effect on NO reduction and on HC oxidation. Their covering of metal surface is high so that the reaction of NO reduction, HC oxidation and NO oxidation are shifted to higher temperatures (Fig. 4, Table 2). Moreover, the N 2 selectivity is higher with olefins than with long chain alkanes. N20 and N 2 formation could depend on the covering of metal by hydrocarbon. Hydrocarbon adsorption strength may be an important parameter in the reduction selectivity. 4. CONCLUSION Under Diesel exhaust gas conditions, only platinum and copper supported catalysts allow high NO reduction activity. Copper practically does not give any N20, but platinum is very selective for N 2 0 formation. Unlike other transition metals, platinum is a good catalyst both for NO oxidation by oxygen and NO reduction by hydrocarbons. In the presence of hydrocarbons, NO 2 is observed at low HC concentration. But NO is not reduced under these conditions. If NO 2 is an intermediate in NO reduction mechanism, it could participate only as adsorbed species. On platinum catalysts, the kinetics of HC oxidation, NO reduction and NO oxidation are strongly dependant on the hydrocarbon adsorption strength. These mechanisms occur at higher temperature with olefins than with long chain alkanes.

222 N20 selectivity is higher with long chain alkanes than with unsaturated molecules. N20 and N 2 formation could depend on hydrocarbon nature and on the metal covering by hydrocarbon.

ACKNOWLEDGEMENTS Part of this work was carried out with the financial support of ECE (Brite Euram project BRE2-CT92-0192) REFERENCES

1. R. Burch, P.J. Millington and A.P. Walker, Applied Catalysis B, 4 (1994) 65. 2. N.W. Hayes, R.W. Joyner, E.S. Shpiro, Applied Catalysis B, 8 (1996) 343. 3. G.R. Bamwenda, A. Ogata, A. Obuchi, J. Oi, K. Mizuno, J. Skrzypek, Applied Catalysis B, 6 (1995) 311. 4. T. Beutel, B. Adelman, W.M.H. Sachtler, Catalysis Letters, 37 (1996) 125. 5. F. Poignant, J. Saussey, J.C. Lavalley, G. Mabilon, Catalysis Today, 29 (1996) 93. 6. C. Gaudin, D. Duprez, G. Mabilon, M. Prigent, Journal of Catalysis, 160 (1996) 10. 7. K.A. Bethke, C. Li, M.C. Kung, B. Yang, H.H. Kung, Catalysis Letters, 31 (1995) 287. 8. T. Tanaka, T. Okuhara, M. Misomo, Applied Catalysis B, 4 (1994) L 1. 9. M. Guyon, V. Le Chanut, P. Gilot, H. Kessler, G. Prado, Applied Catalysis B, 8 (1996) 183. 10. A. Obuchi, A. Ohi, M. Nakamura, A. Ogata, K. Mizumi, H. Ohuchi, Applied Catalysis B, 2 (1993) 71. 11. A. Obuchi, A. Ogata, H. Takahashi, J. Oi, G.R. Bamwenda, K. Mizuno, Catalysis Today, 29 (1996) 103. 12. R. Burch, D. Ottery, Applied Catalysis B, 9 (1996) L 19. 13. G. Mabilon, D. Durand, Ph. Courty, Catalysis and automotive pollution control III, Studies in surface science and catalysis, A. Frennet and J.-M. Bastin (Eds.), Elsevier, 96 (1995) 775. 14. Y.F. Yu Yao, Journal of Catalysis, 87 (1984) 152. 15. G. Mabilon, D. Durand, Catalysis Today, 17 (1993) 285.

223

Mechanistic investigation on the selective reduction of NO with propene in the presence of oxygen over supported platinum S. Eckhoff a, D. Hesse a, J.A.A. van den Tillaart b, j. Leyrer b, and E.S. Lox b

a Institute for Technical Chemistry, University of Hannover, 30167 Hannover, Germany b Automotive Catalysts Division, Degussa AG, P.O. Box 1345, 63403 Hanau, Germany

ABSTRACT The selective reduction of NO with propene in the presence of oxygen over a Pt/alumina catalyst has been investigated using TAP and model gas equipment. Experiments with different gas compositions (stoichiometric and overstoichiometric with respect to the complete oxidation of propene) were carried out at temperatures between 473 and 673 K. Additionally, the NO decomposition on reduced and oxidised Pt/alumina was studied. It is shown that N2 is generated due to NO dissociation and following recombination of Nadatoms. Associatively adsorbed NO needs to be present on the surface to form N20. 1. INTRODUCTION The emission of nitrogen oxides (NOx) from automotive and stationary sources causes serious environmental concern. Automotive exhaust gas aftertreatment systems are commonly based on precious metal catalysts (three way or diesel oxidation catalysts). One undesired effect during NOx reduction with these catalysts is the formation of N20, which is now considered to be an environmental pollutant also [ 1,2]. In this report the generation of N2 and N20 during NOx decomposition or reduction on Pt/alumina is investigated. It has long been established that platinum is active for the decomposition and reduction of NO [3-6] and that this reaction is inhibited by oxygen [7-10]. The formation of N2 is reported to take place over platinum in the presence of NO and 02 according to the following elementary reaction steps [3,4,7]: NO NO* 2 N*

+ +

* ~ * ~ --~

NO* N*

+

O*

(1) (2)

N2

+

2*

(3)

In these reaction equations a free surface site is represented by an asterisk, *. The formation of N20 is not well studied until now [7] but it is reported that N20 is only weakly adsorbed on platinum [11] and that platinum is not an active N20 decomposition catalyst [ 1,16]. Different mechanistic pathways for the formation of N20 over platinum in the presence of NO and 02 are considered in general:

224 2 N* N* NO* N*

+ O* + NO + NO* + NO*

~_~ ~~ ~

N20* N20* N20* N20*

+

2*

+ +

O* *

(4) (5) (6) (7)

It is shown that some of the reaction paths above can be excluded from the results reported in this study.

2. EXPERIMENTAL 2.1. Catalyst preparation and characterization A 1 % - w t platinum supported on alumina catalyst was used throughout this study. This catalyst was prepared by a proprietary incipient wetness method with tetraammine platinum (II) hydroxide, Pt(NHa)4(OH)2, as precursor. After drying in air at 393 K for 2 hours the catalyst was calcined in air for 3 hours at 623 K and reduced at 803 K in flowing hydrogen for 3 hours. The average platinum particle diameter, measured by both CO-chemisorption and TEM, amounted to 2 nm. The total exposed surface area of the platinum amounts to 1.4 mE/g. The BET surface area of the used A1203 was 92 m2/g. 2.2. TAP set-up Most of the experiments described in this study were performed in a so called TAP (Temporal Analysis of Products) apparatus. This apparatus consists essentially of a micro reactor, two high speed pulse valves, and a fast detecting mass spectrometer together with the necessary data acquisition and control systems. The pulse valves can generate up to 40 pulses per second of 1013 up to 1019 molecules with a pulse width of typically 1 ms. With the fast detecting mass spectrometer at the reactor outlet the signals of reactants, products, and intermediates can be monitored in time. A more detailed description is given by Gleaves and co-workers [12]. The TAP reactor was loaded by placing 200 mg of the catalyst granulated to 250-500 ~tm particles between two beds of a-alumina and two stainless steel meshes. In general the amount of molecules introduced per pulse was adjusted in the range of 0.5 - 1 % with respect to the total number of platinum surface atoms. All experiments were carried out with 15NO to differentiate between N2 and CO as well as between N20 and CO2. The NO and N2 responses were corrected to account for fragmentation of N/O in the mass spectrometer to NO and N2. Also the CO response was corrected to account for the CO fragment of CO2. Three different pulse techniques were used in this study: a) single pulse, b) multipulse experiments, where a series of pulses is introduced, and c) pump-probe experiments, where two different pulses are alternately introduced at a user-specified time interval At. All single pulse and pump-probe experiments consisted of 40 pulse cycles of 3 seconds duration. Directly before the measurement 5 initial precycles were given. The responses of the 40 cycles were averaged to improve the signal/noise ratio. Two different gas mixtures were used in the experiments (see Table 1). Argon was used as the internal standard.

225 Table 1. Composition of gas mixtures used in TAP experiments (molecular amounts relative to C3H6) and in model gas experiments (concentrations). mixture TAP experiments Model gas experiments

A istoichiometric)

B (overstoichiometric)

valve A 15NO

valve B 02 / C3H6

1 3

4/ 1 12 / 1

NO

02

[vppm] 600 600

[vppm] 2400 2400

C3H6 ,

[vppm] 600 200

2.3. Model gas test setup The model gas test setup used in this study has already been described in literature [13]. The gas compositions used are given in Table 1. A substrate (NGK, 62 cell/cm2, wall thickness 200 lxm) was coated with the active Pt/A1203 powder (120 g/l) and measured in the model gas setup with a space velocity of 50000 hr 1. Nitrogen was used as the carrier gas.

3. RESULTS 3.1.NO pulses over reduced and oxidised catalyst Figure 1 shows the responses for NO, N2 and N20 during a multipulse NO experiment over a reduced catalyst at 473 K. The catalyst was reduced in-situ in flowing hydrogen at 473 K. No NO2 and 02 was observed during this experiment.

Figure 1. N2, NO and N20 responses during a NO multipulse experiment on a prereduced catalyst at 473 K. Initially, NO adsorbs dissociatively on the reduced surface forming N- and O-adatoms. At this time no NO is detected at the reactor outlet. The N-adatoms recombine to form N2. The O-adatoms remain on the surface as no 02 respons can be observed during this experiment. After a certain induction time, NO is quantitatively converted into N2. The occurrence of an induction time indicates that the concentration of N-adatoms increases until a pseudo steady state is reached. After about 100 NO pulses the signals for NO and N20 increase

226 simultaneously. At this point the active surface is probably almost completely covered with oxygen adatoms from the dissociative NO adsorption. The dissociative NO adsorption becomes increasingly inhibited by the decrease of unoccupied surface sites. Consequently, the N2 formation decreases due to a lower N-adatom concentration. In contrast, the production of N20 increases at this time. This indicates that associatively adsorbed NO is necessary for the generation of N20. Reactions 4) and 5) can therefore be excluded because N20 production would also occur over a reduced surface for these reactions. After about 160 NO pulses the N20 production reaches a maximum. If N20 is formed via reaction 7) the N20 production could indeed be maximal when at this moment the product of the concentrations of associatively adsorbed NO and N-adatoms is maximal. However, when the concentration of associatively adsorbed NO reaches a maximum at this point, N20 formation via reaction 6) can not be excluded. The same trends were observed for a similar multipulse experiment at 673 K. When NO was pulsed at 673 K over an oxidised catalyst, upto 5% conversion into N2 was observed. This observation is in sharp contrast with the observations from Butch et al. [3] who did not observe any N2 production over an oxidised catalyst. Figure 2 shows the normalised N2 formation over a reduced and an oxidised platinum surface. The N2 formation takes place more slowly over an oxidised surface compared with a reduced surface. On a reduced catalyst the N2 signal is close to the Ar signal (N2 comes earlier than Ar because of mass discrimination due to the total diffusion processes) implying a very fast production of N2. NO reacts immediately at contact with the reduced platinum surface. The shape of the N2 peak over an oxidised catalyst resembles closely the corresponding NO peak shape. This indicates that the N2 formation takes place after, on average, very many contacts of NO with the surface.

1.0 (~

0.8

"~

0.6

E: (/)

0.4 0.2

Z

(pre-reduced)

o.o 0.0

0.1

0.2 0.3 0.4 0.5 Time [s] Figure 2. Normalised (peak height=l) N2 formation during a NO pulse over an oxidised and a reduced surface at 673 K. The surface area of the N2 response over a preoxidised surface amounts to only 5% of that on the prereduced surface.

227 To understand the effect of 02 during NO decomposition, the catalyst was first oxidised insitu with 1802. Then NO was pulsed, followed by an 1802 pulse 1.5 seconds later to stabilise the 180-adatom coverage on the platinum. At 513 K the NO left the reactor continuously (baseline increase of NO signal on the MS). At 673 K almost all NO left the catalyst within 5 seconds. At 513 K and 673 K about 10% of the NO coming out of the reactor was NI80. This result suggests that NO adsorbs mainly associatively on an oxidised surface as dissociation and subsequent reassociation should yield a very high isotope exchange. The peak shapes of NO and Nt80 are identical at 673 K. This suggests that the reaction leading to the product N180 is a fast process compared to the total retention processes of NO in the reactor. Otherwise the N180 signal should come somewhat retarded to the NO signal. Although no NO2 could be observed during the TAP experiments, oxygen exchange due to the formation of NOl80 and subsequent decomposition can not be excluded here. However, the formation of N180 by dissociation of adsorbed NO (reaction (2)) and subsequent reassociation of the adsorbed N-adatom with an 180-adatom is more likely (reverse of reaction (2)). 3.2. NO reduction in the presence of 02 Figure 3 shows the responses of the nitrogen containing species during a pump/probe experiment with a NO pulse followed by a propene pulse over a surface preoxidised with 1802"

N2

NO ~ t/) t" --

N2 .z "

2

tO

t"

~ 0.6-

\\

I

~N20 ~ ~ , , ~ ~

1

\

~

0

o

~

T,meIs]

18 I

0.0

,

I

0.5

,

I

1.0

,

I

1.5

,

I

2.0

,

I

2.5

,

I

3.0

Time [s]

Figure 3. Responses of N-containing species during a pump/probe experiment (At=ls) with NO and C3H6 respectively over a platinum surface preoxidised with 1802 at 673 K. The inset shows the normalised responses during the NO pulse. The amount of NISo formed is much higher than the amount formed in the single pulse experiments over an oxidised surface. At 573 and 673 K about 50 % of the outcoming NO is detected as NI80. At 513 K 30 % of the outcoming NO is detected as N180. As the pump/probe pulses are cycled for signal averaging the surface will be partly reduced in this experiment. This result clearly demonstrates that on a partly reduced platinum surface NO will

228 more often adsorb dissociatively than on an oxidised platinum surface as more N2 and N180 is formed. These results do still not unambiguously differentiate between the two proposed mechanisms of NO/NlsO isotope exchange. However, because the oxidation of NO to NO2 is thought to be more probable over an oxidised surface than over a partly reduced surface, these results support the dissociative mechanism for the NO/N180 isotope exchange. As both NO and NlsO are present, one would also expect that N2180 is formed. However, almost no N2180 is detected. At 573 K also little N2180 was observed. However, at this temperature a clear delay of NlSO compared to NO was observed. This indicates that at 573 K the formation of N180 is a slow process compared to both the total diffusion processes of NO through the reactor and to the formation of N20 and N2. As N20 is only formed during the initial phase of the NO pulse, no N2180 will be formed as NISo is formed later in the process. The formation of N~80 at 673 K is probably also slow compared to the formation of N20 and N2, but is faster as the total diffision processes of NO, hence no retardation of Nl80 compared to NO is observed. Figure 4 shows the responses of all components observed during a typical pump/probe experiment. In this experiment propene/O2 from the stoichiometric mixture A was pulsed followed by a NO pulse one second later at 573 K. The sharp 02 peak during the propene/O2 pulse indicates that propene is not completely converted at this temperature. CO2 is formed mainly during the propene/O2 pulse and only a small extra amount is generated during the NO pulse. This increased CO2 formation during the NO pulse is probably caused by the direct reaction of NO with carbonaceous residues on the surface or due to the reaction of O-adatoms, formed by the dissociative NO adsorption, with the carbonaceous residues [14,15]. Some NO comes through the reactor during the NO pulse and a small amount of NO is released during the propene/O2 pulse. N2 and N20 are mainly formed during the NO pulse but small amounts are also generated from the remaining N-adspecies during the propene/O2 pulse. N2 is formed during the propene/O2 pulse in a double peak suggesting two different reaction mechanisms. No propene could be detected at the outlet of the reactor.

2.5 i,---i

~>' t-. Q.) e-

.--.

O2/2

Ar/50

2.0 N2j

1.5 1.0 0.5 0.0

~C3H6 I

0.0

,

I

0.5

,

I

1.0

I

'

1.5

I

Time [s]

......

21.0

i

21.5

i

3.0

Figure 4. Responses of observed components in a pump/probe experiment with NO and O2/C3H6 from stoichiometric mixture A over an oxidised platinum surface at 573 K.

229 0.20

-

0.15

••--•at rl

At = 0 . 0 1 s

"~ ~.,-, r- 9 0.10 0.05

= 0s

V

zxt = l s ~

I

0.00 ~ ~ 0.0

1 0.5

1.0

~ 1.5

2.0

l 2.5

3.0

Time [s] Figure 5. N2 formation in a pump/probe experiment with O2/C3H6 and NO from overstoichiometric mixture B over an oxidised platinum surface at 573 K as a function of the offset time, At, between the two pulses. The Nz response after a single pulses of NO/Ar over an oxidised catalyst is depicted for reference. Figure 5 shows the N2 production at different offset times between the O2/propene and the NO pulse with mixture B at 573 K. It was found that the amount of N2 (and also N20) formation decreases with increasing offset time At from 0 to 1 second for the lean mixture B. The amount and peak shape of the N2 formed during the NO pulse at At = 1 second are identical to those observed when NO is pulsed over a preoxidised surface. This indicates that one second after the propene/O2 pulse no residual carbonaceous species are present on the surface which can reduce NO. When the next propene/O2 pulse enters the catalyst, two seconds after the NO pulse, still some N containing adspecies are on the surface as some N2 (and also N20) formation is visible. At At = 0 and 0.01 seconds more oxygen leaves the catalyst after the propene/O2 pulse then at larger offset times. Also more N2 is formed during the NO pulse at At = 0 and 0.01 seconds. This effect can be caused by a competition between O2 and NO for a direct reaction with propene or reaction products of propene. Another possibility is that 02 and NO compete for reduced adsorption sites. Rottlander et al. [ 14] recently reported similar results with TAP experiments on a Pt/ZSM-5 catalyst. They proposed that carbon containing surface species, formed from propene, are mainly responsible for the NOx reduction at T < 600 K. With the stoichiometric mixture A the amount of N2 and N20 formed was independent from the time interval between the propene/O2 pulse and the NO pulse. Burch et al. [3] observed from similar TAP experiments with lean mixtures of propene/O2/NO that the N2 yield did not change significantly with increasing At.

230 3.3. Correlation between model gas and TAP experiments TAP experiments are performed at operating conditions which are quite far from operating conditions in automotive converters; pulse conditions, low pressure (10 -4 - 10-5 Pa) and low absolute concentrations. To find out in how far conclusions from TAP experiments are still applicable to the automotive converter, correlation experiments were performed on the model gas test setup using gas compositions as given in Table 1. temperature TAP measurements [K] 400 100

450 i

500 ,

550

i

,

600

i

, I 13. ............ ...'"

80 84

.

. "'"":

temperature TAP measurements [K]

700

650 ,

i'

0-4000 1

,"

-.....

~

t

J

i o

40-

---o-- TAP mixture B

.".

O ~ 20Z

xl"

N2

""-.

13.:"

s~o

,

500

,

~,o

,

~o |

,

o! ..... , " ........... |

700

i

7so

~

,

"

"-. "'or'"

" o ....

........ O

,.,..

D 'i ....iiT!ii

~i! ~

.o

:

: : : ::

....................... ?iili

. i

......... "" 0.2 ......................i---i--b-~-~-i F~ iiii!! 0.1 ......................i--!--i+-i-~..................~....r i ii::i:: i ! i::i:. .

0

,

.

.

:

.

:

.

.

:

i

!

i ! i ii

ill i i!

~ ....i ....

: : :

. . . . . .

.

.

.

. . . . . . . . . . . . . . .

.

1

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

1o0

i-iiii!

i i iiili

...................i.... ii-!iii ...................

!.... J...i..L.L.Li

~! i~i~ ! i l iil

Iiiiiiiiiiii[iiiii[iiiiiiiiiiiiiilili

.

10

;...:..;..:.;.;.

!iiiii~

;

.

.

~. . . . .

i ....?-i-i-ii .................. i....

~ i i iilil

0.4

o 0.3 ft..

, .................

A iiiiii!

10o0

. . .

. . .

. . .

: .

.

.

; . . ',. . ;',',: . .

. .

.

.

,,

. .

. .

.

.

:

' . : : :

100o0

Pore Diameter, nm

Figure 2. Mercury intrusion porosimetry of titania-sepiolite 50:50, wt/wt, supports prepared with five different types of titania and heat treated at 500~ Table 2 Textural properties of titania-sepiolite 50:50, wt/wt, supports Mercury Intrusion Porosimetry Titania type A B C D E

Total Pore Volume, mEg "1

0.69 0.62 0.69 0.51 0.61

Surface Area, m2g"1 95 110 97 56 91

N2 adsorption

Pore Diameter, nm 18 18 10

65 40 32 55 67

BET Surface Area, m2g"l 133 111 100 73 113

These results give a picture of the complex influences of the parameters that could affect preparation of catalysts at industrial scale. Firstly, as was observed, the nature of the raw materials strongly influenced the textural properties of the support, affecting the activity of the catalyst. Secondly, not only the BET surface areas of the raw materials and catalyst were of

238 importance when trying to improve its activity, but also its porosity total pore volume, surface area and pore size distribution. These results indicate that a high surface area, large total pore volume and bimodal pore size distribution (mesopores + macropores) should be beneficial to improve the catalyst's activity, but the interaction between all three parameters is not clear and should be further studied. Additionally, the influence of other parameters, e.g. the surface hydration degree, the presence of additives added by the producer, etc., could be of importance and is currently being studied. The influence of the titania content was studied by measuring the activity of three catalysts containing 30, 50 and 70 wt.% of titania type B. The resulting NO~ conversions (not shown) indicated no major differences between the activities of these three catalysts. These results indicated that the Pt dispersion was not adversely affected by a decrease in the available titania surface. Taking into account the previous results, a catalyst with 0.1 wt.% Pt supported on a 50:50, wt/wt, sepiolite/titania A support, designated catalyst A, was selected for the following studies. 3.2. Effect of heat treatment: Pore volume vs. surface area

The heat treatment of support A previous to impregnation of the active phase has a strong effect on the catalyst activity. Figure 3 shows a decrease of the NOx reduction activity when the support is pre-treated at 800~ compared to that pre-treated at 500~ 60 o--*, ~- 50 .o 40

-

(D

> 30

t-

o 20 O

z

/I 111~

10 0 150

175 200 225 250 Temperature, ~

275

Figure 3. NOx conversion of Pt supported on support A treated at (--) 500 and (---) 800~ Although it is well known that the anatase to futile phase change takes place in the range 550-700~ previous studies have shown that the presence of sepiolite retards this process [12]. The XRD spectra of support A treated at 500, 800, 1000, and 1200~ are shown in Figure 4. Up to pretreatment at 800~ the titania phase present is anatase. However, after heating at 1000~ the phase change: anatase --~ rutile was not completed. From calculation of the relative intensities of the principal peaks for anatase and rutile the conversion under these conditions was found to be only ca. 25% [13]. Even treatment at 1200~ was only sufficient to cause an 80% conversion to rutile. The other main component in the support, sepiolite, can also undergo transformation to enstatite above 830~ [14]. The XRD patterns of the support pre-treated at temperatures below 1000~ (Figure 4) showed no enstatite peak. In order to further study if this transformation could be responsible for the effect of heat pretreatment on the activity, the TG-

239

DSC curves of sepiolite and support A from 25 to 1000~ were recorded, and are shown in Figure 5. The peak observed at 830~ in the DSC curve of sepiolite and of the support indicates that the sepiolite to enstatite transformation takes place above the treatment temperature of both catalysts (500 and 800~ and is not affected by the presence of titania.

I --.-E dR ~ it} t"

(d)

(P r ..==.

=>

(c)

~

tli n," .===. N(a) + O(a)

(5)

which is followed by the elementary steps: N(a) + N(a) --->N2

(6)

N(a) + NO(a) --> N20

(7)

The role of CO, 1-12 or hydrocarbon is to scavenge atomic oxygen resulting from the NO dissociation. The observed increase in the selectivity towards N2 is a consequence of increased NO dissociation, i.e. a decreased amount of molecular NO, and an increased amount of atomic N on the surface. Both factors favours reaction (6) over reaction (7). This dissociative mechanism is the generally accepted pathway under ultra high vacuum conditions [18]. However, a recent study by Klein et al [19] has questioned the validity of the dissociative mechanism under atmospheric pressure conditions in favour of a non-dissociative mechanism. A particular difficulty with the non-dissociative mechanism is that it cannot readily accoum for

264 the lack of reactivity of low index planes of Pt. Unpromoted low index planes of Pt, Pt(111), are relatively inert towards NO dissociation, the process we propose as the key reactioninitiating step. Our EP results strongly suggest that the dissociative mechanism holds, even in the high pressure regime. The catalyst film consists of large polycrystalline Pt particles whose surfaces are dominated by low index planes that are inactive for NO dissociation. The low rates observed at high positive catalyst potentials (Na-free system) may be ascribed to defects and high index planes that are inevitably present at crystallite edges. Both Nz and N20 are produced in this region as there is a mixture of molecular NO plus atomic N and O. Na supplied to the Pt surface strongly enhances the overall activity by inducing NO dissociation on the otherwise ineffective low index planes in accord with both theory and experiment. REFERENCES .

2. 3. 4.

10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22.

K.C. Taylor, Catal. Rev.-Sci. Eng., 35 (1993) 457 M. Kiskinova, Studies in surface Science and Catalysis, 70 (1992) 1 J.M. Campbell, Catalysis at Surfaces, NewYork, Chapman and Hall, 1988 I.V. Yentekakis, G. Moggridge, C.G. Vayenas and R.M. Lambert, J. Catal., 146 (1994) 292 I.V. Yentekakis and C.G. Vayenas, J. Catal., 149 (1994) 238 T.I. Politova, V.A. Sobyanin and V.D. Belyaev, React. Kinet. Catal. Lett.,41 (1990) C. Pliangos, I.V. Yentekakis, S. Ladas and C.G. Vayenas, J. Cat.,159 (1996) 189 C.G. Vayenas, S. Bebelis, I.V. Yentekakis, and H.-G. Lintz, Catal. Today, 11 (1992) 303 I.R. Harkness, C. Hardaere, R.M. Lambert, I.V. Yentekakis and C.G. Vayenas, J. Catal., 160 (1996) 471 I.R. Harkness and R.M. Lambert, J. Catal. 152 (1995) 211 A. Palermo, R.M. Lambert, I.R. Harkness, I.V. Yentekakis, O. Marina and C.G. Vayenas, J. Catal., 161 (1996) 471 X. Zhang, A.B. Waiters and A. Vannice, Appl. Catal. B: Env., 4 (1994) 23 7 R. Burch, P.J. Millington and A.P. Walker, Appl. Catal. B: Env., 4 (1994) 65 I.V. Yemekakis, S. Neophytides and C.G. Vayenas, J. Catal., 111 (1988) 152 I.V. Yentekakis and S. Bebelis, J. Catal., 137 (1992) 278 C.G. Vayenas,~S. Bebelis and S. Ladas, Nature, 343 (1990) 625 J.C. Bertolini, P. Delichere and J. Massardier, Surf. Sci. 160 (1985) 531 B.A. Banse, D.T. Wickham and B.E. Koel, J. Catal., 119 (1989) 238 R.L. Klein, S. Schwartz and L.D. Schimdt, J. Phys. Chem., 89 (1985) 238 O.A. Marina, I.V. Yentekakis, C.G. Vayenas, A. Palermo and R.M. Lambert, J. Catal., 166(1997)218 R.M. Lambert, M.S. Tikhov, A. Palermo, I.V. Yentekakis and C.G. Vayenas, Ionies, 1 (1995) 366 I.V. Yentekakis, A. Palermo, N. Filkin, M. Tikhov and R.M. Lambert, J Phys. Chem. B, 101 (1997)3759

CATALYSIS AND AUTOMOTIVE POLLUTIONCONTROLIV Studies in Surface Science and Catalysis,Vol. 116 N. Kruse, A. Frennet and J.-M Bastin (Eds.) 91998 Elsevier Science B.V. All rights reserved.

265

Promoting effect of Zinc in DeNOx reaction over Pt/AI203 A. Bensaddik, N. Mouaddib, M. Krawczyk a, V. Pitchon, F.Garin, G. Maire LERCSI, URA 1498 CNRS, ECPM, Institut Le Bel, Universit6 Louis Pasteur, 4, rue Blaise Pascal, 67070 Strasbourg-France. a Institute of Physical Chemistry, Department of Applied Surface Science. Polish Academy of Sciences, Kasprzaka 44152, 01-224 Warszawa, Poland. ABSTRACT , The selective reduction of NO by hydrocarbon in an excess of oxygen was studied using a platinum catalyst doped or not with zinc. Successive impregnetion or co-impregnation of Pt and Zn on alumina were made. In some cases, in the presence of Zn, the NO conversion is increased in parallel with N2 formation. A better conversion of hydrocarbons was also observed. EXAFS experiments and N20 decomposition experiments have been carried out to explain these observations.

1. INTRODUCTION Organo-zinc compounds are commonly present in nearly all commercially lubricating oils which are intended for vehicle use. Unfortunately, these compounds are involved in the combustion process in the engine and are converted mainly into zinc oxide before entering the catalytic converters. This form of zinc usually deactivates the vehicle exhaust catalysts [ 1]. Catalytic experiments performed with 3.3wt% Pt/AI203 catalyst over which 0.5wt% of zinc was added have shown, for skeletal rearrangement of hydrocarbons, that the activity was nearly completely suppressed. This result indicated that zinc deposition, from Zn(NO3)2followed by air calcination 2 hours at 550~ is not statistically performed onto the whole surface area. If it were the case the activity should have been decreased by about 10% only [2]. Zinc seems to be preferentially located near platinum aggregates. Contrary to these reforming reactions involving only hydrogen and hydrocarbons, we noticed that light-off experiments performed with a complete gas mixture (CO, CO2, 02, NO, HC and H20), either under rich or lean conditions, gave better conversions for the three pollutants on S-[Pt-Zn] catalysts than on platinum catalysts.These very encouraging results for three way catalysts led us to investigate these alumina supported S-[Pt-Zn] catalysts for DeNOx reactions as these catalysts gave promising results for NO reduction under lean conditions.

266 The aim of this paper is to understand the influence of zinc on platinum catalytic behaviour. The added metal can either deactivate or provoke an increase in the catalytic activity of platinum either for reforming reactions or depollution reactions respectively, even when the gas atmosphere is always reductive. We shall study the influence :-i) of the mode of preparation ,-ii) of the zinc loading and -iii) of the kinetic parameters, on the activity of S-[Pt-Zn] catalysts in DeNOx reactions.The catalysts have been characterised by TPR, chemisorption and EXAFS and tested in the reaction of selective catalytic reduction (SCR) using diesel conditions.

2. EXPERIMENTAL

2.1 Preparation The catalysts were prepared by wet impregnation of 7 A1203 (GC0 64 RP, 206 m2g -1) by a solution of H2PtC16 or Zn(NO3)2. The concentrations of these solutions were calculated in order to have l wt%Pt and 0.5wt%Zn. Two sets of catalysts were elaborated : by coimpregnation of both precursor salts (named C) or by successive impregnations (named S). For the latter, the zinc was introduced on a Pt/A1203 catalyst. Between each preparative step, after the solution had been evaporated and the powder dried at 110~ overnight in an oven, the catalyst was calcined under oxygen at 500~ for 4 hours prior to hydrogen reduction performed in situ at 500~ for 4 hours. Before any catalytic test, the catalyst was pre-treated under a flow containing 10% H20 and 6% 02 in nitrogen at 450~ for 30 min. and cooled down to 150~ to start the experiment. 2.2. Characterization BET surfaces were determined by nitrogen adsorption at 77 K, in an automated volumetric set up after a vacuum desorption at 383 K for 1 hour. Dispersion was determined by chemisorption of CO at room temperature using the pulse technique. Prior to chemisorption, the catalysts were reduced under an hydrogen flow at 500~ and flushed with helium at the same temperature for 30 min. In Table 1 are given the metal content, the BET surface and the dispersion of the platinum metal for the fresh catalysts, i.e. catalyst calcined in oxygen at 500~ and reduced in H2 at 500~ for 4h, then purged in nitrogen flow for 30rain. The BET surface is between 148 to 195 m2.g -1 for all the catalysts prepared and the dispersion of the platinum metal, for the fresh catalysts is between 13% and 35%. The BET surface and the numbers of exposed platinum atoms decrease with the amount of added zinc.

26'/ Table 1 : Characterisation of the catalysts Alumina supported Pt wt% catalysts Pt 1.2 S- [Pt-Zn] 1.2 C- [Pt-Zn] 1.2 C- [Pt-Zn] 1* C-[Pt-Zn] 1*

Zn wt%

BET surface

Dispersion %

(mE.g "l ) 0 0.5 0.5 2.7 10

195 174 181 185 148

35 13 29 24 13

* Theoretical values The TPR (Temperature Programmed Reduction) experiments were carried out on some CPtZn and on the S-Pt-Zn catalysts. The samples were heated in 1%H2 in Ar (30 ml/min) from room temperature to 500~ at 8~ while H2 consumption was recorded. From the TPR profiles of Pt and Pt-Zn coimpregnated or successively impregnated samples only one peak was observed at about 280~ even in the presence of zinc. Quantitatively the amount of H2 consumed corresponds to the reduction of PtO2 to metallic Pt. The reduction temperature was the same in all case i.e. in the presence or absence of zinc. This can signify that there is no interaction between platinum and zinc probably due to the small quantity of added zinc in the catalyst (about 0.5% of Zn). The absence of a peak of hydrogen due to the reduction of ZnO and the stability of the temperature of reduction of the platinum may suggest that there is formation of a zinc aluminate. It is important to have an idea on the oxidation state of the platinum in order to know the nature of the adsorbed intermediates. For this reason we have characterised the platinum oxidation state of the S-[Pt-Zn] catalyst using the X-ray atomic absorption technique. The experiments were carded out on the Exafs 4 spectrometer at LURE-DCI, running at 1.85 GeV with an average current of 250 mA. The Exafs data were collected using a conventional step-by-step set up with a channel cut monochromator Si(111) for Pt and two ion chambers as detectors, with Pt foil used to calibrate the monochromator. The spectra were recorded above the Lm edge of platinum (11,560 eV). The results are displayed in figure 1. A simple comparison between the reference spectra of Pt foil, bulk PtO2 and Pt-Zn/A1203 spectrum (Figure 1) allows the conclusion that after the pre-treatment, the platinum is fully oxidised. It is less than likely that it could be reduced during the course of the reaction since this one occurs in a very oxidising media (6% OE). This result is very important in the sense that NO does not dissociate on platinum oxide and therefore implying that one of the intermediate is likely to be an adsorbed hydrocarbon species reacting with NO.

268

2.3. Catalytic experiments 2.3.1. Apparatus The reaction was carried out in a dynamic flow reactor using a synthetic gas mixture. The flow rates were adjusted using Tylan mass flow controllers and the effluents were analysed using IR or UV analysers from the Binos range for NO, NO2 and N20, and FID chromatography for the hydrocarbons. The gas was humidified by passage through a water saturator regulated at 50~ Before the analytical section, a Perma Pure dryer was installed to selectively remove water vapour from the gas stream. The data were collected every 10 seconds on a computer using a "purposely-written" software. 2.3.2. Reactions studied Several types of experiments were performed such as" - Kinetic studies were undertaken with a simplified gas composition (NO 9 1000 ppm, propane 440 ppm, propene 220 ppm, 6%02 and N2) and we mainly studied influence of the space -1

-1

velocity between 60000 h to 15000 h on the catalytic conversion; the catalyst weight ranges between 100 to 400mg. - Reaction tests, for which experimental conditions were as follows. The space velocity -1

was kept at 60000 h ; the catalyst weight was 100 mg, the reaction temperature varied from 150 -1

to 550~ with a temperature ramp of 4 ~ , and the reactant gas mixture was : NO 500 ppm, CO 350 ppm, CO 2 10%, 02 6%, H20 12%, propane 220 ppm and propene 110 ppm.

3. RESULTS AND DISCUSSION

3.1. Influence of the zinc loading and of the method of preparation The conversion of NO as a function of the zinc loading for the C-[Pt-Zn] catalysts is described in figure 2. The activity is almost independent of the zinc content from 0.5% to 10%. Moreover, when compared to a monometallic platinum the presence of zinc inhibits the activity towards NO conversion. The value of the conversion does not exceed 20% in the range of temperature studied with co-impregnated Pt-Zn catalysts while it reaches c.a. 33% with the Pt/A1203. On the other hand, when platinum and Zn are added by successive impregnation there is a noticeable enhancement of the activity below 300~ The maximum of conversion observed is 42% This catalyst is stable with time as evidenced by figure 3. Over a period of 700 min, the conversion remains very stable. This figure represents also the selectivity into the N-containing product when using S-[Pt-Zn] with a space velocity of 30.000 h -1 at 280~ the temperature at which the maximum of conversion was observed.

269

Figure 2: Influence ofthe zinc loading or of the method of prepadon on the NO conversion Conditions: SV:60.000h-', 100 500 ppm, CO 350 ppm, CO, 1OOm. O26%. 12%

PPane~oppm~propene~~oppm

Figure 3: Selectivity and stability versus time on S-[Pt-Zn] catalyst

270

3.2. Influence of space velocity During the catalytic reactions involved in NO reduction versus the temperature we observed the following points: - The NO conversion goes through a maximum versus the temperature. - N20 is formed and its evolution follows a volcano shape curve versus the temperature. The maxima of these two volcano curves are more or less flat. - NO2 is formed and its development follows a plateau versus the temperature. In Table 2, the maxima conversion in NO and N20 in percent are reported as well as the temperature of these maxima. For NO2 formation is reported the temperature at which the plateau is attained and its conversion in percent. The catalyst used was S-[ Pt-Zn]/A1203. From Table 2, we can observe that on S-[Pt-Zn] catalysts more NO is transformed and less N20 and NO2 are formed than on Pt catalyst, except when 200mg of S-[Pt-Zn] is used. The temperatures at which these maxima are observed are about 20~ higher on S-[Pt-Zn] than on Pt but the summits of these volcano curves are more flat for NO conversion on S-[Pt-Zn] which explains why in a range of 20~ we have the same NO conversion. Moreover at 35% NO conversion a larger domain of temperatures is observed as mentioned in Table 3.

Table 2" Conversions versus th e reaction temperature. Catalyst Catalysts weight Max. NO (VVH h "l) conversion (%) temperatureT~ Pt 38% at 300~

100mg

Max. N20 conversion (%) temperatureT~ 17% at 200~

NO2 conversion (%).at T~

8% at 250~

2% at 320~

17% at 210~

3% at 320~

5% at 2600C

(60 000h q) S-[Pt-Zn]

42% at 320~ 42% at 300~ 45% at 260~

Pt 200mg (30 000h 1) S-[Pt-Zn] Pt

58% at 280~ 56% at 260~ 48% at 280~

15% at 225~

18% at 320~

21% at 200~

10% at 280~

60% at 280~

4% at 200~

8% at 300~

400mg (15 000h 1) S-[Pt-Zn]

Table 3: Domain of temperatures for a constant NO transformation of 35%. VVH in h" 1 AT in ~ on S-[Pt-Zn]/A1203 AT in ~ on Pt/A1203 60000 120~ 50~ 15000 21 o~ 1oo~

271 Now we are going to analyse how NO is transformed, at a constant temperature of 300~ versus the space velocity. The nitrogen mass balance is: [NO]0 = [NO]t + 2[N20] + [NO2] + 2[N2] and the conversion is defined as" ([NOlo- [NO]t)/[NO]o = 2[N20]/[NO]o + [NO2]/[NO]o + 2[N2]/[NO]o where [NO]0 and [NO]t are the concentration respectively of NO initially and at time t at one defined temperature. The results reported in Table 4 prove that more N2 is formed on S-[Pt-Zn] catalysts than on Pt catalyst at 300~ Moreover, when the space velocity decreases, N2 selectivity increases mostly at the expenses of N20. This fact would tend to prove than N20 is a reaction intermediate. Table 4: Nitrogen mass balance at 300~ Alumina Catalysts NO supported weight conversion catalysts (VVH h l ) in % Pt 38% 100mg (60 000h ~) S-[Pt-Zn] 42% Pt

N20 Formation in % 5%

NO2 N2 Formation Formatio in % n in % 5% 28%

8%

2%

32%

40%

5%

2%

33%

56%

5%

16%

35%

44%

5%

10%

29%

55%

2%

8%

45%

200mg (30 000h l ) S- [Pt-Zn] Pt 400mg (15 O00h1) S-[Pt-Zn]

3.3. The influence of oxygen. 3.3.1. Effect of the concentration of oxygen The conversion of NO increases with oxygen concentration (Figure 4) but this increase is more marked between 0% and 2% 02 The initial slope from the NO conversion curve versus the percentage of 02 is 1.5 higher on [Pt-Zn] than on Pt catalyst, when the reaction is performed at 250~

272

60

--

50

40

20

10

0 0

I

2

3

4

5

6

Oxygen conc. (%)

Figure 4 : Effect of the oxygen concentration upon NO conversion into N2 on S-[Pt-Zn] 3.3.2. Reaction in the absence of NO In order to investigate the role of hydrocarbon on the NO conversion the activity of Pt and S[Pt-Zn] catalysts was measured using a gas mixture: 500 ppm NO or in the absence of NO in the gas stream, 330 ppm HC, 6% 0 2 and 12%H20. It can be seen from the figures 5a and 5b that the HC conversion was improved by the addition of NO over the S-[Pt-Zn] catalyst; for example, the HC conversion was only 42% at 300~

in the absence of NO, whereas the conversion was

reached about 78% when NO was added. While over Pt catalyst the addition of NO in the feed apparently inhibits the HC oxidation. This enhancement of activity observed on S-[Pt-Zn] can be attributed to the presence of zinc oxide which needs further investigations to precisely define its role. 3.4. The decomposition of N20 This part of the research was undertaken in order to prove whether or not N20 was an intermediate in the reaction of NOn reduction and to demonstrate a possible effect of zinc in the surprising low production of N20. Indeed, it is well known than under TWC conditions one of the by-product when using noble metal type catalysts is nitrous oxide [3]. Two types of experiments were carried out (in both cases, the catalyst was pre-treated under 6 % 02, 12 % H20): i) Feed composition: 0.5 % N20 in nitrogen, ii) Feed composition: 0.5 % N20 + 6 % 02 in nitrogen. The conversions of N20 at 480~ are respectively reported in tables 5 and 6.

273

100

100 90

Pt/Alumina

.-.. 80

/

~

~

- -:

~

---70 e.o

60

r>

50

ul L

e.-

'

__

_

90 8(1

70

~ so e ao4~

o 40 c.) 0 30 'r"9 20

!2w"|' No I

..,.

.. W i t h o u t NO[ 20

10

10

0 150

200

250

300

350

400

450

500

550

0

600

,

150

Temperature (*C)

2{10

,

,,

250

300

350

Temperature

,

u ,,,

400

,

450

500

550

(~

Figure 5b" Conversion of I-IC wltn or without NO on S-[Pt-Zn]/A1203

Figure 5a" Conversion of HC with or without NO on Pt/AI203

Table 5' D.ecomposition of N,O at 480~ .... Conversion % ........ _ ~Catalyst . . . . . . . . . . .

1%Pt

11

S-l%Pt-O.5%Zn C-l%Pt-O.5%Zn C- 1%Pt-2.7%Zn C- 1%Pt- I O%Zn

1 7

Table 6: Decomposition 0fN20 in the presence of oxygen at 480~ Catalyst Conversion % 1%Pt S-1%Pt-0.5%Zn 3 C-l%Pt-0.5%Zn 1 C-l%Pt-2.7%Zn 0 C- 1%Pt- 10%Zn

When N20 is passed on the catalyst, the decomposition reaction never occurs at temperature lower than 430~ The addition of zinc has almost no effect on the activity. The conversions are very low in the temperature range of 430-500~ and are even lower in the presence of 6% oxygen. This suggest several remarks: i) Oxygen inhibits the decomposition reaction, the expression of the reaction rate is [4]: -d[N20]/dt = K x PN20 X P02 "!/2 it) The absence of N 2 0 formation under Lean NOx is not explained by a formation/decomposition mechanism. Nevertheless, we do not exclude the fact that an intermediate of N20 could be formed on the adsorbed state from NO following this scheme, as

274 pointed out by the results in table 4, where the adsoption of N20 from the gas phase during the reaction of decomposition would be strongly unfavoured: 2NO

~

(N20)ads + (O)ads --~

N2 + ~ 02

:r fN20)ga~

4. CONCLUSION From these results we may understand that, surprisingly, under certain conditions that zinc could be a promoter of the platinum rather than a poison for reactions involved in a catalytic exhaust device. Several points have been fotmd: In the presence of Zn, the NO conversion is increased in parallel with the N2 formation. Moreover, the range of temperature in which N2 is formed is larger than in the absence of zinc. Also, this catalyst is very stable in time over a period of 10 hours. From EXAFS characterisation, the presence of platinum oxide was established which has several implications on the possible mechanism. As in the case of Cu-ZSM5 catalyst type, a promoting effect of the oxygen concentration was observed [5] as well as a volcano shape curve [6] for the NO conversion. On the contrary, an inhibiting effect of the partial pressure of oxygen was found in the case of the reaction of N20 decomposition which would indicate that N20 is not a rection intermediate. Another important point was the occurrence of a promoting effect of zinc for the HC conversion in the presence of NO. All these remarks recall the work on several system where an oxygenated HC is involved as a reaction intermediate in the DeNOx process. Zn has a promoting effect in an oxidising atmosphere contrary to the reactivity under hydrogen where inhibiting effects are usually observed[7]. The role of Zn (or ZnO) could be to favour the adsorption and formation of the reaction intermediate. The nature of the interaction between Pt and Zn still need to be elucidated. REFERENCES

1. 2. 3. 4. 5. 6. 7.

W.Fitzgerald, J.V.D.Wilson, SAE 750447 (1975) C.Serre, PhD dissertation, Universit6 Strasbourg (1991) B.K.Cho, B.H.Shanks, J.E.Bailey, J.Catal, 115, 486, (1989) E.R.S.Winter, J.Catal., 19, 32 (1970) C.N.Montreuil, M.Shelef, Appl.Catal., B, 1 (1992) L1 M.Iwamoto, Proc.meet. Cat. Technol.Removal of NO, Tokyo, Jan. 1990, 17 B.Coq, F.Figueras, J.Mol. Catal., 40, 93, (1987)

CATALYSIS AND AUTOMOTIVEPOLLUTIONCONTROLIV Studies in Surface Science and Catalysis, Vol. 116 N. Kruse, A. Frennet and J.-M Bastin (Eds.) 91998Elsevier Science B.V. All rights reserved.

275

Catalytic Properties of Palladium Exchanged ZSM-5 Catalysts in the Reduction of Nitrogen Monoxide by Methane in the Presence of Oxygen: Nature of the Active sites P. G61int , A. Goguet I , C. Descorme I , C. L6cuyer2 and M. Primer I ~Laboratoire d'Application de la Chimie ~ l'Environnement, UMR CNRS 5634, Universit6 Claude Bernard Lyon I, Bat. 303, 43, Boulevard du 11 Novembre 1918, 69622 Villeurbanne Cedex France, Fax : +33 478 94 19 95 2GAZ DE FRANCE, Direction de la Recherche, CERSTA, P.O.Box 33, 93211 La Plaine Saint-Denis Cedex, France, Fax +33 9 149 22 49 67 The catalytic activity of Pd-H-ZSM-5 catalysts containing 0.18 to 1.56 wt.- % Pd in the reduction of NO by CH4 in the presence of excess 02 (lean mixture) was measured and the adsorption of NO was studied by FTIR. For all samples, NO adsorption results in the formation of a Pd + mononitrosyl complex (v NO = 1881 cm l ) and adsorbed NO2 (v NO = 2136 cm -1) interacting with both Pd ions and acidic hydroxyl groups of the zeolite. The formation of NO2 arises from the reduction of isolated Pd 2+ cations and/or Pd 2+ hydroxyl complexes bonded to the oxygen atoms of the zeolite framework into Pd + complexes still anchored to the framework. The amount of Pd complexes increases linearly with Pd content up to 0.6 wt.-% Pd. The catalytic activity measurements indicate that two competitive reactions occur: the reduction of NO by CH4 and the combustion of CH4 by 02. Samples containing less than 0.5 wt.-% Pd exhibit high selectivity for NO reduction and the conversion of NO increases with the amount of Pd nitrosyl complexes detected by IR. For samples with higher Pd contents, the combustion of CH4 is favored. The coexistence of isolated Pd complexes active for the SCR reaction and PdO aggregates active for the CH4 combustion is suggested to explain the catalytic properties. 1. INTRODUCTION In the past few years there has been growing interest in studying the catalytic reduction of nitric oxide by methane in the presence of oxygen over various ion metal-exchanged zeolites including Co, Fe, Ni, Mn, Ga, I n , Pd (1-29). Most of the work was devoted to Co-ZSM-5 because of its high activity (2-14). Attempts to describe the mechanism by which methane reduces NO over these catalysts revealed specific catalytic behaviors depending on the zeolite structure and/or the exchanged metal. For example, Cu-ZSM-5 was shown to form nitro and nitrate species reactive with C3+ alkanes but non reactive with methane (12), which could explain the inactivity of this catalyst in the reduction of NO by methane. The activity of Ga- or In- ZSM-5 catalysts was found highly sensitive to water vapor in the feed, much more than Co-ZSM-5 catalysts (18). Much work is still needed to fully understand the mechanism of the

276 reaction and the specificity of the various zeolite catalysts. Among these, Pd-H-ZSM-5 and Pd-H-MOR catalysts were recently found even more active than Co-ZSM-5 and Co-Ferrierite in the reduction of nitric oxide by methane under lean conditions (24, 25, 29). We present here a study of Pd-H-ZSM-5 catalysts in the reduction of NO by CH4 in the presence of excess 02 coupled with a detailed IR investigation of their reactivity towards NO at room temperature. The surface species formed upon NO adsorption were identified. The influence of the Pd content on the catalytic and adsorptive properties of these catalysts was examined. A relationship between the catalytic activity in the reduction of NO and the formation of surface adsorbed species observed by FTIR upon NO adsorption was obtained. 2. E X P E R I M E N T A L

2.1. Samples preparation The Pd-H-ZSM-5 zeolite samples were prepared by conventional exchange of H-ZSM-5 (PQ Zeolites B.V., CBV 5020, Si/A1 = 25) with aqueous stirred solutions of tetrammine Pd(II) nitrate at 50~ for 6 hours. The amount of Pd(II) salt was adjusted so as to vary the final Pd loading. After exchange, the preparations were thoroughly washed with deionized water, filtered and dried at 120~ overnight. In order to decompose the Pd complexes, the Pdexchanged zeolite samples were subsequently activated under flowing oxygen from 25~ up to 500~ (linear ramp of 0.5 ~ minl). Five samples were prepared, containing respectively 0.18, 0.38, 0.49, 0.66 and 1.56 wt.-% Pd (measured by atomic absorption analysis) and named Pd-H-ZSM-5(x) with x = wt.-% Pd.

2.2. Catalytic activity measurements The experimental details for activity measurements were described elsewhere (25). Briefly, the catalytic activities for the conversion of NO and CH4 were measured using a U-shaped quartz reactor (16 mm ID) operating in a steady-state plug flow mode. Typically, 200 mg of catalyst were activated in-situ in oxygen flow at 500~ (linear ramp rate of 0.5~ purged for one hour at 500~ by flowing helium and cooled to 250~ before being contacted with the reactants. The reaction mixture was adjusted so as to examine the catalytic activity under lean conditions: 2000 vpm NO, 1000 vpm CH4, 6240 vpm 02; helium as balance; total flow rate = 167 cm3/min. [The apparent gas hourly space velocity (GHSV) was 30,000 h l , based on the apparent bulk density of the zeolites, ca. 0.5 g/cm3.] The catalytic activity was measured as a function of temperature in the range 250- 600~ by using a linear heating rate of 1~ The stability of the catalytic activity was examined during two additional hours at 600~ before cooling the sample down to 250~ (linear ramp of 1~ The same sequence under reaction mixture was applied again to check for possible irreversible changes of the catalyst. The effluent gases were analyzed using two gas chromatographs equipped with TCD and FID detectors and NOx infrared analyzers. Carbon and nitrogen balances were checked. The NOx conversion was determined according to the following equation" NOx conversion % = ([NO]0 + [NO2]0 - [NO] - [NO2]) * 100/([NO]0 + [NO2]0) where [NO]0 and [NO2]0 are the inlet concentrations of NO and NO2 respectively and [NO] and [NO2] the outlet concentrations. The NO2 formation was low in the whole range of temperature ([NO2]< 40 vpm), almost independent on the temperature and ascribed to the NO

2'/'/ oxidation in the dead volume of the apparatus. The CH4 conversion was determined from the consumption of CH4.

2.3. X-ray diffraction measurements X-ray diffraction measurements were performed using CuK~ radiation on a Philips PW 1710 diffractometer. 2.4. FTIR measurements The IR studies of NO or NO2 adsorption were performed using self supported samples wafers (18 mm diameter, weight of 30 mg ) introduced into a home made IR cell allowing in situ studies at varying temperatures under controlled atmosphere (30). The samples were pretreated in situ in flowing oxygen at 500~ for 30 min (linear ramp rate of 10 ~ The cell was subsequently connected to a UHV system allowing a base pressure as low as 108 Torr (1 Torr = (101 325/760) Pa) and the sample evacuated at 500~ for 3 hours before being cooled down to 25~ and contacted with NO. The IR spectra were recorded at a resolution of 4 cm -I on a Nicolet Magna-IR 550 FTIR spectrometer. All the reported spectra have been corrected for the baseline (spectrum of the sample activated in situ and evacuated under vacuum at 25~ 3. RESULTS

3.1. Catalytic activity Figure 1A shows the effect of temperature on the conversions of NO and CH4 over Pd-HZSM-5(0.38). Both conversions begin around 300~ and increase with temperature up to 600~ the curves exhibiting an inflection point around 500~ It must be pointed out that NO is converted only to N2 in the whole range of temperature. Over Co-ZSM-5 catalysts, two reactions, NO reduction by CH4 (1) and CH4 combustion (2), have been suggested to occur (2-5): CH4 + 2 NO + 02 "--} CO2 + N2 + 2 H20 (1) CH4 + 2 O2 --+ CO2 + 2 H20 (2) Reaction (1) was based upon the fact that the catalyst was not active for NO reduction in the absence of 02 (2-6) below 500~ The same was observed on Pd catalysts (26, 32). Accordingly, the selectivity towards NO reduction, SSCR, defined as the fraction of methane involved in reaction (1), can be written: SSCR -" 0.5 [NO]o CNO / [CH4]0 CCH4,

where CNO and CcH4 are the conversions of NO and CH4, [NO]o the inlet concentration of NO and [CH4]o the inlet concentration of CH4. In our experimental conditions, 0.5 [NO]0 = [CH4]0, and SSCR= CNO / CCH4. The selectivity for NO reduction over Pd-H-ZSM-5(0.38) was plotted as a function of temperature (figure 1B). The selectivity is high (70%) and constant below 500~ then slightly decreases above 500~ indicating that the CH4 combustion would be slightly favored at high temperatures. At 400~ the selectivities of Co and Pd catalysts for NO reduction are comparable. However, the selectivity of the Pd-H-ZSM-5 catalyst does not vary in the whole range of temperatures, while it decreases rapidly from 75% down to 13% between 400 and 500~ over Co-ZSM-5 (5). Upon ~ lempemtt~ rarnl~ the conversions

278

40

--

-

A

50

_ .

B 0.8-

40

30-

T O

t-

o o 0.6z o

t.-

30 o

ffl L_

> tO

20

v

> t-O

o T

o

$"

~

20 o O z 10

,.

10

u.

250

...

>,

.--- 0.4-O

.,..,.

o9 0.2

-

0

I

I

I

350

450

550

400

Temperature / ~

I

I

500

600

Temperature / ~

Figure 1" Catalytic activity of Pd-H-ZSM-5 ( 0.38 wt.-% Pd ) as a function of temperature. (A) Conversions of NO ( II, [21 ) and CH4 ( O, O ), (B) Selectivity for NO reduction (CNo / CCH4 ). Solid line = 1st run, broken line - 2 nd run.

A

100 --

o~

,,

. t 50

1 -

- 40

.~ 0.8 -

80

=

t-.

- 30 .o ~

.9 60 > tO

o T 0

>t-.-

o.9. o

o 0.6-

v >,

40

-- 20 0o

'~ -= 0.4

20

t

10

co 0.2

/

0

0 250

,

I

I

350

450

550

Temperature / ~

B

o

0 400

I

I

500

600

Temperature / ~

Figure 2: Catalytic activity ofPd-H-ZSM-5 ( 1.56 wt.-% Pd ) as a function of temperature. (A) Conversions of NO ( II, UI ) and CH4 ( Q, O ), (B) Selectivity for NO reduction ( CNO / CCH4 ). Solid line = 1st r u n , broken line - 2 nd run.

279 curves were the same, indicating no irreversible change of the catalyst upon reaction up to 600~ Figure 2A shows the conversions of NO and CH4 over Pd-H-ZSM-5(1.56) as functions of temperature. The NO conversion exhibits a volcano shape curve with a maximum around 500~ At this temperature, the conversion of CH4 reaches 95% and upon increasing further the temperature CH4 is totally consumed and NO conversion declines. The selectivity curve plotted in figure 2B as a function of temperature exhibits a continuous decrease with temperature from 50% at 400~ down to 14% at 600~ the decrease being more severe above 500~ Above 500~ the decline of the selectivity could be ascribed to the disappearance of CH4 through the combustion reaction (reaction 2), as already observed over Co catalysts (4-6). Indeed, the NO reduction is known to be dependent on the partial pressure of CH4 (5, 31). The NO conversion is therefore expected to decrease upon decreasing amount of CH4 due to increasing rate for CH4 combustion with temperature. In the low temperature range, the selectivity of Pd-H-ZSM-5(1.56) appears less than Pd-H-ZSM-5(0.38). Moreover, it decreases with temperature. This might indicate the presence, in Pd-H-ZSM-5(1.56), of catalytic sites only active for the CH4 combustion by 02 in addition to those active for both reactions. Upon repeating the experiment, the Pd-H-ZSM-5(1.56) sample still shows a volcano curve of NO conversion but with much less activity, the selectivity for CH4 combustion being still increased. This would indicate strong irreversible modifications of the catalyst upon reaction at 600~ favoring the combustion reaction. 3.2. XRD The X-ray pattern of the Pd-H-ZSM-5(1.56) catalyst reveals a small peak at 33.9~ characteristic of PdO ([ 101 ] plane), indicating the presence of large particles of PdO. For all the other samples, no peak at 20 = 33.9 ~ could be observed, confirming the high dispersion of palladium in the catalysts. Palladium is supposed to be atomically dispersed in the channels of H-ZSM-5 in the form of Pd(II) or Pd(II) hydroxyl complexes anchored to the zeolite framework (25, 28, 29), presumably at the channels intersections.

3.3. IR Studies of NO adsorption NO was adsorbed at 20~ on all the prepared Pd-H-ZSM-5 catalysts (with varying Pd content). For all the samples, two bands at ca. 2136 and 1881 cm 1 are clearly developing simultaneously when incrementing NO pressure. Figure 3 shows the IR spectra obtained after contacting each Pd catalyst with 0.5 Torr NO at 20~ The 2136 cm -I band, observed also upon adsorption of NO2 on H-ZSM-5, is ascribed to adsorbed NO2 ~+ probably interacting with acidic hydroxyls (28). Upon evacuation under vacuum at increasing temperature, the 2136 cm l band disappears progressively between 50 and 180~ while NO2 is detected in the gas phase, corroborating its attribution to some form of adsorbed NO2 (29). It is noteworthy that NO2 adsorbed on H-ZSM-5 is already removed by pumping off at room temperature. This would indicate that NO2 formed upon reaction of NO with Pd-H-ZSM-5 is much more strongly held to the surface. Therefore it would be ascribed to NO2 ~+ associated with both acidic hydroxyls and Pd cations, in agreement with previous studies (14, 28, 33). The 1881 cm ~ band was attributed to a Pd nitrosyl complex (28). The ratios NO/Pd = 1.5 measured upon NO adsorption and NO2/Pd = 0.5 measured upon thermodesorption in flowing helium were consistent with the reduction of Pd(II) into Pd(I) and subsequent formation of

280 Pd(I) mononitrosyl complexes (28). These complexes are highly stable upon heating under vacuum since the intensity of the 1881 cm 1 band is almost not affected up to 400~ (29). However the nature of this band could be multiple since shoulders at about 1940 and 1840 cm I can be observed too. The 1940 cm "l shoulder behaves exactly as the 1881 cm l one and does not find any ascribment yet. The 1840 cm l band readily disappears upon evacuation under vacuum at 20~ while the intensity of the 1881 cm 1 band slightly decreases too. This would suggest that the 1840 cm "l band and some band masked by the strong 1881 cm -1 absorption characterize NO weakly bonded to the catalyst, tentatively ascribed to NO adsorbed on very small PdO particles. A broad envelope of very weakly intense unresolved peaks between 1670 and 1610 cm "l is observed too, ascribed to nitrate species formed upon reaction between NO2 and the zeolite surface since it readily forms upon contacting H-ZSM-5 with NO2 gas phase at 20~ (28). This ascribment is confirmed by the following. When evacuating Pd-H-ZSM-5(0.49) up to 100~ after NO adsorption, the feature around 1650 cm "1 develops (its intensity keeps weak). At this temperature, adsorbed NO2 is removed, as indicated by the decrease of the 2136 cm -1 band, and it is expected that NO2 moving along zeolite channels reacts partially with energetic sites to form nitrate species. These species progressively decompose into NO in the 150400~ range (29). 1881 cm a

A =0.05 A=0.1

(1)

=o

O t..Q I...

2136 c m l

2200

2000

O

~

d~

1800

1600

1400

Wavenumber / cm 1

Figure 3" IR spectra of Pd-H-ZSM-5 contacted with 0.5 Torr NO at 20~ and containing: (a) 0.18, (b) 0.38, (c) 0.49, (d) 0.66, (e) 1.56 wt.-% Pd.

i

a

.......... "---1

2300

1900 Wavenumber / cm -1

1500

Figure 4: IR spectrmn of Pd-H-ZSM-5(0.49) contacted with 15N180 at 20~

The reduction of Pd(II) upon reaction with NO at room temperature is confirmed by adsorption of 15N180 on Pd-H-ZSM-5(0.49). The IR spectrum in vNO region shown in Figure 4 shows 4 bands at 2093, 2043, 1844 and 1799 cm 1. These bands are shifted from 2136 and 1881 cm "1 towards lower frequencies according to the variation of reduced mass expected for 15N160 and 15N180 respectively. This indicates that Ol6-1abeled NO2 was formed upon reacting Ol8-1abeled NO with ol6-containing Pd species of the catalyst, and therefore explains

281 the reduction of Pd(II) ions. Adsorbed O16-NO2might exchange with O18-NOgas phase either at the Pd or H + center (both sites are available for competitive adsorption of NO and NO2). Therefore as O 16- NO2 forms progressively, the NO gas phase is enriched in 016. At equilibrium, the ratio of O16-to ol8-containing species is the same for adsorbed NO2 and Pd complexes, reflecting the O 16 to O 18 composition of the NO gas phase. In order to know whether the Pd ions or complexes are anchored to the zeolite framework or not, the IR framework vibrations of Pd-H-ZSM-5(0.49) were investigated (Figure 5). After activation under O2, a weak band at 930 cm ~ forms. Upon NO adsorption, the 930 cm 1 band disappear while a new band appears at 980 cm ~ These bands are attributed to asymmetric internal stretching vibrations of T-O-T bonds (T = Si or A1) perturbed by Pd ions. The higher the perturbation, the lower the frequency. Therefore, the 930 cm 1 band could be related to anchored Pd(II) ions or complexes formed upon decomposition of exchanged complexes, and the 980 cm 1 band could be due to Pd(I) nitrosyl entities formed upon NO contact. Similar observations were found on Cu-ZSM-5 catalysts (34). 4. DISCUSSION For all the Pd-H-ZSM-5 samples prepared in this study, IR results indicate the main formation of Pd(I) mononitrosyl species in the presence of NO. These complexes are linked to the zeolite framework and characterized by a sharp intense band at 1881 cm "l. In Figure 6, the integrated intensity of this band (measured with 0.5 Torr NO) is plotted as a function of Pd content. Up to 0.5 wt.-% Pd, a linear relationship is observed. This result is consistent with the existence of isolated Pd(II) ions / Pd(I) nitrosyl complexes as catalytic sites for NO reduction similarly to Co 2+ cations as active sites in Co-ZSM-5 (1, 3). As the Pd loading is further increased, the curve bends over, indicating that an increasing fraction of exchanged Pd does not form Pd nitrosyl complexes. Accordingly, XRD patterns indicate the presence of large PdO particles (and the sample turns to the gray color characteristic of PdO instead of beige pink for low Pd contents). It can be concluded that, in spite of low Pd exchange levels, the ZSM-5 structure cannot maintain Pd cations in highly dispersed state above 15-20 % exchange (equivalent to 0.5-0.7 wt.-% Pd content). The ability of Pd-H-ZSM-5 catalysts to form Pd(I) nitrosyl species was related to their specific behavior of selectively reducing NO to N2 (25). This statement finds support in the curve of NO conversion versus Pd content (Figure 7A). Indeed, for reaction temperatures less than 500~ NO conversion clearly increases with Pd content, in a manner similar to the amount of Pd nitrosyl complexes versus Pd content. Above 500~ volcano shape curves are observed and NO conversion decreases for Pd content higher than 0.5 wt.-%. This can be easily explained by the simultaneous total conversion of CH4. The absence of reductant in the feed is expected to decrease the rate of NO reduction. This implies that CH4 participates to two distinct reactions, SCR reaction and methane combustion by 02, which compete at high temperatures. This competition is confirmed by the selectivity results, which indicates that the combustion is strongly favored above 500~ The question arises to know whether these two reactions are catalyzed by the same types of sites.

282

700

I A = 0.05

600 980 cm -~

r o E t~ .13 I,.,. O

'7,

/~

o (71

930 cm 4

+

500

~E 400 o

.13

, 300

C

c E

200 100

1020

920

I

820

0

Wavenumber / cm -~ Figure 5: IR framework vibrations of (a) activated H-ZSM-5, (b) Pd-H-ZSM-5(0.49) after activation and (c) after subsequent adsorption 0.5 Torr of NO at 20~

4o--

B

600 ~

80 o

E ..O .,.

r-

.9 6 0 -

01"~176

L_

m > 20c O

o 0 z

100 --

~

o--9, 3 0 -

1.5

Figure 6: Integrated intensity of the 1880 cm l band (obtained upon contacting with 0.5 Torr NO) versus Pd content.

A

550 ~

I

I

0.5 1 wt.-% Pd

~

10-

o o

J: ,,x

0 0

40-

20I I 0.5 1 wt.-% Pd

I 1.5

0

I

0

I

0.5 1 wt.-% Pd

I

1.5

Figure 7: Influence of Pd content on catalytic activity of Pd-H-ZSM-5 samples at different temperatures. (A) NO conversion, (B) CH4 conversion.

283 For low reaction temperatures and low Pd contents, the selectivity for NO reduction never reaches unity, which suggests that the two reactions, SCR and combustion, are competing or coupled even at low temperature. On Co-ZSM-5, these reactions were shown to be coupled (13). The striking feature in the case of Pd catalysts is that the selectivity is almost constant up to 600~ and this strongly contrasts with the catalytic behavior of Co catalysts. These two catalysts are suggested to exhibit different mechanisms. On Co- catalysts, NO2, formed by reaction of NO with 02, would initiate both reactions (13), and the bending over of NO conversion curve with temperature would be ascribed to a decrease of NO2 concentration with temperature. This would be related to the decomposition of NO2 into NO + 8902 favored above 500~ The rate determining step of the SCR reaction was attributed to the activation of CH4 into CH3 radicals (1, 8, 11-13). On Pd- catalysts, NO2 is formed too but it does not depend on the equilibrium with NO/O2. Its formation would result in the reduction of Pd(II) cations to the +1 oxidation state. This might explain the specific behavior of Pdcatalysts compared to Co- ones, i.e. the constant and high selectivity of Pd- catalysts for NO reduction. For higher Pd contents, the selectivity for NO reduction clearly decreases and increasing the temperature strongly favors the combustion of methane. This suggests the presence of sites active for methane combustion but not (or little) active for the SCR reaction. This interpretation is also supported by the non linear relationship between Pd and Pd nitrosyl amounts. Since PdO particles do form, it is suggested that Pd catalysts might contain two types of sites: (i) Pd cations atomically dispersed in exchange positions and, upon NO adsorption, forming nitrosyl complexes anchored to the zeolite framework and adsorbed NO2: these sites are thought to be responsible for selectively reducing NO to N2 in the presence of 02 and also catalyzing the combustion reaction at a smaller rate; (ii) PdO aggregates, able to catalyze mainly methane combustion: their size would depend on experimental factors such as Pd exchange level and possibly exchange and/or activation conditions. Unfortunately the latter sites are not revealed by IR study of NO adsorption and further characterization studies are needed. Complementary TPD experiments are under work in order to confirm these statements and evaluate the relative proportions of these two types of sites and characterize their adsorptive properties. REFERENCES

1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11.

J.N. Armor, Catalysis Today, 26 (1995) 147. Y. Li and J.N. Armor, Appl. Catal., B : Environmental, 1 (1992) L31. Y. Li and J.N. Armor, Appl. Catal., B : Environmental, 2 (1993) 239. Y. Li and J.N. Armor, Appl. Catal., B : Environmental, 3 (1993) L1. Y. Li, P.J. Battavio and J.N. Armor, J. Catal., 142 (1993) 561. F. Witzel, G.A. Sill and W.K. Hall, J. Catal., 149 (1994) 229. Y. Li and J.N. Armor, J. Catal., 150 (1994) 376. Y. Li, T.L. Slager and J.N. Armor, J. Catal., 150 (1994) 388. J.N. Armor and T.S. Farris, Appl. Catal., B : Environmental, 4 (1994) L11. Y. Li and J.N. Armor, Appl. Catal., B Environmental, 9 5 (1995) L257. A.D. Cowan, R. Dtimpelmann and N.W. Cant, J. Catal., 151 (1995) 356.

284 12. B.J. Adelman, T. Beutel, G.-D. Lei and W.M.H. Sachtler, J. Catal., 158 (1996) 327. 13. D.B. Lukyanov, J.L. d'Itri, G. Sill and W.K. Hall, l lth International Congress on Catalysis - 40th Anniversary, Studies in Surface Science and Catalysis, 101, J.W. Hightower, W.N. Delgass and A.T. Bell (Eds), Elsevier, Amsterdam, 1996, p. 651. 14. A.W. Aylor, L.J. Lobree, J.A. Reimer and A.T. Bell, l l t h International Congress on Catalysis - 40th Anniversary, Studies in Surface Science and Catalysis, 101, J.W. Hightower, W.N. Delgass and A.T. Bell (Eds), Elsevier, Amsterdam, 1996, p. 661. 15. K. Yogo, M. Ihara, I. Terasaki and E. Kikuchi, Chem. Lett., (1993) 229. 16. E. Y. Li and J.N. Armor, J. Catal., 145 (1994) 1. 17. E. Kikuchi and K. Yogo, Catalysis Today, 22 (1994) 73. 18. T. Tabata, M. Kokitsu and O. Okada, Appl. Catal., B : Environmental, 6 (1995) 225. 19. E. Kikuchi, M. Ogura, I. Terasaki and Y. Goto, J. Catal., 161 (1996) 465. 20. T. Tabata, M. Kokitsu and O. Okada, Catal. Lett., 25 (1994) 393. 21. M. Ogura and E. Kikuchi, 1 l th International Congress on Catalysis - 40th Anniversary, Studies in Surface Science and Catalysis, 101, J.W. Hightower, W.N. Delgass and A.T. Bell (Eds), Elsevier, Amsterdam, 1996, p. 671. 22. A. Fakche, B. Pommier, E. Garbowski, M. Primet and C. L6cuyer, French Patent Application 93 08 006. 23. Y. Nishizaka and M. Misono, Chem. Lea., (1994) 2237. 24. C. Descorme, A. Fakche, E. Garbowski, M. Primet and C. L6cuyer, 1995 International Gas Research Conference, Cannes (France), 6-9 Nov. 1995, Preprints Vol. IV, p. 505. 25. C. Descorme, P. G61in, M. Primet, C. L6cuyer and J. Saint Just, Studies in Surface Science and Catalysis, 97, Zeolites : A refined Tool for Designing Catalyst, L. Bonneviot and S. Kaliaguine, Eds, Elsevier, Amsterdam, 1995, p. 287. 26. C.J. Loughran and D.E. Resasco, Appl. Catal. B : Environmental, 7 (1995) 113. 27. H. Uchida, K. Yamaseki and I Takahashi, 2nd Japan-EC Joint Workshop, JECAT'95, Catalysis Today, 29 (1996) 99. 28. C. Descorme, P. G61in, M. Primer and C. L6cuyer, Catal. Lett., 41 (1996) 133. 29. C. Descorme, P. G61in, C. L6cuyer and M. Primet, Appl. Catal., B : Environmental, (1997), to be published. 30. N. Echoufi and P. G61in, J. Chem. Soc., Faraday Trans, 88 (1992) 1067. 31. C. Descorme, Thesis, Claude Bernard Lyon 1 University, 1996. 32. C. Descorme, A. Fakche, E. Garbowski, M. Primet, unpublished results. 33. T.E. Hoost, K.A. Laframboise and K. Otto, Catal. Lett., 33 (1995) 105. 34. G.D. Lei, B.J. Adelman, J. Sarkany and W.M.H. Sachtler, Appl. Catal., B : Environmental, 5 (1995) 245.

CATALYSIS AND AUTOMOTIVEPOLLUTIONCONTROLIV Studies in SurfaceScience and Catalysis, Vol. 116 N. Kruse, A. Frennet and J.-M Bastin (Eds.) 91998 Elsevier Science B.V. All rights reserved.

285

Influence of the p l a t i n u m - s u p p o r t interaction on the direct reduction of NOx under lean conditions Filip Acke a'b, Bj6rn Westerberg a'c, Lars Eriksson ~'d, Stefan Johansson a'e, Magnus Skoglundh a, Erik Fridell a and Gudmund Smedler a aCompetence Centre for Catalysis, Chalmers University of Technology, S-412 96 G6teborg, Sweden bDepartment of Inorganic Chemistry, G6teborg University, S-412 96 GOteborg, Sweden CDepartment of Chemical Reaction Engineering, Chalmers University of Technology, S-412 96 G6teborg, Sweden dDepartment of Engineering Chemistry, Chalmers University of Technology, S-412 96 G6teborg Sweden eDepartment of Applied Physics, Chalmers University of Technology and G6teborg University, S-412 96 G6teborg, Sweden

ABSTRACT Catalysts containing Pt supported on SiC, A1203 and ZSM-5 were prepared and studied for NO• reduction by C3H6 in 02 excess under transient (temperature ramps) and steady-state conditions. The maximum NO• reduction activity in the heating ramp experiments was similar for Pt/SiC and Pt/ZSM-5, while Pt/A1203 showed higher maximum activity. Both N2 and N20 formation was observed for all catalysts, although the respective amounts varied with the investigated system. Highest N2 selectivity was observed for Pt/A1203. When the NOx reduction activity was studied under steady-state conditions the activity of Pt/A1203 decreased substantially (mainly due to a loss in N2 production). Pt/ZSM-5 became somewhat more selective towards N2 production whereas the activity and selectivity of Pt/SiC remained at about the same values as for the heating ramp experiments. Adsorbed species on the surface of the different catalysts were investigated using in-situ FTIR in order to obtain information about the reaction mechanisms. The adsorption of species on Pt/SiC was negligible, while a number of absorption bands were observed for Pt/A1203 (N and C containing species, and -NCO) and Pt/ZSM-5 (HC).

1. INTRODUCTION Good fuel economy and durability are factors that have made the diesel engine the dominating source of power for the transport industry [ 1]. A setback with diesel technology is the emission of pollutants, in particular nitrogen oxides (NOx) and particulate matter (PM).

286 The emissions of PM can be reduced using filter technology, where the continuously regenerating trap (CRT) system appears promising [1]. The emission levels of NOx are however only affected by the CRT technology to some extent [1 ]. The large oxygen excess in the diesel exhaust obstructs the catalytic reduction of NOx by hydrocarbons. In order to reduce NO• under these conditions, there is a need to develop catalysts that possess high selectivity towards NOx reduction to N 2. Catalysts based on platinum have been pointed out as potential candidates for this process [2-3]. It has been concluded that Pt has the highest NO• reduction activity among the platinum group metals [23]. However, this high activity is accompanied by low N2 selectivity, i.e., large quantities of nitrous oxide (N20) are formed [2]. By appropriate choice of support material, e.g., introducing acidic groups on the surface, the selectivity towards N2 can be enhanced [4]. The objective of this investigation is to study the effect of the platinum support material in the lean reduction of NOx using propene as the reducing agent. For this reaction we describe differences in total activity and selectivity between platinum supported on three different materials with increasing acidity; SiC, A1203 and ZSM-5. The activities of the catalysts are studied in flow reactors under both transient (temperature ramps) and stationary conditions. Adsorbed species on the surface of the catalysts are characterised using in-situ Fourier transformed infrared spectroscopy (FTIR). Different reaction mechanisms and the nature of adsorbed species are discussed.

2. EXPERIMENTAL

2.1. Design of catalysts In accordance with our objective to study the influence of support acidity, our catalyst sample design approach was to keep the total surface area constant at 40 m 2 and total Pt loading constant at 2.0 rag, yielding a surface area based Pt loading of 50 gg/m 2 for all three supports under investigation. Through this catalyst design, we expect to compensate for the different specific areas that would have resulted from a constant support weight, as well as for the different amount of Pt per unit area that would have resulted from a constant Pt loading per unit mass of support.

2.2. Preparation of catalysts The support materials; SiC, T-AlaO3 and H-ZSM-5 (SIO2:A1203 = 34) were initially calcined in air at 600~ for 2 h. The specific surface areas of the calcined support materials were measured using nitrogen adsorption [5] and are included in Table 1. Pt was deposited on the support materials using the method described by Axelsson et al. [6]. A slurry of the respective support material was prepared by dispersing the support in distilled H20 under stirring. Specific amounts of the Pt-solution was added to the respective slurry under continuous stirring in order to obtain a constant Pt-content per square meter support (see Table 1). The three slurries were then freeze dried and finally calcined in air at 550~ for 1 h.

287 Table 1. The nomina! comPosition and BET-surface area of the prepared catalysts. i

Sample Pt/SiC Pt/AI/O3 Pt/ZSM-5

Sample weight [mgJ 1600 200 106

Pt weight {rag] 2.00 2.00 2.01

,

Pt content [mg/g support] 1.25 10.00 18.85

Surface area [mZ/g] 24.9 200 377

Pt/surface area [10 -6 ~/m2] 50 50 50 i

2.3. Flow reactor studies

The flow reactor used in the activity studies is described elsewhere [7]. Briefly it consists of a vertical quartz tube in which the sample is supported on a sintered quartz filter. Gases are introduced via mass flow controllers and the temperature is measured after the catalyst bed. Reactants and products are analysed using a quadrupole mass spectrometer and a photo acoustic FTIR gas analyser. The bed material consisted of a mixture of the powder sample and quartz sand in order to obtain a constant space velocity (25000 h l ) for all tested catalysts. The gas composition used in the experiments was: 10% O2, 405 ppm NO and 911 ppm C3H6, balanced with Ar to yield a total flow of 420 ml/min. The samples were initially reduced in 5000 ppm HE at 400~ for 15 min and stabilised in the reaction mixture at 525~ for 1 h. The samples were then cooled down to room temperature under an Ar flow. At this temperature, the catalyst was exposed to the reaction mixture under 15 min before starting the heating ramp up to 525~ at a constant rate of 6~ The steady-state experiments were performed by subsequently lowering the temperature in steps of 50~ starting from the final ramp temperature and the products were analysed after approximately 90 min. In order to facilitate the interpretation of the flow reactor and FTIR results the model gas was simplified by omitting H20 and SO 2 (which would have been present if a diesel exhaust was used). 2.4. F T I R studies The FTIR experiments were performed using thin discs (approximately 15 mg/cm2) of catalyst in a reaction chamber with CaF2-windows [8]. A disc was fixed in between folded tungsten grids placed in the centre of the reaction chamber. The temperature was measured with a thermocouple, in contact with the grid, and controlled via the voltage applied over the grid. The reaction chamber was placed in a FTIR spectrometer. All spectra were measured with 1 cm/s scan speed and a resolution of 4 cm l. The fresh catalysts were reduced in 30% H2 in N2 (total flow rate of 100 ml/min) at 450~ for 30 min., stabilised in a gas mixture with 5% 02, 1000 ppm NO and 3000 ppm C3H6 in N2 (total flow rate 1000 ml/min) for 30 rain and finally degassed in N2 (1000 ml/min) at 550~ for 30 min. The in situ FTIR measurements were performed with 5% 02, 400 ppm NO and 900 ppm C3H6 in N2 and at a total flow rate of 1000 ml/min (this means that the reaction cell operates as a differential reactor). The experiments started at 450~ and the temperature was then lowered in steps of 50~ with a 5 minute interval. Spectra from an average of 50 scans were taken the last minute of each interval. Reference spectra were taken with pure N2 in an otherwise identical sequence.

288 I

I

Nbal I

I

3500

I

4OO 300

g

3000

8

2500

C) ~

2000 200

100

/~c.~~!~.. _

_

0 ~

100

~

"~%~,

_

'

NO2

-

~klO N2

200

___. I

300

1500

~.

1000

o

500

400 .....

0

500

Temperature (~

Figure 1b. Concentration traces over a Pt/ZSM-5 catalyst in a heating ramp.

-"-I

T

1

I

3500

400

E ~

3OOO 0 0 .

.

.

.

.

.

.

.

.

2500 0

-

o

300

cO ~ c(1) r to

O

2000

o P-I,,

200

1500 m m .

o

100 -

1000

.~.

500

..,..,.-

"o -o

3

l

100

200

300

400

500

Temperature (~

Figure 1c. Concentration traces over a Pt/SiC catalyst in a heating ramp. The Pt/SiC catalyst (Fig. 1c) shows light-off around 220~ and NOx reduction in the same temperature range as for the other catalysts. The N20 formation maximises around 235~ and is of similar magnitude as for the Pt/ZSM-5 catalyst. The NO 2 formation rate increases rapidly around this temperature and shows a maximum at 320~ Adsorption of neither hydrocarbons nor NOn on the Pt/SiC sample is obvious from Fig lc. In Table 2 the NOn reduction efficiency and the selectivity towards N2 and N20 formation are summarised for the flow reactor experiments. Both the NOn reduction activity and the N2 selectivity of the Pt/SiC and Pt/ZSM-5 system appear to be similar, while Pt/A1203 shows a higher peak reduction value for the heating ramp experiments.

289 3. RESULTS 3.1. Flow reactor studies

3.1.1. NOx reduction-Activity and selectivity In the flow reactor study, the NO reduction activities and the N2 and N20 selectivities of the respective powder samples were investigated both by heating ramp (increasing temperatures) and steady-state experiments. The results of the heating ramp experiments are displayed in Figures 1a to 1c, where the outlet concentrations of NO, NO 2, N20 and CO2 are shown as functions of ramp temperature. Included is also the nitrogen balance, calculated as the sum of the concentrations of all detected nitrogen containing species, ([NO] + [NO2] + 2[N20]), and referred to as "Nbal". The by-pass value of all N containing species should be about 405 ppm and deviation from this value is due to adsorption and desorption phenomena and/or N2 formation by NOx reduction. No traces of other nitrogen containing compounds as, e.g., ammonia were detected. 3500 N

400

b

a

l

~

3000 O

0

E

.....

300

_~ i ~ ~ t . - ' ~ -

=~ ==.,-=..-~.=~HNN~,~ 2500 C) O

2000 o O

tO

tO O tO

O

200

m

1500 ~~ ~ 0

1000 ~.

100

500 100

200

300

400

"1:3 "13

3

500

Temperature (~

Figure 1a. Concentration traces over a Pt/A1203 catalyst in a heating ramp. The Pt/A120 3 catalyst (Fig. la) reaches 50% conversion of propene, around 230~ In connection to the light-off, there is a NOx reduction window between approximately 180 and 320~ The formation of N20 has a maximum around 260~ The NO2 formation proceeds above this temperature with a maximum around 370~ Around 150~ the Nbal level is higher than the inlet value of NO indicating desorption of NO adsorbed at lower temperatures. For the Pt/ZSM-5 catalyst (Fig. l b) there is light-off at somewhat lower temperature (210~ and a significant over-shoot in the CO 2 formation just above light-off. This behaviour is probably connected with combustion of hydrocarbons adsorbed at lower temperatures. The NOx reduction window occurs around the same temperatures as for Pt/A1203 but is less pronounced. The maximum in N20 formation is somewhat higher in magnitude than for Pt/A1203 and occurs at a lower temperature (230~ The NO2 formation starts at about this temperature and has a maximum around 340~ There is no desorption of adsorbed NOx below light-off.

290 Table 2. Maximum NOx reduction activity and selectivity (at the temperature for maximum reduction) during heating ramp and steady-state experiments, for a feed containing 405 ppm NO, 10% 02 and 9!1 ppm C3H6 at a flow of 420 ml/min. Sample Heating ramp experiments Steady-state experiments NOx Corresp. Yield [%] NO• Yield [%] red. [%] temp [~ N2 N20 red. [%] N2 N20 Pt/siC 56.1 232 20.8 35.3 50.9 19.2 31.7 Pt/A1203 85.8 250 61.3 24.5 53.1 35.8 17.3 Pt/ZSM-5 58.6 225 23.3 35.3 61.0 30.9 30.1 ii

iiii

i

i

i

i i

Interesting is that the increased reduction activity is accompanied by a high N 2 selectivity. The lowest temperature for maximum reduction is observed for the Pt/ZSM-5 system, followed by Pt/SiC and Pt/AI203. 3.1.2. NO2 formation Comparison of the NO2 formation for the investigated materials, as displayed in Figures la to lc, shows a difference between Pt/SiC on one hand, and Pt/AI203 and Pt/ZSM-5 on the other. The former system shows a fast increase in NO2 formation in the temperature interval of maximum reduction compared with the A1203 and ZSM-5 supported systems. Note the corresponding decrease in the NO signal. 3.1.3. Adsorption of reactants Differences in adsorption behaviour are observed for the investigated systems. No NO or NO2 desorption peaks are observed for Pt/SiC or Pt/ZSM-5, while a clear desorption of NO, with a maximum at 158~ can be observed for Pt/A1203. The Pt/SiC system is also inert towards hydrocarbon adsorption, while Pt/ZSM-5 adsorbs a substantial amount of hydrocarbons. It can be observed that for all tested catalysts, the CO2 formation and the NOx reduction are closely correlated: the maximum in NOx reduction is observed at almost complete hydrocarbon oxidation. 3.1.4. NOx reduction under steady-state conditions When the three materials are tested under steady-state conditions, a different picture is obtained. The results of the NOx reduction activity and the N2 and N20 selectivity, under steady-state conditions, are included in Table 2. The maximum NOx reduction activity and the selectivity towards N2 for the Pt/AI203 system are lower than in the heating ramp experiment. Under steady-state cbnditions, the Pt/ZSM-5 system shows the highest overall NO reduction activity and a somewhat higher selectivity towards N 2 compared with the heating ramp experiments. For Pt/SiC there are no significant differences between the two types of experiments. It was observed that also under steady-state the maximum NOx reduction coincides with almost complete CO2 conversion.

291 3.2. FTIR studies Since all samples only contain small amounts of platinum, we do not expect that absorption of IR radiation due to species adsorbed on platinum sites is strong enough to be detected in the measurements. Thus the absorption bands seen are likely to be connected with molecules adsorbed on the support. Figure 2a shows the spectra of Pt/A1203 when exposed to the reactant gas (see above) at different temperatures. A double band at 2235-2255 crn"1 can be seen at temperatures up to 250~ This feature can be ascribed to isocyanate (-NCO) adsorbed on the support [9-10]. Several bands in the 1200-1700 cm "l region are also observed. Three of them, 1465, 1575 and 1660 cm l , are attributed to carbonate species on the support [11-12]. The remaining bands are attributed to disparate nitrate, nitrite and nitro groups adsorbed on the support [13]. These bands can be observed together with the isocyanate bands up to 250~ The only bands that can be detected above this temperature, are the carbonate bands at 1465 and 1575 cm "l. The interpretation is obvious. Above 250~ the catalyst is ignited and as the reactants are consumed their coverages decrease. The CO2 that is formed is on the other hand still available for adsorption. Below 250~ the catalyst is not ignited and the reactants remain on the substrate.

0.30

Pt/AI20 3

0.25 oc -

0.20

'O

0.15

S~ - H C

r~ = k~c~cO,,~

(1)

S l - H C --> S~ + H C

r2 = kEOl,nc

(2)

S~ - H C --> S~ - H C *

r 3 = k3OI,Hc

(3)

r4 = k4Ol,ttc,

(4)

r5 = k s c Nox O...c.

(5)

S1

-

-

HC* + 0 2

--> S 1 dl- C O 2 -~- HEO

S~ - H C * + N O x --~ S~ + C O z + 1 1 2 0 + N 2

When fitting this mechanism only, large residuals were attained for the hydrocarbon concentration. In order to obtain a better fit a second site with hydrocarbon adsorption and oxidation was introduced. It consisted of the following steps: S 2 + H C ---) S 2 - H C

r 6 ---- k6Cl_1CO2,v

(6)

S 2 - H C --) S 2 + H C

r7 = k702.,c

(7)

S 2 - HC + 02 --, S 2 + CO z + H 20

r8 = k802,,c

(8)

The monolith was modeled with a one dimensional model. The following simplifications have been made in the model: a) b) c) d) e) f)

uniform radial flow distribution negligible radial temperature and concentration profiles no axial diffusion or heat conduction no gas phase accumulation no diffusion resistance in the washcoat transfer of mass and energy between gas and solid is accounted for by coefficients derived from the correlation obtained by Tronconi and Forzatti [7] g) the monolith is treated as a series of continuously stirred tank reactors

323

The following equations were used to model a differential axial monolith segment: Gas mass balance: (9)

Fi,~_ l - Fi, k - k ~ A k (Cg,i,k -- Cs,i, k ) -- 0

Surface mass balance:

kcAk (cg.~.k --C.,..,,k) = ~_~ vi,,,r, mwc,k

(lO)

n

Gas energy balance: F~c p,i ( T~ ,k-, - T~ ,~, ) - hA~, ( T~ ,k - ~,k ) =0

(12)

i

Solid energy balance:

" Ot - hAk (T~'k - ~"k ) + ~-'r"mc'k ( - A H " ) - k f A[ (T"'k - T")

(13)

n

The preexponential factors and the activation energies of the reactions were fitted to the experimental data of the second run of the test cycle. The values of these parameters can be found in table 2.

Table 2. Parameters obtained from fitting the model to experimental data. Reaction Preexponential factor Activation energy number (k J/mole) 1 7.8 x 10 ~ m 3 kg cat. l s l 14 2 9.4 x 101 mole kg cat. -l s -l 51 3 2.4 x 103 mole kg cat. -l s l 69 4 4.0 x 104 mole kg cat. ~ s -I 71 5 2.7 x 108 m 3 kg cat. -l s ~ 97 6 8.4 x 10 3 m 3 kg cat. l s -l 32 7 1.8 x 101 mole kg cat. l s l 26 8 3.9 x 102 mole kg cat. l s l 60

Figure 4 shows the observed and the simulated hydrocarbon concentration at the catalyst outlet during the second run of the test cycle. The standard deviation for the residual is 118 ppm or 18% of the mean HC concentration. The modeled concentration follows the observed

324

with some exceptions. During injection in the 0-10 minutes' interval and during the second injection in the 20-30 minutes' interval the model predicts too low conversion. In the 10-20 minutes' interval the model predicts too low conversion between the injections. There is also a slighter deviation in the conversion during the first and second injection in the 10-20 minutes' and in the 30-40 minutes' interval. There are also deviations at the flanks of the hydrocarbon transients in the 40-50 and the 50-60 minutes' interval. One explanation to the deviations could be that the model treats all hydrocarbon as a single compound. This is a coarse simplification. The diesel fuel itself consists of a variety of larger hydrocarbons which are cracked into shorter ones in the catalyst. All these different hydrocarbons have different adsorption properties and reactivities. An improved model would need to distinguish between different hydrocarbons or at least groups of them. Another improvement would be to account for variations in the oxygen concentration or even include oxygen adsorption in the model.

400

200 E 4000 El. Ex co

3000

c

2000

o

1000

8 0 -r"

20

30

40

50

E

o. (D.

0

-~

-200

32 tO o n,

-400

10

.-.

60

Time (rain)

Figure 4. Observed and simulated hydrocarbon concentration at catalyst outlet during the second run of the test cycle. The simulated curve is offset by 2000 ppm.

Figure 5 shows the observed and the simulated NOx concentration at the catalyst outlet during the second run of the test cycle. The standard deviation for the residual is 33 ppm or 8% of the mean NOx concentration. The model manages to predict the NOx conversion that onsets before diesel injection in the beginning of the 20-30 minutes' and the 50-60 minutes' interval. During injection in the 0-10 minutes' interval and during the second injection in the 20-30 minutes' interval the model predicts too low conversion. This coincides with a predicted too low hydrocarbon conversion. The model also predicts a slightly too low conversion at the beginning and a slightly too high conversion at the end of the 40-50 minutes' interval. In the 50-60 minutes' interval the model deviates at the flanks of the hydrocarbon transients. There are also deviations at the end of each 10 minutes' interval, when the NOx inlet concentration is changed. These deviations could indicate that NOx adsorption and desorption occurs. The agreement between the modeled and the observed NOx concentration is to a large extent influenced by the deviations between the observed and the

325 modeled hydrocarbon concentration. An improvement of the model's ability to predict the hydrocarbon concentration would probably result in better predictions of the NOx concentration. Another improvement would be to distinguish between NO and NO2.

lOO AE 9

50

0

E

n Q.

v

t-

-50 "~

1200

-100

O L_

G) 0 tO

o X

m "10

n,

(---- Simulated

800

(---- Observed

400

o Z

0 0

10

20

30

40

50

60

Time (rain)

Figure 5. Observed and simulated NOx concentration at catalyst outlet during the second run of the test cycle. The simulated curve is offset by 400 ppm.

4. CONCLUSIONS It has been concluded that the reduction of NOx on a high temperature catalyst proceeds via the formation of a hydrocarbon intermediate and the successive reaction between the hydrocarbon intermediate and NOx. When this reaction mechanism was modeled many features of the catalyst behaviour were reproduced.

5. N O M E N C L A T U R E A AP

c Cp Cp F h kc kf

mwc ms N

Channel wall area in monolith Peripheral area of monolith Gas concentration Gas heat capacity Solid heat capacity Molar flow Heat transfer coefficient Mass transfer coefficient Heat loss coefficient Mass of washcoat Mass of solid Number of sites

m2 m2

mole/s J/mole K J/kg K mole/s W/m E K m/s W/m 2 K kg kg mole/kg

326 r t Ta Tg Ts -All 0 v

Reaction rate Time Ambient temperature Gas temperature Solid temperature Heat of reaction Coverage Stoichiometric coefficient

mole/s kg washcoat S

K K K J/mole

Index: i j k n v

Specie number Site number Section of monolith Reaction number Vacant site

ACKNOWLEGDEMENTS The authors would like to thank: Johnson Matthey for supplying the catalysts for this study. AB Volvo for providing admittance to their engine laboratory. Bengt Cyr6n and Martin Bruszt for technical assistance. NUTEK for financial support.

REFERENCES

1. S.L. Andersson, P.L.T. Gabrielsson and C.U.I Odenbrand, AIChE J., 40(11) (1994) 1911. 2. L. Andersson., "Mathematical Modeling in Catalytic Automotive Pollution Control", Ph.D. thesis, Department of Chemical Reaction Engineering, Chalmers University of Technology, Sweden, 1995. 3. C. Havenith, R.P. Verbeek, D.M. Heaton and P. van Sloten, SAE Technical Paper Series 952652 (1995). 4. K.M. Adams, J.V Cavataio and R.H. Hammerle, Appl. Catal. B, 10 (1996) 157. 5. G.P. Ansell, A.F. Diwell, S.E. Golunski, J.W. Hayes, R.R. Rajaram, T.J. Truex and A.P. Walker, Appl. Catal. N, 2 (1993) 81. 6. C.J. Bennet, P.S. Bennet, S.E. Golunski; J.W. Hayes and A.P. Walker, Appl. Catal. A, 86 (1992) L1. 7. E. Tronconi and P. Forzatti, AIChE J., 38(2) (1992) 201.

CATALYSIS AND AUTOMOTIVEPOLLUTIONCONTROLIV Studies in Surface Science and Catalysis,Vol. 116 N. Kruse, A. Frennet and J.-M Bastin (Eds.) 91998Elsevier Science B.V. All rightsreserved.

327

Steady state and transient activity patterns of Cu/ZSM-5 catalysts for the selective reduction of nitrogen oxides Jan Connerton and Richard W. Joyner Catalysis Research Laboratory, Department of Chemistry and Physics, The Nottingham Trent University, Clifton Lane, Nottingham NG11 8NS, UK.

The high activity for NOx reduction by hydrocarbons under lean conditions exhibited by ion exchanged copper/zeolite catalysts is well recognised, as is the high selectivity to nitrogen. This paper reports our continuing studies of this important catalyst system, discussing both steady state and transient kinetics, and relating these to the mechanism of reaction, and in particular the possible role of small, ionic copper clusters. We have studied the kinetics of reaction and determined the turnover numbers of a series of otherwise identical catalysts with different copper contents. The turnover number is roughly constant at copper contents < ca 90% exchange, and then increases by about a factor of two at 100% exchange, remaining constant up to the highest nominal extent of exchange studied, ca 160%. These results suggest that both isolated copper ions and small metal/oxygen clusters, including dimers, catalyse the SCR reaction, with the dimeric species being roughly twice as active per copper/on. In earlier studies we have drawn attention to the possible catalytic importance of hydrocarbon deposits, which can be formed on the catalysts at temperatures of ca 600 K or lower, and we now report non-steady-state results for catalysts with different copper loadings. For each catalyst it was found that a similar amount of reactive deposit was formed, and that it decomposed NO according to rather straightforward kinetics. Significant differences in reactivity were however noted, with the rate of reaction reaching a maximum at 680 K on catalysts with 100% degree of copper exchange, compared to almost 100 K higher for 54% exchanged materials. These results show a variation in the stabifity of the hydrocarbon deposit towards NO, suggesting that the acidity of the host zeolite may be one of the factors determining the way in which catalytic activity varies with copper loading. 1. INTRODUCTION The exceptional activity exhibited by ion-exchanged copper ZSM-5 zeolite catalysts for nitric oxide (NO) decomposition, and for the selective catalytic reduction (SCR) of nitrogen oxides (NOx) in the presence of excess oxygen is well documented [ 1-10]. The nature of the active copper species in the SCR reaction however still remains uncertain. We and others have recognised that there are two different types of copper species within the ZSM-5 zeolite channels [ 11]. Isolated copper ions exist in low symmetry environments, and small clusters, where the copper atoms are linked by extra-lattice oxygen species such as [Cu(II)-O-Cu(II)] 2+ dimers, are also present. Recent studies have also suggested that the isolated copper ions in ZSM-5 occupy two types of sites [ 11], which may have different SCR reactivity. It is likely

328 ~:hat the relative concentration of different isolated copper sites and dimers vary with the zopper content of the zeolite, so this work examines the relationship between the copper zontent of a series of catalysts and the turnover number of the SCR reaction. Dimeric and ~ther copper clusters are likely to be formed preferentially at high copper loadings, so if they are catalytically important we expect to see turnover numbers increasing with copper content. We note that Moretti et al have shown that for NO decomposition the turnover number versus loading curve has an S shape, with a very rapid increase above 2 wt % copper [12 - 14]. In studies of the SCR reaction mechanism much interest now centres on the nitrogen nitrogen bond forming step, with spectroscopic evidence emerging both for oxidised nitrogenous species such as nitro groups [15 - 17] and also for reduced species, including cyanides [17 - 19]. Less attention has been paid to the characterisation of the hydrocarbon moieties which must also be involved, although it is recognised that long lived carbonaceous species can be formed within the pores of the zeolite, with possible mechanistic importance, as discussed by ourselves and others [16,20]. ~ Here we also report on the reactivity of hydrocarbon species deliberately deposited on catalysts with different copper contents. We have previously shown that nitrogen is released from these deposits by reaction with oxygen [ 16], and now show that these deposits are also able to activate NO and oxygen directly.

2. EXPERIMENTAL Catalysts were prepared from H-ZSM-5 obtained from Catal Intemational, Sheffield, with a framework Si/A1 ratio of 25, using conventional ion-exchange techniques [11]. Copper exchange levels ranged between 54% and 160%. Catalytic experiments were carried out using a fixed bed microreactor which has been described previously [11]. Product analysis was performed by a chemiluminescent NOx analyser (Signal Instruments Model 4000) and a gas chromatograph fitted with a thermal conductivity detector (Pye UNICAM PU 4550). All catalysts were activated at 773 K for 1 hour under a stream of 2% oxygen/helium. Steady state rates of reaction were measured over the temperature range 773 - 473 K, using a reactant gas mixture of 2000 ppm NO, 1220 ppm propene and 2% oxygen, balance helium, at a GHSV of 30,000 h l. Turnover numbers are expressed as molecules of NOx (NO + NO2) converted to dinitrogen per copper atom per second. Before studying the reactivity of the carbonaceous deposit, the reactant mixture above was passed over the catalyst, a procedure known to form hydrocarbon or coke deposits [20, 16]. The catalyst was then heated to 773 - 823 K at 10 K min 1 in a mixture of NO (2000 ppm), 02 (2%) balance helium. The extent of NOx conversion was monitored continuously by the chemiluminescent analyser. 3. RESULTS AND DISCUSSION 3.1. Turnover number studies The Cu/ZSM-5 catalysts showed the expected performance in the SCR reaction, as indicated typically in Fig. 1. Selectivity is entirely to dinitrogen, with no nitrous oxide observed at the sensitivity of our GC analysis.

329 !00

9

90-8070=

60--

I,,,,,,,

50-

>

40

o

o

o

30

--

20

-

10

-

0 :z 473

573 Temperature

673

773

/ K

Figure 1. Steady state rate of the SCR reaction with a 164% exchanged Cu/ZSM-5 catalysts, tested under the conditions described in the text. Squares, conversion of NOx to dinitrogen; Diamonds, conversion of propene to carbon dioxide.

Turnover numbers have been determined for the series of catalysts with difference copper contents at 573,598 and 623 K, and the results are shown in Fig. 2. Nitric oxide conversions in these studies range from 3%, for the lowest copper content at the lowest temperature at which turnover numbers are shown, to 45%, observed with the highest copper loading at the highest temperature shown in Figure 2. Despite this considerable range of conversion, a consistent picture emerges, indicating two turnover regimes. At each of the three temperatures studied, the turnover number is approximately constant up to about 90% copper exchange. Above this degree of exchange, the turnover number increases, approximately doubling at the two higher temperatures studied, and then remains constant as the copper content increases further. The decline observed at high degrees of exchange and 623 K is due to the onset of excessive propene oxidation. The simplest interpretation of these results is that two copper species, isolated ions and small clusters, are both active in the SCR reaction. Although more complicated explanations can be envisaged, the approximate doubling of the turnover number between 90% and 100% copper exchange suggests that in this range of copper loading new types of copper entity are created, with greater unit activity than is found in the zeolite at lower copper loading. The constancy of the turnover number at low copper contents suggests that the active species are isolated copper ions, which EXAFS and other techniques show to be present [ 11 ]. A recent careful study by Lamberti et al [21], using a range of techniques including X-ray absorption spectroscopy and IR spectroscopy of adsorbed CO, has suggested that introduction of Cu[I] into a ZSM-5 with a very low aluminium content, from vapour phase copper [I] chloride, generates isolated copper ions in two different sites, as also suggested by others

330 [22]. The present results suggest that these different isolated sites are probably of similar activity in the SCR reaction.

0.002

:~ 0.001

g

E

?.

9

_ _

50

. . . .

'

'

70

90

_

.

_

,. . . . .

.

:

110

....

130

150

.

170

Degree of Exchange / % Figure 2. Turnover number (NO molecules per Cu atom per second) measured as a function of copper content in Cu/ZSM-5 SCR catalysts at 573 K, (triangles); 598 K (squares); and 623 K, (diamonds).

As the copper content is increased, the turnover number doubles, and then becomes constant in the exchange range ca 100 = 165%. This constancy is very significant, as it suggests that there are only two types of active copper entity, and that these continue to increase in concentration up to a maximum copper content studied. Since the catalysts are prepared by exchange and not by impregnation, this indicates that the second type of active species is a dimer or other small metal/oxygen cluster, and not simply isolated copper ions in a different type of site. Stoichiometry dictates that degrees of exchange above 100% cannot be achieved by isolated Cu(II) ions (irrespective of the nature of the site occupied), but must instead involve the formation of metal/oxygen clusters such as [Cu- O - Cu] 2+. The entities which are introduced into the catalyst by copper overexchange are thus dimers or other small dusters and these must therefore by the species with the higher turnover number. It is interesting to note that, since the turnover number is calculated per copper ion, the doubling indicates that the activity of a dimer cluster is about four times that of an isolated copper ion.

331 Our results for the SCR reaction are similar to but much less dramatic than those reported by Moretti et al [ 12 - 14] for the NO decomposition reaction. These authors observed an Sshaped relationship between activity and copper content. As in this study, three regions of activity were reported, below ca 80% exchange - where the catalyst activity was negligible, between 80 and 100% exchange - where the catalyst activity increased by nearly a 100 fold and above ca 100% exchange, where the activity again remained almost constant. From this it was concluded that not all of the copper sites are equivalent in their NO decomposition activity, and that the high catalytic activity of Cu-ZSM-5 is due to the very last fraction of copper exchanged in the zeolite framework (20% of the total copper at most) immediately below 100% exchange. Li and Hall had earlier reported a two fold increase in activity for NO decomposition as the copper exchange level was increased form 76 to 166% [4].

3.2. REACTIVITY OF CARBONACEOUS DEPOSITS We now report on the reactivity of carbonaceous deposits which can be laid down on the Cu/ZSM-5 catalysts by exposure to the reaction mixture at low temperatures. As in our earlier study [16], carbonaceous material referred to for simplicity as coke was deposited by exposing the catalysts to the reaction mixture at 473 K. The reactivity of material deposited was then examined by exposing the catalyst at 473 K to the same concentration of NO and oxygen as present in the reaction mixture, (but no propene), and then heating to ca 920 K at 10 K min ~. The conversion of NO was monitored by the chemiluminescent detector, which has a response time of < 1 s. Fig. 3 shows the results of ramping experiments carried out on two catalysts, with respectively 54% and 100% copper exchange. The shapes of the curves are similar, as is the total amount of nitric oxide converted, in each case corresponding to a minimum of 25 mg g-l catalyst of coke. This amount is also close to our previous observation of 30 mg of carbonaceous deposit per gram of catalysts [16]. The reactivity is, however, significantly different for the two catalysts studied here, with maximum reactivity being observed at a much lower temperature at the higher degree of exchange. A simple kinetic model has been formulated in an effort to understand the differences in reactivity between catalysts with different degrees of copper exchange. The model assumes: 1. That the rate of reaction is first order in the amount of hydrocarbon deposit remaining on the catalyst as the temperature is increased: 2. That the rate of reaction is first order in the NO concentration, taken to be the mean of the NO concentration on entering and leaving the catalyst bed: 3. That the reaction is zero order in oxygen, since this is present in substantial excess at all stages of the experiment: 4. That the reaction is described by a single activation energy. 5. That the rate of reaction may be taken as constant over a 5 K interval, and that the rate at any temperature is obtained by numerical integration over all of the 5 K intervals up to that temperature, from the initial temperature of 473 K. The model has 3 disposable parameters, the activation energy for reaction, and two constants which have the nature of pre-exponential factors. One of these normalises the calculated NO concentrations to those which are observed, while the other relates to the consumption of the hydrocarbon deposit.

332

2500 T I

2OOO -

E

1500 -

X

0 z

1000 -

500 -

0

03 l'.-,q,-

03 ("4 t..O

03 1",.t..O

03 ("4 (,.0

03 I".-r

03 t'4 P'-.-

I

03 r~

I

O3 t"N O0

I

cO !".-.. oo

I

I

Temperature/K

2 5 0 0 --

2000

E 1500 X

0 z

1000 I

500

,

03 r'-~1"

I

03 04 If)

!'

I

03 r'-LO

r

I r

I 1'~

I ~

I-

I::

o3

TemperaturelK

Figure 3. Observed and calculated NOx concentrations during heating in NO - oxygen after formation of a hydrocarbon deposit, as described in the text. A) 54% Cu exchanged catalyst: B) 100% Cu exchanged catalyst. Squares, experimental: Circles, calculated.

333 As well as the experimental results, Fig. 3 shows the NO concentrations calculated from the model, and very reasonable agreement can be seen given the simplicity of the modelling approach. The model indicates that the greater activity of the catalyst with the higher copper content is mainly due to a lower activation energy, 82 kJ mol "1, compared with 92 kJ mollfor the catalyst with lower copper content. The model also shows that all of the hydrocarbon deposit is consumed during each of the ramping experiments. These results show a variation in the stability of the hydrocarbon deposit towards NO, which is inversely dependent on copper content. This suggests that the residual acidity of the zeolite could be a factor in determining how catalytic activity varies with copper loading. The results imply that having too many Bronsted zeolite acid sites inhibits the NOx reduction reaction. We have shown previously that relatively few Bronsted acid sites remain in the fully exchanged catalysts which are most active for the NOx reduction reaction [23], and which appear to form the most active coke. The present study compares a catalyst of low degree of exchange with one having little residual Bronsted acidity. Where these materials are pre-treated under conditions which allow the acid sites to have maximum influence in hydrocarbon activation, namely at low temperature, the result on the more acid catalyst is a hydrocarbon deposit which is less active in NOx reduction.

ACKNOWLEDGEMENTS We are grateful to Johnson Matthey PLC for their support of this work, through a PhD studentship granted to JC. We also acknowledge very useful discussions with Drs. Jack Frost, Alan Diwell, Raj Rajaram, Janet Fisher and Andy Walker of Johnson Matthey, as well as Dr. Olga Tkachenko of the Zelinsky Institute of Organic Chemistry, Moscow, who also prepared some of the samples used in this study. Experimental assistance was provided by Messrs. Chris Angell and Tim Shaw as part of their MSc projects. The late Professor Efim Shpiro also contributed much to this programme of study. REFERENCES ~

2. 3. 4. 5. 6. 7. 8.

9. 10. 11.

12.

M. Iwamoto, Stud.Surf.Sci.Catal.,54 (1990) 121. M. Iwamoto, H. Yahiro, S. Shundo, Y. Yu-u and M. Mizuno, Appl.Catal., 69 (1991) L 15. M. Iwamoto and H. Hamada, Catal. Today 10 (1991) 57. L. Li and W.K. Hall, J Catal., 129 (1991) 202. J. Valyon and W.K. Hall, J.Phys. Chem., 97 (1993) 1204. W. Held, A. Koenig, T. Richter and 1 Pupper, SAE Tech. Pap. Ser., 1990 900496 M. Iwamoto, H. Yahiro, Y. Mine, S. Kagawa, Chem. Lett., 1989, pp. 213. C.N. Montreuil and M. Shelef, Appl. Catal., B 1 (1992) L 1. J.O. Petunchi and W.K. Hall, Appl. Catal., B2 (1993) 303. K.C.C. Kharas, Appl. Catal. B, 2 (1993) 207. W. Griinert, N.W. Hayes, R.W. Joyner, E.S. Shpiro, M.R.H. Siddiqui and G.N. Baeva, J. Phys. Chem., 98 (1994) 10,832; B. Wichterlova, J. Dedecek, Z. Sobalik, A. Vondrova and K. Klier, J. Catal., 169 (1997) 194. G. Moretti, Catal. Lett., 23 (1994) 135.

334 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23.

M.C. Campa, V. Indovina, G. Minelli, G. Moretti, I. Pettiti, P. Porta and A. Riccio, Catal. Lett., 23 (1994) 141. G. Morretti, Catal. Lett., 28 (1994) 143. T.Tanaka, T. Okuhara and M. Misono, Appl. Catal. B, 4(1994) L 1. N.W. Hayes, W. Griinert, G.J. Hutchings, R.W. Joyner and E.S. Shpiro, JCS Chem. Commun., 1994 pp 531. N.W. Hayes, R.W. Joyner and E.S. Shpiro, Appl. Catal. B, 8 (1996) 343. F. Radtke, R.A. Koeppel and A. Baiker, JCS Chem.Commun., 1995, pp 427. A.W. Aylor, L.J. Lobree, J.A. Reimer and A.T. Bell, Stud. Surf. Sci. Catal., 101 (1996) 661. G.P. Ansell, A.F. Diwell, S.E. Golunski, N.W. Hayes, R.R. Rajaram, T.J. Truex and A.P. Walker, Appl. Catal. B, 2 (1993) 81. C. Lamberti, S. Bordiga, M. Salvalaggio, G. Spoto, A. Zecchna, F. Geobaldo, G.Vlaic and M. Belltreccia, J. Phys. Chem., 101 (1997) 344. I.C. Hwang, D.H. Kim and S.I. Woo, Catal. Lett., 42 (1996) 177. J. Connerton, M.B. Padley and R.W. Joyner, JCS Faraday Trans., 91 (1995) 184

CATALYSIS AND AUTOMOTIVE POLLUTIONCONTROLIV Studies in Surface Science and Catalysis,Vol. 116 N. Kruse, A. Frennet and J.-M Bastin (Eds.) 91998 Elsevier Science B.V. All rights reserved.

335

Selective reduction of nitrogen oxide with hydrocarbons and hydrothermal aging of Cu-ZSM-5 catalysts P. Denton a, Z. Chajar a, N. Bainier-Davias b, M. Chevrierc, C. Gauthier d, H. Praliaud a, M. Primet a aLaboratoire d'Application de la Chimie ~ l'Environnement, Unit6 Mixte CNRS-UCB n~ Universit6 Claude Bernard Lyon I, 43 Bd du 11 novembre 1918, 69622 Villeurbanne Cedex, France, Tel (33) 4 72 43 15 87, Fax (33) 4 78 94 19 95. bRenault Automobiles, Direction de la Recherche, 9-11 Av. du 18 Juin 1940, 92500 Rueil Malmaison, France ~ Automobiles, Direction de l'Ing6nierie Mat6riaux, 8-10 Av. Emile Zola, 92109 Boulogne Billancourt Cedex, France dRenault Automobiles, Centre de Lardy, 1 All6e Comuel, 92510 Lardy, France

ABSTRACT This paper deals with the hydrothermal deactivation, under an air + 10 vol. % H20 mixture between 923 and 1173 K, of Cu-MFI solids, catalysts for the selective reduction of NO by propane. Fresh and aged solids were characterized by various techniques and compared with a parent H-ZSM-5 solid. The catalytic activities were measured in the absence and in the presence of water. The differences between fresh and aged Cu-ZSM-5 catalysts (destruction of the framework, extent of dealumination...) were shown to be small in spite of the strong decreases in activity. Cu-ZSM-5 is more resistant to dealumination than the parent H-ZSM-5 zeolite. The rate of NO reduction into N2 increases with the number of isolated Cu2+/Cu+ ions. These isolated ions partially migrate to inaccessible sites upon hydrothermal treatments. At very high aging temperatures a part of the copper ions agglomerates into CuO particles accessible to CO, but these bulk oxides are inactive. Under catalytic conditions and in the presence of water, dealumination is observed at a lower temperature (873 K) than under the (air + 10 % H20) mixture, because of nitric acid formation linked to NO2 which is either formed in the pipes of the apparatus or on the catalyst itself. 1. INTRODUCTION The selective reduction of NO by hydrocarbons in oxygen rich atmosphere has been reported for zeolite-based catalysts, especially Cu-ZSM-5 solids (1), but their low thermal stability limits their use for treatment of emissions from Diesel and lean-burn engines. An

336 understanding of the deactivation of Cu-ZSM-5 solids could, however, facilitate the improvement of their stability and the search for a more durable catalyst. A number of causes of deactivation have been invoked: structural collapse, dealumination (2), partial dealumination with the corresponding loss of exchangeable sites (3, 4), agglomeration of copper ions with the formation of CuO clusters (5, 6), migration of copper ions to inaccessible sites (7) and change in the nature of Cu species (8), especially change in coordination (9). In this work the catalytic activities of fresh and hydrothermally treated Cu-MFI solids (Si/Al = 19, 78, 130, 151, 319) are measured in the absence and in the presence of water. Fresh and aged Cu-MFI catalysts and a parent H-ZSM-5 solid are characterized by various techniques in order to understand the modifications of the copper ions and of the zeolite itself, as well as the relationship between these modifications. 2. EXPERIMENTAL The main starting material was a commercial H-ZSM-5 zeolite from Degussa with a Si/A1 ratio of 19 (1-14.8A14.8Si91.2O192) (crystals of size inferior to 0.6 ~tm with some spherical aggregates of diameters ranging between 1.0 and 4.0 ~tm). Copper was introduced by conventional ion exchange with an aqueous solution of copper nitrate. After calcination under an oxygen flow at 773 K (heating rate 1 K min1) the solids contained 1.2, 1.77 and 1.96 wt. % Cu. If one assumes that one Cu 2§ ion replaces two protons, 1.2 wt % Cu corresponds to an exchange level of 44 %. For comparison some other MFI zeolites with high Si/A1 ratios (78, 130, 151, 319), either commercially available or hydrothermally synthesized, were also considered. In those cases copper (1 or 4 wt % atter calcination) was introduced by impregnation. All the prepared solids showed the XRD patterns of the parent zeolite and the CuO phase was not detected. Thermal treatments were performed at 923, 973, 1073 and 1173 K in the presence of water (10 vol. % H20 in air) for 24 hours (total flow rate 10 1 h-1for 5 g solid). Catalytic measurements were made using 100 mg catalyst diluted with 400 mg of inactive otA1203 in a fixed-bed flow reactor. The typical gas mixture consisted of 2000 vpm NO, 2000 vpm C3Hs, 10 vol. % O2, balance He, without or with 10 vol. % water (total flow rate 10 1 h-l). In the absence of water in the reactant mixture, the temperature was increased from 300 to 773 K (or 873 K) (heating rate 2 K min-1) and then decreased to 423 K. Water was then added at 423 K and the temperature increased and decreased again as above. The analysis was performed by gas chromatography with two columns (porapak and molecular sieve) and a TCD detector for CO2, N20, 02, N2 and CO, and with a porapak column and a flame ionisation detector for hydrocarbons. Moreover, on-line IR and UV analyzers were used for NO, NO2, CO2, and N20 analysis. The NO conversion was calculated from the N2 production and the nitrogen balance was checked. The solids, particularly the 1.77 %-Cu-H-ZSM-5(19) and the parent H-ZSM-5, were characterized before and alter steam treatments by various techniques: powder X-ray diffraction (Siemens diffractometer with CuKcx radiation), SEM (Hitachi $800 with a 10 nm resolution), N2 adsorption (BET and pore volume with a laboratory-made automatic apparatus), FTIR spectroscopy (framework vibrations with KBR dilution and CO probe molecule in an in situ cell), 29Si and 27A1MAS NMR (BRUKER DSX 400).

337 For FTIR spectroscopy of adsorbed CO the calcined samples (thin discs of known weight) were degassed at 773 K for 1 hour and the background spectrum was recorded after cooling at 300K. After introduction of CO (around 50 Torr) at 300 K as well as after evacuation at 300 K the spectra were recorded as a function of time, as already described (10, 11, 12). Spectra were recorded on a Nicolet 550 spectrometer (2 cm-1 resolution). The optical densities of the bands were normalized by taking into account the amount of copper. 3. RESULTS

3.1.Catalytic activity The main products are N2, CO2 and NO2. The formation of CO and NzO is negligible. In the presence of oxygen, NO2 is formed at 298 K in the pipes of the apparatus (13). The fresh Cu-H-ZSM-5 (Si/AI = 19) solids are active for SCR in the presence of excess 02. In the absence of water in the mixture, no deactivation is observed upon increasing and decreasing the temperature, whether the temperature reached is 773 or 873 K (Fig. 1, Table 1). Upon addition of water to the stream the activity of the fresh Cu-H-ZSM-5(19) solids clearly decreases (Figure 1) but this effect is fully reversible if the water is suppressed and if the temperature does not exceed 773 K, as already noticed (14). Most probably an competitive adsorption between H20, NO or C3Hs, is invoked. When the temperature reaches 873 K in the presence of water the catalyst is deactivated irreversibly (Table 1). Kharas et al (5) have also noticed that a working temperature of 873 K induces a deactivation contrarily to 773 K. Furthermore, Yan et al (15) have noticed that the deactivation is faster with a complete [hydrocarbon, NO, 02, H20] mixture than if a component is missing. In fact, we observed more deactivation under catalytic conditions and in the presence of water than under the air + 10 % H20 mixture at the same temperature (873 K). We will explain this phenomenon later on, in paragraph 3.2.

100

NO/N2 conversion ( % )

without H20 50

423

623

823

T(K)

Figure 1. Conversion of NO into N2 as a function of the temperature with the Cu(1.96 %)-HZSM-5(19) solid. The reaction is performed up to 773 K in the absence and in the presence of water in the feed. The arrows indicate the increase and decrease in the reaction temperature.

338

The activity strongly decreases when the aging temperature reaches 923 K (Table 1). When water is added to the mixture the activity decreases again but to a lesser extent than in the case of the fresh solid. With the solids previously aged at 923 and 973 K (air + 10 vol. % H20 mixture) an irreversible deactivation under the reactant mixture is still observed if the temperature reaches 873 K. After aging at 1073 or 1173 K the activity is very weak, even in the absence of water in the feed.

Table 1. Conversion of NO into N2 for the flesh and aged Cu(1.96 %)-H-ZSM-5(19) solids in the absence and in the presence of water in the feed. The reaction temperature reaches 773 K or 873 K. Ts0 light-off temperature in K (temperature to reach 50 % conversion). C0nv623, c0nv773, c0nv873: conversions at 623, 773, 873 K. Max: maximum NO conversion at T(K). Ts0 Max C0nv623 C0nv773 C0nv873 Fresh solid Up to 873 K without H20 with H20 ~a) suppression H20

598 _ 593

66 % at 638 K 30 % at 823 K 58 % at 633 K

63 2 56

58 35 30

Up to 773 K without H20 with H20 , suppression 1-120

593 593

70 % at 623 K 45 % at 683 K 68 % at 623 K

65 28 65

60 37 59

493 -

54 % at 810 K 40 % at 810 K 45 % at 813 K

20 8 15

50 36 44

23 15 18

-

24 % at 833 K 15 % at 773 K 20 % at 830 K

12 6 10,

21 15 18

10 8 9

38 13 17

Aged solids Up to 873 K Aged 923 K without H20 with H20 (a) suppression H20 Aged 973 K without H20 with HRO (") suppression HzO

_

~a) Because of the deactivation observed in the presence of water in the feed stream at 873 K, the values reported here are measured during the decrease in temperature.

339

3.2. Physicochemicai characterizations

3.2.1. After aging treatments at 923, 973, 1073 and 1173 K. The modifications of the Cu(1.77 %)-H-ZSM-5(19) solid upon aging have been compared to those of the parent H-ZSM-5 zeolite. We detect no significant modification of the SEM pictures and no modification of the X-ray diffractograms, i.e., no destruction of the zeolite framework and no loss of cristallinity, even after the treatment at 1173 K. However traces of CuO with X-ray peaks at 2.52 and 2.32 A are detected after aging at 1073 or 1173 K. The nitrogen adsorption isotherms are characteristic of microporous solids. The aging treatments cause a clear decrease in the micropore volume and in the microporous surface. The decreases are nevertheless smaller for the Cu-ZSM-5 solid (variation A = 0.04 after aging at 1173 K) than for the parent zeolite (A= 0.07) (Table 2). An apparent BET surface area has been reported though the BET theory is not applicable to microporous materials since the pore condensation isotherm is interfering with the multi-layer adsorption isotherm. Table 2. Apparent BET surface area (SBETin mE/g), micropore volume (laVol in ml/g) and microporous surface (laS in m2/g) for the fresh and aged H-ZSM-5(19) and Cu(1.77 %)-H-ZSM-5(19) solids Solid. . . . . SBEa' ........ ~Vo1 laS H-ZSM-5 fresh H-ZSM-5 aged 973 H-ZSM-5 aged 1073 H-ZSM-5 aged 1173 Cu-ZSM-5 fresh Cu-ZSM-5 aged 973 Cu-ZSM-5 aged 1073 Cu-ZSM-5 aged 1173

347 330 323 299 343 297 314 295

0.15 0.13 0.11 0.08 0.15 0.11 0.11 0.11 .

.

.

.

.

.

.

.

.

292 257 222 150 287 216 232 215

Table 3. Variations of the FTIR band at 1227-28 cml as a function of the treatments for the H-ZSM5(19) and Cu(1.77 %)-H-ZSM-5(19) solids. H-ZSM-5 Cu-ZSM-5 Fresh Aged 923 K Aged 973 K Aged 1073 K Aged 1173 K

1228 1227 1232 1235 1235

1227 1228 1230 1231 1230

340 It is known (16) that there is a linear relationship between the IR wavenumbers of the T-O vibrations of the zeolite lattice and the aluminum content. The aging treatments shift the FTIR bands towards higher wavenumbers which indicates a partial dealumination of the lattice (Table 3). The variations remain however small because of the low initial Si/A1 ratio (for the 1227-28 cm ~ band and after aging at 1173 K: Av = 7 cm~ for H-ZSM-5 and 4 cm ~ for CuZSM-5). Let us recall that, when the Si/AI ratio decreases from 319 to 19, the 1235 cm~ band shifts to 1220 cml (A= 15 cm-1). From the 2VAlM R spectroscopy it is possible to follow the amounts of lattice and extra-lattice A1 in the flesh and aged samples studying the signals corresponding to tetrahedral Td AI (at around -5 5 ppm) and octahedral Oh A1 (at around 0 ppm) (17).

7

H-ZSM-5

I

I

80

I

40

100

I

I

i

0

100

0 K

Cu-ZSM-5

,,I

80

,,

I

40

.....

S

I . . . . .

80

I

40

.

.

.

.

.

.

ppm

Figure 2.27A1 NMR signals for the flesh, 973 K-aged and 1073 K-aged H-ZSM-5 solids and for the fresh and 1073 K-aged Cu(1.77 %)-ZSM-5(19) solids.

341 Both the flesh samples show only the signal of T~ AI. After the hydrothermal treatments the signal of Oh extra-lattice A1 appears in the case of the parent H-ZSM-5 zeolite (Figure 2). The quantitative determination is not very accurate, but however, it may be noticed that the quantity of Oh AI reaches approximatively 30 % after aging at 1073 or 1173 K, instead of 0 for the flesh solid. The decrease in Td A1 is not really clear. With Cu(1.77 %)-H-ZSM5(19), no Oh A1 appears after aging at 923, 973 and 1073 K, contrarily to H-ZSM5, for which extra-lattice A1 appears as soon as 923 K. A small peak at 0 ppm (Oh A1) is observed only after aging at 1173 K. From the 29Si NMR signal it is theoretically possible to discriminate Si linked to 4 Si (and 0 AI) at around -111 ppm from Si linked to 3 Si (and 1 AI) at around -105 ppm but a problem arises from the contribution of SiOH at -106 ppm (17), which prevented meaningfull interpretation of the spectra. By studying the properties of Cu/AI203 and Cu-ZSM-5 solids with electronic and vibrational spectroscopies, we have already concluded that Cu 2§ and Cu ~ are not detected by the FTIR spectroscopy of the adsorbed CO probe molecule. The IR bands belong to CO adsorbed on Cu + ions, these surface ions being generated by the reduction of Cu 2§ ions under vacuo and/or by the CO probe itself (12). Furthermore the zeolite framework acts as a host for isolated Cu "§ ions (10). In fact for a ZSM-5 zeolite a ZO-(CuOH) § species may be formed during the exchange process, where ZO- represents the zeolite framework; Cu2+(OH) is thus linked to only one AI atom (18). We have already shown (10) that, for various flesh solids, the NO reduction rate into N2 (activity expressed as moles NO transformed into N2 per gram Cu and per second) correlates with the optical density (normalized by taking account the weight of copper) of the 2152-57 cm~ band, i.e., with the number of superficial isolated copper ions accessible to CO. Furthermore the reduction into Cu ~ and the agglomeration of the Cu "+ isolated ions into bulk oxides induce strong decreases in the reaction rate (10,12). From the present work, after aging at 923 K or 973 K, the IR spectrum of the adsorbed CO is the classic one (Figure 3), characteristic of isolated Cu + ions (10). The band at 2177 cm-1 vanishing upon evacuation at 300 K is assigned to the vs mode of the dicarbonyl Cu+(CO)2. The second band due to the vas mode overlapping with the band due to the Cu+CO species is not detectable. The band at 2151 cm-1 (2157 cm ~ after evacuation at 300 K) is assigned to the Cu+CO species, the Cu + ions being isolated. By comparison with the flesh solid the intensities of the bands decreases during the aging treatments. For instance, for the Cu(1.77 %)-ZSM5(19) solid and after treatment at 923 K, the absorbance of the 2151-2157 cm1 band, and therefore the number of the isolated Cu n* ions accessible to CO, has decreased by a factor of 3 to 4 (3 in the presence of CO, 3.8 after evacuation at 300 K). It can be concluded that, atler aging at 923 K (or 973 K), some copper ions become inaccessible but the spectrum is not qualitatively modified, namely an agglomeration is not detected. After aging at 1173 K a vCO band is detected at 2138 cm~ (2139 cm~ after evacuation at 300 K) (Figure 3). This band may be attributed either to CO adsorbed on isolated Cu + ions in a new environment or to non-isolated Cu + ions resulting from a partial reduction of a bulk CuO (11), which supposes a previous copper agglomeration. It may be noticed that partial

342 reductions of a model CuO oxide and of a high-copper loaded-Cu/AlzO3 solid in which CuO has been detected by XRD lead to a vCO band at 2135-2125 cm 1 (11).

A g e d 973 K

A g e d 1173 K i,-,

9,

2200

2100

I

I

2200

2100

I

’

,,,

cm-~

Figure 3. Infrared spectra of CO adsorbed on the Cu(1.77 %)-ZSM-5(19) solids previously calcined under oxygen at 773 K and evacuated at the same temperature. (a) upon contact with 50 Torr of CO at 300 K for 20 h, (b) previous sample evacuated at 298 K for 4 h. 3.2.2. After treatment at 823 K under catalytic conditions in the presence o f water. When the temperature reaches 873 K, the deactivation observed under catalytic conditions in the presence of water is not only a hydrothermal effect and it is not due to coking. In spite of the relatively low temperature, a dealumination is observed leading to the appearance of the 27Al NMR signal of octahedral AI (appearance of 10 to 20 % of extra-lattice A1), to a shift of the FTIR band from 1227 to 1230 cm1 and to decreases in the microporous volume and in the microporous surface from 0.15 to 0.11 ml/g and from 287 to 215 m2/g, respectively. This dealumination occurs at 873 K under the reaction mixture (in the presence of water) but not under an air + 10 % H20 mixture at the same temperature. This is due to the presence of an acidic component, HNO3, produced by the reaction of NO2 (and 02) with H20. It may be also noticed that, in the presence of water, the quantity of disappeared NO exceeds the quantities of N2 and NO2 formed (by a factor of 10-20 % for temperatures under 623 K) and the pH of the trapped water reaches 1. We have already noticed (13) a NO2 formation in

343 the pipes of the apparatus, at room temp6rature and in the absence of catalyst. The acid formation is linked either to this NO2 formation or to a NO2 formation on the catalyst itself. This formation increases when the NO reduction into N2 decreases.

4. DISCUSSION AND CONCLUSION Most of the physico-chemical measurements (XRD, SEM, N2 adsorption, framework vibrations) show little difference between fresh and hydrothermally treated (air + 10 % H20 mixture) Cu-ZSM-5 solids. There is no clear destruction of the zeolite framework; the decrease in micropore volume remains moderate and it is difficult to observe dealumination in the aged solids, even alter treatment at 1073 or 1173 K. Significant changes in catalytic activity are observed, however, even after treatment at 923 K, and the activity becomes negligible at~er treatment at 1073 or 1173 K. From the infrared spectroscopy of adsorbed CO it appears that aging treatments, as low as 923 K, lead to a migration of the active isolated copper ions to inaccessible sites. In these conditions an agglomeration is not detected but, after aging at 1173 K, an agglomeration is evidenced both by XRD and by the infrared bands of CO adsorbed on partially reduced bulk CuO oxide. These accessible copper oxide crystallites are probably located at the external surface of the zeolite and are inactive. In fact, the activity remains correlated to the number of Cu2+/Cu+ isolated ions deduced from the infrared spectra of adsorbed CO and located in the zeolite structure. This correlation holds whatever the treatment and whatever the Si/AI ratio (Table 4).

Table 4. ,Ratios of the activities of the fresh and 923 K-aged solids and ratios of the optical densities of the fresh and 923 K-aged solids. Activities measured at 623 K and with 10 vol. % 02 and without water. Optical densities (O.D.) of the vCO bands at 2151-2157 cm1 (isolated Cu§ ions) under CO and after evacuation. Two solids are considered, the Cu(1.3%)-MFI(21) one .and the Cu(4%)-MFI(130) one. .... Cu(1:3 %)-MFI(21) Cu(4 %)-MFI(130) Activity fresh solid/activity aged solid 3.7 2 O.D. fresh solid/O.D, aged solid CO, 298 K, l h 3.2 1.4 Vacuo, 298 K, lh 3.5 2.5

We were expecting that a loss of the isolated copper ions would be linked to a loss of exchangeable sites via dealumination. The possibility of a local dealumination, not evidenced by the physico-chemical characterizations performed here, cannot be excluded. Under catalytic conditions and in the presence of water, a dealumination is observed at relatively low temperature (873 K) and the deactivation of the solid is thus stronger than under the (air + 10 % HE0) mixture at the same temperature. This dealumination is attributed to the formation of acid resulting from the reaction of H20 with NO2 formed in the pipes of the apparatus. NO2 could also be formed on the catalyst.

344 In this work it is also shown that Cu-ZSM-5 is more resistant to dealumination by steaming than the parent H-ZSM-5 zeolite (Tables 2 and 3, Figure 2). Such a phenomenon has already been reported for Cu, Zn (4, 5, 20) and Cr (19). It may be supposed that the presence of another cation would neutralize part of the dealumination and would thus impede the migration of the exchanged copper. We have for this reason studied the effect of cocations on the thermal resistance of Cu-ZSM-5 solids. Among the cocations studied, the silver cation, active by itself, with a relatively weak affinity for water and an oxide unstable enough to probably allow an NO dissociation, is the most promising. The thermal stability is, however, a function of the order of exchange and of the preparation procedure. In conclusion the differences between flesh and aged solids (destruction of the zeolite framework, extent of dealumination...) are small in spite of the strong decreases in activity. Furthermore Cu-ZSM-5 is more resistant to dealumination by steaming than H-ZSM-5. The rate of NO reduction into N2 is correlated with the number of isolated Cu2+/Cu§ ions located in the zeolite structure; this number decreases with the aging treatments. A partial migration of copper to inaccessible sites seems more important than the degradation of the zeolite itself. At higher aging temperatures a part of the copper ions agglomerates into CuO particles accessible to CO but these bulk oxides are inactive.

REFERENCES

.

3. 4.

.

10. 11.

M. Iwamoto and H. Hamada, Catal. Today, 10 (1991) 57. M. Iwamoto, Catal. Today, 29 (1996) 29. M.D. Amiridis, T. Zhang and R.J. Farrauto, Appl. Catal., B, 10 (1996) 203 A.P. Walker, Catal. Today, 26 (1995) 107 J.O. Petunchi and W.K. Hall, Appl. Catal., B, 3 (1994) 239 M. Shelef, Chem. Rev., 95 (1995) 209. R.A. Grinsted, H.W. Jen, C.N. Montreuil, M.J. Robosz and M. Shelef, Zeolites, 13 (1993) 602. K.C.C. Kharas, H.J. Robota and D. Liu, Appl. Catal. B, 2 (1993) 225. W. Joyner and E.S. Shpiro, Symposium NOx reduction, 207th National ACS Symp. San Diego, CA, Division Petroleum Chemistry. Preprints, vol. 39, n~ February 1994, p. 103. T. Tanabe, T. Ijima, A. Koiwai, J. Mizuno, K. Yokata and A. Isogai, Appl. Catal., B, 6 (1995) 145. S. Matsumoto, K. Yokota, H. Doi, M. Kimura, K. Sekizawa and S. Kasahara, Catal Today, 22 (1994) 127. A.V. Kucherov, C.P. Hubbard, T.N. Kucherova and M. Shelef, Appl. Catal., B, 7 (1996) 285 A.V. Kucherov, C.P. Hubbard and M. Shelef, J. Catal., 157 (1995) 603. A.V. Kucherov, J.L. Gerloch, H.W. Jen and M. Shelef, Catal. Today, 27 (1996) 79 Z Chajar, M. Primet, H. Praliaud, M. Chevrier, C. Gauthier, F. Mathis, Appl. Catal., 4 (1994) 199. Z. Chajar, M. Primet, H. Praliaud, M. Chevrier, C. Gauthier, F. Mathis, CAPOC III, Studies in Surface Science and Catalysis, Vol. 96, A. Frennet and J.M. Bastin eds., Elsevier, 1995, p. 691.

345 12. 13. 14.

15. 16. 17. 18. 19. 20.

H. Praliaud, S. Mikhailenko, Z. Chajar, M. Primet, submitted to J. Catal. Z. Chajar, M. Primet, H. Praliaud, M. Chevrier, C. Gauthier, F. Mathis, Catal. Letters, 28 (1994) 33. Y. Zhang, T. Sun, A.F. Sarofim and M. Flytzani-Stephanopoulos, Symposium NOx reduction. 207th National ACS Symp. San Diego, CA, March 1994, Division of Petroleum Chemistry, Preprints, Vol. 39, n~ February 1994, p.171 J.Y. Yan, G.D. Lei, W.M.H. Sachtler and H.H. Kung, J. Catal., 161 (1996) 43. E.M. Flaningen, Zeolite Chemistry and Catalysis, ed. J.A. Rabo, ACS Monograph 171, Washington D.C., 1976. P. Budi, E. Curry-Hyde and R.F. Howe, Catal. Letters, 41 (1996) 47. G. Centi and S. Perathoner, Appl. Catal., A, 132 (1995) 179 and J. Catal., 152 (1995) 93. R.L. Keiski, H. Raisanen, M. HarkOnen, T. Maumula and P. NiemistO, Catal. Today, 27 (1996) 85. T.Tabata, M. Kokitsu and O. Okada, Catal. Today, 22 (1994) 147.



347

T r a n s i e n t kinetic s t u d y on N O d e c o m p o s i t i o n o v e r C u - Z S M - 5 catalysts Z. Schay, I. Kiricsi a and L. Guczi Department of Surface Chemistry and Catalysis, Institute of Isotopes of the Hungarian Academy of Sciences, P. O. Box 77, H-1525 Budapest, Hungary aApplied Chemistry Department, J6zsef Attila University, Rerrich B. t6r 1, Szeged, H-6720, Hungary ABSTRACT Decomposition of NO was studied on Cu-clinoptilolite and Cu-ZSM-5 zeolites of different Si/AI ratio and copper ion exchange rate. During the first contact at 600oC with NO an irreversible oxygen uptake by the catalysts was observed. On the long pulse of the NO concentration in the transient stage carried out under isothermal condition, overshoots in N 2 and 0 2 concentration were observed at the leading and falling edge, respectively. In the TPD experiments after NO adsorption a surface complex formulated as NO 3 is decomposed at about the reaction temperature into equimolar amounts of NO and 0 2. The role of the surface complex in the NO decomposition is discussed. 1. INTRODUCTION Since the discovery of the activity of ion-exchanged Cu-ZSM-5 zeolites in the catalytic decomposition of NO by Iwamoto and co-workers [1 ], the nature of the active sites and the reaction mechanism is still conflicting. Even review articles do not consider changes in the state of the catalyst during the start up period [2, 3]. Although the catalysts have been characterized by sophisticated techniques before the NO decomposition, still little effort has been made to study the catalysts after the reaction [3]. Nevertheless, there are some pieces of evidence available about the change of the structure of catalyst itself under reaction conditions. Previously we have reported in a transient kinetic study an overshoot in the formation of N 2 and 0 2 at the beginning and at the end of a long NO pulse, respectively [4]. In the present study we demonstrate that transiem stages in the N 2 and 0 2 formation are also present in other Cu-zeolites, as well as the very first pulse of NO at the highest reaction temperature results in a drastic change in the state of the catalysts. 2. EXPERIMENTAL Two copper containing ZSM-5 catalysts with Si/AI ratio of 24 and 66 and Cu/AI ratio of 0.5 and 1, respectively, have been prepared by conventional ion exchange of sodium with copper. The third catalyst was a c o ~ ion exchanged dinoptilolite, a natural zeolite of small pore size.

348 Transient kinetic studies were performed in a fixed-bed quartz tubular flow reactor of 4 mm inner diameter. 0.2-0.4 g catalyst of 0.5-0.25 mm sieve fraction was placed between quartz wool plugs. The catalysts were activated by heating it at 600oc in a stream of 25 cmJ/min argon for lb. The flow was then switched for 25 cm3/min 2 vol. % NO/Ar and after having reached a steady state (in about 10-40 rain) the was switched back to Ar. The effluent was analyzed by a QMS in multiple ion detection mode. A gas inlet system consisting of a heated stainless steel capillary differentially pumped by a rotary pump and linked via an orifice plate to a turbomolecular pump made the QMS signal proportional to the gas concentration in the effluent and ensured a high stability of the QMS calibration. The m/e values 18, 28, 30, 32, 38, 44 and 46 were recorded for measuring H20, N 2, NO, 02, Ar, N20 and NO2 concentrations, respectively. In Figures 1-3 only m/e values are given in which significant changes were observed. For calibration a mixture of 0.9 vol. % N 2 and 0.9 vol. % 02 in argon was used. To study the catalytic activity the catalysts were cooled in 2 vol. % NO/Ar mixture from 600oc to 300oc in about 30oc steps. In temperature programmed desorption (TPD) 25 cm3/min argon was used as carrier gas and a heating rate of 20~ was applied. If not stated otherwise, before TPD experiments the catalyst was cooled in 2 vol. % NO/Ar from reaction temperature to 200~ and purged with argon for 5 rain. An KRATOS XSAMS 800 XPS machine equipped with an atmospheric reaction chamber was used to characterize the valence state and surface composition of copper in the catalysts before and after the NO decomposition reaction. The binding energies were determined relative to Si 2p at 103.2 eV. For the surface composition signals of Cu 2p, O 1s, C Is, Si 2p and AI 2p were considered using the sensitivity factors given by the manufacturer. 3. RESULTS During the first contact of the catalysts with NO at 600~ an overshoot in N 2 formation along with a significant uptake of the 02 have been observed (see Fig. 1). At the same time the leading edge of the NO signal has been significantly leveled off and some N20 formed.. On switching back to argon, all signals with the exception of the 02 signal of very low intensity in Figure l a returned sharply to their background. A second contact to NO resulted in a fast response in all signals without any overshoot or delay and without any N20 formation. Even an extended purge in argon at 600~ was not able to restore the characteristic features given in Figure 1. This indicates an irreversible change in the catalysts during the first contact to NO. The amount of oxygen uptake is given in Table 1. Note that all quantities are about an order of magnitude less than those which correspond to the copper content in the catalysts. The starting temperature for the NO decomposition lies in the range between 340-360oc. On increasing the temperature an Arrhenius type temperature dependence has been observed up to 450~ with an apparent activation energy of about 90 kJ moleq. Between 500-550~ there was a maximum in the NO conversion followed by a marked decrease above 600oc. The highest conversion of about 50% was observed for the Cu-ZSM-5 Si/AI = 24 Cu/AI = 0.5 catalyst. The highest conversions for the other Cu ZSM-5 and Cu-clinoptilolite catalysts were about 10 % and 8 %, respectively. The transient measurements under isothermal condition at 400~ are shown in Fig. 2. The catalysts were cooled in argon from 600~ and at 400~ the flow was switched for a NO/At mixture for 5-10 min. At the beginning of the NO signal an overshoot in the N 2 concentration

349

%

NO x 0.1 ~

2 r/)

Cr

N2

02 x 10

4

o

~

-

~"~i~ J

-- "':TF . . . . . . x.,.,-.t-V -~.' .,.~=.~ V,,v v -"

i

J

'"J

~

0

I

~'

i

N20

a)

~',-

%..;~ . ................... V'~C,,' l

i

I

500

i

i

i

"

I

I

1000

t

i

i

I

i

4

&

2

r.~

O~

I

/

0

10

-

8

-

I

i

i

i

i'"

.,,'_/_

,,~..,'"-.v"v"-.,"

/_

t

~

NO

,~

'1'"

i

i'

,

,

I

500

i

i

,

i

I

1000

i

i

i

,

I

i

1500

i

i

,

Time ~)'s'

j2

E

64

NOxO.l

J

,.o

&

/

r

----/ .......... : ~"./ / A\ ........ ~"............................................................................................. " V ....... " ~ "

I1-"-::;:'~"'

o

/

02•

l

b)

N2

/

i

Time (s)

6 ,.o

E ~,

i

1500

~. -

02x 2 NO

x

C)

0.2

t#b

2 -

2

, ,

Ot 0

I

0

500

I ........ I

I

I

I

1000

Figure 1. First contact with NO at 600 ~ a) Cu-clinoptilolite b) Cu-ZSMS, Si/AI=66, Cu/AI=I c) Cu-ZSM5, Si/AI=24, Cu/AI=0.5

I

1500

I

I ....... w

I

I

w

I

I

Time (s)

;

350

/NO • 0.1

E

o,,

i~

4 2

9i , . = .

--~i I

0 {:7

"

0

L...... 'V1 1

/02

K/" : ..... \-, /

f'-'-----~ / S

-

/

_. / rx..,

[

II " - ./ 'r i

i

a) dr,

--.

"’" ""-" 4 " ' ' x

L*i

"'I

x lO

i

i

.... i

500

'

"

I

1000 Time (s)

..

E

..o

03

6-

/i 02X10 ii l .t ~ '

4

./

~, ~ .

I

N 2

i

f\ i \

I

I#H~ ~ rJ

..'.-~._.'.I -i

"

I

NO•

',,/

L__~..,,.,. l~ ~ . i : - ' ,..

~

I

I

''1

I

0

b)

~.

"'

I

'

v

i

--"...,-, v . V'" " " " i

I

500

1

i

1000 Time (s)

_

i

_

o,

r

r.t)

12

x4

i.

c)

\ NO x 0.1

_

~

% "%....

I

0

I'

'

i

I

i

i

i"

500

I

I

....

I

J

1000 Time (s)

Figure 2. Isothermal transients 400 ~ a) Cu-clinoptilolitc b) Cu-ZSMS, Si/AI=66, Cu/AI=I c) Cu-ZSMS, Si/AI=24, Cu/AI=0.5

331 10,.o

E

"d t~

8 a)

6

NO~/~O

%

2

4

j

t~

r/)

2 0

. . . .

...

0

-.~

.

,-.-..-:,-.

. . . .

400 I 350

I

250

,

.

.

.

.

800 I 450

.

.

.

0

~ ~

54

4

3 eta t~

" ' ; ' : ' ~ . " ~ ' ~ ' Y : ."'."7 . . . . . . . . . . . . . . . . .

400 I 350

I

250

_

E

y .t

0

(s) (~

1,1 d~

/ jo

_

t

800 (s) 450 "'~'t ~ !

d)

NO

o...

1 0

2

m2 N o'1

2

,,

0

.

.

.

.

.

.

I

.

250

.

-,

~

400

,-

~,

I

350

.

.

.

.

.

.

.

.

.

800 I 450

,

.

(s) (~

.

o

.

.

.

.

.

.

.

.

,

.

.

.

.

.

.

.

.

.

400

,

.

.

.

.

.

.

800

I

I

I

200

300

400

Figure 3. NO TPD after cooling in NO/Ar from 600 ~ a) Cu-clinoptilolite b) Cu-ZSMS, Si/AI=66, Cu/AI=I c) Cu-ZSMS, Si/AI=24, Cu/AI=0.5 d) as c) but cooled in Ar and dosed by NO at 200 ~

.

.

.

.

.

.

(s) (~

.

352 was observed together with a delay in the NO and 0 2 signals. The 0 2 signal was delayed to much lesser extent than that observed at the first contact with NO at 600~ At the end of the NO pulse, an 0 2 overshoot was observed. In contrary to the first NO pulse at 600oc these transients characteristics were the same on repeated NO pulses indicating reversible changes in the catalysts during the NO decomposition. The TPD experiments shown in Fig. 3. were started at 200oc. Although at room temperature the NO molecule is adsorbed and its desorption starts already at about 25~ [6, 7], we believe that this interaction with the surface entirely differs from that measured under reaction conditions. Thus, the catalysts were treated in 2 vol. % NO/Ar mixture at the reaction temperature, cooled down to 200~ in the same mixture and the TPD started at that temperature to obtain information about the intermediates formed on the catalysts. It is remarkable that N 2 is produced in a very small quantity, no N20 is found among the desorption products (the trace for N20 is omitted for the clarity of the figure) and the NO and 0 2 peaks appear nearly at the same temperature for the Cu-ZSM-5 catalysts. In the TPD after room temperature NO adsorption a similar high temperature NO + 0 2 TPD spectrum was reported by Li and Armor after oxidative pretreatment of Cu-ZSM-5 [6]. On Cu-clinoptilolite a broad low temperature NO desorption peak was observed without any N 2 or 0 2 desorption. The high temperature NO peak together with the 0 2 peak are shifted by about 50~ towards higher temperature compared to that observed on Cu-ZSM-5 catalysts. When the latter sample was cooled first in argon, then contacted with NO at 200oc, the NO TPD drastically changed Table 1 Oxygen retardation and TPD Catalyst

Cu content (Ixmol/g)

02 1 uptake

TPD 2

(gmol/g) NO low temperature

NO high temperature

02 high temperature

peak

peak

peak

0maol/g)

(lamol/g)

(~tmol/g)

Cu-

clinoptilolite Cu-ZSM-5 Si/Al=66 Cu/AI=I Cu-ZSM-5 Si/AI=24 Cu/AI=0.5

350

12

45

52

42

450

35

4.2

3.8

4.1

320

27

6.5

70

60

28 3

26 3

30 3

1) First contact with NO at 600 oC 2) 2) Cooling in 2 vol% NO/Ar from reaction temperature to 200 oC 3) 3) Cooling in Ar and dosing with NO at 200 ~

353 resulting in the TPD profile similar to that found for the Cu-clinoptilolite catalyst. The low temperature peak increased and simultaneously the high temperature peak decreased (see Fig. 3c and d). This means that at 200~ about half of the NO adsorbs in molecular form, whereas during cooling in NO the amount of the molecular form drastically decreases together with a slight increase in the total amount of NO. In Table 1. the amounts of NO and 0 2 detected in the TPD are presented. It is worth mentioning that the NO to 0 2 ratio is in all cases close to one indicating the decomposition of a NO 3 instead of a Cu-NO 2 type surface complex proposed in ref. [6]. Note also that above 400~ the highest amount of NO + 0 2 is released and the low temperature NO peak is the smallest for the most active Cu-ZSM-5 catalyst (Si/A1 = 24, Cu/AI = 0.5). The total amount of NO is always much less than the copper content indicating that only a small fraction of the copper adsorbs NO under reaction conditions. XPS results shown in Fig. 4. demonstrate how the copper sites change in the reaction. Although the binding energy of Cu 2p at 934.2 eV is unchanged being characteristic of Cu 2+, there is a drastic drop in the surface copper concentration from about 3 at. % to 0.9 at. % as well as a change in the structure of the satellite peak. This is indicative of a change in the environment of the copper sites [8] as the satellite structure originates from the paramagnetic properties of Cu 2+. x 10"s . 934.2 eV 5.6 5.4 5.0 Q o

4.8

-

~

4.6 ""J

it i

4.4

b) "'~,%

4.0 9

965

'.

955

1

' d

945

935

.

925

Binding energy (eV)

Figure 4. Cu XPS signals on Cu-ZSM5 Si/AI=24 Cu/AI=0.5 a) as prepared b) 600 ~ Ar lh followed by 450 ~ NO/Ar lh

354 4. DISCUSSION Although the three catalysts are quite different in respect to the copper content and zeolite structure, they show similarities in the NO decomposition. The catalyst activities are developed during the first contact with NO when the oxygen atoms resulted from NO decomposition are trapped and some extra-lattice oxygen is formed. Most probably during the first heating in argon a part of Cu 2+ ions is reduced to Cu + [6, 9] and on contact with NO they are reoxidized. Simultaneously, trace amount of N20 is formed, a situation similar to the room temperature NO adsorption [6]. The copper ions either migrate inside the zeolite cages or form larger oxide particles at the external surface of the zeolite crystals. This structure is rather stable as on extended purge in pure argon at 600oc there is no additional oxygen trapped or N20 formed when the catalysts are again contacted to NO at 600oc. This indicates that in argon Cu 2+ is not reduced into Cu+ ions in the active catalysts. In the 350-450~ temperature range the catalysts are active and oxygen rich intermediates or poison are developed. The poisoning is evidenced by the N 2 overshoots shown in Fig. 2. As the oxygen rich surface species are formed the N 2 evolution decreases. The peak in the N 2 concentration appears considerably above the state characteristic of steady state. The oxygen rich surface species are stabilized by NO, as on switching off the NO flow they decompose resulting in an overshoot in the 0 2 signal. The presence of this oxygen rich surface species is also evidenced by TPD results. The high temperature the desorption peak at 450-500~ this complex decomposes into NO and 0 2 without formation of N 2. If NO molecule were adsorbed on the surface at this temperature, it should have decomposed, at least partially into N 2 and 02. The amount of the complex depends on the "prehistory" of the catalyst. When NO is adsorbed at 200~ most of it is in molecular form and desorbs at about 300~ Only a part of NO forms oxygen rich complex which decomposes at 450-500~ When the catalyst is cooled in NO, there is practically no molecular NO adsorption and only the oxygen rich complex forms. The amount of this complex is considerably less than the copper content of the catalysts and there is no correlation between this and the initial oxygen uptake. On the Si/AI = 24 Cu-ZSM5 catalyst the nature of this complex was studied by FT-IR [5]. In agreement with the present study it was shown that only a small fraction of the copper ions is in interaction with this complex. In the present work we have shown that the same effect is measurable also on other Cu-zeolites. We propose that the intermediate is also the same as it was suggested in [5], namely a Cu2+(O)(NO)(NO2) type complex. The amount and the bond strength of the complex determines the catalytic activity.

ACKNOWLEDGEMENTS The financial support of this research by Grant OTKA T-017047 is acknowledged.

355 REFERENCES

1. 2. 3. 4. 5. 6. 7.

M. Iwamato and H. Hamada, Catal. Today, 10 (1991) 57 W. K. Hall and J. Valyon, Catal. Lett., 15 (1992) 311 G. Centi and S. Perathoner, Appl. Catal. A, 132 (1995) 179 Z. Schay and L. Guczi, Catal. Today, 17 (1993) 175 Z. Schay, H. Kn6zinger, L. Guczi and G. P~-Borbdy, Appl. Catal. B, to be published Y. Lee and J. N.Armor, Appl. Catal.,76 (1991) L1 G. P. Ansell, A. F. Diwell, S. E. Golunski, J.W. Hayes, R. R. Rajaram, T.J. Truex and A.P. Waker, Appl. Catal. B 2 (1993) 81 8. W. Grunert, N. W. Hayes, R. W. Joyner, E. S. Shpiro, M. R. H. Siddiqui, G. N. Baeva, J. Phys. Chem., 98 (1994) 10832 9. Y. Lee, W. K. Hall, J. Catal. 129 (1991) 202


CATALYSIS AND AUTOMOTIVEPOLLUTIONCONTROLIV Studies in Surface Science and Catalysis, Vol. 116 N. Kruse, A. Frennet and J.-M Bastin (Eds.) 91998 Elsevier Science B.V. All rights reserved.

357

Stability of cerium exchanged zeolite catalysts for the selective catalytic reduction of NOx in simulated diesel exhaust gas W.E.J. van Kooten, H.P.A. Calis and C.M. van den Bleek Delft University of Technology, Department of Chemical Technology and Materials Science, Julianalaan 136, 2628 BL Delft, The Netherlands

This paper describes the activity and the stability of several Ce exchanged zeolite SCR catalysts. NH3 is used as the reducing agent. CeNa-MOR is very active and reaches NOx conversions up to 100%, at a GHSV of 43000 hl and temperatures between 300 and 500~ The stability however, especially when SO2 is added, appeares to be poor. CeH-ZSM-5 on the contrary is less active but shows SO2 resistance, at least for the relatively short time it is investigated (37 hours) with SO2 concentrations up to 450 ppmv. CeH-ZSM-5 extruded with 50 wt% alumina suffers from irreversible deactivation when the catalyst is exposed to SO2 concentrations higher than 300 ppmv.

1. INTRODUCTION Earlier research [ 1-6] has shown that Ce exchanged zeolites are very promising catalysts for the selective catalytic reduction (SCR) of NOx using NH3 or urea as reducing agent because of their high activity and selectivity. These catalysts perform very well over a large temperature range, i.e. 300 to 600~ They can be operated with an e x c e s s amount of NH3 (up to 30% excess), the excess NH3 being converted to N2 rather than NOx or N20. The Ce zeolites are only slightly active in the oxidation of SO2 to SO3, which is advantageous for a diesel deNOx catalyst because SO3 adds to particulate emission, corrosion and formation of salts, which plug the catalyst. These features make the Ce zeolite catalyst in combination with urea as reducing agent a suitable candidate for a deNOx process for stationary diesel engines, such as marine diesel engines. Regarding the composition of diesel exhaust gases (containing amongst others water and SO2), developing a stable, zeolite based diesel exhaust deNOx catalyst is a challenging task. Zeolites can show dealumination under hydrothermal conditions accompanied by a loss of active material; furthermore SO2 can also cause deactivation. Many authors already have reported on the hydrothermal stability of zeolite SCR catalysts [e.g. 7-9] and also some papers exist on the stabilization with respect to hydrothermal deactivation of zeolite SCR catalysts by the choice of proper cations [ 10-13]. A small number of articles describes the influence of SO2 on zeolite SCR catalysts [ 14-17]. The current paper gives the results of measurements on both the short term hydrothermal stability and the influence of SO2 on CeNa-MOR and CeH-ZSM-5 zeolite catalysts. For application of zeolite SCR catalysts a monolith type reactor can be used, which is however

358 relatively expensive. Cheaper low pressure drop reactors, such as a Radial Flow Reactor (RFR) or a Lateral Flow Reactor (LFR) can also be used. For the RFR and the LFR as well as for a traditional fixed bed reactor, the zeolite should be extruded with a binding material to obtain proper, mechanically strong particles. To investigate the influence of a binding material, some zeolites were extruded with alumina as this is an often used binding. Results are shown of measurements on the activity and stability of an (50/50 wt%) extruded CeH-ZSM-5 with A1203.

2. EXPERIMENTAL CeNa-MOR was prepared by exchanging Na-MOR (PQ Zeolite, CBV-10A) with an aqueous solution of Ce2(SO4)3 at 80~ CeH-ZSM-5 was prepared by first exchanging Na-ZSM-5 (Uetikon, PZ-2/40) with an aqueous NH4NO3 solution at 80~ and next with a Ce(Ac)3 solution in water also at 80~ The ion-exchange process is subject of current research and will be described elsewhere [ 18]. After ion exchange CeNa-MOR and CeH-ZSM-5 contained 3.2 wt% (i.e. 41% ion exchanged) and 0.64 wt% (i.e. 23% ion exchanged) Ce respectively, as determined with ICP-AES. The zeolite samples were pelletized and crushed to a sieve fraction of 0.8-1.0 mm particles. Furthermore CeH-ZSM-5 was extruded with A1203 (50/50 wt%). The extrudates were also crushed to a sieve fraction of 0.8-1.0 mm. The Ce zeolites were tested for their catalytic activity for the SCR reaction in repeated temperature program runs: from 200~ up to 600~ and back to 200~ in steps of 50 or 100~ with a 2 hours dwell at each temperature level. The activity measurements were performed in a tube reactor made of quartz under plug flow conditions at a GHSV of 43000 hl, using 0.45 gram of catalyst particles. The standard feed composition was: 900 ppmv NO, 900 ppmv NH3, 5 vol.% 0 2 , 0 , 7 or 10 vol.% H20, 0, 100, 300 or 450 ppmv SO2 and balance nitrogen. Before the gases entered the reactor, they were mixed in a stainless steel gas mixing chamber (150~ water was added to this chamber by using a peristaltic pump. The stainless steel sampling lines had also a temperature of 150~ An ECO-physics CLD 700 EL-ht NOn-analyzer, based on the chemiluminescence principle, was used to monitor the NOx conversion. At each temperature the NOn concentrations of the inlet gases and outlet gases of the reactor were analyzed. The ammonia concentration was analyzed by a microwave process gas analyzer (Siemens, M52033-A901). The N20 formation was examined using ECD Gas Chromatography. In the temperature range of 200600~ the maximum amount of N20 produced was 1.5 ppmv at a temperature of 300~

3. RESULTS 3.1 SCR activity in presence of H20 The deNOx activity test sequence was the same for all three samples (CeNa-MOR, CeH-ZSM5 and CeH-ZSM-5/A1203), see Table 1. In this paper we only show the activity curves of the 'dry' and the 'wet' experiments as those activities are more representative for the catalyst activity than the activity during the pretreatment . The activities at ascending temperatures during the pretreatment were always a little higher (0-10%) than at the descending temperature phase of the pretreatment. The activity at the descending temperature phase of the pretreatment (step 2) always coincided with the ascending temperature activity of the dry experiment (step 3).

359 Table 1 Description of the test sequence of all catalYStsamples. Step # Temperature program Gas composition 1 200 to 600 ~ No H20, no SO2 2 600 to 200 ~ No H20, no SO2 3 200 to 600 ~ No H20, no SO2 4 600 to 200 ~ No H20, no SO2 5 200 to 600 ~ 10 vol.% H20, no SO2 6 . . . . . . 600 t o 200 ~ .......! 0 vol.;% .H20, n o S O 2 "-"

100

g

8o

o

60

~

o

~

T

.

.

.

,

,.,,,

,

.

40 2o

0

200

300 -

I

400 ~-

9 Dry gas, Tup Wet gas, T up

500

600

Temperature (~ O Dry gas, Tdown E] Wet gas, T down

Fig. 1 NOx conversion as a function of temperature for CeNa-MOR. (GHSV = 43000h", 900 ppmv NO, 900 ppmv NH3, 5 vol.% 02, 0 vol.% H,O (dry), 10 vol.% H:O (wet) and balance nitrogen)

100 l

o

i

80

O

o 0

d

60 40

2: 20 200

I

300

4

I

’

400

i,

I

- ~

500

..

,.J

,.,

.

,,,.,

Experiment . . . . . designation Pretreatment, Asc. T Pretreatment, Desc.T 'Dry' exp., Asc. T 'Dry' exp., Desc. T 'Wet' exp., Asc. T. 'Wet' exp., D.esc: T

I

600

Temperature (~ Fig. 2 NO, conversion as a function of temperature for CeH-ZSM-5. (GHSV - 43000h", 900 ppmv NO, 900 ppmv NH3, 5 vol.% 02, 0 vol.% H20 (dry), 10 vol.% H20 (wet) and balance nitrogen) (for legend see Fig. 1)

360 Figs. 1 to 3 show deNOx activity versus temperature for the three zeolite samples (CeNa-MOR, CeH-ZSM-5 and CeH-ZSM-5/AI203) during steps 3 to 6 indicated in Table 1. Fig. 1 shows that CeNa-MOR in the dry run possessed high activities already at low temperatures. A maximum conversion of 100% was reached between 300 and 500~ The descending temperature curve is shifted to higher temperatures, particularly in the low temperature region. When 10 vol.% water was added to the gas stream, the wet ascending temperature curve more or less coincided with the dry descending temperature curve. Again high conversions were reached. The wet descending temperature data showed a deactivation, chiefly at temperatures of 300 to 400~ Further experiments indicated that this deactivation was only partly reversible. The decrease in activity of all curves at high temperatures is probably caused by the oxidation of a small part of the ammonia, which is a common phenomenon for deNOx catalysts. Fig. 2 displays the results for CeH-ZSM-5. Though the catalyst only contains 0.64 wt% Ce, the activity reached peak conversions higher than 80%. At high temperatures there is a small NOx conversion decrease. The presence of water shifted the maximum conversion to higher temperatures in both the ascending and descending temperature curve. No further deactivation was noticed, as a second wet experiment (not shown here) gives the same activities as the first 'wet' experiment which is shown in Fig. 2.

100 80 60 40

20 0 200

300 ....

400

500

600

, Temperature (~

Fig. 3 NOx conversion as a function of temperature for CeH-ZSM-5/A1203. (GHSV = 43000h !, 900 ppmv NO, 900 ppmv NH3, 5 vol.% 02, 0 vol.% H20 (dry), 10 vol.% H~O (wet) and balance nitrogen). (for legend see Fig. 1) Fig. 3 shows that the extrudate CeH-ZSM-5/A1203, which in fact is a diluted catalyst, had only reasonable activity at temperatures higher than 400~ in the dry experiment. The ascending and descending temperature curve exhibited about the same data. The presence of water shifted the curve to higher temperatures. The low temperature activity of the catalyst was very poor. In the presence of water the conversion at 300 ~ reached only about 10%. A reversible deactivation for the wet descending temperature line was found.

361 3.2 S C R activity in the presence of SO2 Fig. 4 shows the influence of 50 ppmv SO2 on the NOx conversion for CeNa-MOR. The experiment was carried out with 7 vol.% water at a constant temperature of 387 c C. Directly from the start of the SO2 addition, the NOx conversion decreased and continued decreasing as long as SO2 was supplied. After stopping the SO2 addition, the deNOx activity did not recover significantly.

,-, 100 150 ppm SO=

~

80

-

-

(400"C)

0

d Z

6o

T

40

CeNa-MOR (387"C)

50 ppm SO= .........

,

I'

t

t

5 Time (h)

10

Fig. 4 The influence of SO: on the NO~ conversion of CeNa-MOR and CeH-ZSM-5 at 4000C. (GHSV = 43000h t, 900 ppmv NO, 900 ppmv NH3, 5 vol.% O:, 7 vol.% H20 for CeNa-MOR, 10 vol.% H:O for CeH-ZSM-5, SO2 as indicated in the figure and balance nitrogen.) 100 =

Q

500

90

400 ~

. ....~

Q

'-.'

80

o

70

200

o" 7;

6O

100 o

;> CJ

T 5o o

...-4

j

300

;o

20 Time (h)

2 00

2=

9

T

Fig. 5 The influence of SO: on the NO~ conversion of CeH-ZSM-5 at 500~ (GHSV = 43000h "t, 900 ppmv NO, 900 ppmv NH3, 5 vol.% 02, 10 vol.% H~O, SO~ as indicated in the figure and balance nitrogen.)

362 In the same figure the activity of CeH-ZSM-5 is plotted during a similar experiment with an even higher SO2 concentration of 150 ppmv and with 10 vol.% water at a temperature of 400 * C. The NOx conversion before S02 addition was about 80 % (which is in agreement with the data of Fig. 2) and upon SO2 addition it dropped to about 70% NO• conversion in about half an hour. No further activity decline was observed within 4 hours. Removal of SO2 from the feed restored the initial catalyst activity within about 1 hour. Further experiments revealed that the catalyst was resistant to SO2 concentrations up to 450 ppmv maintaining a NOx conversion of 70% at 400 o C during 16 hours. Repeating this experiment at 500" C, we found a rather surprising effect. The presence of SO2 in the feed gave rise to an increase in the NOx conversion as shown in Fig. 5. Going from 0 to 100 ppmv SO2, the NOx conversion increased from about 80% to about 90%. Further increase of the SO2 concentration did not show any significant influence. After making the feed SO2 free, the NOx conversion decreased to 80%, the value it was before SO2 exposition. Thus the CeH-ZSM-5 catalyst used, did no show any irreversible deactivation during the 3 experiments with SO2 in the feed, which lasted a total time of 37 hours. ~" 50

500 ,~,

o 40

400

r/l

300 .~

> 30

200

0

~ 2o

tl.,)

I00

Z

T

10

+

0

10

t

20 ; time (h)

30

4O

0

= Q

d

T

Fig. 6 The influence of SO2 on the NOx conversion of CeH-ZSM-5/A1203 at 500~ (GHSV = 43000h 1, 900 ppmv NO, 900 ppmv NH3, 5 vol.% 02, 10 vol.% 1-/20, SO2 as indicated in the figure and balance nitrogen.) o

The same type of experiment was carried out on CeH-ZSM-5/AI203 at 400 C in presence of 10 vol.% water. In Fig. 6 the results are plotted. The initial catalyst activity was lower than for the others and started at about 40% NOx conversion. Increasing the SO2 concentration to 100 or even 150 ppmv caused no significant deactivation on this time scale. A further increase of the SO2 concentration to 300 ppmv and 450 ppmv caused a significant decrease in NOx conversion. The deactivation was slow though it went on until SO2 was no langer added to the feed, at t = 18 hours. Only a slight recovery of the catalyst activity could be noticed.

4. DISCUSSION 4.1 Influence of H20 CeNa-MOR showed a very high activity but suffered from deactivation in 10 vol.% H20. Partly, the deactivation could be caused by dealumination. Especially the low temperature deNOx activity of the zeolite is positively influenced by the presence of Ce cations [1,5]. Thus

363 dealumination mainly caused deactivation in the low temperature region, which indeed can be seen in Fig. 1. Dealumination is a common problem for zeolite catalysts, depending not only on the type of zeolite, the gas feed and the temperature but also on the kind of cations. Some rare earth metals (such as lanthanum) are thought to inhibit dealumination [ 13] but so far this could not be proven for cerium cations. This study also can not confirm that Ce is an inhibitor for dealumination, at least not for CeNa-MOR. The activity of CeH-ZSM-5 is not very high but seems to be stable. Thus unlike CeNa-MOR, CeH-ZSM-5 did not suffer much from dealumination. More severe conditions are necessary to test the long term hydrothermal stability of CeH-ZSM-5. Increase of the amount of Ce to about 100% of the Ion Exchange capacity would surely benefit the low temperature activity [18]. Extrusion of this catalyst with A1203 has a clear effect on the activity. In the SCR reaction in the absence of H20 there is no deactivation but the low temperature activity has decreased compared to the non extruded zeolite. This is due to the dilution of the catalyst with A1203, a material which has no low temperature activity (A1203 started to show deNOx activity itself from about 400~ under our test conditions.). The high temperature activity of the extrudate still is reasonably good compared to non extruded CeH-ZSM-5 and CeNa-MOR. The experiment with H20 in the feed results in a clear reversible deactivation. Generalizing, the effect of water in the feed is twofold. Water in the feed causes a shift of the conversion maximum to higher temperatures (clearly visible in Figs. 2 and 3). This temperature shift is noticeable already at low temperatures. The shift is caused by a reversible inhibition of active sites by water. Water in the feed can also cause irreversible deactivation by dealumination (CeNa-MOR) which only occurs at high temperatures and results in a decreased deNOx activity mainly in the low temperature region. 4.2 Influence of S02 In general, addition of S02 to the feed can cause several problems. Oxidation of S02 to S03 adds to the emission of particulates and moreover 803 reacts with H20 to give HESO4,which causes corrosion in the exhaust pipe. Oxidation of SO2 to sulfates, e.g. ammonium sulfates, is also highly undesirable, because of possible blocking of pores of the catalyst resulting in catalyst deactivation. It is known [ 19] that ammonium sulfates decompose easily at moderate temperatures depending on the concentration of NH3 and SO3. We did not find any indication on the oxidation of SO2 to SO3 or sulfates in the sampling lines (stainless steel, 150~ Our results show that CeNa-MOR was more susceptible to deactivation than CeH-ZSM-5. CeNa-MOR probably formed sulfates which clog the catalyst. The influence of SO2 on CeHZSM-5 was quite different. Upon SO2 addition a 10% activity decay occurred in one discrete step. The catalyst regained its full activity within one hour of SO2 free feed. This could point to a reversible site-blocking effect. The formation of a sulfate salt is less likely because there was no continuing deactivation during SO2 exposition. The results of the experiment at 500~ confirm this hypothesis. At this temperature even an increase in deNOx activity was found. We think that it is due to an effect that is comparable to the influence of water. Fig. 2 shows that the influence of H20 on the NO• conversion at 400~ was slightly negative but that at 500~ and 600~ a small increase was obtained, i.e. a shift of the conversion maximum to higher temperatures. Others also have reported that a zeolite of the type ZSM-5, used as SCR catalyst, could withstand SO2 at temperatures higher than 400~ [20]. In case of the extrudate (Fig. 6) a deactivation at SO2 concentrations higher than 300 ppmv on the time scale of hours could be noticed which is probably related to a reaction of the A1203, as

364

dealumination mainly caused deactivation in the low temperature region, which indeed can be seen in Fig. 1. Dealumination is a common problem for zeolite catalysts, depending not only on the type of zeolite, the gas feed and the temperature but also on the kind of cations. Some rare earth metals (such as lanthanum) are thought to inhibit dealumination [13] but so far this could not be proven for cerium cations. This study also can not confirm that Ce is an inhibitor for dealumination, at least not for CeNa-MOR. The activity of CeH-ZSM-5 is not very high but seems to be stable. Thus unlike CeNa-MOR, CeH-ZSM-5 did not suffer much from dealumination. More severe conditions are necessary to test the long term hydrothermal stability of CeH-ZSM-5. Increase of the amount of Ce to about 100% of the Ion Exchange capacity would surely benefit the low temperature activity [18]. Extrusion of this catalyst with A1203 has a clear effect on the activity. In the SCR reaction in the absence of H20 there is no deactivation but the low temperature activity has decreased compared to the non extruded zeolite. This is due to the dilution of the catalyst with A1203, a material which has no low temperature activity (A1203 started to show deNOx activity itself from about 400~C under our test conditions.). The high temperature activity of the extrudate still is reasonably good compared to non extruded CeH-ZSM-5 and CeNa-MOR. The experiment with H20 in the feed results in a clear reversible deactivation. Generalizing, the effect of water in the feed is twofold. Water in the feed causes a shift of the conversion maximum to higher temperatures (clearly visible in Figs. 2 and 3). This temperature shift is noticeable already at low temperatures. The shift is caused by a reversible inhibition of active sites by water. Water in the feed can also cause irreversible deactivation by dealumination (CeNa-MOR) which only occurs at high temperatures and results in a decreased deNOx activity mainly in the low temperature region. 4.2 Influence of SO2 In general, addition of SO2 to the feed can cause several problems. Oxidation of SO2 to SO3 adds to the emission of particulates and moreover SO3 reacts with H20 to give H2SO4, which causes corrosion in the exhaust pipe. Oxidation of SO2 to sulfates, e.g. ammonium sulfates, is also highly undesirable, because of possible blocking of pores of the catalyst resulting in catalyst deactivation. It is known [ 19] that ammonium sulfates decompose easily at moderate temperatures depending on the concentration of NH3 and SO3. We did not find any indication on the oxidation of SO2 to SO3 or sulfates in the sampling lines (stainless steel, 150~C). Our results show that CeNa-MOR was more susceptible to deactivation than CeH-ZSM-5. CeNa-MOR probably formed sulfates which clog the catalyst. The influence of SO2 on CeHZSM-5 was quite different. Upon SO2 addition a 10% activity decay occurred in one discrete step. The catalyst regained its full activity within one hour of SO2 free feed. This could point to a reversible site-blocking effect. The formation of a sulfate salt is less likely because there was no continuing deactivation during SO2 exposition. The results of the experiment at 500~C confirm this hypothesis. At this temperature even an increase in deNOx activity was found. We think that it is due to an effect that is comparable to the influence of water. Fig. 2 shows that the influence of H20 on the NOx conversion at 4 0 0 ~ C was slightly negative but that at 500~C and 600~C a small increase was obtained, i.e. a shift of the conversion maximum to higher temperatures. Others also have reported that a zeolite of the type ZSM-5, used as SCR catalyst, could withstand SO2 at temperatures higher than 4 0 0 ~ C [20]. In case of the extrudate (Fig. 6) a deactivation at SO2 concentrations higher than 300 ppmv on the time scale of hours could be noticed which is probably related to a reaction of the A1203, as

365 the non-extruded ZSM-5 clearly shows other behavior. The formation of pore plugging sulfates is the probable cause for the observed deactivation. Using the catalysts at low SO2 concentrations of about 20 ppmv (which is the SO2 concentration in the exhaust when standard European diesel fuel containing about 0.05 %S is used), a deactivation is expected on the long term. So the use of a 50/50 wt% extrudate with alumina seems no suitable choice for diesel deNOxing. Extrudates with less alumina or with other binding materials should be investigated.

5. CONCLUSIONS In absence of H20 and SO2, CeNa-MOR is a very active SCR catalyst. CeNa-MOR however suffers from deactivation under hydrothermal conditions and in presence of SO2. CeH-ZSM-5 is slightly less active but is a much more stable catalyst both in the presence of H20 and SO2. Application of this latter zeolite using a 50/50 wt % extrudate with A1203 seems not appropriate due to poisoning by SO2 at concentrations higher than 300 ppmv. To maintain the good properties of CeH-ZSM-5, extrudates with less alumina or other binding materials should be used.

ACKNOWLEDGMENT The authors would like to thank Dr. J. Nieman of AKZO NOBEL for arranging the manufacturing of the extrudates and for helpful discussion.

REFERENCES

.

,

~

9. 10. 11. 12.

E. Ito, R.J. Hultermans, P.M. Lugt, M.H.W. Burgers, M.S. Rigutto, H. van Bekkum and C.M. van den Bleek, Stud. Surf. Sci. Catal. 96 (1995) 661. R.J. Hultermans, E. Ito, A. Jozsef, P.M. Lugt and C.M. van den Bleek, Stud. Surf. Sci. Catal. 96 (1995) 645. E. Ito, R.J. Hultermans, P.M. Lugt, M.H.W. Burgers, M.S. Rigutto, H. van Bekkum and C.M. van den Bleek, Appl. Catal. B. 4 (1994) 95. E. Ito, C.M. van den Bleek, H. van Bekkum, J.C. Jansen, R.J. Hultermans and P.M. Lugt, Delft University of Technology, Dutch Patent Appl., NL 93.02288 (1993), PCT patent appl. no. PCT/NL95/00001. E. Ito, Y.J. Mergler, B.E. Nieuwenhuys, H.P.A. Calis, H. van Bekktun and C.M. van den Bleek, J. Chem. Soc., Faraday Trans. 1996, Vol. 92 (10),1799. E. Ito, Y.J. Mergler, B.E. Nieuwenhuys, H. van Bekkum and C.M. van den Bleek, Microp. Materials, 4 (1995) 455. R. A. Grindsted, H.W. Jen, C.N. Montreuil, M.J. Rokosz and M. Shelef, Zeolites 13 (1993) 602. J.Y. Yan, G.D. Lei, W.M.H. Sachtler and H.H. Kung, Jnl. Catal. 161 (1996) 43. R.F Howe and R.H. Meinhold, Jnl. Catal. 161 (1996) 338. J.N. Armor, T.S. Farris, Appl.Catal. B: Env. 4 (1994) L 11. J.Y. Yan, W.M.H. Sachtler and H.H. Kung, Catal. Today 33 (1997) 279. Y. Zhang, M. Flytzani-Stephanopoulos, Jnl. Catal 164 (1996) 131.

366 13. 14. 15. 16. 17. 18. 19. 20.

P. Budi, E. Curry-Hyde and R.F. Howe, Catal. Lett. 41 (1996) 47. X. Feng, W.K. Hall, J. Catal. 166(2) (1997) 368. M. Iwamoto, H. Yahiro, S. Shundo, Y. Yu-u, N. Mizuno, Appl. Catal. 69 (1991) L15. S.-W. Ham, H. Choi, I.-S. Nam, Y.G. Kim, Catal. Today, 11 (1992) 611. Y. Li, J.N. Armor, Appl. Catal. B: Env. 5 (1995) L257. B. Liang, W.E.J. van Kooten, O. Oudshoorn, H.P.A. Calis, C.M. van den Bleek, to be published. H. Bosch and J. Janssen, Catalysis Today 2 (1987), 436. J.R. Kiovsky, P.B. Koradia and C.T. Lim, Ind. Eng. Chem. Prod. Res. Dev. 19 (1980) 218.


367

Study of Copper - and Iron - containing ZSM-5 zeolite catalysts: ESR spectra and initial transformation of NO J. Varga a, J. Hal~sz a, D. Horv~tha, D. M6hn a, J. B.Nagy b, Gy. Sch0bel a and I. Kiricsi a aApplied Chemistry Department, J6zsef Attila University, Rerrich t6r 1 H-6720 Szeged, Hungary bLaboratoire de RMN, Facult6s Universitaires Notre-Dame de la Paix, rue de Bruxelles 61, B-5000 Namur, Belgium

ABSTRACT For a better understanding of the first steps of the reaction of NO over Cu-ZSM-5 and FeZSM-5 zeolites the following measurements were performed: (i) the products of the gas phase interactions were followed by MS, (ii) the valence state and coordination of transition metal ions in zeolites by ESR spectroscopy. Catalysts were prepared both by conventional and solidstate ion-exchange methods and pretreated in vacuum, in oxidative and in reductive atmosphere. The conventional ion-exchanged samples are more active in NO decomposition than the solidstate exchanged ones. Over the reduced catalysts the first step consists in N20 formation and the oxidation of Cu + ~ Cu 2+ (Fe2+ ~ Fe 3+) followed by N20 reduction to N2 (in these conditions 02 release was not detected). These results are consistent with a simple redox mechanism in which NO adsorption and decomposition on active (reduced) sites leads to the formation of gaseous nitrogen and adsorbed oxygen. Over oxidized samples NO transformation is slower than over reduced ones, and the first step is to be the reduction of metal ions (Cu2+, Fe 3+) with NO as reducing agent.

1. INTRODUCTION The removal of nitrogen oxides from mobile and stationary sources remains an important environmental technology task for solving acid rain and other pollution problems [1-3]. Transition metal exchanged MFI zeolites, particularly Cu-ZSM-5, were proven to be active in catalytic reduction of NO (and NO2) by ammonia [4] or hydrocarbons [5,6] as well as in the direct decomposition of NO to molecular nitrogen and oxygen in "lean-burn" engine conditions [7,8]. Numerous ESR measurements were carried out to identify the oxidation states and coordinations of transition metals - especially Cu - in ZSM-5 zeolite structure [9-12]. Moreover Kucherov et al. performed in situ ESR studies to clarify the redox properties of copper exchanged zeolites [ 13,14]. Although much effort was devoted in the recent years to understand the mechanism of these reactions, involving the explanation of the exact role of the catalyst, the information on the character of the metal active sites and their interactions with the zeolite structure and the reactant

368 molecules are contradictory and incomplete. In the present work, the role of copper and iron content of the catalysts (prepared by conventional and solid-state ion-exchange methods) was investigated in the initial elementary steps of the direct decomposition of NO by ESR spectroscopy and catalytic measurement.

2. EXPERIMENTAL Cu2+- and Fe3§ containing ZSM-5 zeolites were prepared by conventional and solid-state ion-exchange methods described in ref. [ 15]. The Si/AI ratio of parent Na-ZSM-5 was 40. The samples were characterized by X-ray diffraction, IR spectroscopy, thermal analytical method and BET measurement. Related data are shown in Table 1. Table 1

Sample characterization Sample

Cu-ZSM-5

Exch. method

Conventional

Fe-ZSM-5

Solid-state

Conventional

Solid-state

Metal ion content (wt %)

1.14

3.05

1.19

5.11

Ion-exchange degree %

93

250

54

190

BET area m2/g

370

318

367

318

ESR spectroscopy is a powerful tool to monitor the valence state and coordination of cupric and iron ions in zeolites. After different pretreatment procedures (listed in Table 2) and/or NO adsorption spectra were recorded on a Bruker BER-420 spectrometer with a TE~o2 rectangular cavity - at both ambient temperature and 77 K. Table 2

Pretreatment procedures and designation of ESR samples Pretreatment

Designation

I.

One hour evacuation at ambient temperature

II.

Evacuation at 773 K for 2 hours

V

III.

Oxidation by 100 Torr (13.33 kPa) 02 at 573 K for 2 hours

O

IV.

Reduction by 100 Torr (13.33 kPa) 1-12at 373 K for 2 hours

R

V.

Heating in air at 773 K for 2 hours

L

The kinetic measurements to study the transformation of nitrogen oxides were carried out in a recirculatory batch reactor with mass spectrometric analysis (see details in [16]). To eliminate the

369 effect of the gas phase oxygen for the initial steps of NOx reactions the measurements were carried out in conditions where the molar ratio of active centres (supposedly metal ions) to reactant was as high as 3. The catalyst samples (0.5 g) were heated at 773 K in vacuum (0.1 Pa) for two hours in every case followed by oxidative - at 573 K in 100 Torr (13.33 kPa) 02 for 2 hours - or reductive - at 373 K, 100 Torr (13.33 kPa) H2 for 2 hours - treatment. The gas-phase concentrations of reactants (NO, NO2 and N20) and products were measured by mass spectrometry. Mass numbers 30 of NO, 46 of NO2, 44 of N20, 28 of N2 and 32 of O2 were used for the quantitative analysis. It is important to emphasize that no detectable 02 was found in NO decomposition at the reaction conditions used.

3. RESULTS AND DISCUSSION The catalytic transformations of nitrogen oxides are considerably affected by the oxidation state of the metal ions occupying exchange position in the zeolite. The ESR technique is a useful method to follow the oxidation states and changes of catalysts in a different way, however, the real reactions could be characterized by measurements in actual conditions. 3.1. ESR Measurements 3.1.1. Characterization of Cu-ZSM-5 samples For copper in Cu-ZSM-5 zeolites three different coordinations are given in the literature [914,17]: square planar (g = 2.27), square pyramidal (g = 2.33), and octahedral (g = 2.38). Additional exchange ions present in zeolites, or water content of the samples influence these potential distributions. Furthermore, a given coordination can be realized by involving different number of framework oxygens, exchange cations or other extra framework ligands such as water or hydroxyl groups. Considering the general rules the following main features can be drawn (more detailed explanation is found in [18]). In hydrated CuL and CuS samples (see designations in table 3) the Cu 2§ ions occupy octahedral coordinations. Evacuation causes partial or full dehydration, and a decrease of symmetry from water molecules assisted as ligands of copper ions. CuLV and CuSV spectra can be considered as superpositions of two spectra - one for square pyramidal and one for square planar coordinations. Oxidation treatment caused no change in the spectra just as it was expected. Reduction had to be made cautiously as Cu2+ can be readily reduced by hydrogen above 573 K. This treatment at 373 K resulted in an increase in the symmetry of copper ions. The signal intensities of spectra CuLR and CuSR are lower due to the fact that Cu + and Cu ~ are ESR silent species. By reduction water molecules can be formed, which can be coordinated to the remaining 2+. Cu Ions forming octahedral complexes, different from that of the hydrated samples. Upon NO adsorption some broadening and simultaneously, some increase in the intensity of the signals took place, which indicate a complex redox transformation on the surface metallic sites of the catalysts. In the interaction between NO and Cu-ZSM-5 mainly Cu 2+ ions of lower symmetry (square planar or square pyramidal) are involved.

370 3.1.2. Characterization Fe-ZSM-5 samples The general rules listed in case of Cu-ZSM-5 are still valid for Fe-ZSM-5. In zeolite structure four main coordinations for the Fe species can be found: octahedral (g - 2.06), oxo-hydroxo species (g = 2.89), tetrahedral (g = 4.34), and distorted tetrahedral samples (g = 5.4 - 6.8) are mentioned in the literature [ 19]. The ESR parameters of samples are collected in Table 3. Table 3

ESR parameters determined for different Fe-ZSM-5 samples Sample*

gB

g

g

g

FeL

2.442

5.383/5.403

4.380

2.904

FeLNO

2.442

6.534/5.442

4.385

2.894

FeS

2.433

5.405

4.338

2.896

1.999

FeSNO FeLV FeLVNO FeLL FeLLNO FeSL FeSLNO

2.435 2.658 2.579 2.436 2.438 2.447 2.443

6.908/5.719

2.892

6.248/5.815 6.302/5.743 6.22/5.943 5.861

4.346 4.340 4.347 4.320 4.324 4.341 4.336

2.907 2.899 2.907 2.893

2.058 2.004 2.007 2.062 2.066 2.064 2.061

FeLO

2.384

6.269/5.812

4.304

2.899

2.159

FeLONO

2.385

6.249/5.838

4.318

2.897

2.157

FeSO FeSONO FeLR FeLRNO FeSR FeSRNO

2.444 2.443 2.442 2.441 2.439 2.440

6.194/5.940 5.430 6.749/5.345 5.384 6.286/5.869 5.876

4.339 4.313 4.312 4.284 4.336 4.340

2.394 2.885 2.891 2.897 2.897 2.900

2.022 2.022 2.062 2.004 2.055 2.054

*Designation of samples:

g

1. chemical symbol of exchanged metal (Cu, Fe) 2. ion-exchange method used (Liquid or Solid state) 3. designation of the pretreatment procedure (see in table 2) 4. +NO if spectrum was taken after NO adsorption on the sample

As can be concluded from the spectra of Fig. 1 (FeL and FeS, respectively) both peaks at g - 2.06 and g - 4.34 are more intense for samples prepared by solid state ion-exchange than for those prepared by conventional method. These deviations can be explained by the different quantities of iron-content. Indeed, the spectrum of sample pretreated in vacuum (FeLV) is extremely simple. During the pretreatment some extent of the iron present was reduced to Fe ~ ferromagnetic properties of which disturbed the magnetic field applied.

371

FeL

FeS

FeLL

FeSL

FeLV-C----~

FeSV

FeLO

~,vf.~

FeSO

FeL

f

/4 FeSR

f f

Fig. 1" The ESR spectra of Fe-ZSM-5 samples prepared byconventional (L) and solidstate (S) ion-exchange method.

FeSO

FeSONO~.,//q

Fig. 2. The effect of NO adsorption on the ESR spectra of Fe-ZSM-5 samples.

372 Upon NO adsorption (Fig. 2) the intensities of the distorted tetrahedral peaks decreased which lead us to the conclusion that adsorption took place on these sites, hi situ adsorption measurements showed increases in line intensities indicating the oxidation of iron ions I ' I i I ' I ' I 3.2. Catalyticmeasurements100 ~k_~,T_ -(3- N2 The transformation of NO and NO2 as main _~ 80 ~ -0- NO components of NOx pol,~, lution in exhaust gases, ~ ~-\ -El- N20 and N20 as stable inter~ 60 mediate in these reactions were investigated ~ 40 over the Cu- and Fe-conr taining ZSM-5 zeolite catalysts. ~ 20

t

3.2.1. NO

tion

decomposi-

0

0 10 20 Time (rain) For the NO decomposition reaction the CuFig. 3: Decomposition of NO over reduced Cu-ZSM-5 ZSM-5 prepared by con(prepared by conventional ion-exchange) at 573 K. ventional ion-exchange method and pretreated in reductive atmosphere proved to be an effective catalyst, as can be seen in Fig. 3, where the results obtained at 573 K are presented. The first and very fast step is the NO transformation into 1'420 over the reduced Cu centres: 2 N O + Cue0 "~ N 2 0 + 2 CuO,

which is followed by the relatively slow decomposition of N20 into nitrogen and oxygen. At the end of the reaction only nitrogen can be detected in the gas phase, the oxygen reacts with Cu § centres resulting in Cu2+. Over catalyst pretreated in oxidative atmosphere similar reaction can be observed (see Fig. 4), however, the rate is much lower. The quasi steady-state N20 concentration can be explained by the reaction of Cu2§ with NO in forming Cu § ions, i.e. the NO is capable to reduce the oxidized metal ion. As can be seen in Fig. 5 and 6, over catalysts prepared by solid-state ion-exchange the rate of NO conversion is the same, however, the N2 formation is lower both in the reduced and oxidized samples corresponding to those prepared by conventional method. Considering the kinetic behaviour, similar results could be obtained in the reaction over FeZSM-5 samples, however, the conversion of NO was lower at a notable extent.

373 I

100

~

I

J

I

.... I

.... I" "

-0- N2

8O ~

7

-@- NO -E]- N20

o 60

4a O

r~ 20 v

20

0

40

~e

60

(rain.)

Fig. 4: Decomposition of NO over oxidized Cu-ZSM-5 (prepared by conventional ionexchange) at 573 K.

100t I

I ''

I

"1"'

!

I

J

I"

l'

I

I

~80 ~9

- CO2 + ~ N 2

klXCOsXO2s

k4xcOs1"4xo2s 0"3XNOs0"13 -3.74 x 105

,

F I ( T s x s) = Ts (1+ k a l x c o s + ka2XCHxs

)2

r4 =

F2(%,x~)

0.7 (1+ ka3xcos2XCHxs 2)(1+ ka4xNO ~ )

F2 (Ts, Xs ) = Ts-O.17 (Ts + ka5xco s )2 k i=kOiexp -~s) kaj = kaoj exp Table A2 Rate constants Reaction i

R'rs

1

Activation energy Eai (J.mo1-1) 1.040 x 105

6.699 x 1013 mol.K.s-l.m -2 PtRh

2

1.210 x 105

1.392 x 1015 mol.K.s-l.m -2 PtRh

3

1.040 x 105

6.699 x 1013 mol.K.s-l.m -2 PtRh

4

7. 177 x 104

3.061x 1012 mol.Kl-S3.s-l.m-2 PtRh

Table A3 Adsorption constants Specie j Adsorption heat AHaj (J.mol-1)

Rate factor koi

Adsorption factor kaoj (dimensionless)

1 2 3

-7. 990 x 103 -3.000 x 105 -9.650 x 104

6.550 x 101 2.080 x 103 3.980

4

3.100 x 104

4.790 x 105

408 Appendix

B: Theoretical

relation

k0(Ea, TI~)

The light-off phenomenon is a safe runaway. There is a lot of articles devoted to the prediction of runaway condition in the chemical engineering literature. Villermaux [10] showed that runaway occurs in a batch reactor at a time given by:

tRoWo2

(B1)

to = ATadEa / R

where To is the initial temperature and tR0 a characteristic reaction time defined below. If we assume that plug flow prevails in the monolith and that chemical expansion is negligible, equation (B1) applies when the mean residence time of the fluid is to. tR0, to, and ATad are given by: AWad = ( - AH) C0A pCp

, to =

aV , C0A ~ -Q- tR~ = r 0

(B2)

where CA0 is an arbitrary reference concentration. The reaction rate ro [mol.s-l.m "3 of fluid] is given by:

(Ea)

r 0 = koSpt exp - ~ - 0

X0AsX0Bs

(B3)

where Spt is the area of noble metal per unit volume of monolith. Combining (B1) to (B3) gives:

k 0 --R--exp - R ~ o

= eVSpt (-AH)X0AsX0Bs

(B4)

It r e m a i n s to define the reference composition, X0As, X0Bs, and t e m p e r a t u r e TO. The light-off phenomenon takes place over a narrow temperature range. It is thus concluded that To = TLO. A few degrees below the light-off temperature, the reaction is sufficiently slow to assume t h a t the composition of the fluid in contact with the wash-coat is close to the inlet composition. (B4) thus becomes:

a0ex/ / R

RTLo

Spt eV (-AH) XAinXBin

which is identical with (6). (7) results from the fact t h a t the real LO temperature inside the monolith is the observed temperature plus the adiabatic temperature rise.

CATALYSISAND AUTOMOTIVEPOLLUTIONCONTROLIV Studies in Surface Science andCatalysis,Vol. 116 N. Kruse, A. Frennet and J.-M Bastin (Eds.) 91998Elsevier Science B.V. All rights reserved.

409

NO + CO --> ~/2 N2 + CO2 differentiated from 2NO + CO --> N20 + CO2 over rhodia/ceria catalysts using lSN180 and 13C160 reactants or timeresolution of products Joseph Cunningham a, Neal J. Hickeyb, Frank Farrell c, M. Bowkerd and Colin Weekse. a'b'CChemistry Department, University College Cork, Ireland. d'eCatalysis Research Centre, University of Reading, UK.

ABSTRACT Profiles versus Ramp-temperature for isotope exchange between 15Nl80 and 0.5% RhO• CeO2 or Rh203 materials within a recirculatory reactor system indicate unique low-temperature lability of ~60-surface species at rhodia-ceria perimeter positions upon preoxidised 0.5% RhO• and its absence there from after LTR. Onset temperatures and relative efficiencies are likewise compared for conversion of 15N180 plus 13C160 mixtures to isotopicaliy distinguishable forms of N20, N2 and CO2 products over preoxidised and prereduced materials. Complementary insights into the time-sequence for appearance of N20 and N2 products in the gas phase over the materials at selected temperatures in the range 125298~ are provided by results carried out in an alternative micro reactor system which allowed introduction of individual 10s pulses of CO into a continuous flow of NO plus helium over preoxidised or prereduced aliquots of the materials.

1. INTRODUCTION Synergisms between Rhodium and Ceria components of three-way catalytic converters (TWC' s) are widely considered important for continuing efficient operation of the latter, with the zero-valent metallic state being favoured for the dispersed Rhodium component. Transitory existence of at least some part of the dispersed rhodium component in oxidised R h n+ form (hereinafter denoted by RhO• is, however, probable at various times during TWC operation and most especially during engine >[1,2]. Twin objectives of the experiments here performed in a recirculatory reactor system were: (a) to determine what catalytic activity and selectivity is associated with 0.5% RhO• in respect of the conversions NO + CO --> 1/2 N2 + CO2 and 2NO + CO --> N20 + CO2 across such warm-up temperature range, and (b) to compare such activities/selectivities with those of the same catalysts after prereducing, firstly, the rhodium component to Rho, (LTR) and secondly, the surface-ceria component to CeO2.x (HTR). In view of previous reports of facile oxygen isotope exchange between N180 and the lattice oxygens of some metal oxides

410 [3,4] and in order to gain background information necessary for meaningful interpretation of present studies of the NO + CO reactions, preliminary studies of the interactions of 15N180 alone over the CeO2, RhaO3 and 0.5% Rh/CeO2 materials in their pre-oxidised LTR and HTR states were undertaken. Subsequently, an equimolar mixture of 15N180 and 13C160 was used for mass spectrometric based comparisons of the relative efficiencies and selectivities of the various materials in promoting 15N180 + 13C160 ~ 1/2 15N2+ 13CO2and/or 2 15NlSo + 13C160 15N2180 d- 13C160180 conversions in the recirculatory reactor system. Such experiments clearly could not provide direct insights into aspects of TWC operation during switching between exhaust gases with 'lean' and 'rich' stoichiometry- e.g. the time-sequence for isothermal evolution of N20 and other products from such reactions at warm-up temperatures. Such insights were sought instead using an alternative experimental arrangement which allowed injection of individual 10s duration pulses of normal CO into a continuous flow of helium plus normal NO over the catalysts and was equipped with fast MS detection to delineate the time sequence and profiles for product evolution. 2. EXPERIMENTAL Origin and properties of the Rh203 and CeO2 powders have elsewhere been described in full, as also the wet impregnation of CeO2 with non-aqueous Rhodium III acetyl acetonate (5,6). In addition to prior, ex-situ, calcinations in a flow of pure, dry O2 at 823K for 6-15 h, each sample introduced into the high-conductance section of the recirculatory reactor system received an in-situ calcination for 3 h at 823K under 100 torr 02 recirculating through a liquid N2 cooled tap. This was followed by cooling to RT in O2 and evacuation of the 02. Samples in that condition are termed preoxidised, whereas those subsequently subjected to in-situ reduction by H2 at 423K or at 773K are designated by LTR and HTR respectively. Comparisons were then made at several temperatures between activities of those variously pretreated materials in promoting: (a) isotope exchange between 5 mbarr 15Nl80 and 1602" lattice oxygens to yield 15N160 at various temperatures, and (b) in similar manner, conversions of equimolar (15Nl80 + 13C160) mixtures to yield various possible isotopic forms of N20, NO and CO2.

3. RESULTS AND INTERPRETATION

lSNlSo conversions on pre-oxidised samples: Some adsorption of the 15NlsO together with small extent of conversion(s) were measured upon introducing the gas, at room temperature (RT) and 5 mbarr pressure into the recirculatory reactor system (total volume ~ 1.5 dm 3) wherein was positioned a 300 mg aliquot of pre-oxidised CeO2, or Rh203 or 0.5% RhOx/CeO2. The most unique and clear-cut of these conversions of 15N180 at room temperature was a limited yield of 15N160(g) over the 0.5% RhOx/CeO2 aliquot only. This was reminiscent of reports that oxygen isotope exchange (0.i.x) occurred with surprising ease at RT between 14NlSO(g) and the 1602-(s) anions of iron or nickel oxides. In those cases such 0.i.x was envisaged to proceed via triatomic surface intermediates resulting from additional coordination of chemisorbed NO to 1602-{s)of the metal oxide [3,4]. If the 15N180 --> 15N160 exchange here observed at RT only over pre-oxidised 0.5% RhOx/CeO2 is assumed to have

411 occurred via a similar mechanism, then an essential role for 1602" at microinterfaces between dispersed RhOx and the CeO2 support in the formation of such surface intermediates seemed probable in view of additional present observations that neither CeO2 alone nor Rh203 alone yielded any detectable 0.i.x at RT when in pre-oxidised condition equivalent to that of 0.5% RhOx/CeO2. A further difference between mass spectra of the gas phase present over the preoxidised materials after 30min recirculation of 15NlSO over each at RT was the detection of a gradual but very limited increase in signal level @ m/e - 30 above background over CeO2 and Rh203 but not over 0.5% RhOx/CeO2. Thus the RT o.i.x, activity detected over the latter was replaced by limited conversion of 15N180 to 15N2 as the preferred RT chemical conversion over pre-oxidised CeO2 and Rh203. Changes from the gas phase composition attained by 0.5h contact at RT between 15N180(g) and the pre-oxidised CeO2, Rh203 or 0.5% RhOx/CeO2 samples, were monitored by MS at 30 see. intervals, whilst ramping temperature of the reactor at 10~ min -1. As illustrated by fig. 1a, readily detectable rates of decrease of 15NlSO(g) from its RT - pre-equilibrated value were observed to onset at 120~ over pre-oxidised 0.5% RhOx/CeO2 (cf fig.la) or over CeO2 (not shown), together with comparable increases in 15N160(g). In both cases, equality of the 15NlSO(g) and 15Nl60(g) signals was reached at ramp temperature ca. 530~ Clearly, activation of the 15N180(g) ~ 15Nl60(g) isotopic exchange to similar extents was the predominant effect over both oxidised ceria-based materials at 120 - 530~ Such similarity in 0.i.x activity across that temperature range was in sharp contrast to observations above that only pre-oxidised 0.5% RhOx/CeO2 was effective in promoting a limited amount of 0.i.x at RT. That contrast could be rationalised on the basis of (i) limited availability only on preoxidised RhO2/CeO2 of labile 16on"or 1602n" species capable at RT of undergoing exchange with 15N180 in low activation energy events; and (ii) shared capability of pre-oxidised CeO2 surface regions on both materials at 120-580~ to promote 0.i.x events requiring significant thermal activation. Since no significant decrease in 15N180 or increase in 15N160 was detected at RT or upon applying the temperature ramp to ~5N180 in contact with pre-oxidised Rh203, neither type of site for 0.i.x appeared to be available on pre-oxidised Rh203. Comparisons of 15N180conversions over LTR samples (cf. fig. lb): On the basis of TPR profiles measured upon pre-oxidised samples [5,6], the following redox conditions were expected for the LTR materials: (a) extensive reduction of the rhodium content of 0.5% RhOx/CeO2 to metallic rhodium, allied to limited reduction of adjacent ceria by hydrogen spillover, but with much unreduced CeO2; (b) reduction of CeO2 to very limited extent, since the > feature of CeO2 in TPR is usually delayed until ca. 450~ [7]. In line with (b), no significant differences were detected in the interaction of 15N180 with LTR-CeO2 relative to that with pre-oxidised CeO2. Behaviour similar to that reported for NO over metallic rhodium [8] was to be expected over LTR 0.5%RhOx/CeO2 within the context of (a) above. Results in Fig. l b consistent with this include more extensive decrease in 15NlSO(g) during 1 hour contact with LTR- 0.5% RhO• at RT than with the pre-oxidised material, allied to small but clearly detected yields of 15N2and 15N2180. These could arise from some re-oxidation of mainly metallic rhodium surface, 15N 18O + Mm ~ 15N(s)+ JsO/Mm, followed by reaction of 15N~s)with one another to give 15N2, or with adsorbed 15N180 to yield 15N2180. Observations in Fig. l b that increase in the latter became noticeable at 150~ under temperature ramp, whereas the yield of ~5N2 did not increase until 300~ also resembled temperature dependences reported for N20 formation and dissociation over metallic rhodium

412

t6

15

t8

la

10

0t

. . . . . t00

,

300

16 [ 151.,,N],180

500 ,,

t,4,

lb

12 [

t5N 2lgO

0

,

,,

t4. "-"~..~ 8

lc

.................................

} 0 ~

tS'N't60 - x;0 .....

I "-.ti0

'

' .~(i,,

-:

[9]. A further experimental observation, which was not expected but which could be rationalised within the context of 16on" or 1602n" species having been removed from rhodia/ceria contact perimeters by prior LTR of RhOx/CeO2, was the non-appearance of detectable 15NlSO(g) ---> 15Nl60(g) conversion over the LTR 0.5% RhOx/CeO2 samples either at RT or under temperature ramping [cf. fig. lb with 1a]. Comparisons of lSNlsO conversions over HTR CeO2 and 0.5% Rh/CeO2(ef. fig. lc): The principal additional redox changes expected to result from such pre-treatment under H2 at 550~ followed by evacuation at 550~ to remove any H20 formed, were more extensive reduction of surface/sub-surface regions, allied to increased numbers and types of oxygen anion vacancies in the reduced ceria support [10]. Concentrations of the latter seemed likely to be greatest in regions of HTR 0.5% Rh/CeO2. adjacent to rhodium metal particles, as a consequence of hydrogen spillover [11]. However, the MS studies indicated zero formation of 15Nl60(g) at RT or at ramp temperature < 250~ over either material, despite significant decreases in 15NlSO(g) accompanied by smaller invariant yields of tSN2(g) and 15N2180(g). No further increase in those limited yields of 15N2(g) or 15N2180 occurred until ramp temperature ca. 200~ over HTR-CeO2 or 300~ over HTR-0.5% RA/CeO2 (cf. data for RhO• in Fig. l c). However, those likewise represented the respective temperatures for onsetplus-continued growth of 15Nl60(g), allied to decreases in 15NlSO(g), thereby demonstrating recurrence of temperature-activated 15NlSO(g) ---> 15Nl60(g) (o.i.x.) over both HTR materials. This contrasted markedly with absence of such 0.i.x over the same materials when in the LTR condition (fig. l b). Apparently, the much enhanced extent of ceria reduction after HTR - with resultant large increases in concentrations of surface and sub-surface oxygen anion vacancies and of co-ordinatively unsaturated

Fig. 1" T-ramp induced changes in gas phase composition, subsequent to RT equilibration between 5 mbarr 15N 180 and 300 mg of 0.5% RhOx/CeO2 when in preoxidised (la); LTR (lb) or HTR (lc) condition. [Vertical scale MS peak heights in mutually consistent a.u.; bottom scale ~

413 Ce 3+ ions adjacent to oxygen anion vacancies [ 10] - had created new ceria-related active sites which opened up new thermally activated pathways for the 15NlsO -~ 15N160 isotope exchange at such defect sites. ISNlsO plus 13C160 over pre-oxidised materials: The trace amounts of 15N2and 15N160 produced over pre-oxidised CeO2 by RT contact with equimolar 15N180 + 13C160 were similar to those observed from RT contact with 15NlsO only. Under T-ramp the first significant difference from the latter to emerge in contact with the equimolar mixture was a delay of ca 100 ~ in onset of ramp-induced increase of 15N160(g) from its initial trace level. Such delay until 250~ could be understood in terms of competition by 13C160 against 15N180 for reaction with 1602"s species rendered labile on ceria at ~160~ and above. The fact that onset of ramp-induced increases in either 13C1602(g) or 13C160180(g) did not become apparent until ~450~ made it clear that CO2 products from CO oxidation on preoxidised ceria were retained by the CeO2 surface until such temperatures. Decreases in 15NlSO(g) at 260-450~ over preoxidised ceria appeared to originate from the isotopic exchange process 15NlgO(g) ~ 15N160(g), since 15N160 was the only 15N-containing species observed to increase at 260 ~ 450~ Subsequently, the 15Nl60(g) signal levelled off and then decreased at 450 ~ 650~ consistent with 15N160 species then reacting with chemisorbed CO or CO2. Isotopic analysis of the then-observed increase in gas-phase CO2 products showed AI3cI602/AT to be three fold greater than AI3C160180/AT, consistent with 15N180 + 13C160 --~ 13C160180 + 1/215N2being less efficient than reaction of adsorbed 13C160 with lattice 1602" or with 15N160(s) to yield 13C1602. No evidence for 15NE160(g) or 15NE180(g) was found over the pre-oxidised CeO2. Over pre-oxidised Rh203 the most notable difference observed under the equimolar 15NlSO h- 13C160 mixture was that uptake of 15NlSO(s) commenced at ~280~ which was ca. 220~ lower than onset of a much smaller decrease over the same material under 15N180 only. Furthermore that onset at ~280~ was accompanied by a parallel decrease in 13C160(g) and by onset of increases in CO2 products, thereby indicating that the pre-oxidised Rh203 sample promoted reaction(s) between 15NlSO(g) and 13C160(g) at temperatures 280 ~ 600~ Since no 15N2160or 15N2180 was detected, reaction with 1"1 stoichiometry appeared to be favoured so that 13C160180 could be expected as the primary product from 15N180 + 13C160 ~ 1/215N2 -t- 13C160180. Increases in 13C160180 and 15N2were observed but isotopic composition of the CO2 product - with 13C1602(g) increasing at similar rate to that for 13C160180(g) and 13C1802(g) increasing at only one-quarter of that rate pointed to more efficient reaction of CO with lattice ~602 of the rhodia surface than with lSo from 15Nt80. Comparison between the composition profiles versus ramp temperature observed over preoxidised 0.5% RhOx/CeO2 when in contact with 15N180 + 13C160 (el. fig. 2a) rather than with 15N180 only (cf. fig l a) reveals the absence of detectable 15N180 ~ 15N160 in the former. Furthermore, the profiles in fig. 2c confirm the lack of any evidence for the oxygen isotope exchange process over the 0.5% Rh/CeO2 sample when in the HTR condition in contact with the equimolar mixture, thereby doubly emphasising the inhibitory role of 13C160 co-reactant upon the 0.i.x process observed over the same material equivalently pre-treated but in contact only with 15NlsO. Blockage or removal by CO of the sites or species active for o.i.x, on 0.5% RhOx/CeO2 is implied by these results, but details thereof are as yet unclear. Other notable differences in conversions of the equimolar 15NlsO + 13C160 mixture over pre-

414

'iI ,cl,o

2a)

12-

8-

4- t'~'N2 ~

I~C160~ .

100 25-

300

500

13C~60

2b

20

IO-

13C~60~

" _1~2~, 'S'"~ 20-

100

................... .......... 9 ...... 300

500

. . . . . ~3C160

-

16-

12"

-

15~ 2

100

,

..--

.-.............."'"

c Q o,

300

500

N .o

oxidised CeO2 or Rh203, were the definite detection of 15N2180 over RhOx/CeO2 when ramp temperature reached 150~ and the subsequent doubling of that signal at 150-> 350~ without significant increase in 15N2. For preoxidised materials this low temperature, formation of 15N2180 was observed only over 0.5% RhOx/CeO2 which implied that selective promotion of the conversion 215N180 + 13C160 --> 15N2180 +13C160180 required synergism between the RhOx and CeO2 components -such as could be envisaged at contact perimeters between them. Remarkably, the onset temperature for release into the gas phase of CO2 products from conversions of the equimolar mixture over the preoxidised 0.5% RhOx/CeO2 were halved relative to those observed over preoxidised ceria oxidised 0.5% RhOx/CeO2, relative to the abovenoted conversions over pre-oxidised CeO2 or Rh203, were the definite detection of 15N2180 over RhOx/CeO2 when ramp temperature reached 150~ and the subsequent doubling of that signal at 150 ~ 350~ without significant increase in 15N2.above (i.e. to 225~ in Fig. 2a from 450~ over CeO2). Onset and growth of CO2 products in the gas phase over preoxidised RhOx/CeO2 at 225--->450~ were thus markedly enhanced relative to those over preoxidised ceria and significantly relative to preoxidised rhodia Evidently the 0.5% RhOx content of RhOx/CeO2 was especially effective in facilitating release of CO2 product to the gas phase, yielding much more 13C1602 than 13C160180. Whilst MS peak heights for both those species continued to increase 300-450~ that for 15N180 levelled off and began to decrease at T>300~

Fig. 2. Changes in gas phase composition following RT introduction of 5 mbarr each of 15NlsO and 13C160 and T-ramp to 700~ over 0.5% RhOx/CeO2 when pre-oxidised (2a); LTR (2b); and HTR (2c). (Vertical scale MS peak height in mutually consistent a.u., except for small upward displacements of CO2 profiles to avoid overlap) Bottom scale ~

415

This would be consistent with catalysed conversion NO + CO changing over from 2:1 to 1:1 stoichiometry, [11 ] or possibly with onset of catalysed N20 dissocation [9]. 15N180 + 13C160 over L T R materials: Over LTR 0.5% RhOx/CeO2 in presence of the equimolar gas mixture, onset of a low yield of 15N2180 was observed at ramp temperature -~ 100~ with subsequent growth at 100-300~ The contrast between this and strong predominance of 15N160 production from 15N180 alone over LTR CeO2 in that temperature range implied that the 13C160 component of the equimolar mixture, allied to 0.5% Rh upon LTR CeO2, somehow blocked/removed the ceria-related sites responsible for o.i.x, on LTR CeO2 and replaced them by sites active for 215N180 + 13C160---~ 15N2180§ 13C160180. Evidence in support of this was that the constant rate of decrease of 15NlSO(g) observed across the ramp temperatures 120 --~ 600~ was approximately twice the observed rate of decrease in lacl60(g) (cf. fig. 2b). Escape of the CO2 products into the gas phase once more again become evident at ramp temperatures > 230~ The fact that 13C1602(g) increased three-fold faster than 13C160180(g) at 230 --~ 600~ pointed again to substantial isotope exchange, 13C160180(S) + 1602"(S) --} 13C1602(S) + 1802"(S) before escape. 15N180 plus 13C160 over HTR materials: The above noted, mechanistically significant ratio of 2:1 between (-AISNISO/AT) and (-AI3CI60/AT) over LTR 0.5% Rh/CeO2 was not reproduced when a similar run was carried out over the same 0.5% Rh/CeO2 material following its re-oxidation for lh in O2 at 550~ plus lh reduction in HE at 550~ and lh evacuation at 550~ On the contrary, the MS peak heights for 15N180 and 13C160decreased in parallel across 150 -~ 600~ ramp temperatures with equal slopes (cf. fig 2c), suggesting that 15NlSO(g) § 13C160(g) ---} 1/215NE(g)+ 13C160180(g) was favoured over this HTR sample rather than reaction with 2:1 stoichiometry. Support for this came from observations of increasing release of CO2 and N2 products in approximately a 2:1 ratio at ramp temperatures > 200~ A much smaller increase in 15N2180, from its trace level after RT contact with the equimolar mixture, was observed at ramp temperatures 100 --~ 300~ after which it decreased. Conversions from pulsing 12C160 into continuously flowing 14N160: For all samples investigated by this method one of two pretreatment regimes was employed: (i) pre-oxidation conducted at 300 ~ in 100% 02 for 1 hr; or (ii) pre-reduction conducted at 200~ in 100% H2 for 1 hr (LTR). The results depicted are those obtained from the fourth pulse of CO injected, by which time stable pulse profile were established which could be fully reproduced from the fifth and subsequent pulses. Such measurements thus provided insights into the time sequence for CO-pulse initiated growth of product-related MS peaks (upward displacements) and subsequent flushing away, by the continuous NO/helium flow, of gas phase products and any unconverted CO. 0.5% Rh / CeO2 : Figure 3a depicts the time-profiles of responses observed at m/e = 44 over pre-oxidised 0.5% RhOx / CeO2 at various temperatures between 100 ~ and 175 ~ Across this temperature range onset of activity was observed as low as 125 ~ whilst by 148 ~ and 175 ~ there is a clear division of responses into a prompt peak and a slow eluting peak. Time-profiles of responses observed at m/e = 44 for the same sample under the same conditions at high temperatures between 175 ~ and 297 ~ are depicted in Figure 3b.

416

i

i ,._

i .....

440 _

_

_

.

.

.

9

.

" .

i-

.

:

.

480 .

i

l.-.c I-’

.

.

.....

440

"

Time!s ~ _

~"

-

,

55_.0 ,

.

-.

.

' .

~-

.

,--

.

560.

-~

480 520 Time/s

.

.

_

.

.

.

_

"~,~-~

9 560

[ .o--~s !O~ I

-9

~

- - ~ ' i

440

" . '

~

-..

j

.

r-

480 520 Time&

r

i

- - - r

560

Temperature increase in this range results in decreasing magnitude of the prompt response whereas the slow component remains high and very broad. The following points emerge from cross-correlation between these m/e = 44 time-profiles and those observed at other m/e values over the same sample from the same pulse: (1) Decreases at m/e = 30 were observed indicative of 14N160uptake during the CO pulse. However time-profiles of that uptake did not exhibit a nature but more closely corresponded to the slow m/e = 44 response. (2) The response profiles observed at m/e = 12 showed that at and above 148 ~ there were no features paralleling the prompt feature in Figures 3a and 3b, thereby eliminating CO2 as their source and indicating that the prompt feature at m/e = 44 was exclusively 14N2160; (3) Profiles of small responses at m/e = 28 closely resembled that of the prompt m/e = 44 signal, an observation which when allied to the absence of prompt signal at m/e = 12, pointed to a disproportionately small yield of prompt N2 product at 148 and 174~ relative to the signal size expected if the overall uptake of CO and NO had reacted with 1"1 stoichiometry over the preoxidised RhOx/CeO2. Between 175 ~ and 297 ~ the 14N2 peak progressively increases in magnitude and broadens resulting in some overlap with the slow rn/e = 44 component. The overall behaviour of the prompt 14N2160and 14N2responses were compatible with the low-temperature activity of preoxidised RhOx/CeO2 having high initial selectivity towards 14N2~60 but with some increase in selectivity towards 14N2 at high temperature. The slow component of 14Nl60 loss (see below) likewise appeared to favour 14N2160.

Fig. 3: MS-response time profiles @ m/e = 44 upon injection of the fourth 10s duration pulse of CO into continuous flow of NO/helium over preoxidised 0.5% RhOx/CeO2 after 440s onstream therein: 3a and 3b, fixed P(NO) = 15.2 torr and T=100 --~ 175~ or 175-+297~ resp.; [3c, T fixed at 162~ P(NO) varied as indicated]

i

417

Figure 3c illustrates the time-profiles obtained at rn/e = 44 over oxidised 0.5% RhO• / CeO2 from isothermal introduction of identical CO pulses when different partial pressures of 14N160 were established in the continuous flow of NO/helium over the sample before admitting CO pulses. Decreasing that variable clearly resulted in decreases in magnitude of the prompt and slow responses and in a shift to longer times for maximum of the slow component. Data for uptake at rn/e = 30, recorded simultaneously with the data in Figure 3c, generally mirrored that observed for the m/e = 44 slow component in Figure 3c but with no prompt uptake feature being observed. Overall behaviour of the slow-eluting features, appeared consistent with contribution towards slow formation of products by post-CO-pulse surface processes involving 14N160 from the continuous flow reacting with 12C160 retained on the surface from the pulse. Size variations of the prompt, m/e = 44, component evidenced in Fig. 3c may be understood in terms of different extents of interaction of an incoming CO pulse with different surface coverages, 0NO, of the preoxidised RhOx/CeO2 material established under NO/helium flows having different PNo before CO-pulse injection. Prior formation of N20 from NO over preoxidised RhOx/CeO2 at 162 ~ being unlikely (of. Fig. 1a), these prompt effects of CO in each incoming pulse appeared to include CO(g) + NO(ads) --->CO2(ads) + N(ads) and facilitation of N(ads) + NO250~ and came to resemble that of HTR CeO2.x This pointed to HTR induced production of labile oxygen species associated with defect sites on ceria. Delayed onset-temperatures, allied to substantial overall inhibition of the above-noted levels of 0.i.x towards 15N180 alone, were evidenced whenever equimolar ISNl80 + 13C160 was introduced over equivalently pre-treated 0.5% RhOx/CeO2, thereby pointing to efficient scavenging of labile 160-containing surface species by 13C160, especially at T < 200~ Yields of 15N1180 observed at those temperatures over 0.5% RhOx/CeO2 in LTR and HTR condition could be understood, as an indirect consequence of such scavenging: by virtue of surviving ISN fragments reacting via 15NlSO + lSN ~ lSN21sO, thereby redirecting selectivities for lSNlSO conversions away from o.i.x, and towards N20 formation. Contact perimeters between RhOx and

418 CeO2 appeared the likely locations for such selectivity modifications, since equivalent effects were not observed over CeO2 alone or Rh203 alone. Evidence yielded by the pulsed experiments for predominance of N20 in the 'prompt' product detected from CO pulse contact with NO//0.5% RhO2/CeO2 could likewise be understood in terms of the ~2C~60 pulse having scavenged oxygens from the laN~60-covered surface, thereby facilitating N20 formation through ~4N + ~4N160 --->N20 reaction events. ACKNOWLEDGEMENTS Mobility of researchers between the laboratories involved has been aided by support under EC contracts SC1 CT91, 0904 and ERB CH RX CT 92 0065 and Eolas/British Council Grants '91 and '92. UCC workers also gratefully acknowledge the access given to pulsedreactant equipment at University of Reading and the expert assistance there received in applying it for the present studies. REFERENCES

1. (a) B. Harrison, A. F. Diwell and C. Hallet Plat. Met. Rev., 32 (1988) 73 (b)K.C. Taylor, fatal. Rev. Sci. Eng., 35 (1993) 457. 2. C.H.F. Dedon, D.N. Belton and S.J. Schmeig, J. fatal., 15 (1995) 204. 3. K. Otto, M. Shelef and J.K. Kummer, Z. Phys. Chem., N.F., 72 (1970) 316. 4. K. Otto and M. Shelef, J. fatal., 35 (1974) 460. 5. Trovarelli fatal. Reviews, 38 (1996) 439. 6. F.D. Farrell PhD Thesis NUI (1996); J. N. Hickey PhD Thesis NUI (1997) 7. J. Cunningham, D. Cullinane, F. Farrell, M.A. Morris, A. Datye and D. Kalakkad

NO• .... trap + HEAT ....... > NOx + trap

(3)

lean NOx catalyst NOx + HC + 02 ,> N2 (N20) + CO2 + H20

(4)

533 A fresh trap was initially treated with 1000 vppm of NOx in background gas at 300 ~ as shown on the left side of Figure 3. Following a N2 purge for 10 minutes a NOx desorption profile was generated by the injection of 6500 vppm Cl (propylene) in background gas at a trap inlet temperature of 300 ~ The profile is shown on the right side of Figure 3. In a vehicle the NO• would be adsorbed during a normal driving cycle whenever the exhaust temperature is between 150 and 500 ~ The injection of fuel is controlled so that the downstream catalyst is at about 210 ~ optimum for NOx reduction. The amount of NOx desorbed depends on the intensity of the exotherm which depends on the amount of hydrocarbon injected and oxidized. This is shown in Figure 4 where the amount ofNOx desorbed increases with an increasing amount of hydrocarbon fuel. Repeat desorption runs, after adsorption of NOx as in Figure 3, indicate that the trap is completely regenerated after an injection of 10,000 vppm hydrocarbon. The bulk gas phase temperature remains

Figure 3 NOx A D S O R P T I O N 1000 PPM NOx IN BACKGROUND GAS,

DESORPTION OF NOx BY HC INJECTION

300 C, SV= 25,000/h 400

PPM NOx REMOVAL 350

GAS PHASE NOx (PPM)

350 300

3O0

250

250

200

200

150

150 100

100

~__

~o 0

....

5

, ,,

10

TIME (MIN)

5O ,,,

0

i 20

ii 21

ii 22

23

30

35

TIME (MIN)

fairly constant because the desorption of NOx is endothermic which counters the exothermic heat of the oxidation reaction. The downstream lean-NOn catalyst temperature is at 210 ~ which gives a reduction of about 60% of the inlet NOx at a 4:1 C1/NO ratio. Additional hydrocarbon (HC2) must be available to the NO• reduction catalyst. This can be added immediately before the leanNOx reduction catalyst or alternatively the trap system can be under-designed to allow for sufficient hydrocarbon slip or breakthrough to accomplish the downstream NOx reduction. Results for a combination trap/lean-NOx system are shown in Figure 5. A continuous flow of 250 vppm NOx in background gas is passed over the trap at 300 ~ The rising portion of the curve on the left side shows the adsorption of NOx. At the maxima 7000 vppm Cl, as propylene, (HC 1 in Figure 2) is injected desorbing all of the adsorbed NOx as evident by the equal areas

534

F i g u r e 4" N O x DESORPTION BY HYDROCARBON INJECTION NOx DESORBED (PPM)

800 700

10,000 PPM C1 (C3He)

600

INJECTED

500 400

~,000

300

4,000

200 100 0

1,000 11

o

0.5

1

1.5

2

2.5

3

3.5

4

TIME (MINUTES)

Figure 5" T R A P + Pt LEAN NOx CATALYST % NOx REMOVAL 100

LEAN-NOx CATALYST C I l N O = 4"1 ( B A S E D ON

LEAN-NOx

CArA,XSr / ~

80 60

/~

F~D NOb

A

40 20 0 -20

TRAP INLET =300C C A T A L Y S T IN=210 C

-40 -60

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

TIME (ARBITRARY) N O x = 250 PPM IN B A C K G R O U N D G A S ( C O N T I N U O U S ) HC1 = 7000 PPM C1 ( C a l l s ) I N J E C T E D A T M A X l M I M S

18

535 above and below the zero point. When the cycle is repeated but with secondary HC2 injected at a 4:1 CI/NO (based on feed NO) the entire profile is elevated reflecting an average reduction of about 45%. Optimization of the Cl to NO ratio to account for the extra NO resulting from the desorption would further increase the NOx reduction possible. Naturally a more selective lean NOx catalyst would also be beneficial. The proprietary architecture of the trap system was evaluated for thermal stability as shown in Figure 6. Adsorption experiments were conducted with 250 vppm NO• in background gas (minus the SO2) at a ramp rate of 10 ~ The system maintains good adsorption capacity after aging at 700 ~ in air with 10% H20 for 100 hours and 50% of the original capacity is maintained after aging at 800 ~ This was encouraging because diesel engines are not expected to exceed 700 ~ exhaust temperatures in use.

70

F i g u r e 6" T H E R M A L A G I N G OF T R A P : C H A N G E IN ADSORPTION CAPACITY NOxADSORPTION

60 50 40 30 20 10 100

150

200

250

300

350

400

450

500

TEMPERATURE ADSORPTION:250

P P M N O • IN B A C K G R O U N D

GAS

AGING: AIR+10% H20 FOR100 HOURS

The presence of SO2 deactivated the trap as shown in Figure 7. The trap was initially treated with 1000 vppm NO• in background gas (minus the SO2) for 15 minutes. The desorption profile generated by injection of 6500 vppm Cl, as propylene, is shown as the fresh catalyst. After aging in 150 vppm SO2 in air and 10% H20 at 500 ~ for 24 hours the trap effectively loses all of its NOx adsorption/desorption capacity. Even with a calcination at 800 ~ in sulfur free air only about 30% of the initial capacity could be recovered. The absence of regeneration is primarily due to retained SOx which occupies NOx adsorption sites although some thermal deactivation at 800 ~ also occurs.

536 Figure 600

7:

AGING

IN SULFUR

DIOXIDE

NOx DESORBED (PPM)

500 400 300 200 100 0

0.5

1

1.5

2

2.5

3

3.5

4

TIME (MINUTES)

INITIAL DOSING: 1000 PPM, 300 C, B A C K G R O U N D G A S A G I N G : 150 PPM SO2, 500 C IN 10%H20,10%O2, N2, 24h NOx D E S O R P T I O N : Cl = 6500 PPM (C3He)

4. CONCLUSIONS 1) A system designed to manage NOx between 150 and 500 ~ in a lean bum environment has been explored as an alternative to current lean NOx catalyst limitations. 2) NOx desorption can be accomplished by the exotherm generated by the oxidation of injected hydrocarbon within the trap while always maintaining the environment lean. 3) The trap inlet injection temperature must be controlled to be compatible with the maximum activity of the downstream lean-NOx catalyst. 4) The oxides of sulfur compete more strongly than NOx for the trap sites leading to a permanent loss in NOx adsorption capacity. 5) A SOx tolerant trap component is needed for this approach to have commercial significance. REFERENCES 1. R.M. Heck and R.J. Farrauto, Catalytic Air Pollution Control: Commercial Technology. Van Nostrand Reinhold, NY, NY 1995, Chapter 7. 2. R.J. Farrauto and K.E. Voss, Applied Catalysis B: Environmental 10:1-3 (1996)29 3. Y.K. Lui, J. Dettling, O. Weldlich, R. Krohn, D. Neyer, W. Engeler, G. Kahman and P. Dore, SAE 962048 (1996) 4. M. Iwamoto, Catalysis Today 28 (1996)29 5. M. Amiridis, T. Zhang and R. Farrauto, Applied Catalysis B: Environmental 10:1-3 (1996) 203 6. K.M. Adams, J. Cavataio and R. Hammerle, Applied Catalysis B: Environmental 10:1-3 (1996)157 7. J. Feeley, M. Deeba and R. Farrauto, SAE 950747 (1995) 8. M. Miyoshi, S. Matsumoto, K. Katoh, T. Tanaka, J. Harada, N. Takahashi, K. Yokota, M. Sigiura and K. Kasahara, SAE 950809 (1995)


537

Investigations of NOx storage catalysts Erik Fridell a, Magnus Skoglundha, Stefan Johansson a'b, Bj6rn Westerberga'c, Anders T6rncrona~d and Gudmund Smedler a aCompetence Centre for Catalysis, Chalmers University of Technology, S-412 96 G6teborg, Sweden bDepartment of Applied Physics, Chalmers University of Technology and G6teborg University, S 412-96 G6teborg, Sweden CDepartment of Chemical Reaction Engineering, Chalmers University of Technology, S 412 96 G6teborg, Sweden dDepartment of Engineering Chemistry, Chalmers University of Technology, S-412 96 G6teborg, Sweden

ABSTRACT NOx storage catalysts are used to reduce nitrogen oxides from lean-bum vehicles. The nitrogen oxides are stored in the catalyst during lean conditions and subsequently released and reduced during short periods of rich conditions. In the present study, we systematically investigate the sequence of elementary steps in the NOx reduction cycle, and the extent to which these steps influence the maximum NOx reduction potential of the catalyst. As a model system, we use barium oxide as the NOx storing compound in a Pt/Rh/A1203 system. Kinetics of NO oxidation, NO and NO2 adsorption, NO and NO2 release and reduction are studied under controlled conditions with systematic variations of temperature, gas composition, and storing/release times. The transient experiments comprise a storing phase using a lean NO/C3H6/O2/N2 gas mixture, and a regenerating phase where the 02 flow is turned off. Experimentally, a significant amount of NOx is found to be stored in the Ba-containing material. A maximum in NOx storage is observed around 380~ For most of the experiments, there are clear NO and NO2 desorption peaks upon switching from the storing to the regeneration phase. TPD studies of NO and NO2 reveal a significant difference between prereduced and pre-oxidised samples where the former produce predominantly N2 and N20 at around 200~ while NO and 02 desorb from the latter around 500~ In situ FTIR spectra show nitrate peaks in the region 1300 - 1400 cm -~ when NOx is stored under lean conditions.

1. INTRODUCTION There is currently a great interest in improving automobile fuel economy and to reduce emissions of carbon dioxide into the atmosphere. One step in this direction is to use a technique where a gasoline engine is operated at lean-bum conditions. Using this concept it is possible to improve fuel economy significantly [1, 2] compared to the normal stoichiometric operation. A setback with lean-burn technology is that the common three-way catalyst is

538 unable to efficiently reduce nitrogen oxides (NOx) at oxygen excess resulting in significant emissions of NOx from the vehicle. One concept to solve this problem is the NOx storage catalyst. It was first presented in several patent applications by Toyota and later by Miyoshi et al. [2], Takahashi et al. [3] and B6gner et al. [4]. NOx reduction capability of 90% [3] has been reported from vehicle tests. The NOx storage catalyst usually contains noble metals for reduction and oxidation and a NOx storage compound (an alkaline earth compound) supported on alumina [2]. During leanburn conditions, NOx is being stored in the catalyst. To regenerate the catalyst, short periods of rich conditions are employed during which the stored NOx is released and subsequently reduced on the noble metal surfaces. The objective of this work is to investigate some key steps in the reaction mechanisms of catalytic reduction of NOx under oxygen excess by a systematic variation of the essential components for NOx storage and reduction. We describe the function of a model system for NOx storage catalysts containing Pt, Rh, A1203 and barium oxide as the storage component. The catalyst was made as simple as possible, yet containing the essential constituents for NOx storage and reduction, in order to investigate the mechanisms of the NOx storage concept. Also catalysts without noble metals or barium, respectively, were prepared and studied. The reaction mechanisms were investigated by transient reactor studies, temperature programmed desorption (TPD) and in situ Fourier transformed infrared spectroscopy (FTIR). TPD measurements were performed for both pre-reduced and pre-oxidised samples in order to elucidate the different surface conditions during the different phases of the transient storage reduction cycles. The reaction sequence and the nature of the stored NOx compound are discussed.

2. E X P E R I M E N T A L P R O C E D U R E

Flow reactor studies were performed using monolith samples with an alumina washcoat, a storage compound (BaO) and noble metals (Pt, Rh). In order to elucidate the importance of the various ingredients, samples were also prepared without storage compound or noble metals, respectively. For the FTIR studies, similar powder catalysts were pressed into thin discs. Three different monolith samples were prepared according to a procedure described elsewhere [5]; the first with 160 mg alumina, 40 mg barium oxide, 4.0 mg Pt and 2.0 mg Rh, the second with 160 mg alumina and 40 mg barium oxide and the third with 200 mg alumina, 4.0 mg Pt and 2.0 mg Rh. We do not know the exact nature of the barium compound in the catalyst but refer to it as BaO in this paper. The fresh catalysts were reduced in H2 at 450~ for 35 minutes and stabilised in the lean gas mixture (see Table 1) at 550~ for two hours. The flow reactor used in most experiments is described elsewhere [5]. Briefly, it consists of a horizontal quartz tube encased in a divisible tubular furnace. The catalyst is sealed in the middle of the heated zone with quartz wool and the gases are introduced via mass flow controllers. Reactants and products are analysed on-line with respect to N20, CO, CO2 (IR), and NO, NO2 (chemiluminescence) The TPD measurements were performed with the same monolith samples in a different quartz flow reactor described elsewhere [6-7]. The samples were either prereduced (4% H2 in Ar) or preoxidised (5% 02 in Ar) at 500 ~ for five minutes. They were then exposed to 1%

539 of NO or NO2 in Ar at a flow of 100 ml/min, for 5 minutes in room temperature prior to a heating ramp (40~ in Ar flow. The FTIR-experiments were performed with thin discs (approximately 15 mg/cm2) of catalyst in a reaction chamber with CaFE-windows [8]. Two different catalysts were used; one with Pt/BaO/A1203 (2% Pt, 20% BaO) and one with Pt/A1203 (2% Pt). The fresh catalysts were initially reduced in 30% H2 in N2 (total flow rate of 100 ml/min) at 450~ for 30 minutes, stabilised in a gas mixture with 5% 02, 1000 ppm NO and 3000 ppm C3H6 in N2 (total flow rate 1000 ml/min) for 30 minutes and finally degassed in N2 (1000 ml/min) at 550~ for 30 minutes. All FTIR-experiments were performed at 380~ at a total flow rate of 1000 ml/min with a scan speed of 1 cml/s and a nominal resolution of 4 cm -l. Table 1. Gas concentrations in the different exPeriments.., Experiment NO (NO2)[ppm] Flow reactor, lean phase 600-1100 Flow reactor, rich phase 600-1100 FTIR, lean phase 400 FTIR, rich phase 400 , ,

"

C3H6 [ppm] 900 900 900 900

,,

02 [ % ] ' 8.0 0.0 5.0 0.0

3. RESULTS 3.1. General

In order to obtain effective NOx storage the catalyst is used in connection with transients in the gas composition, i.e., during lean-burn periods NOx is stored in the catalyst and during short periods of rich conditions, the stored NOx is released and reduced. 1000

400 E i3_ o E O

~

’ I

:

800

Iklt'~ --~.......~-----~----._.,~ii., H

,

V

No

:

I

, I

200

I --

I

I

0

200

,

’

l jr !

I

I

i

I

I

I

I

I

I

I

I

I

I

!

I !

I

I

-

I

!

I

400

I

I

’___I’ ....

60O

I

,

t

,4--1

r (1) O r" O O

’ I

I

I

_._

I

400

time

I

i

I

.

600

I

I

'

800

(s)

Figure 1. The NO and NO2 concentration traces during a transient with the gas compositions quoted in Table 1 over a Pt-Rh/BaO/A1203 catalyst. The dashed vertical lines mark the switches in gas composition where the rich phase is indicated by the double arrow.

540 Fig. 1 shows the NO and NO2 concentration traces in the product gas over a PtRh/BaO/A1203 catalyst, at an inlet temperature of 400~ during such transients with the conditions described in Table 1. The difference between the lean (NOx storage) phase and the rich (regenerating) phase is that the oxygen content in the gas is turned on and off, respectively (and compensated by changes in the N2 flow to maintain a constant flow rate). When 02 is switched on, at 320 s, for example, there is a relatively slow increase in both the NO- and NO2 signals. The reason for this slow increase is that NOx is being stored in the catalyst. The amount of stored NOx is obtained by integration of the area related to storage in this type of transients. For the experiment in Fig. 1, the ratio between stored NOx and barium is approximately 1:14. During the rich phase, all NOx is being effectively reduced by the propene. Also the NOx stored in the catalyst during the previous lean phase is released and reduced. Break-through peaks can be seen for both NO and NO2 when switching from lean to rich conditions. These desorption peaks will be commented upon below. The reduction of NOx under the lean phase is negligible at 400~ and the gas mixture used here (see Fig. 3 below). The formation of nitrous oxide (N20) can be observed both when switching from lean to rich conditions and vice versa (not shown). It is well known that N20 may be formed when switching from an oxidised to a reduced surface, for example, during light-off [9] (see also Fig. 3 below). The HC concentration in the product gas (not shown) is zero during the lean phase and reaches about 750 ppm during the rich phase. An HC break-through peak is present when switching from rich to lean conditions. There is a temperature decrease within the catalyst of about 15~ during the rich phase compared to the lean phase at the present conditions due to the lack of combustion during the rich phase (the magnitude of this decrease depends on the space velocity). When performing similar cycles with a catalyst containing no specific NOx storage component, i.e., a Pt-Rh/A1203 catalyst, no NOx storage can be observed. Break-through peaks of NO and NO2 similar to those in Fig. 1 can be observed also with this sample. For a sample without Pt-Rh but with BaO, no storage (with either NO or NO2 in the feed gas) and only minor reduction of NOx (during the rich phase) is observed. 3.2. Influence of temperature The NO and NO2 signals during storage transients are measured at different temperatures using the conditions in Table 1 (with NO in the feed). The amount of NO• stored during each lean phase is presented vs. temperature in Fig. 2 for cycles with 240 s in the lean phase followed by 60 s in the rich phase. In the temperature interval studied, a maximum in NOn storage is found at around 380~ and very low values above 500~ are measured. Also shown is the corresponding storage values when using NO2 in the feed gas rather than NO. The latter shows similar appearance as with NO.

541 ..... ...-..%

.~

20

-sL. ~3

15

~3 L. O

10

-!-3 03

O Z

X

I

9

.4"..

I

".o,.

I

I

*o.~

"..... N O 2 in feed

__

X - ~

5 I

,.

I

300

,

350

I

I

I

400

450

Inlet Temperature

500

(~

Figure 2. The NOx storage vs. inlet gas temperature with either NO or NO2 in the feed (compositions are given in Table 1). The CO2, NO, NO2, N20 signals and the catalyst temperature, during a heating ramp with the lean gas mixture (i.e., without transients) over the Pt-Rh/BaO/A1203 catalyst, are shown in Fig. 3. Observe that there is a maximum in NO oxidation to NO2 around the temperature (375~ where there is a maximum in NOx storage (see Fig. 2). Light-off can be observed around 250~ which is in the same range where reduction to N20 and N2 take place. Similar experiments with the Pt-Rh/A1203 catalyst, with the same noble metal load, give a significantly more pronounced light-off behaviour. This may imply that either the dispersion is lower when BaO is present, that BaO partially covers the noble metal, or that BaO in the close vicinity of Pt/Rh inhibits the capacity of the noble metal to catalyse oxidation reactions. 600 -I

0

50o -

I

Pt/Rh/BaO/AI203 catalyst NO in feed - 9149 9 .:'" ;~_

..'\

400 -

F-~--" 300 E

~

..-" .,t 9

9149 ,,oo

o~"

".j

*.

NO

"',~. ~" 9 .,... , ;

9

I

~

I

.

. ,, 3000 ---~" t ....

--

_./- -

9

B

/

~

~

~

_...-:'0-

...#%"

.."" .000 . . 99~1147 96-1 4 9 99

~.....~.., _.... ~ " .-". .:':':: . :;:::.............

i It

9

2500

_ --

2000

C)

r~ O

9

Cat. Te m

~

- 1500 O

200

/' " o..""" \o2

C 0

o

o~

o

'b. 9 o9

99

_

& s

-,..,

9

o..._..

, .t. ~ ; ~ ; ",,

100 ~ ; I

~.1 . . . . . . . . 100

:.

"'.

."

o"

,,

." ,, .?,. ,. ,.

"";"- ,..". 999

":'~1-

200

"

"-............-.-".-

1000

%" "23

3

500

.~,~N20

" ,~.. I

300 Inlet T e m p e r a t u r e

.... (~

Figure 3. Concentration traces during a heating ramp (5~ catalyst in the lean gas mixture.

;,," I" -,-',- . . . . . . 400

I--

500

over the Pt-Rh/BaO/A1203

542 3.3. Influence of gas composition The storage capacity was measured as a function of 02 and C3H6 content in the feed gas. At very low 02 concentrations (i.e., a net reducing gas) there is no net storage can be observed. The storage yield in lean gas mixtures increases slowly with increasing 02 content. The magnitude of the NO and NO2 break-through peaks (see Fig. 1) also increase with increasing 02 concentration (much more rapidly than the storage yield). Regarding the C3H6 concentration, there is a maximum in storage around 500 ppm C3H6. There is storage observed even if the C3H6 is turned off but the regeneration and reduction is then much slower. This variation is probably connected both with the different temperatures in the catalysts caused by the different reaction heat dissipated at the different C3H6 flows and with the somewhat lower oxygen coverage at high propene concentration. There are NOx breakthrough peaks of similar magnitude observed at all C3H6 levels examined. 3.4. Influence of phase duration The influence of the storage and regeneration times on the NOx storage over a PtRh/BaO/AI203 catalyst at 400~ is studied. The catalyst is completely regenerated after 40 s, i.e., longer regeneration times do not influence the amount of NOn being stored. The storage component is saturated after about 200 s during these conditions. 3.5. Break-through peaks There are break-through peaks in NO and NO2 when switching from the storage phase to the regenerating phase. Regarding these peaks the following observations were made: 1) they show a significant increase with increasing temperature (below 500~ 2) they show an increase in magnitude with increasing oxygen content in the lean phase, 3) when the propene fraction in the feed gas was varied from 100 to 1300 ppm, the magnitude of the peaks remained constant, 4) peaks are also seen for a catalyst without BaO, 5) the storage time does not influence the peak heights, 6) the amount of NOn in the peaks for the data in Fig. 1 corresponds approximately to 3.10 -6 moles for each cycle. 3.6. TPI) studies TPD measurements were performed for the NOx-storage catalyst with Pt-Rh/BaO/A1203 and for samples without the storage medium (BaO) or noble metals, respectively 9TPD experiments were performed by exposing pre-reduced and pre-oxidised catalysts, respectively, to either NO or NO2 at room temperature. i i i I , , , , -t I i

30X10"3

NO-TPD prereduced Pt/Rh/BaO/A~O 3 sample

'..

i

.... ;

2O =.. < .-- 15 O

~.:-

--

i.

0

i

ii

.

..

....

X ,Oo

"'.. .... .....

i .... I 300 .....200 T e m p e r a t u r e (~ ,,.

100

-

NO

~.......’-t: 9

~

-

N2

5)

473

H2

0.5

5

H 2, 373 K

0 (120-260)

a Adsorbate nature and adsorption temperature are reported. b The apparent Pd dispersions were obtained by back extrapolation of the linear part of the adsorption isotherm to zero pressure. Equilibrium time 1 min. The assumed reaction stoichiometries are: 2Pd + H 2 .... > 2 Pd-H Pd + CO .... > Pd-CO 2Pd-H + 3/20 2 .... > 2 Pd-O + H20 c equilibrium time l h For the catalysts reduced at 473 K, an increase of the metal loading leads to a decrease of the H/Pd ratio. This picture is reversed after reduction at 1000 K. The relative efficiency of the H 2 treatment at 1000 K in blocking the H 2 chemisorption is illustrated in Figure 3. For comparison the data for Pd/A1203 are included. H 2 chemisorption on noble metals (NM)/CeO 2 has been extensively studied (18) and special attention was generally given to the so called strong metal-support interaction (SMSI), e.g. the suppression of H 2 and CO chemisorption after a high temperature (usually 773 K)

563 reduction. This phenomenon which is attributed to the migration of the reduced support to cover the metal particles such as depicted in the scheme (19), was observed on Rh/CeO 2 only above 973 K (20). After reduction at 473 K, the Pd dispersion decreases with Pd content. The independence of the H 2 adsorption stoichiometry was reported for Pd/La203 for a range of Pd-loadings 0.25-8.80 wt% (21). Accordingly, the decrease of the H/Pd with Pd-loading suggests an increase of the Pd particle size.

CeO2_x Pd

7;-/7///'/////77///,I. ":". "

A reasonable interpretation of the variation of the H/Pd ratios after a reduction at 1000 K invokes a SMSI type phenomenon. Attribution of this effect only to a Pd sintering may be ruled out by the smaller relative decrease of H/Pd observed in the case of A1203 as support, compared to Pd/Ceo.6Zro.402-A1203. There are, indeed, some reasons which suggest that the SMSI type effect may be operative in our case: i) the reduction temperature (1000 K) is comparable to that observed for Rh/CeO 2 (20), ii) high surface area which favors the reduction process, and iii) an easy Pd encapsulation is expected for small Pd particles. 0.6

,--

0

100%

. ~ ,4...o

0.5

0

80%

t"q ,--,

60%

0.4

-r-~

0.3

o

0.2 0.1

r

0

tl} t/}

J 1

o

n 2

Pd loading (wt%)

3

"-1

40% 20%

0% 0

1

2

3

Pd loading (wt%)

Figure 2. H/Pd ratios measured over the 0.7, 1.4, 2.1 and 2.8 wt% Pd-loaded Ceo.6Zro.402A120 3 catalysts after a reduction at 473 (e) and 1000 K (m). Figure 3. Relative decrease of H 2 chemisorption induced by a reduction at 1000 K, (R) Pd/Ce0.6Zr0.402-A1203, (A) 0.7% Pd/A1203. Accordingly, Lavalley et at. showed that efficient Pd encapsulation was induced by a reduction of highly dispersed Pd supported on a high surface area CeO 2 (22). The reduction at 1000 K causes the sintering of the CeO2-ZrO 2 supports (7,23) which would favor metal encapsulation. However, the ZrO 2 may also well participate in determining the chemisorption properties of the present system. Infact, Lee et al. (24) found that the addition of ZrO 2 to

564 Pd/A120 3 catalyst resulted in a reduction of Pd 3d5/2 binding energy relative to that observed in metallic Pd suggesting that Pd had become negatively charged. Summarizing, the H 2 chemisorption studies here reported show a partial suppression of the H 2 adsorption after a high temperature reduction suggesting that, in addition to a metal sintering, significant metal particle encapsulation occurs. A possible methodology for the determination of the true H/Pd ratios is also disclosed. 3.3. NO reduction by CO. The catalytic activity in the reduction of NO by CO was investigated in a flow reactor in the range of temperatures 433-573 K. All the samples show a very high activity at moderate temperatures (< 500 K). Indeed, gas hourly space velocities as high as 1 x 106 h -1 and reaction temperatures of 433 K had to be employed to measure the reaction rates in absence of diffusional limitations. The reaction rates were measured under isothermal conditions after aging of catalysts in the reaction conditions for at least 15 h. During this period slow and partial deactivation of the freshly reduced catalysts is observed to reach a steady activity. Afterwards, the catalysts were subjected to a thermal cycle similar to that depicted in Figure 4 for the 0.7 wt% Pd/Ceo.6Zro.402-A1203. A light-off type behavior is observed. However, during the run-up part of the cycle, a partial deactivation occurs as denoted by the peak of activity centered at about 500 K (compare Figure 5). This effect is reversible and, remarkably, immediately after the end of the thermal cycle, the catalyst slowly regains its initial high activity upon aging at 473 K. No such behavior is observed either during the run-down part of the cycle or using 0.7 wt% Pd/A1203 as catalyst. All these results clearly point out an "active" state of the catalyst at low temperatures which is reversibly deactivated above 500 K.

800

100

700

80

,,g v

v

60 E 600

ID

40

tO

-= O

o

"5 500

20

n,' 400 0

1000

2000

3000

Reaction Time (min) Figure 4. NO-CO reaction catalyzed by 0.7 wt% Pd/Ceo.6Zro.402-A1203, ( l ) NO and (e) CO conversions, (-) temperature. Figure 5 compares the N20 and N 2 formation over the Pd/Ceo.6Zro.402-A1203 and Pd/A1203 catalysts. The peak in the N20 formation at about 650 K which is present on both

565 the catalysts is associated with the shift of the selectivity from the N20 production to the N 2 formation commonly observed over supported platinum group metals. The peak at 500 K observed in the picture 1 of Figure 5 is attributed to the low temperature "active" state of the Pd/Ceo.6Zro.402-A1203 catalyst which is promoted by the presence of the mixed CeO2-ZrO 2 oxide. The reaction rates measured in steady conditions at 433 K are reported in Table 2, e.g. under conditions when the "active" state of the catalyst is present. The activity of the catalysts is compared after a reduction at 473 and 1000 K. As a general trend, the reaction rate increases with metal loading. This increase is more pronounced on the catalysts reduced at 473 K where a four-times increase of the Pd-loading, from 0.7 to 2.8 wt% induces an eighttimes increase of activity. Also the TOFs slightly increase as the Pd particle size increases, indicating a slight structure sensitivity of the reaction, even though, there are some scattered data after reduction at 1000 K. The overall picture clearly indicates a favorable effect of the Ceo.6Zro.402 on the rate of the NO-CO reaction which does not appear to be related to a stabilization of higher dispersions induced by the support. Consistently, even after the high temperature reduction, despite a considerable decrease of the number of active sites, either due to a sintering and/or encapsulation of the metal particles, high TOF are observed on the Ce0.6Zro.40 2 containing catalysts compared to Pd/A1203. All the evidence points out an active role of the Ceo.6Zro.402 support which improves the catalytic efficiency of the supported metal. Consistently, recently we found evidence for a low temperature CeO2-ZrO 2 promoted catalytic cycle which favors the reduction of NO on Ce 3+ sites. In agreement with an active role of the support in the catalytic cycle, a very slight structure sensitivity is observed. Consistently with our observations, Gorte et aL recently reported that an interaction of NO with a reduced Pd/CeO 2 catalyst leads to enhanced N20 and N 2 production compared to a Pd/A1203 (10).

4.0E-05

o

~o 6.0E-05 >= 4.0E-05

E

8 2.o5-o5

O o

..,.,

o

~ 2.0E-05

E

1.0E-074O0

500

600

700

Temperature (K)

800

n~ 1.0E-07 400

500

600

700

800

Temperature (K)

Figure 5. Reaction rates vs temperature for ( I ) N20 , (11) N 2 formation and (A) NO conversion measured in the run-up cycles over (1) 0.7 wt% Pd/Ceo.6Zr0.402-A1203 and (2) 0.7 wt% Pd/A1203.

566 Table 2 Steady state reaction rates measured at 433 K over Pd/A1203 and Pd/Ce0.6Zr0.402-A120 3. Pd loading (wt%)

Reduction Temperature 473 K 1000 K Reaction rate a TOF b Reaction rate a 0.7 c 0.7 2.8 0.3 0.7 d 2.8 7.2 0.5 1.4 d 5.3 9.2 3.4 2.1 d 19 25 5.0 2.8 d 22 22 4.2 a MolesN O converted gcatalyst-1 s- 1 . 106; b MolesN O converted m~ exposed-1 s-1 * 102; c Pd/A1203; d Pd/Ce0.6Zr0.402_A1203.

3.4.

TOF b 2.3 8.8 20 14 7.9

In situ i.r. spectra of 2.8 wt% Pd/Ce0.6Zr0.402-AI203 In order to obtain further indication on the role of the support in the catalytic reaction, we investigated the interaction of the 2.8 wt% Pd/Ce0.6Zro.402-A1203 with both CO and NO under reaction conditions using an i.r. flow cell. After a thermal pretreatment at 773 K in flow of He, the interaction of flowing NO/CO with the 2.8 wt% Pd/Ce0.6Zr0.402-A1203 generates the spectra reported in Figure 6. Two sets of prominent band are observed in the 2500-2300 and 2300-2200 cm -1 regions (Figure 6.1) which are associated respectively to CO 2 and N20 produced during the reaction, both in the gaseous phase and adsorbed on the support. It is worthy of note that formation of CO 2 starts immediately, at about 313 K, while formation of N20 is observed above 400 K. The strong band at 2168 cm -1 may be associated to CO chemisorbed on the Ce 4+ sites (25,26). Its intensity is highest at 313 K and it disappears at about 400 K. The weak broad band at about 2140 cm -1 is a low frequency end of the roto-vibrational band of the CO (the high frequency one is covered by the band at 2168 cm -1) due to a non perfect subtraction of the gaseous phase. Its intensity slightly decreases with temperature as the CO conversion increases. At about 450 K, a new band at 2121 cm -1 suddenly forms. This band could be attributed to the presence of Ce 3+ sites (25,26). Its intensity decreases as the temperature approaches 473 K. As far as the species adsorbed on the metal are concerned (2000-1700 cm1), the weak band at about 1950 cm -1 which is formed above 400 K, is associated with bridged CO species bonded to Pd. The two weak broad bands at about 1900 cm -1 are due to gaseous NO, while the strong band at 1757 cm -1 is associated with NO linearly bonded to Pd. The intensity of the latter peak is highest at about 400 K. The peak maximum gradually shift from 1757 to 1745 cm -1 at 473 K. Above 473 K (Figure 6.2), no evidence for CO adsorption on the support is found, while the Pd is essentially covered by adsorbed CO as denoted by the broad weak band at about 1900 cm -1. Note the higher intensity of the peaks in the 2300-2200 cm -1 region compared to those at 2500-2300 cm -1.

567

T /

~-

l'lA

473 -

I

t

II

I

i

.2A

I

313--

1 i

2500

i

2

2 00

24oo

i

Wavenumber (cm-1)

,,,

1700

....

2 00

,

2 00 "1 2 00

, 1 00 Wavenumber (cm-1)

1700

Figure 6. I.r. spectra of 2.8 wt% Pd/Ceo.6Zr0.402-A1203 in flow of NO/CO mixture (1% each in He) measured increasing linearly the temperature (1 K min -1) from (1) room temperature to 473 K and, after aging at 473 K for 10 h, (2) from 473 K to 523 K. For the description of the indicated bands, compare text. Summarizing, the experiments reported in Figure 6, show an important evolution of the interaction of the gaseous reactants with both the support and the supported Pd. The coincidence of the lighting temperature for the catalytic NO/CO reaction and the disappearance of the CO adsorbed on the Ce4+ with consequent observation of Ce 3+ sites suggests that all these phenomena are related to a catalytic cycle mediated by the Ceo.6Zro.40 2 mixed oxide. Above 473 K, limited evidence for interaction with the support is found. Accordingly, we associate the high temperature activity to a metal catalyzed cycle where the contribution of the support is minimal. A possible reaction scheme which account for the above finding is reported in the following scheme. According to this scheme, reduction of NO to give N20 easily occurs at the Ce 3+ by a redox process involving the Ce4+/Ce3+ redox couple. Formation of the Ce 3+ sites which is a crucial step of the reaction may occur either by a direct interaction of CO with the support or by a reverse spillover of the lattice oxygen to the metal surface as found by Cordatos and Gorte (10). Two reasons suggest that the latter hypothesis might be more reliable. The interaction of CO with the Ceo.6Zro.402 leads to formation of carbonates. Their desorption which leads to creation of the Ce 3+ sites, generally occurs at high temperatures. On the contrary', upon desorption of NO from the Pd surface, CO starts being adsorbed on the metal and at the same time the catalytic reaction starts as denoted by the formation of N20 (Figure 6.1).

568 CO

Pd

Pd --- CO

C02 ~

C2 +

2 NO

C 4 -----+ 0

N20

4. C O N C L U S I O N S The present work reports a promoting effect of the Ceo.6Zro.40 2 mixed oxide on the catalytic reduction of NO by CO at moderate temperatures. This effect is attributed to the Ce4+/Ce 3+ redox couple which efficiently reduces NO. The interaction of the Pd/Ceo.6Zro.402-A120 3 catalysts with H 2 reveals that considerable metal particle encapsulation occurs after a high temperature reduction. This suggests that the choice of an appropriate metal particle size may be an important factor to avoid the catalyst deactivation. Finally, a methodology for the determination of the H/Pd ratios is discussed. Acknowledgments. Magneti Marelli D.S.S., Ministero dell'Universit~t e della Ricerca Scientifica (MURST 40% and 60%, Roma), CNR (Roma) and Universitg di Trieste are acknowledged for financial support. 5. R E F E R E N C E S

1.

K.C. Taylor, "Catalysis-Science and Technology" (J.R. Anderson and M. Boudart, Eds.), Chap.2, Berlin, Springer-Verlag (1984).

2.

K.C. Taylor, Catal. Rev. -Sci. Eng. 35 (1993) 457.

3.

B. Harrison, A.F. Diwell, and C. HaileR, Plat. Met. Rev. 32 (1988) 73.

4.

J.G. Nunan, H.J. Robota, M.J. Cohn, and S.A. Bradley, J. Catal. 133 (1992) 309.

5.

G. Ranga Rao, J. Ka~par, R. Di Monte, S. Meriani, and M. Graziani, Catal. Lett. 24 (1994) 107.

6.

G. Ranga Rao, P. Fornasiero, R. Di Monte, J. Ka~par, G. Vlaic, G. Balducci, S. Meriani, G. Gubitosa, A. Cremona, and M. Graziani, J. Catal. 162 (1996) 1.

7.

P. Fornasiero, J. Ka~par, and M. Graziani, J. Catal. 167 (1997) 576.

8.

P. Fornasiero, R. Di Monte, G. Ranga Rao, J. Ka~par, S. Meriani, A. Trovarelli, and M. Graziani, J. Catal. 151 (1995) 168.

569 9.

P. Fomasiero, G. Balducci, J. Ka~par, S. Meriani, R. Di Monte, and M. Graziani, Catal. Today, 29 (1996) 47.

10. H. Cordatos and R.J. Gorte, J. Catal. 159 (1996) 112. 11. M. Yashima, N. Ishizawa, and M. Yoshimura, J. Amer. Ceram. Soc. 75 (1992) 1541. 12. J.A. Anderson, "Structure of Metallic Catalysts", London, Accademic Press, (1975). 13. S. Bernal, F.J. Botana, J.J. Calvino, M.A. Cauqui, G.A. Cifredo, A. Jobacho, J.M. Pintado, and J.M. Rodriguez-Izquierdo, J. Phys. Chem. 97 (1993) 4118. 14. S. Bernal, J.J. Calvino, G.A. Cifredo, A. Laachir, V. Perrichon, and J.M. Herrmann, Langmuir 10 (1994) 717. 15. S. Bemal, J.J. Calvino, G.A. Cifredo, J.M. Gatica, J.A.P. Omil, and J.M. Pintado, J. Chem. Soc. Faraday Trans. 89 (1993) 3499. 16. A. Badri, C. Binet, and J.C. Lavalley, J. Chem. Soc. Faraday Trans. 92 (1996) 4669. 17. V. Ragaini, R. Giannantonio, P. Magni, L. Lucarelli, and G. Leofanti, J. Catal. 146 (1994) 116. 18. A. Trovarelli, Catal. Rev. -Sci. Eng. 38 (1996) 439. 19. D.E. Resasco and G.L. Hailer, J. Catal. 82 (1983) 279. 20. S. Bemal, F.J. Botana, J.J. Calvino, G.A. Cifredo, and J.A. Perezomil, Catal. Today 23 (1995)219. 21. R.F. Hicks, Q.J. Yen, and A.T. Bell, J. Catal. 89 (1984) 498. 22. A. Badri, C. Binet, and J.C. Lavalley, J. Chem. Soc. Faraday Trans. 92 (1996) 1603. 23. P. Fornasiero, G. Balducci, R. Di Monte, J. Ka~par, V. Sergo, G. Gubitosa, A. Ferrero, and M. Graziani, J. Catal. 164 (1996) 173. 24. B.Y. Lee, Y. Inoue, and I. Yasumori, Bull. Chem. Soc. Jpn. 54 (1981) 3711. 25. C. Binet, A. Badri, and J.C. Lavalley, J. Phys. Chem. 98 (1994) 6392. 26. C. Morterra, V. Bolis, and G. Magnacca, J. Chem. Soc. Faraday Trans. 92 (1996) 1991.



571

Comparative sulfur storage on Pt catalysts : effect of the support (CeO2, ZrO2 and CeO2-ZrO2) P.Bazin a, O. Saur a, J.C. Lavalley* a, A.M. Le Govic b and G.Blanchard b a Laboratoire Catalyse et Spectrochimie, UMR 6506, ISMRA-Universit6, 6, Boulevard du Mar6chal Juin, 14050 CAEN-C6dex (France) b Rh6ne-Poulenc-Recherches, 52 rue de La Haie-Coq, 93308 Aubervilliers (France)

The effect of the support on the nature, amount and reducibility of the sulfate species formed from SO2 oxidation on Pt catalysts has been studied by IR spectroscopy and thermogravimetry. Only surface species are observed on pure zirconia while bulklike species are also formed on the CeO2-ZrO2 mixed oxide (Ce/Zr=64/36) as on ceria. It appears that the H 2 reducibility of the bulklike species does not depend on the support, CeO2 and CeO2-ZrO2. By contrast, on Pt catalysts, surface sulfate species which first appear from SO2 oxydation are more easily reduced on CeO2-ZrO2 than on CeO2. This shows the role of the surface sites for the formation and reduction of sulfate species.

1. I N T R O D U C T I O N Sulfur is present in all commercially available gasoline. It interacts with Three Way Catalysts (TWC), deactivates their reactivity and also leads to H2S emission during rich operations by reduction of sulfates stored on the catalysts during stoichiometric or lean conditions (1, 2). In previous works, we studied ceria sulfation and the effect of platinum on both formation, from SO2 and oxygen, and reduction by H2 of sulfates (3, 4). Ceria is a key component of TWC for the treatment of exhaust gas from automobiles due to its oxygen capacity storage (OCS) (5). High surface area ceria samples are essential to obtain a significant OCS since the redox processes essentially occur on the surface. Moreover, this high surface area has to be maintained at high temperature. Recently, it has been claimed that zircinia mixed with ceria stabilized the surface at high temperature (6,7) and, in the new generation of TWC, it is possible that ceria will be replaced by mixed oxides such as CeO2ZrO2. The purpose of this work is to compare the nature, amount and H2 reducibility of the sulfate species formed from SO2 oxidation on Pt catalysts supported on CeO2, ZrO2 and a CeOz-ZrO2 solid solution. Two techniques have essentially been used : thermogravimetry and IR spectroscopy, both in static conditions. * To whom correspondence should be addressed

572 Previous works showed the importance of the surface in the formation and reduction of sulfate species (3,4). So, in addition to high surface area samples, the study has been extended to highly calcined samples (900~ in order to study the ageing effect.

2. EXPERIMENTAL 2.1.Catalysts preparation The cerium-zirconium mixed oxide (Ce/Zr = 64/36, atomic ratio) as well as the pure ceria samples (both without and with Pt) were obtained by a Rh6ne-Poulenc proprietary process. The purity of the cerium used was higher than 99.5%. For the study of the interaction between Pt and cerium-zirconium mixed oxides, a hexachloroplatinic solution was used as a source of Pt. The OH groups on the surface of the mixed oxides were exchanged by PtC162- ions in water. The material obtained was dried in an oven at 120~ and calcined in a muffle furnace at 500~ For aged catalysts, the calcination was performed at 900~ for 6 hours. The Pt loading was 0.5% (weight %).

2.2. Catalysts characterization X-Ray diffraction was used to identify the major phases and to measure the lattice parameters of the cubic phases. A Siemens D500 diffractometer was used for all XRD measurements. The mean crystallite diameter was measured from XRD pattern using line broadening Warren et al's formula (8). The surface area of the oxides was measured by the BET method and are reported in table 1. Tablel Surface area of the samples Samples

Pt/ZrO2

CeO2

Pt/CeO2

Aged Pt/Ce02

CeO2 ZrO2

Pt/CeO2 -

Aged

ZrO2

Pt/CeO2ZrO2

Surf.Area (m 2 g,1)

38

2.3. Infrared study

170

147

22

137

107

32

For infrared studies, powders were pressed (1 ton.cm 2) and used as wafers of c.a.lO mg.cm ~ They were activated at 450~ under oxygen atmosphere for two hours and then evacuated at the same temperature. Sulfation was performed by adding 600 ~tmol.g-1 of SO2 and a large excess of 02 (Pe - 6.5 kPa) in the cell and then heating at increasing temperature. For the study of sulfate reduction, a large excess of H2 (Pe -~ 13 kPa) was introduced at room temperature (r.t.) in the cell and then the wafer was heated at increasing temperature until

573 disappearence of the sulfate absorption bands. All spectra were recorded at r.t. after evacuation of the sample at 400~ using a Nicolet Magna-500 spectrometer.

2.4. Thermogravimetric study For gravimetric measurements, the powders were pressed (c.a.400mg), activated and sulfated in the same conditions as for the IR study. A McBain thermobalance was used. The temperature was increased from r.t. to 450~ (0.5~ l ) and then kept at 450~ until a plateau was obtained for the weight. For the study of sulfate reduction, a large excess of H2 (Pe - 13 kPa) was introduced at r.t. in the thermobalance and then the sulfated sample was heated under H2 atmosphere at increasing temperature (0.5~

3. RESULTS

3.1. Formation and nature of sulfate species Previous IR studies of metal oxides sulfation have evidenced two types of sulfate species, surface species characterized by one or more bands in the 1410-1370 cm -1 frequency range, and bulklike species leading to wide bands in the 1200-1000 cm -1 range (3, 9-12). In a recent paper (4), it has been shown that Pt does not affect the sulfate formation nor the nature of adsorbed species. Spectra reported in Fig. 1 show that only surface sulfate species are present on Pt/ZrO2 (Fig.la) while both species are formed on Pt/CeO2 (Fig. 1b) and Pt/CeO2-ZrO2 (Fig. 1c).

,, iX:)

A b s d

t',q v-.

/

O r b a n ec ~ 4 0 5 ~ ~

I

,,

-c~ ~ ~ ~I

~176

a

....

1500

I

13'00

I

11'00

I

Wavenumber (cm -1)

"

960

t

I

Figure 1. IR spectra of sulfate species formed on various samples after evacuation at 400~ a) Pt/ZrO2; b) Pt/CeO2 ; c) Pt/CeO2-ZrO2; d) aged Pt/CeO2-ZrO2.

574 Moreover comparison of spectra l c and ld shows that the amount of bulklike species is higher while surface species are less numerous on the aged samples. As observed on ceria (3), the relative amount of the two types of sulfate species depends on the sample surface area.

65 60 q

-"

CeO 2 (a)

--~

Pt/(Ce/Zr)O 2 (e)

"

(Ce, Zr)O 2 (b)

~

Pt/CeO 2 - 900~

Pt/ceo~ (c)

-+-

Pt/(Ce, Zr)O~-900~ (g)

-o-

55 -

(f)

_

- - v - - p t / Z r O 2 (d)

_

50 _

_

45 _

_

~.40

:

35 N 3o ~25-

2o 10 5 00

100

200

temperature /

300

400

~

Figure 2. Mass gain of various samples (CeO2, CeO2-ZrO2, Pt/CeO2, Pt/ZrO2, Pt/CeO2-ZrO2, aged Pt/CeO2, aged Pt/CeO2-ZrO2) heated under SO2 + 02

In figure 2, we report the mass gain of different samples heated under SO2 (600 ~mol.g -1) + 02 (- 6.5 kPa) atmosphere versus temperature. The results on ZrO2 and Pt/ZrO2 show a small weight gain (- 13 mg. g-l) confirming that surface sulfate species are only formed. The curves are very similar for CeO2 and CeO2-ZrO2 and all the SO2 amount is oxidized at E] 200~ on the high surface area catalysts in the conditions used. No effect of platinum is clearly detected on the sulfate formation on Pt/CeO2. The difference between CeO2-ZrO2 and Pt/CeO2-ZrO2 shown in figure 2 may be due to a surface area effect (Table l).Indeed, in the case of the aged catalysts, the total SO2 oxidation rate is drastically reduced. This suggests that the oxidized species are formed on the surface and then migrate into the bulk.

575 In order to understand the sulfate formation mechanism, IR spectroscopy measurements have been performed.Fig. 3a shows the spectra of SO2 species formed from SO2 adsorption at r.t. on CeO2-ZrO2. Strong bands due to sulfite species are observed at 1005, 919, 830 cm-1; sharp bands at 1334 and 1144 cm -I are assigned to physisorbed SO2. After heating up to 150~ new bands appear, in particular at 1378 cm "1 (Fig. 3b), showing that surface sulfate species are formed. Concomitantly, the absorbance of the bands due to sulfite species decreases (negative absorbance near 800 cm -1, Fig.3e). Bands due to sulfate bulklike species (near 1170 cm -1) appear only at 250~ while the intensity of the band assigned to surface species (1389 cm -1) still increases (Fig.3c, 3f). Further heating at 350~ (Fig.3d, 3g) mainly provokes bulklike sulfates formation. This confirms that surface sulfate species first appear at the expense of sulfite adsorbed species. Diffusion of sulfate into the bulk then occurs and is favored by heating at higher temperature.

A

/

T,

\~ ~

0.5

5"-

A

'

sd

: Oct,

v 1388

I a

>(c-b) f

c b

-(b-a) ,

,

'14'00'

,

,

.

,

i

,

.

.

I

12'00 1000 800 Wavenumber (cm-')

,

,

I '

'

i

,

1400

,

,

!

,

'~

,

!

,

'

,

,

i

,

1200 1000 800 Wavenumber (cm")

Figure 3 - IR spectra of adsorbed species on CeO2-ZrO2 after SO2 + 02 addition in the cell at a) r.t. b) 150~ ; c)250~ ;.d)350~ ; e, f, g ) subtractions of spectra (a) from (b), (b) from (c) and (c) from (d), respectively.

3.2. R e d u c t i o n o f s u l f a t e s by H 2

Previous studies have been reported on adsorbed sulfate species on pure zirconia showing that they are thermally stable until 600~ (12). Thermal stability of adsorbed sulfate species on ceria or on Pt/CeO2 has also been previously studied (3,4) whereas studies are in progress on CeO2-ZrO2 mixed oxides. All these results show that sulfate groups do not decompose under vacuum before 400~ on zirconia or ceria based catalysts. Under H2 atmosphere, most sulfates adsorbed on oxides are less stable than under vacuum (3,9,11,12). In figure 4, we have reported the variation of the sulfated samples mass versus temperature during treatment under H2 atmosphere for CeO2 and CeOa-ZrO2 compounds with or without

576 Pt. Reduction begins at the same temperature for the CeO2 and CeO2-ZrO2 samples (- 380~ and the curves are very similar, the average temperature of reduction being about 430~ However, we can note that the total mass loss is larger for the CeO2-ZrO2 sample than for pure CeO2 although they initially contain the same amount of sulfates. Addition of platinum to ceria and ceria-zirconia samples decreases the sulfate reduction temperature of about 80~ It is worthwhile noticing that the mass of the sulfated Pt/CeO2ZrO2 begins to decrease between 150 and 350~ while it seems stable in the case of the Pt/CeO2 sample up to 200~ The mass loss is also more important for Pt/CeO2-ZrO2 than for Pt/CeO2 (Fig.4). As for Z r O 2 (results not shown), the sulfate reduction, in the conditions used, begins near 380~ and mainly occurs near 450~ Addition of Pt makes the reduction easier; in agreement with (13), it begins at 300~ and still occurs at 480~

5O 40 9

i~~.

~--~

30

' 2o 9

IIII

1/11

>

///I

7, lo

III

-10

Oxidation and evacuation at 450~ ,,,,,,,,,

0

i,,,,,,,,,

100

i,,,,,,,,,i,,,,,,,,,

i,,,,,,,,

200 300 400 temperature / ~

,1 . . . . . . . . . . . . . . . . . . . . . . .

500

Figure 4 - Variation of the weight of various samples (CeO2, Pt/CeO2, CeO2-ZrO2, Pt/CeO2ZrO2),heated under H2 atmosphere, at increasing temperature.

Experiments on ceria catalysts under H2 f l o w at 400~ have shown that sulfate reduction leads to H2S formation (3,4) .However, H2S titration showed that the amount evolved is far lower than expected. This can be explained by the remaining of sulfur species on the samples. To confirm such a result, the reduced samples have been treated by O2 at 450~ in the thermobalance. This treatment leads to a mass gain for all samples (Fig.4). This is due to the

577 reoxidation of sulfur species, which remain on the samples after H2 reduction, into sulfate as shown by the IR spectra reported in Fig.5 which evidence for sulfate formation.

tt~

0.4

~" co

o o

T-I t~

A b

r

co

t'~

v.-

v-I

S O

b a

n C e

'

'14'oo'

'

'12'oo'

'

'lo'oo'

Wavenumber (cml)

'

'8do

Figure 5 - IR spectra of species formed from the reoxidation at 450~ of the reduced sulfated samples'a) Pt/CeO2, b) Pt/CeO2-ZrO2). The IR study of sulfate reduction by H2 was also carried out on ceria and ceria-zirconia samples. It confirms the gravimetric results: -platinum favors the sulfate reduction, decreasing the reduction temperature for all samples, -addition of ZrO2 to CeO2 gives rise to a higher amount of sulfate species which are reduced between 250 and 350~ It has recently been published that ceria reduction leads to Ce 3+ ions giving rise to an IR band near 2120 cm -1 (14). This band appears during the H2 reduction of all sulfated samples (Fig.6) showing that ceria reduction concomitantly occurs with the sulfate reduction. It has been shown that the introduction of Z r O 2 i n t o C e O 2 strongly modifies its reduction behaviour in comparison with pure ceria (15). The gravimetric curves in figure 4 confirm that ZrO2 enhances the ceria reduction. Moreover the study of the intensity of the 2120 cm- 1 band (Fig. 6) also shows that the Pt/CeO2-ZrO2 samples are more easily reduced at a given temperature than Pt/CeO2 even in presence of adsorbed sulfate.

578

0.02

A b c S 0

n

e ' 2200' ' 2100' '

' ' 2200" 21'00'

' ' 2~00' 2;00' ' ' 2300' 21'00' Wavenumber (cm -l)

' '2200' 2~00'

'

Figure 6 - Variation of the intensity of the 2120 cm -1 band during heating, under H2 atmosphere, sulfated samples a) Pt/CeO2, b) Pt/CeO2-ZrO2, c) Aged Pt/CeO2-ZrO2

Study of the IR spectra of sulfate species during their reduction at increasing temperature in the case of high area samples(Fig.7) allows us to compare the reducibility of surface and bulklike species. It appears that, without platinum, heating under H2 atmosphere

A

0.5

S O

0.5

S

a

a

n

n

C

C

e

e II Ot'~ (D v--

~

15'00 afle~

'

13'00 ' 11'00 ' 960 Wavenumber (cm-l)

15'0o

'

13'00

'

0

11'oo

'

9oo

Wavenumber (cm-1)

Figure 7 - IR spectra of sulfate adsorbed on A) CeO2, B) CeO2-ZrO2 and heated under H2 atmosphere at" a) 250~ b) 350~ c) subtraction of spectra (a) from (b).

579 up to 350~ only some surface sulfate species are reduced on ceria (negative absorbance at 1404 and 1370 cm -1) (Fig. 7A). On CeO2-ZrO2, at this temperature, a higher amount of surface sulfate is decomposed and few bulklike species are also reduced (Fig.7B).

4. CONCLUSION Only surface sulfate groups are observed on zirconia by $02 oxidation, even in presence of platinum, while bulklike species are also formed on CeOz-ZrO2 mixed oxides, with or without platinum, as on pure ceria (3,4). Then the sulfate poisoning can be as important on Pt/CeO2 -ZrO2 catalysts as on Pt/CeO2. As previously reported for pure ceria (4), platinum favors the H2 reducibility of both sulfate species on all studied catalysts. In this work, we show that sulfate reduction occurs at a lower temperature on Pt/CeOz-ZrO2 than on Pt/CeO2. So, it appears that addition of zirconia, not only enhances ceria reduction (15) but also sulfate reduction. This result relative to the sulfate reducibility on Pt/CeO2-ZrO2 can be considered as a favorable factor for the COS recovery. We also show that the surface sulfate species first appear and are the first ones to be reduced by H2. Their formation but also their reduction involve the surface sites. Diffusion of the oxidized species into the bulk during their formation or towards the surface during their reduction then occurs. The textural properties of the ceria catalysts then play an important role and have to be stabilized.

REFERENCES

1. D.R. Monroe, M.H. Krueger, D.D. Beck, and M.J. D'Aniello, Stud. Surf. Sci. Catal. 71 (1991) 593. 2. A.F. Diwell, S.E. Golunski, J.R. Taylor and T.J. Truex, Stud. Surf. Sci. Catal., 71 (1991) 417. 3 . . M . Waqif, P. Bazin, O. Saur, J.C. Lavalley,G. Blanchard and O. Touret, Appl. Catal. B, 11 (1997) 193 4. P. Bazin, O. Saur, J.C. Lavalley, G. Blanchard, V. Visciglio and O. Touret, Appl. Catal. B,in press. 5. B. Harrison, A. Diwell and C. Hallett, Platinum Metals Rev., 32 (1988) 73. 6. M. Pijolat,M. Pfin, M. Soustelle,O. Totaet and P. Nortier, J. Chem. Sot., FaradayTrans., 91 (1995) 3941. 7. G. Sauvion, J. Caillod and C. Gourlaoen, Rh6ne Poulenc, Eur. Pat.,0207857, (1986); T. Ohata, K. Tsuchitani and S. Kitaguchi, Nippon Shokubai Kagaku, Jpn Pat.,8890311, (1988); N.E. Ashley and J.S.Rieck, Grace W R and Co-Conn, US Pat., 484727, (1991). 8. B.E. Warren and B.L. Averbach, J. Appl. Phys., 595 (1950) 21. 9. O. Saur, M. Bensitel, A.B. Mohammed Saad, J.C. Lavalley, C.P. Tripp and B.A. Morrow, J. Catal., 99 (1986) 104. 10. M. Bensitel, M. Waqif, O. Saur and J.C. Lavalley, J. Phys. Chem., 93 (1989) 6581. 11. M. Waqif, O. Saur, J.C. Lavalley, Y. Wang and B. Morrow, Appl. Catal., 71 (1991) 1373. 12. M. Bensitel, O. Saur, J.C. Lavalley and B. Morrow, Mat. Chem. Phys., 19 (1988) 147. 13. C. Morterra, G. Cerrato, S. Di Ciero, M. Signoretto, F. Pinna and G. Strukul, J. Catal., 165 (1997) 172. 14. C. Binet, A. Badri and J.C. Lavalley, J. Phys. Chem., 98 (1994) 6392. 15. P. Fornasiero, G. Balducci, J. Kaspar, S. Meriani, R. di Monte and M. Graziani, Catal. Today, 29 (1996) 47.



Oxygen storage capacity in Perovskite-related oxides stoichiometric oxygen in three-way catalysis

581

9The role of over-

N. Guilhaume and M. Primer Laboratoire d'Application de la Chimie/l l'Environnement Universit6 Claude Bernard Lyon I, UMR CNRS 5634, Bat. 303 43 Boulevard du 11 Novembre 1918 F-69622 Villeurbanne Cedex, France.

Two

perovskite-related catalysts doped with small amounts of noble metals, and La2Cuo.gPdo.~O4+8, present high activity for three-way catalysis reactions including the simultaneous removal of CO, NO and C3H6. These catalysts exhibit oxygen storage properties, evidenced in step-change experiments where CO was oxidised by the lattice oxygen of the solids, at low temperature (400~ These properties allow them to compensate for the fluctuation of the feedstream stoichiometry, when the activity is evaluated under cycling conditions. LaMn0.976Rh0.02403+8

1. INTRODUCTION Because the exhaust composition in automobile catalytic converters fluctuates between oxidising and reducing conditions, oxygen storage is a particularly important property for three-way catalysts, which allows the solid to compensate for these fluctuations. This role is ensured by ceria, an essential additive in automotive emissions control catalysts. Among the different roles played by ceria in enhancing the catalytic performances (oxygen storage, promotion of reactions involving water, stabilisation of noble metals dispersion and of 7-A1203 [ 1-5]), the most significant one appears to be its ability to store and release oxygen, based on the easy and reversible change in the oxidation state of cerium between Ce4+ and Cea+. In a previous work, we have shown that the introduction of small amounts of noble metals in the perovskite-related oxides LaMno.976RM.024Oa+8 [6] and La2Cuo.gPd0.104+8 [7] leads to solids with high activity for the simultaneous removal of CO, NO and C3H6. These catalysts showed no deactivation when tested under periodically rich/lean conditions with low frequencies, suggesting that the lattice oxygen of the catalysts can compensate for the fluctuations in the composition of the reactant mixture. In the present study, we examined the response of these catalysts to step-changes in the composition, and how it is related to oxygen storage properties in the conditions of three-way catalysis.

582 2. RESULTS AND DISCUSSION

2.1. Experimental The experimental details concerning the preparation and characterisations of the catalysts have been described previously [6-7]. The main characteristics of the solids are presented in table 1. They were prepared by calcination of polyacrylamide gels in which the metal salts were incorporated. This method allows to obtain the mixed-oxide phases at moderate temperature (700~ with rather high surface areas. Table 1 Characteristics of the two per ovskite-type catalysts. Catalyst . . . . Calcination X-ray diffraction pattern temperature LaMno.976Rho.o2403+~ 700~ LaMnO3.,5 La2Cuo.aPdo.lO4+8 700~ La2CuO4+~(T)

Amount of noble metal 1 wt.% 2.41 wt.%

Specific surface area (m2/g) 27 15.8

The catalytic testing equipment is described in ref. [7]. The simulated exhaust compositions chosen for light-off experiments, under stationary and cycling conditions, are reported in table 2. In cycling tests, we chose a very low frequency (0.1 Hz) compared with the real cycling frequency in an engine (around 1 Hz), because the design of our apparatus and the flow rate are such that at 1 Hz the two streams mix together and the catalyst is submitted to an average composition. When cycling at 0.1 Hz, the streams reaching the catalytic bed correspond to 8085% of the two individual net-oxidising or net-reducing compositions. Table 2 Simulated exhaust gas composition used in stationary and cycling Light-off tests (total flow rate 10 l.hq). Composition (ppm) a Stationary . Cycled c O2 5600 3377 7823 CO 6200 10781 1619 NO 1000 1000 1000 C3I-I6 667 667 667 Sb 1 0.462 2.184 a The effective compositions were adjusted with an accuracyof + 1% around these values. b Stoichiometric factor S= (2 [O21+ INO])/([CO] + 9 [c~l). ~The average composition is stoichiometric, and identical to that used under stationary conditions. The step-change experiments were performed under isothermal conditions (400~ with single components (CO as the reducer, 02 or NO as the oxidiser) diluted in nitrogen. This temperature was chosen since it corresponds to 100% conversion of NO, CO and C3I-I~ in light-off experiments for the two catalysts. Two series of tests were performed, the "oxidation step" corresponding to: (1) dwell under CO, (2) flushing by nitrogen, (3) dwell under Oz (or NO), (4) flushing by nitrogen and (5) dwell under CO. The reverse experiment

583

(O2/N2/CO/N2/O2) corresponds to the "reduction step". Before starting the experiment, the catalyst was stabilised under the stream corresponding to the first dwell for one hour at 400~ 2.2. D y n a m i c evaluation of the redox properties of the catalysts

In order to evaluate how the catalysts change under oscillations in the feedstream composition, we examined the response of the solids to step-changes in the composition of the gas phase. Single components, CO and 02 (or NO) were sent alternatively on the solids at 400~ separated by intermediate flushing by nitrogen to remove the gas phase and weakly adsorbed species. 2.2.1. LaMn0.976Rho.02403+~catalyst . O x i d a t i o n s t e p (Fig. 1A) During the first dwell under CO (following one hour stabilisation under CO), no oxidation into CO2 is observed since the catalyst has been fully reduced. After 15 min flushing with nitrogen, the N2/O2 transition is accompanied by a very small COz peak (40 ppm at the maximum) which probably corresponds to the oxidation of traces of carbon formed during the first dwell under CO. The catalyst is then oxidised in oxygen for 30 min, flushed with nitrogen and CO is introduced again. A large CO2 peak appears immediately, whose trace has not completely returned to the baseline after 30 min dwell under CO. 6000 E

N~ '. I

c(

4000

i

i

I

I I ,

CO

I

co2

e

|

E O

!

I

'

i

02

O

E 2000 O

0

15

30

45

60

75

90

105

Time (min) 6000

' N2J'!"

~" C:: .o

'

; ..,.,_.,_._...,-.1.

! i

ii..

CO,,

= 4000

CO

i

,.

"

ii

~2,

*~9

N2

,,

E 2000 0

O= I

'I

(O

"1

i

0

15

,i L _ 30

. I

45

,

::

v ~.

,,

60

,§ 75

(B) I

!

g0

105

Time (min)

Figure 1. Step-change activity of LaMno.976Rho.02403+8under successive streams of CO and O2 at 400~ (100 mg catalyst, total flow rate 10 1.h~). (A) oxidation step; (B) reduction step.

584 . R e d u c t i o , s t e p (Fig. 1B) The reverse experiment shows the same CO2 peak upon introduction of CO on the solid previously oxidised. These results show that : 9 The large CO2 peak corresponds to CO oxidation by the lattice oxygen of the catalyst, and not to the CO dismutation reaction (2 CO ~ CO2 + C) since there is nearly no CO2 formed upon introduction of oxygen on the solid reduced under CO. 9 This large CO2 peak is highly reproducible : in several cycles (up to 5 cycles were performed) of step-change experiments with changing the CO and 02 compositions and increasing dwells lengths, the same response of the catalyst is obtained. Integration of this peak gives an average amount of 55 + 6% lamoles CO2 for 100 mg catalyst, which corresponds to 0.133 mole CO2 per mole catalyst. This corresponds well to the amount of over-stoichiometric oxygen which can be accommodated in this Lal~Ino.976Rh0.02403.15 perovskite (0.15 mole [O] per mole catalyst). The CO2 evolved represents the oxygen storage capacity of this catalyst 9 The same experiments were performed with NO as the oxidiser instead of oxygen. The same CO2 amount is evolved upon introduction of CO, while a small amount of N20 is formed when NO is introduced on the reduced catalyst, showing that the lattice oxygen can be replenished by NO as easily as by 02. 2.2.2. La2Cu0.9Pd0.104§ catalyst Similar tests were performed on the La2Cuo.9Pd0.104+~ catalyst (Fig. 2). The same profiles are obtained as in the case of the LaMn0.976Rh0.02403+8catalyst, although the small CO2 peak due to the oxidation of carbonaceous deposits is somewhat bigger than previously (440 ppm CO2 at the maximum). The main difference with the previous catalyst is that the amount of CO/evolved on introduction of CO on the oxidised solid is much larger (about 400 lamoles), and corresponds in this case to 0.75 to 0.8 mole CO2 per mole catalyst. According to several authors, the La2CuO4 structure can also accommodate overstoichiometric oxygen [8] up to La2CuO4.13[9]. However the CO~ peak formed on the oxidised solid cannot be due only to the removal of the over-stoichiometric oxygen, since it corresponds to a nearly total reduction of Cu2§ and Pd2§ into the metals. This indicates that this catalyst undergoes much deeper reduction that the lanthanum manganite, probably because the copper can be reduced into the metal in the present conditions, while it is unlikely that the surface manganese ions should be reduced more than in a Mn2+ state during CO oxidation.

2.3. Characterization of the reduced and oxidised states of the catalysts The X-ray diffraction patterns of the fresh catalysts correspond to those of LaMnO3.15 in the case of LaMno.976Rh0.02403§ and to a tetragonal form of La2CuO4 in the case of La2Cu0.gPd0.~O4+~. After successive step-change experiments under CO/O2, the solids were stabilised under CO or 02 at 400~ cooled at room temperature under the same atmosphere and flushed with nitrogen before recording the X-ray diffraction patterns. The phases identified by XRD are presented in table 3. In both cases the diffraction lines are broad and noisy. The structure of the Rh-doped lanthanum manganite is modified when stabilised in the reduced state: the XRD patterns corresponds now to the stoichiometric form LaMnO3.00 (orthorhombic, ICDD n~ 35-1353), which is slightlydifferent from that of (hexagonal, rhombohedrally

585 distorted, ICDD n~ 32-0484). This corresponds very well to the amount of CO2 evolved in the step-change experiments (0.133 mole per mole catalyst), and shows that the structure can easily switch between LaMnO3.00 and LaMnO3.~s depending on the reducing or oxidising stoichiometry of the feedstreams. Several authors have studied the structure of the LaMnO3+a perovskite according to its oxygen content, which may simply depend on the calcination temperature (the solid tends to loose oxygen on increasing the calcination temperature, starting from rhombohedral-hexagonal LaM/103.09 at 800~ to reach the orthorhombic form LaM/IO2.99 at 1300~ [10]). The oxygen non-stoichiometry in LaMnO3+a has been shown to influence the activity of the catalysts in various reactions like NO reduction by CO and 1-12 [11] or N20 decomposition [ 12].

12000

"~o

, CO i

8000

I' '

I

ii

;

;:

~4000

8

I

CO

1

CO2 ~j~

;

;

!, I

i

;

I x'k:

0

...... N2

I

;1 0

, ,

'

!'

-9

'

m

N2 ,

: ....

i5

30

45

,: ~

:"

:

60

75

(A) _

|

!

!

90

105

T i m e (min)

12000

I

'

I CO2

E

c}..

v

8000

,_5.c~

N2

I

N2

m

|

I

m|

i

I:

i ~

i

I

,,

|

, m

o

E

0

o2!

o

I I'

4000

|

o (0

i i

m

s

, 0

.

,

|

15

,L

I: !" ,!

,

-I ....

30

45

60

02

I 75

i

!

|

90

105

T i m e (min)

Figure 2. Step-change activity of La2Cu0.gPdo.lO4+a under successive streams of CO and 02 at 400~ (200 mg catalyst, total flow rate 10 l.hl). (A) oxidation step; (B) reduction step.

586 Table 3 Phases identified in the X-ray diffraction patterns of the catalysts in the fresh state and after step-change experiments, when stabilised in the oxidised or reduced state at 400~ Sample

Fresh catalyst

LaMno.976Rho.02403+~i

LaMnO3.15

La2Cuo.9Pdo.iO4+~

La2CuO4 (T)

Catalyst after step-change tests Reduced state Oxidised state LaMnO3.00 LaMnO3.15 La2CO5 Cu (very weak) Cu20 (traces)

La2CO5 CuO (very weak)

The XRD patterns of the La2Cu0.9Pd0.104§ catalyst after stabilisation under oxidising or reducing feedstreams explain the results of the step-change experiments: under our conditions (CO, 400~ the catalyst is not stable, leading to the formation of lanthanum oxycarbonate and reduced copper (and probably palladium). It must be reminded that the La2CuO4 structure, often called 'perovskite-type', consists in (LaCuO3) perovskite blocks separated by (LaO) § layers. It is also well established that lanthanum oxide carbonates rapidly even at room temperature, leading to the formation of LazCOs. This carbonate is quite stable (up to 1000~ under CO/). It is not decomposed during the oxidising dwell under oxygen at 400~ while the copper (and probably also the palladium, although not depicted by XRD) is fully oxidised into CuO. The same phenomenon has been observed with LazCuO4+8by other authors, in the case of the simultaneous NOx reduction and soot oxidation reactions [13]. We suppose that this collapse of La2CuO4+6 under CO is connected with the presence of (LaO)+ sheets which have a strong affinity for CO2, and destabilise the structure: CO is oxidised on surface Cu 2+ and Pd 2§ ions, leading to reduced copper and palladium while the CO2 formed is incorporated into the La203 lattice until saturation and total collapse of the mixed oxide structure. This does not occur in the pure perovskite La(MnRh)O3+8, where the lanthanum ions are surrounded by manganese only, and manganese ions cannot be reduced more than in a Mn 2+ state under these conditions. We had observed previously [7] by XPS and 1R study of CO adsorption that the La2Cu0.gPdo.lO4+~catalyst was deeply modified atter light-off tests in the presence of CO, NO and C3I-I6, even though the reactant gas mix is stoichiometric: it is transformed into highly dispersed copper and palladium species in various oxidation states (Cu 2+, Cur, Pd 2+ and Pd~ in a lanthanum oxycarbonate matrix. This instability is shown here to be connected with the reaction of the solid with CO. It is interesting to note that, since the lanthanum oxycarbonate is not decomposed under oxygen at 400~ and cannot thus be taken into account for the formation of carbon dioxide, this means that the amount of CO2 evolved in the step-change experiments (about 0.75 to 0.8 mole per mole catalyst) correspond to the nearly total and reversible reduction or oxidation of the two transition metals (Cu and Pd), at a mild temperature.

2.4. Evaluation of the catalytic activity under cycling conditions The ability of the catalysts to supply oxygen for the oxidation of CO and HCs when the exhaust is fuel-rich, and to remove excess oxygen when it is lean is evidenced in the light-off tests under cycling conditions. We chose a low cycling frequency (0.1 Hz) between two

587 strongly oxidising (S=2.184) and reducing (S=0.462) compositions, which minimises the mixing of the two feedstreams before they reach the catalytic bed: we checked that under these conditions the catalyst is in contact with 80-85% of the initial compositions. The light-off activities for CO, NO and C3H6 conversions are shown in Fig. 3. For comparison, the results obtained with a classical Pt-Rh/CeO2-AI203 (1.13 wt.% Pt, 0.19 wt.% Rh, 19.3 wt.% Ce) are included. 100

100

-

=> 50

,

,

.._

_

g

~

"~

~

5o

tO

8 o

o 150

250

350

450

Temperature (*C) 100

.

o 9 50.} d

.

.

.

.

350

450

.

;'//

o 250

250

Temperature (*C)

,, 150

150

350

Temperature (*C)

450

Figure 3. Light-off activity for CO, NO and C3H6 conversion in cycling conditions over LaMno.976R~.02403+8 (m, full line), La2Cu0.9Pd0.104+8 (B, dashed line) and Pt-Rh/CeO2-AI203 (C, thick line) catalysts. (Conditions : v=0.1 Hz, total flow rate 170 ml.mn~, 100 mg catalyst A, 200 mg catalysts B and C, G.H.S.V.: 13000hl).

Clearly, the Mn and Cu-based catalysts are very active for CO and HC oxidation reactions, a behaviour which is well known for these two metals, their activity being higher than that of the reference Pt-Rh catalyst. The assistance of small amounts of noble metals (1 wt.% Rh or 2.4 wt.% Pd respectively) improves the activity in NO reduction, which is total around 400~ Monceaux et al. [ 14] also observed a pronounced promoting effect of Pt and / or Rh dopes on the three-way catalytic activity of La0.sSr0.2MnO3+8 perovskites. Rh was shown to be particularly important to obtain a nearly total conversion of NO at 500~ with a high space velocity (100 000 h1) under stationary conditions. The interesting point is the comparison of the light-off activity under stationary or oscillating compositions (table 4). It is clearly seen that, despite the low cycling frequency chosen (the rich and lean feedstreams are switched for each other every 5 seconds), the catalysts show no loss of activity when compared with the light-off tests under stationary composition.

588 Table 4 Comparison of the light-off activity under stationary and cycling conditions, expressed as the temperatures (~ for 10, 50 and 80% conversion of CO, NO and C3H6. LaMno.976Rho.02403+~i

LaECUo.9Pdo.lO4+~

Pt-Rh/CeO2-Al203

Stationary CO

Tlo Tso Tso

190 230 245

673 K, so that the conversion observed for CuA below that temperature must be attributed to copper. The CO conversion curve corresponding to this sample shows that the reaction starts at 323 K and increases slowly up to ca. 573 K; above this temperature a much faster increase is observed. This variation in the conversion rate may be associated to changes in the nature of the copper entities during the reaction. The curve corresponding to Cul0CA shows a marked increase in conversion, showing that addition of 10% CeO2 to the catalyst decreases the isoconversion temperature at 50% conversion ('I"50)from 643 K for CuA to 443 K for Cul0CA (ATs0=-200 K). A higher increase in the Ce content up to 39 % further decreases Tso (but the additional change is comparatively lower: 100 T5o=383 K) and the conversion curve slope increases for any .-. 80 conversion level (below 80 %). The CuC sample is the most .~ 60 active one, presenting ca. 20% conversion already at RT. --- C u A o 40 Logarithmic plots (not 9 ~ CA shown) of the conversion data r~ 20 ~ Cu39CA vs. T -1 in the conversion + CuC interval 15-35% yield nearly 0 straight-line sections the slopes 273 373 473 573 673 773 of which correspond to apparent T (K) activation energies of 8.5, 10.0 and 12.7 Kcal/mol for samples Cul0CA, Cu39CA and CuC respectively. Tests on the xCA Figure 1. CO conversion profiles of the Cu-containing catalysts. r~

>

.

.

.

.

i

,

,

,

f'

,'

,

,

!

,

,

,

i

,

,

,

!

- -

594 or CeO2 supports (not shown) yield Ts0 values ca. 250 K higher than the corresponding Cucontaining catalysts. 3.3. E P R results

EPR spectra of the starting calcined samples show significant differences between the Cu-containing samples, Figure 2. Thus, CuC shows mainly an axial signal, showing four-line hyperfine splittings in each of its features, with parameters g ll =2.265, g . = 2 . 0 4 0 , All = 16 x 10-3 cm -1 and A ~ = l . 3 x 10-3 cm -1, signal A, due to isolated Cu 2§ ions [9] in a ceria environment; other smaller peaks at the magnetic field zone typical of the g II component of these signals reveal the presence of smaller amounts of similar isolated Cu2§ ions, the differences between them being most likely due to small variations in the coordination environment of the corresponding ions. On the other hand, for CuxCA or CuA, at least two different signals can be discerned: one presents g ll =2.321 and g . = 2 . 0 5 7 and four-line hyperfine splittings with All---17.1 x 10-3 cm -~ and A~= 1.9 x 10 -3 cm ~, signal B; this signal is similar to that found earlier in other Cu/A1203 samples and attributed to isolated Cu 2§ ions in a square pyramidal environment [10]. The other signal is significantly broader and presents extremes at g=2.24 and g=2.05, signal C; it must be ascribed to Cu 2§ ions into an oxidized copper-containing dispersed phase (like copper oxide or aluminate), the higher linewidth being due to dipolar interactions between Cu 2+ ions. The fact that these species are detected in the ceria-free sample suggests they are due to A1203-based Cu 2§ species. Small changes are however observed between the spectra of CuxCA samples and CuA, indicating the presence of other smaller Cu 2§ signals in CuxCA samples, whose parameters cannot be measured with certainty, and which might be related to Cu 2§ species interacting with the ceria component of these samples. It is noteworthy that double integration of the spectra shows that the fraction of total copper detected as Cu2§ increases as CuC (25 %) < Cul0CA (53 %) - Cu39CA < CuA (63 %), suggesting that copper-ceria interactions favour formation of diamagnetic species (reduced copper or antiferromagnetically coupled Cu 2+ ions). Introduction of a small amount of 02 at 77 K on reduced ceria-containing samples 2.265 2 040 leads to the formation of superoxide species I I II I ~ (formally Of-Ce 4+) [5, 11], as shown in the spectra of Figure 3; these were obtained by subtraction of spectra after and before adsorption, in order to cancel contributions of Cu 2§ signals. The characteristics of these species change as a function of the amount of ceria present in the catalyst, which I reveals changes in the environments of the corresponding superoxide species [5]; Table 1 summarizes the characteristics of these signals. Thus, CuC shows signal O1, 2oo G ~1~2.057 presenting a g tensor close to that shown by some of the signals observed in outgassed pure ceria [11], except for a slightly Figure 2. EPR spectra of the starting calcined lower gz value, which might be due to the samples.

595 Table 1. Summary of O2-Ce4§ signals detected by EPR in this study. More details on axis assignment or signal attributions can be found elsewhere [5, 11]. signal O1

EPR parameters

proposed assignment

g,=2.031-2.029, gx=2.014-2.013, gy-- 2.011

O:-Ce 4+ located on relatively large tridimensional ceria particles (3D-Ce) .

,

02

g,=2.028-2.027, gx=2.017, gy=2.011

.

.

gj_=2.027-2.026, gll =2.012

.

Of-Ce4+ at the surface of bidimensional ceria patches (2D-Ce) dispersed on A1203

,,

03

.

.

.

.

.

.

.

.

.

.

.

O,-Ce 4+ located at ceria-alumina borderline locations (like 2D-Ce edge sites) .

.

.

.

.

influence of carbonate species in the environment of the adsorption center. For the CuxCA samples, the maximum of the first derivative spectra is shifted to lower g values which is due to the overlapping of increasing contributions of signal 02 and a smaller one of 03, as the ceria loading decreases. Thus, for reduced Cul0CA, most of the O~-Ce 4+ species are formed at 2D-Ce patches where the Ce cations have aluminium cations as close neighbours, indicating that the amount of 3D-Ce particles is small. Comparison of the contributions, evaluated by computer simulations, of O2-derived signals (obtained after 02 adsorption at 77 K on the samples reduced in CO) as a function of T~, between Cul0CA and 10CA is presented in Figure 4. It shows that the presence of copper favours the formation of O2 species at the surface of the 2D-Ce patches but hinders the formation of these species at the edges of these bidimensional patches and at the 3D-Ce particles; some effect of copper in promoting ceria dispersion (but blocking edge sites at the 2D-Ce patches) should not be 1.0

--

I'

CulOCA:

CulOCA_J. \ 2 or2 .... 2-o2

I j

I OCA:

xl

c u c .........--J 3

'1

Y,

I

open symbols/;21 full symb

,.,.,o 0.5

2.029 5O G 0.0

1

373 r Figure 3. EPR spectra after oxygen adsorption at 77 K on the samples reduced at Tr =573 K.

473 (K)

573

Figure 4. Comparison of the contributions of superoxide signals vs. temperature of reduction in CO for Cul0CA and 10CA.

596 discarded to explain this latter point. 3.4. FTIR results Figure 5 shows IR spectra in the CO stretching region after CO adsorption at RT on the initial calcined samples. Sample CuA displays two very weak bands at 2113 and 2097 cm -~. As the Ce amount is increased an important increase of bands at 2106-2098 cm ~ is produced. Besides, for CuC, other small bands or shoulders are detected at 2115 and 2051 cm ~. A recent literature report has shown that changes in the stability of the different stable oxidation states of copper can be produced upon interaction with CO-O2 mixtures [12]. Thus, in order to examine the changes produced in the copper species characteristics upon interaction of the calcined samples with a CO-O2 mixture, the evolution of carbonyl bands has been monitored for CuA and Cul0CA, as shown in Figure 6. Two different temperatures were selected for these experiments at 373 and 573 K. The samples were heated using a 10 K rain -~ ramp until the corresponding temperature is reached, and subsequently cooled to RT, in a stoichiometric flow similar to that used for the catalytic reactivity tests shown above. Then, after prolonged outgassing at RT, a known CO pressure is admitted in the cell. Two bands are mainly formed on CuA at 2114 and 2097 cm -~. Upon increasing reaction temperature, an increase of the intensity of these bands is produced with hint of a shoulder

~ ,061 i/

98

/

CuC

Cu39CA

~/

/~

/

,,-\

/i/

\

\.

T

.1. a.u.

2051 xl

2 x3

2113 2097 CuA

~

x8 1

2150

2100

2050

Wavenumbers (cm- 1) Figure 5. FTIR spectra following adsorption of 10 Torr of CO at RT on the calcined samples. (Ordinate axis in absorbance units).

597 appearing at 2125 cm -~, Figures 6a-b. Certain differences, in respect to the CuA sample, are observed for the Cul0CA sample contacted with the mixture at 373 K, since maximum absorption is detected in this case at 2100 cm -t, with a shoulder at higher wavenumber, Figure 6c. In order to check whether new bands are formed in comparison to the CuA sample treated in the same conditions, the spectrum of Figure 6a has been partially subtracted from that of Figure 6c. The criterion used to perform this subtraction operation is to assume that the bands observed for CuA in Figure 6a are present as overlapping components in the spectrum of Figure 6c; thus the spectrum 6a subtracted from spectrum 6c is gradually increased until the point where some amount of negative band begins to be detected, the final result being shown in Figure 6e. It shows that two new bands at 2104 and 2089 cm1 can be present for the Cul0CA sample treated at 373 K. For the Cul0CA sample treated at 573 K, Figure 6d, the main bands are detected at 2097 and 2114 cmt, the spectrum being quite similar to that observed for CuA treated in the same conditions, except for a somewhat higher absorption at about 2100 cm -t. It is generally acknowledged that carbonyl bands at wavenumbers lower than about 2110 cm ~ are due to carbonyl species adsorbed on metallic copper particles [13], the variations in their wavenumbers being related to changes in the nature of the exposed metallic copper faces [13]. On this basis, bands appearing at 2097 cm -1 in the CuA or CuxCA samples can

2104- .. /-'-'-._~'

e)

x2

2114 2097

I

d) Cul0CA

0.03 a.u.

c) CulOCA

b) CuA a) CuA

~ '

2150

.

~ +

i

2100

+

2050

Wavenumbers (cm- 1) Figure 6. FTIR spectra after adsorption of 10 Torr of CO at RT on the samples treated in CO+O2 at 373 K (a,c) and 573 K (b,d). (e) Partial subtraction (c)-0.6(a). (Ordinate axis in absorbance units).

598 be ascribed to carbonyl species adsorbed on metallic copper, while the low wavenumber of the smaller bands at 2089 and 2051 cm-~ shown by ceria-containing samples can be attributed to similar species, in which the copper particles are affected by interactions with the basic ceria support, as proposed for a Cu/MgO catalyst [14]. Some doubts can exist in respect to the nature of the adsorption center for the bands observed in ceria-containing samples at 2106-2098 cm 1. Although they are in the wavenumber range of Cu~ carbonyls, an alternative assignment might be made on the basis of the existence of interactions with the support [14] and the possibility that they are due to Cu § carbonyls cannot be fully discarded. On the other hand, bands appearing at 2113-2115 cm -~ can be attributed to Cu § carbonyls, on the basis of previous reports [13-15] or to metallic copper carbonyls adsorbed on open sites of the particles [13].

4. DISCUSSION As pointed out in the introduction, it is recognized that ceria can adopt at least two different configurations when dispersed on alumina [4, 5]; well-dispersed 2D-Ce patches, showing important interactions with the alumina support and relatively large 3D-Ce particles. Characterization of the former is difficult; it has been claimed that some small peaks in Raman spectra might be due to these species, but there are doubts in respect to their possible attribution to other species like transitional alumina phases [4]. In our Raman spectra it was not possible to discern unambiguously other peaks in addition to that of large ceria particles located at ca. 460 cm -~. This can be due either to the non-existence of well-defined Raman peaks for these highly dispersed ceria entities or to the fact that in our case relatively mild thermal pretreatments have been performed, leading to a poor crystallization of this superficial phase (so called CeA103 precursor [4]). For similarly prepared CeO2/A1203 samples, more reliable XPS results [4] point towards a predominance of 2D-Ce entities for ceria loadings lower than ca. 3/~mol/m2 CeO2/A1203 while larger 3D-Ce particles would predominate for higher loading; it is worth recalling that even for the lower ceria loading used in that work (0.58 ~mol/m2 CeOE/A1203) the presence of 3D-Ce entities was noticed [4]. In our case, 10CA is close to the breakpoint (3.2/~mol/m2 CeO2/A1203) while 39CA is well above this point and thus one would expect the predominance of large ceria particles for this latter. In this sense, both Raman and XRD evidence the presence of 3D-Ce particles which seem to increase both in amount and in size as ceria loading increases from 10% to 39%. On the other hand, EPR spectra after oxygen adsorption following a method developed in [5] show the formation of characteristic superoxide signals due to the presence of 2D-Ce entities in the catalysts based on aluminasupported ceria, which give a higher contribution for the lower ceria loading. In such situation, when copper is deposited on these supports one may expect the presence of at least three kinds of copper entities differing in their interactions with the underlying support and which may be designated as copper-alumina, copper-2D-Ce and copper-3D-Ce. The chemical behaviour of the first and the last of these entities can be assumed in principle to be close to that of copper supported on the pure supports. For the calcined samples, in the case of the CuA catalyst two different Cu E* entities have been identified on the EPR spectra differing in their dispersion degree, with a fraction of Cu remaining undetected. On CuC, the observable species is clearly different from those

599 detected on CuA; besides, an important part of the copper is not detected by EPR. For CuxCA samples, the spectra show important similarities with that of CuA, thus suggesting that a substantial part of the copper in these samples interacts with alumina; however, a lower amount of Cu 2+ is detected for these samples, approaching in this sense the behaviour of CuC; this points to an interaction of part of the copper in these samples with the ceria component. The non-detection by EPR of part of the copper present in the calcined samples may correspond to the formation of EPR-non-observable Cu 2+ species (due e.g. to diamagnetic coupling) or to the stabilization of reduced states of copper. Previous reports propose, on the basis of XPS results, that ceria can stabilize copper, already for preoxidized samples, as Cu + species [3]. The important increase in the amount of reduced copper carbonyls (giving a band at 2106-2098 cm-~) when ceria loading is increased, as shown in Fig. 5, suggests that this would be the case as well in our samples, even in the calcined state (this being then the reason for the lower EPR intensity); thus, formation of these carbonyl bands must be taken as indicative of the existence of direct copper-ceria contacts in these samples. Further evidences on the existence of these interactions are obtained by comparative analysis of the O2-Ce4+ species obtained for Cul0CA and the corresponding copper-free 10CA support, Figure 4. Thus, the relatively lower intensity of superoxide species at 2D-Ce entities for Cul0CA treated at T~

2 K2Cu2(MoO4)3+ CuMoO4 -FCuCl2

(4)

2 KCI + K2M03Olo + 2 Cu3Mo209

648 The activity of K2CuCI,2H20 was determined to be similar to the activity of the KCI/CuMoO4 mixture, as well as to" the activity of the sublimed material out of a Cu/K/Mo/CI-ZrO2 catalyst. Oxidation temperatures of catalytic mixtures of CuMoO4 (or CuWO4 or Cu2V2OT)with KCI, CsCI, and LiC1 (without pre-calcination), were also detemained to be in the range of 655-665 K, whereas addition of K2CO3 did not significantly enhance the activity of CuMoO4 [13]. Apparently a chloride ion is essential for the high activity, and the presence and formation of copper chlorides results in the high soot oxidation activity of Cu/K/Mo/CI- and related catalysts. The only function of the Cuanion (i.e. molybdate, tungstate or vanadate) is to react with potassium and to s t a b l e the system. 1000

Oxidation Temperature (K)

900 ............................................................................................................................................ BaCI~........ 9

9

HgCI2 8oo

~eCaCI2 ................................................... c i S c i ~ . ~

...............................

700

Cu(ll)Ci2

600

'

500 500

/

~

FeOCI

~

9

~

o-

/

..............~

.... MoO3

.

.

.

.

.

.

.

.

.

.

.

.

.

NiCI2

e//~

Cu(I)Cl

I

600

i

700

I

800

,

I 900

t

1000

I

1100

~

1200

9 1300

Melting or Decomposition Point (K)

Figure 2. Correlation between the melting point of metal chlorides and the corresponding soot oxidation temperature. Solid dot: well defined melting point. Open dot: decomposition or oxidation takes place before melting. The horizontal solid line indicates the non-catalytic soot oxidation temperature. The activity of various metal chlorides, expressed by the soot oxidation temperature, is shown in figure 2. The non-catalytic soot oxidation temperature is indicated by the horizontal solid line. Several metal chlorides have hardly any effect on the soot oxidation temperature. HgCl2, BaCI2 and CaCI2 only show a small catalytic effect. Upon heating, COC12.6H20 and NiCI2.6H20 lose crystal water; in air they are (partially) converted into the corresponding oxides. MoC15 is completely oxidized in air at relatively low temperatures. The activity of oxidized MoCls is equal to MoO3 in 'tight contact'. FeCIs.6H20 and hydrated BiCI3 are fist converted into FeOCl and BiOCl, respectively [16,17]. The catalytic activity of these oxychlorides are given in figure 2. C u C I 2 and

649 PbCb are quite active, and CuCI even has a higher activity: the soot oxidation temperature is lowered by 285 K. Figure 2 also shows a correlation between the 'loose contact' activity and the melting point of a transition metal chloride. Unfortunately, several metal (oxy) chlorides do not have a well defined melting point in air. As previously discussed, they are (partially) transformed before they melt into the corresponding oxide and 02 or decompose otherwise. Therefore, the decomposition temperatures given by Knacke [18] are plotted in the figure, except for Mo, whose melting point of the oxide (MOO3[18]) was used. These metal chlorides are indicated with an open circle in figure 2, while a solid circle indicates that the metal chlorides (and MOO3) have a well defined melting point. Metal (oxy)chlorides with high melting points (CaCb, BaCI2, NiCI2 and COC12)are less active in 'loose contact' than metal chlorides with relatively low melting points and high vapor pressures. Although HgCb does have a well defined melting temperature, it is not on the curve, obviously because HgCI2 has evaporated before it can exert its catalytic influence. Milling a metal chloride and soot hardly effects the catalytic soot oxidation temperature. The observations suggest that the high catalytic activity of several metal chlorides in 'loose contact', is due to an in-situ distribution of the chlorides over the soot surface (resulting in 'tight contact'). Whether the in-situ 'tight contact' formation occurs by 'wetting' or gas phase transport has yet to be established. Xie et al. [19] have investigated the spreading (or 'wetting') behavior of many inorganic salts on several carrier materials (like Al203, TiO2 and activated carbon), and found that CuCb was able to wet the surface of alumina [19]. Previously, it has been shown that gas phase transport of copper chlorides also occurs [ 11]. Weight (mg)

60-1

|

Heat Flow (mJ/sec)

160 -120

"-

..~, i " \

T5 -80

| 58-1

I,

/

..... ---

-'--f--'f--'~-'V. .....

I /

\

~.. :,'k.

-40 "'"--. A

-0 --40

9od8O

5 Temperature (K)

Figure 3. TG/DSC analysis of the Cu2OCh catalyzed soot oxidation (curves A), the CuCI catalyzed soot oxidation (curves C) and the oxidation of CuCI without soot (curves B). TG: dashed lines, DSC: solid lines.

650 It should be mentioned, however, that the activity of the (oxy)chlorides of Cu, Pb, Fe and Bi is even higher than that of their corresponding oxides in 'tight contact' [11]. This might be the result of an even better contact obtained by 'wetting' or condensation than obtained after ball-milling, but a chlorine ion might also effect the activation of oxygen and the redox properties of transition metal oxides. Therefore, a thorough TG/DSC and DRIFT analysis of CuCl and FeOCI was carried out, in order to reveal the mechanism by which transition metal (oxy)chlorides are active in soot oxidation. The TG and DSC profiles of soot oxidation catalyzed by Cu~OC12 (A) and CuCI (C), and the oxidation of CuC1 in air (4 mg, without soot, but diluted with SiC (1:15)) ~), are shown in figure 4. The CuCI and Cu~OCI2 catalyzed soot oxidation temperatures are located at approximately 595 K (T1) and 625 K (T5), respectively. After 100 % soot conversion, an increase in weight can be observed in the range of 660 K - 690 K, accompanied by two heat effects (T2 and T3): one of which is positive, due to oxidation, and one a superimposed negative heat effect, due to melting. Similar heat effects are observed for pure CuCI. The interpretation is the (re)oxidation of CuCI into Cu2OC12, which is thermally stable in air up to 740 K (T4). At this temperature of 740 K a weight decrease is observed, due to decomposition of the oxychloride, yielding CuO and gaseous chlorine. In the TG profile of soot oxidation catalyzed by Cu2OC12, an increase in weight can be observed after complete soot conversion at 650 K - 680 K. Apparently, even in the presence of 20% 02 in N2, an in-situ conversion of Cu2OC12 into CuCI during soot oxidation has taken place. This is corroborated by a TG analysis of heating CuCI in air, which revealed a weight increase at the same temperature where the weight increase of the Cu~OC1Jsoot sample atter 100 % soot conversion (figure 3) occurred. Also the heat effects are similar. Furthermore, the TG profile of heating Cu2OCI~ and soot in nitrogen showed a weight decrease around 600 K, indicating carbothermic reduction of Cu2OC12by soot [11,13]. Therefore, we conclude that during soot oxidation Cu2OC12 is reduced to CuCI.

A b

1605

0.04

905

810 -"

Soot T

2000

1500

"r

1000

Wavenumber (cm "1)

Figure 4. DRIFT analysis of partial converted soot samples. Non-catalytic (3% soot conversion), CuCI catalyzed (50 %), and BiOCI catalyzed (20 %).

651 DRIFT spectra of a partially converted CuCl/soot mixture (50~ conversion), and a FeOCl/soot mixture (60% conversion) are depicted in figure 4. The spectrum of soot after an identical heat treatment is shown for comparison. The DRIFT spectra contain three main absorptions located at 1738 cm"l, 1607 cml and centered around 1257 cm"l. The 1607 cm"l absorption is caused by aromatic stretching vibrations of the soot, which are enhanced by polar functional groups like quinone [20]. The other two absorptions have been assigned to oxygen complexes formed on the soot surface: lactones ( 1738 cm"l) and ether-like complexes (1257 crn'l), respectively [21,22]. Clearly CuCI causes an enhancement of the amount of surface oxygen complexes. Pure CuCI has no infrared absorptions in the spectral region recorded (400-4000 cm'~). However, the absorptions located at 905, 855 and 810 cm~ can be ascribed to water adsorbed on CuCI [23]. A strong band below 600 cml is indicative for Cu2OC12. As this band is not present in the spectrum displayed in figure 6, the DRIFT analysis shows that transformation of CuCI into Cu2OC12 does not occur during catalytic soot oxidation at 550 K. The DRIFT spectrum of soot partially converted in the presence of FeOCI is also included in figure 6. FeOCI catalyzes the formation of surface oxygen complexes. Similar absorptions as for the CuCI sample can be observed. The absorption band at 810 cm"l can be ascribed to FeOCI [24]. Apparently, during soot oxidation FeOCI is not reduced into FeCI2 or other iron chlorides, as was confirmed by a TG/DSC analysis [ 131. The following catalytic cycle is proposed for the activity of CuCI and other transition metal (oxy)chlorides. CuO + ~ Ch

I

CuCI-O* C

C0/C02

C-O,* - ' - - " ~ R3

02 CuCI 5

CuCh Starting soot oxidation with CuCI, the first step is oxygen activation on the surface of CuCI ~1). Transfer of activated oxygen (indicated by O*) occurs according to reaction R2. The DRIFT analysis shows that this results in the formation of Surface Oxygen Complexes (SOCs), indicated by C-O,*. Decomposition of the oxygen complexes results in the formation of CO and CO2. (Surface)

652 oxidation of CuCI has already been proposed in the seventies as an important step in the catalytic conversion of HCl to C12,called the Deacon reaction, and also in the oxy-chlorination of e.g. ethene [25]. Bulk oxidation of CuCI does not occur during soot oxidation, as was discussed previously. Instead, carbothermic (bulk) reduction of Cu2OC12 occurs around 600 K (if this compound is applied as the starting material), yielding CuCI and CO and CO2. Cu2OC12carbothermally reduces at lower temperatures than CuO (600 K vs. 685 K), and even in the presence of oxygen (figure 4). This indicates that a chlorine ligand affects the reducibility of CuO. Once CuCI is formed, it exerts its activity according to scheme (1). Starting soot oxidation with CuCI2 results in a higher soot oxidation temperature, than starting with CuCI or Cu2OC12, because decomposition of CuCb (reaction R5), which is apparently essential for an efficient activation of oxygen (on CuCI), takes place at temperatures higher than 700 K. Reaction R4 indicates the possible decomposition of the active (oxy)chloride into the oxide and chlorine by reaction with the activated oxygen. For CuCI this reaction does not occur in the temperature region, where it is catalytically active (550-650 K). However, for other metal chlorides bulk oxidation occurs at relatively low temperatures, even in the presence of soot, as was discussed previously. At these low temperatures transfer of activated oxygen to the carbon (soot) surface is not fast enough to compensate for reaction R4. As FeOCI was not carbothermally reduced, it is suggested that FeOCI activates and transfers oxygen, just like CuCI. Interestingly, the application of FeOCI leads to the formation of surface oxygen complexes, whereas catalytic soot oxidation by Fe203 does not [19]. This is another indication that chlorine chemically affects the catalytic soot oxidation activity of metal oxides and that the previous scheme also holds for FeOCI. Evaluating the experimental results, the high activity of metal (oxy)chloride based catalysts in the oxidation of soot is induced by a high mobility or volatility of the metal chlorides. Chlorine modification of transition metal oxides might also induce beneficial oxygen activation properties and/or transfer of activated oxygen to the soot surface. A priori, it is to be expected that a major problem of the application of a soot oxidation catalyst based on copper (oxy)chloride, such as the Cu/K/Mo/(CI) catalyst, will be deactivation by evaporation and/or decomposition of the active compound. Although copper chlorides can be reformed by reaction of KCl with CuMoO4 and a high stability has been reported [7,8], Neet~ has demonstrated that the 'loose contact' activity of the Cu/K/Mo/CI-ZrO2 catalyst significantly decreases after treatment in air at 975 K for 24 hours [12]. Moreover, CtrtK/Mo/Cl catalysts supported on ZrO2 and TiO~ were tested in the exhaust gas of a diesel engine, and a profound deactivation was observed [12]. Another aspect of soot oxidation catalyzed by (copper)chloride based catalysts worthwhile mentioning is the possibility that carbon-chlorine bonds are formed by an oxychlorination like reaction. When this reaction is followed by reaction with oxygen, very toxic compounds can be formed. Luijk et al. [26] have demonstrated that during oxidation experiments of an activated carbon catalyzed by CuCb, a relatively small bum-off of the chlorinated carbon surface gives rise to the production of chlorinated compounds such as chlorobenzenes and chlorophenols. Especially chlorophenols are very reactive precursors in the formation of polychlorinated dibenzo-pdioxines at carbon surfaces. In this respect chlorine containing soot oxidation catalysts are less attractive for practical applications. Since the results described in this paper show a correlation between the melting point and catalytic activity of transition metal chlorides, a possibility to eliminate the contamination of the environment by diesel soot particulate is to use oxidic cocktails with low melting points. Also other solutions to overcome the contact problem of transition metal

653 oxides, such as the application of diesel fuel additives [27] or the use of the NOx/soot reaction [28] need to be evaluated. 4. CONCLUSIONS 9Molybdates of copper and potassium have a moderate activity in (diesel) soot oxidation, even in 'tight contact'. 9The high soot oxidation activity in loose contact of catalytic systems containing an alkali metal chloride (KCI or CsCI or LiCI) and CuMoO4 (or CuWO4 or copper vanadates) can be ascribed to the formation of volatile copper chlorides. 9The metal (oxy)chlorides ofCu, Pb, Fe, and Bi appear to be more active in the oxidation of soot than their corresponding oxides in 'loose contact'. 9The high activity of metal chlorides can be explained by the in-situ formation of intimate contact between the soot and the active metal chloride by 'wetting' or through the gas phase. 9The high catalytic activity of metal chlorides and metal oxychlorides can be further explained by the activation of oxygen, followed by a transfer of the activated oxygen to the soot surface, resulting in the formation of surface oxygen complexes (SOCs). Decomposition of SOCs results in CO and CO2 formation. 9Although copper chlorides can be formed by reaction between KCI (which serves as a chlorine supplier) and CuMoO4, the application of the C ~ o / C I - Z r O 2 catalyst and other chloride based catalysts is questionable, because loss of activity due to evaporation and decomposition of the active species will occur eventually. REFERENCES

1. J.P.A. Neefl, Fuel Proc.Techn., 47 (1996) 1. 2. J.P.A. Neefl, O.P.v. Pruissen, M. Makkee and J.A. Moulijn, in A. Frennet, and J-M. Bastin, (Editors), Preprints of the Third International Congress on Catalysis and Automotive Pollution Control, April 20-22, 1994, Brussels, Belgium, 1994, p. 355. 3. A.F. Ahlstr6m and C.U.I. Odenbrand, Appl. Catal., 60 (1990) 157. 4. Sri Rahayu, W.L. Monceaux, B. Taouk and P. Courtine, in A. Frennet, and J-M Bastin, (Editors), Preprints of the Third International Congress on Catalysis and Automotive Pollution Control, April 20-22, 1994, Brussels, Belgium, 1994, p. 365. 5. A. L6we and C. Mendoza-Frohn, Appl. Catal., 66 (1990) L11. 6. P. Ciambelli, V. Palma and S. Vac.c,aro, Studies Surf. Sci. Catal., 71 (1991) 323. 7. P. Ciambelli, V. Palma and S. Vaccaro, Catal. Today, 17 (1993) 71. 8. P. Ciambelli, M. D'Amore, V. Palrna and S. Vacxaro, Combustion and Flame, 99 (1994) 413. 9. Y. Watabe, C. Yamada, K. Irako and Y. Murakami, Catalyst for use in cleaning exhaust gas particulates, European patent application 0092023, (1983). 10. S. Yuan, P. M6riaudeau and V. Perrichon, Appl. Catal. B: Env., 3 (1994) 319. 11. G. Mul, J.P.A. Neeft, F. Kapteijn, M. Makkee and J./L Moulijn, Appl. Catal. B, Env. 6 (1995) 339. 12. J.P.A. Neefl, G. Mul, M. Makkee, J.A. Moulijn, Appl. Catal. B: Env., (1996) accepted. 13. G. Mul, F. Kapteijn and J.A. Moulijn, Appl. Catal. B: Env., (1996) accepted. 13b.A. Goreaud, M. Goreaud and L. Walter-L6~, Bull. Soc. Chim. France, (1970) 2789.. 14. P.A.Neefl, M. Makkee, J.A. Moulijn, Appl. Catal. B: Env. 8 (1996) 57.

654 15. J. van Doom, J. Varloud, P. M&iaudeau and V. Perfichon, Appl. Catal. B: Env. 1 (1992) 117. 16. Gmelins Handbuch der Anorganischen Chemic, 8. Auflage, Verlag Chemic, GMBH.,Weinheim, Fe, 59 B (1932) 279. 17. Gmelins Handbuch der Anorganischen Chemic, 8. Auflage, Verlag Chemic, GMBH., Weinheim, Bi, 19 Erg. (1964) 689. 18. O. Knacke, O. Kubaschewski and K. Hesselmann, Thermochemical properties of inorganic substances, Vol. I and 1I, Springer-Verlag, Berlin, 1991. 19. Y-C Xie and Y-Q Tang, Adv. Catal., 37 (1990) 1. 20. C. Morterra and M.J.D. Low, Spectrosc. Lett., 15 (1982) 689. 21. Q.-L. Zhuang, T. Kyotani and A. Tomita, Energy & Fuels, 8 (1994) 714. 22. G. Mul, F. Kapteijn and J.A. Moulijn, in Proceedings of 22nd Biennial Conference on Carbon, San Diego, USA, 1995, pp. 554-555. 23. C.F. Ng, K.S. Leung and C.K. Chan, J. Ca'tal. 78 (1982) 51. 24. R.A. Nyquist and 1LO. Kagel, Infrared Spectra of Inorganic Compounds, Academic Press, New York, 1971. 25. J.A. Allen and A.J. Clark, Rev. Pure and Appl. Chem. 21 (1971) 145. 26. R. Luijk, D.M. Akkenn~ P. Slot, K. Olie and F. Kapteijn,Environ. Sci. Technol., 28 (1994) 312. 27. J.P.A. Neett, 'Catalytic oxidation of soot-Potential for the reduction of diesel particulate emissions', Ph.D. Thesis TU Deltt (1995), chapter 9. 28. G. Mul, F. Kapteijn, J.A. Moulijn, in preparation.


655

Copper catalysis for particulate removal from diesel exhaust gas. Copper fuel additives in combination with copper coatings. J.P.A. Neeft, S.J. Jelles, M. Makkee, and J.A. Moulijn Delft University of Technology, Section Industrial Catalysis, Julianalaan 136, 2628 BL, Delft, The Netherlands

An exploratory study was carried out with respect to the performance of a copper fuel additive in combination with monolithic wall flow filters for the removal of soot from diesel exhaust gas. Cordierite filters, copper coated cordierite filters, and silicon carbide filters were studied. Model experiments have been performed to investigate the influence of contact between soot and catalyst on the oxidation rate. The observation that the effect of a copper coating on the filter performance is marginal compared to the influence of the copper additive is supported by the model experiments. The contact between soot and catalyst is a crucial parameter for the performance of a catalyst. The contact between a copper coating applied to a filter and soot collected on this filter is not sufficient to achieve continuous, complete soot oxidation under realistic diesel engine exhaust conditions. I. INTRODUCTION Diesel engines are reliable, fuel efficient, and relatively clean. Since the application of three-way catalysts for otto engines, the emissions of diesel engines are subject to discussion. The emissions that are the main topics of concern are those of NOx and particulates. In this paper, the removal of particulates from diesel exhaust gas will be discussed. Particulates in diesel exhaust gas, often called 'soot', are agglomerates of spherical particles, 10 to 30 nm in size, and composed around graphitic spheres. In and on these spheres, hydrocarbons, water, and sulfate are condensed. In the hydrocarbon fraction among others polynuclear aromatics (PAH) are present and these are thought to have carcinogenic properties. The most convenient way to remove particulates from the exhaust gas is combustion. Diesel particulates are, however, relatively unreactive and oxidation occurs moderately at normal exhaust temperatures for both passenger cars and heavy-duty trucks. In the non-catalytic as well as the catalytic oxidation of soot the reaction rate under all circumstances is rather slow in comparison to the residence time in the exhaust system. To achieve complete oxidation, the reaction

656 conditions have to be optimized. The residence time has to be increased by using a filter. The reactivity of the particulates can be increased by either increasing the temperature of the filter system or by activation with a catalyst. Since the increase of temperature is an economically unattractive option, a catalytic solution is preferred. Straightforward application of an oxidation catalyst to a flow through monolith will result in the oxidation of adsorbed hydrocarbons and partly the graphitic nuclei of the soot. This system is able to reduce particulate emissions up to 50 % and is considered to be an intermediate solution [1]. The ultimate goal will be the complete removal of the soot. In general, in soot oxidation two solids, namely catalyst and soot, and a gas phase are involved. The interaction between the catalyst and the soot under practical conditions is poor. To achieve sufficiently high oxidation rates, this contact between catalyst and soot has to be improved [2]. Up to now, the most effective way to achieve sufficiently good contact between the catalyst and the soot is the application of a metal fuel additive [2]. The high activity can be explained as follows. The organo-metallic compound in the additive decomposes and the organic part is completely oxidized in the engine combustion chamber. The residual metallic part is enclosed in the soot at the moment of nucleation and subsequent particulate growth. This results in a high dispersion of the metal in the soot and thus in the ultimate contact between catalyst and soot. Metal based fuel additives, such as Ce [3], Cu [4, 5] , Fe [6], and Mn [7, 8, 9] have been studied for use in combination with a particulate filter. In general, a wall flow monolith is applied as a particulate filter in this kind of operations. In this study the performance of a copper fuel additive in combination with a cordierite wall flow monolithic filter is compared with that of copper coated wall flow monolithic filters. Furthermore, the performance of a silicon carbide wall flow monolithic filter was compared with that of a cordierite one. 2. EXPERIMENTAL 2.1.

Materials

Cordierite wall flow filters were made from segments of 20 mm in diameter and 40 mm in length cut out from an EX 47 / 100 filter, supplied by Corning Glass Works, and consisted of about 36 channels. From these segments, the alternate channels were plugged using ceramic glue (Ceramabond 569 supplied by Gimex B.V.). Silicon carbide filters with identical dimensions were supplied by Stobbe Engineering. Four different types of copper coating were applied to cordierite filters. Type 1 coating was applied by impregnation of a cordierite filter with a saturated Cu(NOa)2 solution, followed by drying and calcination for I hour at 575 K. Type 2 coating was similar to type 1 coating, but instead of one impregnation the filter was sequentially impregnated three times, each time followed by drying and calcination. Type 3 coating was prepared by impregnation of the cordierite filter

657 with a Cu(OH)2 slurry, followed by drying and calcination. With the calcination, the copper compounds are decomposed to copper oxide. Type 4 coating was a proprietary copper coating from ECS and was a gift from Lubrizol. The fuel used was reference diesel CEC-RF-03-A-84 (Haltermann). The lubrication oil used was Valvoline 10W40. The copper additive used was a gift of Lubrizol. 2.2. Procedures 2.2.1. Model experiments Four types of contact were studied. Printex-U, a model soot, was mixed with CuO using a spatula, resulting in 'loose contact' or using a ball mill, resulting in 'tight contact'. Further, Printex-U was impregnated with a solution of Cu(NO3)2 followed by decomposition of the nitrate at 675 K into CuO by heating in nitrogen. A fourth sample was soot collected from the engine described in p a r a g r a p h 2.2.3. with 100 ppm copper in the fuel. With these four samples, flow reactor and TGA experiments were performed. For experimental details see [2] and [10]. 2.2.2. Characterization of soot Soot samples were collected on a glass fibre filter with the engine running on fuel containing 100 ppm copper and analyzed by TEM and XRD. TEM analysis has been carried out with a Philips CM30, high resolution transmission electron microscope operating at 300 kV. This apparatus is integrated with energy dispersive analysis of X-rays (EDX). Alumina sample grids were used for the detection of copper in the soot particulates. XRD has been performed with a FR552 Guinier camera. 2.2.3. Performance m e a s u r e m e n t s The engine used was a Y a n m a r L90E 4 kW, direct injected, naturally aspired, air cooled, one cylinder diesel engine, and had a built-in generator set. The electrical power generated was converted into heat by a resistance bank and was used to control the engine output power. All experiments were performed at an engine load of 3 kW. The engine rotational speed was set at 3000 rpm. At the applied load of 3 kW, the particulate emission is 4.7 g/h [11]. The filters were glued onto a quartz tube with an outer diameter of 20 mm and a wall thickness o f - 1 . 5 mm using Ceramabond 569. The quartz tube was then placed in the oven, as illustrated in Figure 1. A side stream of the engine exhaust gas, generally ~12 1/min, was pumped through the filter. The oven temperature was controlled. The t e m p e r a t u r e within the filter was normally about 20 K below the oven temperature. Before each experiment, the pressure drop over the clean, unused trap was measured with air at a filter temperature of~575 K. The total gas throughput was measured with a wet gas meter. A guard filter, placed downstream of the test filter (item 9 in Figure 1) was used to check for leakage. If, after the experiment, the guard filter was not sufficiently clean, the experiment was considered to be a failure.

658 During the experiment, the pressure drop over the filter was continuously measured and recorded in combination with the temperatures upstream, downstream of, and within the filter. At the end of a test, the filter was usually regenerated by heating the oven with a controlled rate of 25 K/min to a temperature of 875 K, resulting in the complete combustion of the collected soot on the filter. During these regenerations the pressure drop over the filter and the temperatures within and downstream of the filter were recorded.

Figure 1. Experimental set-up XRF analyses have been performed with a Philips PW 1400 spectrometer to get a n indication of the amount of copper present in different trap types. Five traps were analyzed. The matrix of l~Iter test experiments is shown in Table 1. These tests were performed to compare the performance of catalytic coatings, of fuel additives, and of the combination thereof. These filter experiments were all performed with the oven temperature set at 650 K.

659 Table 1. Matrix of experiments (referred to by the run number) performed for comparison of filters. The oven temperature was set at 650 K for all experiments. a dditive concen tra tion filter type

0 ppm

100 ppm

blank cordierite

12

3-6, 14, 20

blank silicon carbide

11

10, 18, 19

cordierite, coating #1

1


2


7


22

200 ppm

23

The influence of the oven temperature on the collection behaviour has also been investigated. Several experiments have been performed with blank traps of cordierite and of silicon carbide, at different collection temperatures. For all these experiments, the additive concentration was 100 ppm. These experiments are listed in Table 2. Table 2. Matrix of experiments (referred to by the run number) performed with different external filter temperatures. The additive concentration was 100 ppm for all experiments. Oven t e m p e r a t u r e (K) l~lter type

blank cordierite

650

blank silicon carbide

3, 6, 14, 20 10, 18, 19


23

3.

675

700

725 13, 21

17

16

15 24, 25

RESULTS AND DISCUSSION

3.1. Model experiments

From the model experiments performed by Mul [10], comparison of contact modes results in an order of activity of i m p r e g n a t i o n > tlght contact > loose contact. Results from Neeft et al. [12], give an order of activity of addltlve >> t i g h t contact > loose contact. The difference in activity is several orders of magnitude. From Kissinger plots, derived from TGA/DTA data, the activation energy Ea and the pre-exponential factor ko can be determined. For soot with copper from an additive and soot with CuO in tight contact, the Ea is the same but the ko differs [11]. This indicates that the copper in its environment of soot is

660 the same in both cases, but the number of sites of interaction differs. In practice, the contact between soot and a catalytic coating applied on a filter, resembles loose contact. 3.2.

Soot characterization

Figure 2a shows a TEM micrograph of a collected soot sample generated with 100 ppm copper in the fuel. The agglomerates and the individual soot spheres of 30 40 nm can be distinguished. Copper particles should appear as dark spots. Although EDX analysis shows t h a t in the background copper is present all over the sample, dark spots are not observed. From the line resolution of the microscope of 0.16 nm, it can be concluded t h a t the copper in the particles is almost atomically dispersed in the soot and t h a t the sizes are generally smaller t h a n 1 nm. the sample contains at least 1 wt % copper. These results, therefore, confirm the TEM/EDX results. The copper is highly dispersed present within the soot. 3.3. Elemental composition of the filters after use

In Table 3, the analyzed traps are listed in combination with the amount of exhaust gas t h a t had been pumped through the filter and the copper concentration in the fuel used. Table 3.

sample 1

The history of several traps and their results of XRF analysis. life* (1)

fuel (ppm copper)

MgO (wt%)

A1203 SiO2 (wt%) (wt%)

CuO (wt%)

trap type blank cordierite (fresh) blank cordierite

-

-

9.2

48.9

38.9

-

4100

200

7.6

52.4

35.3

0.1


3800

100

8.5

42.4

29.0

17.3


3100

100

8.2

45.3

30.6

12.1


5000

100

7.6

50.2

33.1

5.3

*Life : Amount of exhaust gases in litre passed over the filter device XRF is a semi-quantitative technique and it gives a good impression on the relative amounts. The amount of copper present in the blank, cordierite trap (sample 2) is roughly two orders of magnitude lower t h a n the amount t h a t is deposited on such a trap by impregnation. The numbers in Table 3 show t h a t the copper coated filters contain high amounts of copper. In view of the fact that cordierite is a meso/macro porous material, the copper dispersion is expected to be low and reproducibility is also expected to be poor. The low number for exCu(OH)2 coating #3 (sample 5 in Table 3) is not surprising in view of the low solubility of Cu(OH)2 in the impregnation solution.

561

Figure 2a. TEM micrograph of soot particulates from a 100 ppm fuel additive run. No individual copper particles can be observed. - - " " - - : 100 nm

Figure 2b shows a micrograph of a r a t h e r unique spot in the sample, where amorphous copper particles around the soot can be distinguished by EDX. In the lower right corner, a large lump of copper can be seen. Both phenomena are probably the result of sintering of copper from soot, which has been accumulated in the exhaust system. Figure 2b. TEM micrograph of soot particulates from a 100 ppm fuel additive run. Copper particles are dark. -----:

100 nm

With XRD analysis no copper reflections were found. The smallest crystalline particles t h a t can be detected with the used camera are typically 5- 10 nm, when

662 the sample contains at least 1 wt % copper. These results, therefore, confirm the TEM/EDX results. The copper is highly dispersed present within the soot. 3.4. Filter performance tests The pressure drop over the clean, unused filters was around 20 - 30 mbar for silicon carbide and around 50 mbar for non-coated EX 47. For coatings of type #1 and type #2, the clean pressure drops were around 80 mbar. For type #3 coating, it was around 50 mbar and for the type #4 coating it was around 130 mbar. The reported pressure drops have been corrected for the contribution of the clean filter. In Figure 3, the pressure drop curves are presented for experiments performed with EX 47 filters at a temperature of 650 K. These curves show the

500 q no copper

,~'~ 400

/ r u n 12

~o

.-..-..run 14 ~ 3

///

"~ 300

/ //~/t///

200

__ l OOpp m

~

/copper

/ ..... .~--

run8

100

0 0

........

2 0 0 p p m copper

!

+

~-

I

2

4

6

8

.........

t 10

time (hours)

Figure 3. The influence of the additive concentration on the pressure drop. For every run, a fresh EX47 trap was used. The oven t e m p e r a t u r e was 650 K. dramatic influence of the additive concentration on the system performance. Without the additive the pressure drop increases steeply, whereas in the presence of the additive a constant level is reached. At higher copper concentrations, the pressure drop is lower. The experiments performed with 100 ppm copper in the fuel (3, 14, 20) show t h a t the reproducibility in these pilot bench experiments is remarkably good.

663

=o I t t

-I

~,

l OOppm copper

m

"u

~3oo

~ 2oo

~

100

0 0

2

4

time (hours)

Figure 4. The effect of regenerations on pressure drop curves. For all runs, the same EX47 trap was used. The filters were regenerated after each run. The oven temperature was 650 K, the additive concentration 100 ppm.

500

no copper

]'

i......................................

!

r'~ 400

1

i

run 12/ ....

E X 47

sic

/

/run

11

run 14 -----'--

/ |=

~ ~

run 19 - - - -

.

.

.

.

.

.

E X 47

_

l OOppm copper

_

__J

200

~

100

0

.................................. 0

-t- . . . . . . . . . . . . . . . . . . . . . 2

-4- . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

A

4

6

t ~ e (hours)

Figure 5. The influence of the filter material on the pressure drop. Thin lines represent runs with an EX47 trap. Thick lines represent runs with a SiC trap. The oven temperature was 650 K. In Figure 4, the influence of regeneration on the performance of the system is shown. The difference between the curves is not significant. This implies that there is no plugging of the filter by the copper collected on the filter within the experimental time. On the other hand, it is clear t h a t the copper that is built up

654 on the filter during a collection period, does not exhibit measurable activity after a regeneration probably due to sintering during the regeneration procedure. During the regeneration process of filters, the combustion of the soot collected on the filter proceeds very rapidly and causes a temperature runaway. The t e m p e r a t u r e in the filter can be as high as 1200 K during these events. This imposes high demands on the filter. Exceptions for this observation are silicon carbide filters and cordierite filters with coating #4. Regenerations of these filters are not accompanied by thermal runaways. Silicon carbide has a better thermal conductivity and this property possibly has a strong impact on the regeneration behaviour. In Figure 5, the pressure drop curve is plotted for the two different trap materials used. The build-up of pressure drop occurs less steep with silicon carbide filters t h a n with EX 47 filters. In presence of 100 ppm copper in the fuel, the performance of the two different trap materials are almost identical. 500

no copper

"~ 400 ~o

run 12/,'-~..run 22 S c ~

no coatina

3X) ~ / / ~ I

~'/

~

~

~ I~// o 100 ~

0

o

__

,.

no coaung

run7 run2

y /

0

run 14

~

coating#3 a

t

i

n

g

~

.

.

run 3

~ 2

l OOppm

oating #

.

.

.

copper

.

no coating

~-...... 4

200ppm copper

6

t~e (hours) Figure 6. The performance of copper t e m p e r a t u r e was 650 K for all runs.

coatings.

The

oven

The effect of a copper coating on the performance of the system is shown in Figure 6. The activity of a copper coating is very low compared to that of the copper additive. In all cases, the copper coated traps show a slightly lower equilibrium pressure drop, which might be favourable. It should, however, be t a k e n into account t h a t the pressure drop over clean, copper coated traps is in most cases higher t h a n t h a t of clean cordierite traps. Therefore, the reduction of the pressure drop by a coating is only relative. In an absolute sense, a copper coating results in a higher equilibrium pressure drop. In summary, coating of a trap in additive based systems will only be useful when the manufacture of the coated traps is done in such a way t h a t the coating does not contribute to the pressure drop.

665 All pressure drop curves in Figure 6 show a maximum except for the curve for the uncoated cordierite filter with 100 ppm copper (run 14). This maximum in pressure drop suggests that there is a build-up of activity on the filter. Two explanations can be given for this increased activity. A possible explanation is the formation of small copper particles that are stabilized by the soot. These particles have a high mobility and can be active [11]. A second explanation is also based on the formation of small copper particles. These particles can act as a catalyst for the oxidation of NO into NO2. The formed NO2 subsequently oxidizes the soot, forming NO. This catalytic cycle with NO as a key intermediate is described for platinum by Hawker [13] and for copper, chromium and molybdenum by Mul [10]. Although complete oxidation of diesel soot with only an NO-oxidation catalyst such as platinum may prove not to be possible, the catalytic cycle with NO as intermediate should be considered crucial for every catalytic process for oxidation of diesel soot.

NIO~

"~400

o

300

200 ~.~ r~ o

100

................

0

0

4. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

1

2

3

4

5

6

time (hours)

Figure 7. The performance of several filters at elevated temperature. The oven temperature was 725 K for all runs, the additive concentration 100 ppm. To get information on the temperature dependence, also experiments at high collection temperature were performed. Figure 7 gives results for 725 K and 100 ppm copper additive. Spontaneous regenerations can be observed with the cordierite filter without coating. These regenerations occurred stochastically, and are reproducible. The peak temperature measured in the filter during these regenerations were as high as 1200 K. With silicon carbide traps or a cordierite trap with coating #4, this phenomenon was not observed and stable operation seemed to be possible. An important difference between silicon carbide and cordierite is the thermal conductivity. Probably due to the high thermal conductivity of silicon carbide, hot spots in the filter can be avoided and

666 runaways do not occur. The stable operation characteristics of the copper coated cordierite trap could be explained by catalytic activity of the copper coating or by the increased thermal conductivity. The latter explanation is preferred by comparing the smooth regeneration behaviour of silicon carbide traps one of cordierite with coating #4, although a catalytic influence is not excluded.

4. CONCLUSIONS The additive concentration is of great influence on the pressure drop build-up with any filter coating or trap material. Copper coated EX47 filters reduce the pressure drop build-up over a filter, but this effect is limited compared to the effect of a higher additive dose. Further, the gain in corrected pressure drop is reduced by the fact that the clean pressure drop is higher for coated filters. The limited influence of the copper coating on the pressure drop is confirmed by model experiments. The activity of copper as an additive is several orders of magnitude higher than that of copper in loose contact, the latter resembling the contact between soot and a coating on a filter. Silicon carbide is a promising material for use in diesel filters. NO can play an important role in the catalytic oxidation of soot from diesel exhaust gas. REFERENCES

1. 2. 3. 4. 5.

J.P.A. Neeft, M. Makkee and J.A. Moulijn, Fuel Process. Technol. 47 (1996) 1 J.P.A. Neeft, M. Makkee and J.A. Moulijn, Chem. Eng J. 64 (1996) 295 K. Pattas et al., SAE Paper 920363 (1992) J. Saile, G.J. Monin and D.T. Daly, IMechE, (1993) 171 D.T. Daly, D.L. McKinnon, J.R. Martin, and D.A. Pavlich, SAE Paper 930131 (1993) 6. E. Miiller, B. Wiedemann, A.W. Preuss, and H.K. Sch~idlich, ATZ 91, (1989) 674 7. A.G. Konstandopoulos et al., SAE paper 880009 (1988) 8. E.D. Dainty et al., SAE Paper 870014 (1987) 9. B. Wiedemann and K.H. Neumann, SAE Paper 850017 (1985) 10. G. Mul, Catalytic diesel exhaust purification. Thesis (1997) ll.J.P.A. Neeft et al., IMechE (1996) 233 12.J.P.A. Neeft, M. Makkee and J.A. Moulijn, Appl. Catal. B:Environ 8 (1996) 57 13.P.N. Hawker, Platinum Metals Rev. 39 (1995) 2


667

Supported liquid phase catalysts: A new approach for catalytic oxidation in diesel exhaust particulate emission control. S.J. Jelles, B.A.A.L. van Setten, M. Makkee, and J.A. Moulijn Delft University of Technology, Section Industrial Catalysis, Julianalaan 136, 2628 BL, Delft, The Netherlands

An exploratory study was carried out with respect to the performance of molten salts as diesel soot oxidation catalyst. The activity of two binary eutectic salts, with a low melting point, was measured and compared with the activity of two very active solid single oxides. Also the influence of NO on the oxidation rate was investigated. It was observed that the molten salt catalysts are highly active compared to the solid single oxide catalysts, probably as a result of the increased contact area due to wetting of the soot by the mobile catalyst. The oxidation rate is strongly increased by the presence of NO in the gas phase. 1.

INTRODUCTION

In the abatement of the emissions of diesel engines, particulate removal is an essential step. Catalysis is a promising technology but a fundamental problem is the poor contact between particulate and catalyst in filters, which are currently under development. Diesel fuel additives are at this moment the best way to achieve optimal contact between catalyst and soot. The organo-metallic compound in the additive decomposes in the engine combustion chamber and the metallic part is integrated within the soot at the point of nucleation. Copper and ceria are successfully used as a fuel additive. Topics complicating the introduction of fuel additives on a large scale are numerous. The additive should be added to the fuel at some point, leading to either an additional distribution network for the fuel or a complicated dosage system. Another point is the emission of the metal or the catalytic side reactions which might occur. For example, polyaromatic hydrocarbons from the soot and chlorine, present in traces in the fuel, could be converted into dioxins by the catalyst. Copper is an active catalyst for this reaction. Finally, the additive could cause engine wear or oxidation of the fuel in the tank. In summary, catalytic solutions other than fuel additives could be favourable. The application of a catalytic system which becomes liquid at the average lower exhaust temperature, can have the possibility to achieve sufficient contact between catalyst and soot. The interaction between catalyst and soot is than no longer based on the poor interaction between two solids, but a good contact between a solid and a liquid. The use of a liquid catalyst is new in the field of exhaust gas cleaning and there are many unanswered questions and unsolved problems. In this exploratory work, it will be shown that the application of molten eutectic salt mixtures as liquid catalysts might result in sufficiently

668 active oxidation catalysts for continuous diesel soot removal. The most relevant topics will be discussed. 2. BACKGROUNDS 2.1. Contact between soot and catalyst In earlier work, [1, 2] it has been shown that the contact between soot and catalyst is crucial for the performance of the catalyst. McKee et al. reported that eutectic salt mixtures can be more active than their individual pure components in the catalysed gasification of graphite and coal char [3]. McKee also reported that the activity of a eutectic mixture increases dramatically above the melting point [4] and argued that this was caused by migration of the small salt droplets over the carbon surface. In diesel soot oxidation, the situation can be similar, but more complex. Migration of the catalyst over the soot surface can easily lead to loss of catalyst in the exhaust gas stream. Ideally the wetting of soot should occur without the loss of contact between the salt and the material by which the liquid catalyst is supported. In Figure l a, the interaction between a solid fractal soot particle and a solid catalyst interface is schematically drawn. It is clear that the number of contact points between the fractal soot and the catalytic surface are small, compared to the size of the soot particle.

Figure 1.

Simplified view of the contact between a fractal soot particle and the catalyst, a) normal solid catalyst, b) liquid catalyst

The contact between a liquid catalyst and a soot particle is schematically drawn in Figure lb. Because of the wetting capacity of the liquid catalyst, the contact area between soot and catalyst can be significantly higher than in the case with solid catalysts. It is possible that the wetting of the soot particle by the molten salt imposes a limitation on the transport of oxygen to the soot particle, as suggested by McKee [3]. On the other hand, an oxidation cycle with the catalyst as an intermediate, can be the dominating mechanism. This would imply that the complete coverage of the soot with a film of the liquid salt is perhaps favourable. 2.2. Particle capture The technology used for the application of a molten salt catalySt is a dominant issue. The soot particle has to be separated from the gas stream and brought into contact with the liquid catalytic phase, whereas the liquid phase should not be blown out or evaporated to the gas phase. The surface structure that is in contact with the gas stream should be large to increase

669 the contact area between particulates and catalyst and the structure should be porous to retain the catalyst. It is clear that a normally used filter, such as a wall flow monolith, is not preferred for the application with a molten salt, because the particles are collected on the filter in stead of in the filter. An interesting material can be a ceramic foam. Their normally lower filter efficiency can possibly be compensated by the wetting of the soot particles by the molten salt, resulting in high sticking efficiency of the soot onto the filter surface.

2.3. Catalyst selection The temperatures, where the system should operate range from 425 K to 1250 K. The water partial pressure in the exhaust gas (up to 10 %) and the gas velocity in the exhaust system (-~2.5 m/s) are high. This imposes a number of requirements that the molten salts have to fulfil, apart from intrinsic catalytic activity. The number of possible salt mixtures are of course large and rational design procedures are called for. The number of possibilities can be reduced as follows. Several molten salts are volatile under the applied conditions. Chlorides for example are known for their high vapour pressure. Nitrates, usually forming low melting point eutectic mixtures, are unstable and will decompose. Another point of interest is the reaction of the original salts into, for example, carbonates and sulfates with components in the exhaust gas, e.g. carbon dioxide or sulfur dioxide, respectively. This could influence the composition of the melt and, thereby, increase the melting point and subsequently reduce or even destroy the activity. The intrinsic activities of the catalytic material are also of importance. From earlier studies and literature, several components can be selected as promising candidate, for example vanadium, cobalt, and molybdenum [1].

2.4. Catalyst testing The soot oxidation activity of a molten salt can be studied using for example a thermobalance (TGA) or a micro-flow reactor. The long term stability of the salt cannot be tested straightforward using soot oxidation with this kind of equipment. Long period tests using a diesel engine have some drawbacks. Engine experiments are complicated and not always reproducible. Further, separate parameters, such as soot production, cannot be set without changing other parameters, such as exhaust temperature and the gas-phase composition of the exhaust stream, such as NOx, H20, and SO2 concentration. Therefore, a dedicated experimental set-up has been developed and is currently being evaluated. It consists of a feeding system, with which a constant amount of model soot can be introduced into the gas stream. The soot loaded gas stream is led towards a catalyst system. The pressure drop over the catalyst filter and the production of CO and CO2 are measured. In this way, the performance of the catalyst can be determined quantitatively. The feed gas composition can be varied and components such as water, SO2, and NOx, can be added without changing the amount of soot.

670 3. E X P E R I M E N T A L

3.1. Catalyst preparation The eutectic mixtures discussed here were CsVO3 - MoO3 (75.5 mol% - 24.5 mol%) (A) and CSEMoO4 - V205 (49 mol% - 51 mol%) (B), with a eutectic melting point of 635 K and 625 K, respectively. C s V O 3 was prepared by sintering of stoichiometric quantities of Cs2CO3 and V205 at 725 K overnight, followed by melting at 925 K and cooling down during several hours. The obtained crystals were white. Cs2MoO4 w a s prepared by sintering of stoichiometric quantities of Cs2CO3 and MoO3 at 725 K overnight, followed by melting at 1200 K and cooling down in several hours. The obtained crystals were white. The eutectic mixtures were prepared by heating of physical mixtures of stoichiometric quantities of the composing salts at 100 K above the eutectic melting point for 1 hour, followed by cooling down. Differential thermal analysis, (DTA) performed with a STA 1500 from Polymer Laboratories confirmed the presence of a eutectic composition. The eutectics were milled and sieved.

3.2. Flow reactor experiments The flow reactor equipment used is described in detail by Neeft [5]. As a model soot, Printex-U, a gift from Degussa, was used. Details about the properties of Printex-U are given by Neeft [5]. Catalyst samples, sieve fraction 50 - 215 p m , and Printex-U were mixed with a mass ratio of 4:1 using a spatula. From these samples, 100 mg were used in the flow reactor. The activity of the eutectic mixtures A (CsVO3 - MoO3) and B (Cs2MoO4 - V205) and of copper oxide, and molybdenum oxide, all in loose contact with the soot, was measured at a temperature of 650 K. The gas composition was 10% oxygen in argon. The influence of NO on the oxidation rate with catalyst B was investigated at a temperature of 675 K, with 10 % oxygen and 1000 ppm NO in argon.

3.3. TGA experiments A STA 1500 thermobalance from Polymer Laboratories was used to investigate the activity of the eutectic mixtures. The ratio catalyst : Printex-U used was 4 : 1. The samples were diluted with silicon carbide. A heating rate of 5 K/min was applied in synthetic air. In a Cahn TG 131 thermobalance, two samples were analysed. One sample was blank cordierite EX47 (Coming) and one sample was cordierite impregnated with CsVO3 - MoO3. The load of salt on the cordierite was 1 g/g. From both samples, a segment of lxl cm and 1 channel (-~0.5 cm) thick was cut. Both segments were then covered with about 40 mg PrintexU. The soot covered segments were placed in the thermobalance. The sample was flushed with air. The temperature was increased from room temperature to 625 K with 5 Fdmin and was sequentially heated up to 675 K with 0.5 K/min, kept there for 6 hours, and heated with 1 FUmin up to 875 K. The mass signal and the temperature signal were simultaneously recorded.

671 4. RESULTS AND DISCUSSION

4.1. Flow reactor experiments

4.1.1. Comparison with single oxide catalysts In Figure 2, the oxidation rate of Printex-U at 650 K is plotted as a function of conversion for the eutectic mixtures CSEMoO4- 7 2 0 5 and C S 7 0 3 - MoO3 and for the single oxides MoO3 and CuO, all in loose contact. The experiments were performed under identical conditions. It is clear that over the whole conversion range the oxidation rate is 2-3 times higher for the eutectic mixtures than for the single oxides. The difference in oxidation rate increases with higher conversion. In contrast to copper and molybdenum, the rate of oxidation in the eutectic mixtures remains more or less constant up to nearly complete conversion. This means that the contact between the soot and the catalyst is maintained during oxidation.

8.0.E-05

, i

.L,t"

E

6.0.E-05

i

4.0.E.05

~

g

!

",

", Cs2MoO4.V205 ",

i 2.0.E-05

"--

- ~ ~ . . ~ . ~ C

uO

" * " ~ .....

MoO3 O.O.E+O0

. 0.1

.

. 0.3

. 0.5 conversion

0.7

0.9

(-)

Figure 2. Oxidation rate (in mg C combusted / mg C initially present/s) of Printex-U at 650 K. The gas phase was 10 % 02 in argon. The experimental time on stream was the same for each experiment. The ratio soot : catalyst was 1:4 (g/g).

4.1.2. The influence of NO In Figure 3, the influence of NO on the oxidation rate of Printex-U is shown. The eutectic mixture used for this experiment is CsVO3 - MOO3. The influence of NO on the oxidation rate is significant, although the shape of the oxidation curve is not changed.

672

1.5E-04

...........................................................................................................................................................................................................

1.0E-04

o

~

o

,,~

"- o o

E

=,

~

with NO ",..

O~ v

E .~

~

5.0E-05

o o

,,o

without

0.0E+00 0.20

NO

I

i

i

0.40

0.60

0.80

1.00

c o n v e r s i o n (-)

Figure 3. The influence of the NO on the oxidation rate (in mg C combusted / mg C initially present / s) of Printex-U at 675 K catalysed by CsVO3 MoO3. The gas phase was 10 % O2 in argon. The NO is probably enhancing the oxidation of soot by an oxidation cycle. This cycle is discussed by Mul [2] and Hawker [6]. Mul proposed a cycle as shown in figure 4. The NO is oxidised catalytically into NO2, whereas the soot is oxidised by NO2 uncatalysed. The oxidation by NO2 can also be catalysed. This could be a possible application of two different catalytic systems. One metal then catalyses the oxidation of NO to NO2 and the other metal catalyses the soot oxidation. Recent work with fuel additives supports the existence of such a mechanism [7].

02

I catalyst

c NO

NO2

I Icatalyst I ] Figure 4. Catalytic cycle with NO as intermediate [2]

co + co2

673 4.2. T G A

experiments

In Figure 5, the TGA-DTA curves for the eutectic mixture Cs2M004 - V205 are shown.

1.2 1.0 "7", m 0.8

15 "--"

i. . . . . .

:= _ - ~ N ~ \

.....

-

, relative mass --heatflow

o.6 4.a

r

.....

0.4

/!i___

0.0 -0.2 400

................. mittin eutect of c i 500

1

600

! 700 temperature

~r'

g

o

0

a)

0.2

10

--,,.,,..' i

1

800

900

-10 1000

(K)

Figure 5. TGA-DTA curves for the eutectic catalyst mass ratio catalyst" soot was 4 19 (g/g).

Cs2M004

- V205

9The

Around the melting point of the eutectic, 620 - 640 K, the DTA signal shows a small dip for all mixtures, representing the fusion enthalpy. From this point on, the catalyst immediately shows combustion activity, as can be seen from the rising DTA signal. In Figure 6, the mass loss and temperature are plotted against the time for the two cordierite samples that were analysed with the Cahn TG 131 thermobalance. It is clear that in the isothermal part at 675 K, the mass loss of the impregnated cordierite sample occurs significantly faster than that of the blank cordierite sample. The majority of the soot is burnt at this temperature. For the blank, not impregnated, cordierite sample, the major part of the soot only bums at higher temperature. It should be noted that the circumstances for contact between soot and catalyst were not optimal under these conditions. The soot was deposited on the segments when they were at ambient conditions. Therefore, the catalyst was in the solid state and no wetting could have occurred. This means that significantly higher oxidation rates could be observed under realistic conditions. Although detailed studies are called for, these results show the potential of eutectic catalysts. The difference between the rates for the blank and the impregnated samples is not very large although the molten salts clearly show a higher oxidation rate.

674

1000

~-'- -'-~

-10

~__,_.

..~.. _~_. - 9~

.

v ry~ ~O -20

.

.

.

.

,~'~'-"~---~-" / /

.

.

.

.

.

.

.

.

.

.

.

.

.

.

~.~,,~ s

.st ~t

. ~ J"

_ . . . . . . . . .

800

.~

e"

impregnated ~.-- cordierite ~

\~ \ blank ~-cordierite

600

400 -30

-

~ - ~

~1\ 200

-40

I ..................................

0

200

I .......

400

I

600

800

time (minutes) Figure 6.

TGA-curves for CsVO3 - MoO3 impregnated and blank cordierite segments, covered with Printex-U.

5. CONCLUSIONS The use of a Supported Liquid Phase Catalyst offers an opportunity for continuous removal of soot from diesel exhaust gas. Molten salt mixtures show a higher activity compared to solid metal oxides. This high activity can be ascribed to a better contact between the liquid catalyst and the soot. The contact between soot and catalyst remains intact during oxidation. The oxidation rate of soot is further enhanced by NO, which is present in diesel exhaust gas and this effect is, therefore, favourable. For successful application, a technology has to be developed to bring the soot into contact with the catalyst without losing the catalyst during operation. Ceramic foams impregnated with a molten salt or dedicated catalyst/filter systems are thought to be promising candidates for this application.

REFERENCES 1. 2. 3. 4. 5. 6. 7.

J.P.A. Neeft, M. Makkee and J.A. Moulijn, Chem. Eng. J. 64 (1996) 295 G. Mul, Catalytic diesel exhaust purification, PhD thesis (1997) D.W. McKee, C.L. Spiro, P.G. Kosky, and E.J. Lamby, FUEL 64 (1985) 805 D.W. McKee, Carbon, 25 (1986) 587 J.P.A. Neeft, Catalytic oxidation of soot, PhD thesis, 1995 P.N. Hawker, Platinum Metals Rev. 39 (1995) 2 S.J. Jelles, M. Makkee, J.A. Moulijn, unpublished results

AUTHOR INDEX

675

Abi-Aad E.

625

Bowker M.

Abouka'is A.

625

Brungard N.L.

165

Acke F.

285

Burch R.

199

Andersson B.

317

Calis H.P.A.

357

Angelidis T.N.

155

Capelle S.

625

Arendarskii D.A.

507

Cataluna R.

591

Avila P.

233

Cauqui M.I.

611

Azelee W.

495

Cerrato G.

601

BNagy J.

365

Chajar Z.

335

Baiker A.

61

Chevrier M.

135,335

Bailey C.

495

Ciambelli P.

307, 635

Bainier- Davias N.

335

Cochardo P.

611

93

Conesa J.C.

591

Bazin P.

571

Connerton J.

327

Bennett R.

431

Corbo P.

307

Bensaddik A.

265

Corro G.

93

Bernal S.

611

Cort6s Corberan V.

591

Blanchard G.

571

Courcot D.

625

Blanco G.

611

Cunningham J.

409

Blanco J.

233

Daniell W.

495

Boix A.

175

Dathy C.

419

Borovinskaya I.P.

477

Day J.Paul

353

Botas T.A.

73

Deeba M.

529

Bourges P.

213

Denton P.

335

t

Barbier J.

409,431

676

Gauthier C.

Descorme C.

275

Di Monte R.

185,569

Gelin P.

275

Donnerstag A.

125

Gilot P.

297

Drewsen A.

113

Goguet A.

275

Dubien C.

399

Gonzalez-Velasco

Duprez D.

549

Granger P.

Eckhoff S.

223

Graziani M.

Eigenberger G.

125

Grigoryan E.H.

477

Emons H.

155

Gubitosa

569

Eriksson L.

285

Guczi L.

347

Essayem N.

137

Guilhaume N.

581

Farrauto R. J.

529

Guti6rrez-Ortiz

Farrell F.

409

Guyon M.

297

Feeley J.

529

Halfisz J.

367

Ferrero A.

569

Harrison P. G.

495

Ferret R.

73

137, 335,549

73 419 185,569

73

Hecq W.

Filkin N.C.

255

Herzog P.

35

Fonda E.

185

Hesse D.

223

Hickey N. J.

409

Hoebink J.H.B.J.

389

Irusta S.

175 507

Fornasiero P. Fr6ty R.

Fridell E.

185,569 137 113,285,537

Fritz A.

243

Ismagilov Z. R.

Frost J.C.

379

Isomura N.

83

Gambino M.

307

J~irSs S. G.

465

Garin F.

27, 265

Jelles S.J.

621,655,667

677

Jobsson E. Johansson S.

113 285,537

Le Govic A.M.

571

Leason P.

513

Jones I.

431

Leclercq G.

419

Joyner R.W.

327

Leclercq L.

419

Kapteijn F.

645

Lecomte J.J.

419

L~cuyer C.

275

Leyrer J.

223

Kaspar J.

185,569

Kessler H.

297

Kirchner T. Kiricsi I.

25 347, 367

Lindner D. Lioutas Ch.B.

155 495

Kisenyi J.M.

513

Lloyd N.C.

Knapp C.B.

233

Lox E.S.

Kiinig A.

125

Lunati S.

Kiippel R.A.

61

51

Mabilon G.

51,223 213 213,419

Koryabkina N.A.

507

Magnacac G.

Koutlemani M.M.

a55

Maire G.

Krawczyk M.

265

Makkee M.

621,645,655,667

Kreijveld R.J.M.

389

Marecot P.

93

51

Marin G.B.

389

Kreuzer T.

601 147, 243,265

Kucherov A.V.

441

Marques Da Silva A.

Kiinkel C.

317

Martinez-Arias A.

591

Massardier J.

103

93

Lafyatis D.S.

79

Lakeev S. G.

441

Mathis F.

137

Lambert R.M.

255

Merzhanov A.G.

477

Lavalley J-C.

571

'Migliardini F.

307

Le Chanu V.

297

Mir6 E.

175

678

Moraweck B.

103

Primer M.

Morgan G.

513

Pritchett D.

513

Morterra C.

601

Qian M.

485

Mouaddib N.

265

Rajaram R.R.

379

Mouaddib-Moral N.

549

Renouprez A.

103

Rodriguez-Izquierdo

611

Rogemond E.

137

51

Rohe R.

147

645

Russo P.

635

621,645,655

Salasc S.

137

Moulijn J. A. Mubarak A.T. Muf~mann L. Mul G. Neeft J. P.A.

621,645,655,667 495

137,275,335,581

Nibbelke R.H.

389

Saur O.

571

Odenbrand I.

317

Schay Z.

347

Palermo A.

225

Schiibel G.

367

Palma V.

635

Schweich D.

399

Pang X.Y.

485

Searles R.A.

23

Papadakis V.G.

155

Shelef M.

441

Perrichon V.

a37

Shikina N.V.

a07

Pettersson L.J.

465

Shinjoh H.

Petunchi J.

175

Shkrabina R.A.

507

Pintado J.M.

611

Sklavounos S.A.

155

Pitchon V.

83

147,243,265

Skoglundh M.

113,285,537

Poulston S.

431

Smaling R. M.

165

Prado G.

297

Smelder G.

285

Praliaud H. Prigent M.

103,335 419

Sobukawa H. Soe K.

83 513

679

Soria J. Sugiura M. Sung S. Tagliaferri S.

591 83

165 61

Taha R.

549

Thorm~ihlen P.

113

Tikhov M.S.

255

Tooby C.

513

Tiirncrona A.

113,537

Trillat J.F.

103

Vaccaro S.

635

van den Bleek C.M.

357

van den Tillaart

223

van Kooten W.E.J.

357

van Setten B.A.A.L. van Yperen R..

621,667 51

Varga J.

367

Vassallo J.

175

Vlaic G.

185

Voulgaropoulos A.

155

Wahlberg A.M.

465

Walker A.P

379

Walsh M. Watling T.C.

199

Weeks C.

409

Westerberg B. Yates M.

Yentekakis l.V.

285, 317, 537 233 255

Zeng H.C.

485

Zillikens M.

513


681

LIST OF PARTICIPANTS

ACKE

Filip

Chalmers University of Technology

Sweden

ADELMAN

Bradley Jay

Universit6 Pierre et Marie Curie

France

ANDERSSON

Lennart

Volvo Car Corporation

Sweden

ANDORF

Renato

Daimler- Benz A.G.

Germany

ANDRE

Didier

Total Raffinage Distribution

France

ANGELIDIS

Thomas

Aristotle University

Greece

ASUQUO

Raymond

Chemopetrol A.S.

Czech Republic

AUERNHAMMER

Markus

Swiss Federal Institute of Technology Switzerland

BAIKER

Alfons


BAILEY

Craig

Nottingham University

United Kingdom

BARBIER

Jacques

Universit6 de Poitiers

France

BASHFORD- ROGERS Graham

Delphi Automotive Systems

Luxemburg

BASTIN

Jean-Marie

Universit6 Libre de Bruxelles

Belgium

BAZIN

Philippe

ISMRA

France

BELOT

G.

PSA Peugeot Citroen

France

BENNETT

Paul

BP Oil

United Kingdom

BERGER

Marc

Universit6 Pierre et Marie Curie

France

BERNAL

Serafin

University of Cadiz

Spain

682 BERTILSSON

Tommy

Scania

Sweden

BONNEFOY

Frederic

Bosal

Belgium

BORDET

Antoine

Johnson Matthey Ltd

United Kingdom

BORNER

Robert

Universit6 libre de Bruxelles

Belgium

BOURGES

Patrick

Institut Frangais du P6trole

France

Royal Institute of Technology

Sweden

BOUTONNET KIZLING Magali BOWKER

Michael

University of Reading Whiteknights United Kingdom

BRINKMEIER

Clemens

Universit~itStuttgart

Germany

BROECKX

Willy

Texaco

Belgium

BROWN

David

Zeton Altamira

BRUSSELAERS

Union Mini6re Hoboken

Belgium

BUESS- HERMAN

Claudine

Universit6libre de Bruxelles

Belgium

BUGLASS

John

Shell International Chemicals BV

The Netherlands

BURCH

Robbie

University of Reading, Whiteknights United Kingdom

BURGER

Beate

University of Stuttgart

Germany

Delft University

The Netherlands

CALLS - VAN KOOTEN CAPANNELL1

Gustavo

Universita Genova

Italy

CAUVEL

Anne

Katholieke Universiteit van Leuven

Belgium

CERRATO

Guiseppina

University of Turin

Italy

CHANDES

Karine

E.C.I.A.

France

CIAMBELLI

Paolo

University of Salerno

Italy

683 CONESA

Jos~ C.

Consejo Superior de Investigaciones Cientificas

Spain

CONNERTON

Jan

Nottingham Trent University

United Kingdom

CORBO

Pasquale

Instituto Motori CNR

Italy

Combo

Griselda


France

COURCOT

Dominique

Universit6 du Littoral

France

COVEY

David

Shell Additives International LTD

United Kingdom

Cox

Julian

Johnson Matthey Technology Centre

United Kingdom

DATH

Jean-Pierre

Fina Research S.A.

Belgium

DAVIAS

Nathalie

Renault S.A.

France

DAVIES

Michael J.

Rover Group Ltd

United Kingdom

DAY

J. Paul

Coming Incorporated

NY, USA

DE DEKEN

Jacques

Catalytica Inc.

CA, USA

DECKER

S6bastien


Belgium

DEMEL

Yvonne

Degussa AG

Germany

DEMIDDELEER

Jean-Pierre


Belgium

DREWSEN

Astrid


Sweden

DUBIEN

Cecile

CNRS

France

DUESTERDIEK

Thorsten

Volkswagen AG

Germany

DUPREZ

Daniel


France

DURAND

Daniel

Institut Fran~ais du P6trole

France

ECKHOFF

Stephan

University Hannover

Germany

684 EL OUADI

Brahim

Universit6 de La Rochelle

France

ERIKSSON

Lars


Sweden

ESPRIT

Marleen

Union Mini~re

Belgium

EUSDEN

Alan T.

Coming GmbH

Germany

EVANS

Julia

Johnson Matthey Technology Center United Kingdom

FARRAUTO

Bob

Engelhard Corp.

Iselin NJ, U.S.A

FAVENNEC

Jean

Rosi

France

FONTAINE

Jean-Luc

Universit6 de Liege

Belgium

FORNASIERO

Paolo

University of Reading, Whiteknights

United Kingdom

FRANCIS

Agna


Belgium

FRENNET

Alfred


Belgium

FRENNET

Elsie


Belgium

FRIDELL

Erik


Sweden

FUCALE

Michele

Fiat Auto

Italy

GABRIELSSON

Par

Haldor Topsoe A/S

Denmark

GAGNERET

Philippe

Allied Signal

France

GALTAYRIES

Anouk

Facult6s Universitaires Notre Dame de la Paix

Belgium

GARIN

Francois

Institut Le Bel

France

GELIN

Patrick

Universit6 Claude Bemard Lyon I

France

GLOCKER

Reiner

Condea Chemie Gmbh

Germany

GOERIGK

Christian

Porsche AG

Germany

685 GONZALEZVELASCO Juan R.

Universitad del Pais Vasco

Spain

GRAFFAGNO

Giovanni

Johnson Matthey Ltd

Italy

GRANGE

Paul

Universit6 Catholique de Louvain

Belgium

GRANGER

Pascal

France

GRAZIANI

Mavro

Universit6 des Sciences et Technologies de Lille University of Trieste

GRIGORYAN

Eduard

Inst. of Structural Macrokinetics RAS Russia

GUBITOSA

Giuseppe

Magneti Marelli DSS

Italy

GUILHAUME

Nolven

Universit6 Claude Bemard Lyon I

France

GUYON

Marc

Renault S.A.

France

HALASZ

Janos

Joseph Attila University

Hungary

HALPIN

Eibhlin

University of Reading, Whiteknights

United Kingdom

I-IAN

HYUN-Sik

Heesung Engelhard

Korea

HANNA

Josh

Coming Incorporate

NY, USA

HANSEN

Poul Lenvig

Haldor Topsoe A/S

Denmark

HANSSON

Maria

Volvo Car Components Corporation

Sweden

HARMSEN

Italy

Eindhoven University of Technology The Netherlands

HARRISON

Philip G.

Nottingham University

United Kingdom

HARTOFELIS

Christopher

Coming GmbH

Germany

HARTWEG

Martin

Daimler-Benz AG

Germany

HAWKER

Pelham


HECQ

Walter


Belgium

686 HENN

Jtirgen

Coming GmbH

Germany

HERKT-BRUNS

Christian

ULB

Belgium

HERZOG

Peter

Gesellschaft fiir Verbrennungskraftmaschinen und Messtechnik m.b.H.

Austria

HICKEY

Neal

University College Cork

Ireland

HJORTSBERG

Ove


Sweden

HODJATI

Shahin

ECPM- LERCSI

France

HOEBINK

J.H.B.J.


HOLMGREN

Anna


Sweden

HOLY

Gerhard

Gesellschaft ftir Verbrennungskraft maschinen und Messtechnik m.b.H.

Austria

HUBERT

Claude


Belgium

IRUSTA

Silvia

Fac d Ingenierio Qca UNL Argentino Argentina

ISMAGILOV

Zinfer

Boreskov Institute of Catalysis

Russia

JANNES

Georges

Institut Meurice

Belgium

JAPENGA

Marten

Zeton Altamira

The Netherlands

JARAS

Sven


Sweden

JAYAT

Francois


Belgium

JELLES

Sytse

Delft University

The Netherlands


Sweden

JOBSON JOHANSEN

Keld

Haldor Topsoe A/S

Denmark

JOHANSSON

Stefan


Sweden

JONES

Hannah


687 Friedrich

Emitech GmbH.

Germany

KASPAR

Jan

University of Trieste

Italy

KEENAN

Matthew

Rover Group Ltd/Warwick University

Untied Kingdom

KESSLER

Henri

Ecole Nationale Sup6rieure de Chimie France de Mulhouse

KEUNG

Michael

Condea Chemie GmBH

Germany

KHARAS

Karl C.C.

ASEC Manufactoring

Tulsa Oklahoma, U.S.A.

KIM

[3Seok Jae

Hyundai Motors co

Korea

KIRCHNER

Thomas

Bayer AG

Germany

KISENYI

Jonathan

Ford Motor Co

United Kingdom

KNAPP

Carlos

Inst. de Catalisis CSIC

Spain

KONRAD

Brigitte

Daimler-Benz AG

Germany

KONSTANDOPOULOS Athanasios

CPERI

Greece

KOPPEL

Ren6


KOSTERS

Martina

Universitiit Hannover

Germany

KREUZER

Thomas

Degussa AG

Germany

KRIJNSEN

Henrike

Delft University

The Netherlands

I~USE

Norbert


Belgium

KRUTZSCH

bernd

Daimler Benz A.G.

Germany

KUMBERGER

Otto

BASF Aktiengesellschaft A.G.

Germany

KWON

Yeong

Esso Petroleum Co

United Kingdom

LAURELL

Mats


Sweden

KAISER

688 LAVALLEY

Jean-Claude

ISMRA

France

LECLERCQ

Lucien

Universit6 des Sciences et Technologies de Lille

France

LECLERCQ

Ginette


France

LEDUC

Bemard


Belgium

LEHMANN

Ulrich

Condea Chemie GmBH

Germany

LEMAIRE

Jacques

Rh6ne-Poulenc Chimie S.A.

France

LI

Xinsheng


Belgium

LICKES

Jean-Paul

ULB

Belgium

LJUNGSTROM

Sten


Sweden

LOENDERS

Raf


Belgium

LOFBERG

Axel


France

Lox

Egbert

Degussa AG

Germany

MABILON

Gil

Institut Fran~ais du Pdtrole

France

MAKKEE

Michiel

Delft University of Technology

The Netherlands

MARECOT

Patrice


France

MARET

Dominique

E.C.I.A.

France

MARTENS

Johan


Belgium

MARTIN

Brigitte

Institut Frangais du P6trole

France

MARTIN

Ashley

Johnson Matthey Ltd

United Kingdom

MASSARDIER

Jean

Inst. de Recherches sur la Catalyse CNRS

France

MASUDA

Masaaki

NGK Europe GmbH

Germany

689 MAUNULA

Teuvo

Kemira Metalkat

Finland

MEDVEDYEV

Valentyn


Belgium

MERKLE

Friedebald

Degussa AG

Germany

Mel Chemicals

United Kingdom

MOLES MOLLER

Thomas

Condea Chemie Gmbh

Germany

MONTICELLI

Orietta


Belgium

MONTIERTH

Max R.

Coming GmbH

Germany

MOSCHOUDIS

Nikos

Aristotle University Thessaloniki

Greece

MURTAGH

Martin

Coming Incorporate

NY, USA

MUSSMANN

Lothar

Degussa AG

Germany

NIBBELKE

R.


NIEUWENHUYS

Bemard

Leiden University

The Netherlands

NOBILE

Cosimo

Politecnico di Bari

Italy

O'MEARA

Rainaldo

Johnson Matthey Ltd

United Kingdom

PAOLO

Ingemar

Chemical Engineering II

Sweden

PARMENTIER

B6atrice


Belgium

PENTENERO

Andr6

Universit6 Henri Poincar6

France

PERRICHON

Vincent

Universit6 Claude Bernard Lyon I

France

PETTERSSON

Lars J.


Sweden

PHILLIPS

Paul

Johnson Matthey Ltd

United Kingdom

PITCHON

V6ronique

LERCSI

France

690 PONCELET

Georges

Universit6 Catholique de Louvain

Belgium

PRIGENT

Michel

Institut Fran~ais du P6trole

France

PRIMET

Michel

Universit6 Claude Bernard Lyon I

France

PRIN

Marie-Agnes Pechiney

France

RICHTER

Thomas

VolkswagenAG

Germany

RIED

Thomas

TH- Darmstadt

Germany

RINGQVIST

GOran

Degussa Norden AB

Sweden

ROBOTA

Heinz J.

Allied Signal

USA

ROHE

Renaud

LERCSI

France

ROUSSEAU

Paul

Touring Secours, FEDERAUTO, Moniteur Automobile

Belgium

Russ

Gerald

Adam Opel AG

Germany

RYOO

Wan-Hyang

SABATINO

Luigina

Eniricerche S.p.A.

Italy

SALAMATI

Hassan


Grand Duch6 de Luxembourg

SALIN

Laurence

PSA- Universit6 de Paris 6

France

SCHARR

Detlef

Daimler Benz AG

Germany

SCHAY

Zolt~n

Institute of Isotopes

Hungary

SCHIMMER

Peter

Degussa AG

Germany

SCHMITT

Dietmar

TH- Darmstadt

Germany

SCHMITZ

Guy


Belgium

SCHMITZ

Eric

ULB

Belgium

S-Korea

691 SCHNEIDER

Stephanie

Automobiles Peugeot

France

SCHULZ

Philippe

Elf Antar France

France

SCOTT

Stephen

NGK Europe GmbH

United Kingdom

SEARLES

Robert

Automobile Emissions Control by Catalyst

Belgium

SEGUELONG

Thierry

Rh6ne-Poulenc Recherches S.A.

France

SEXTON

Brian


Luxemburg

SHELEF

M.

Ford Motor Co.

MI, USA

SHINJOH

Hirofumi

Toyota Central

Japan

SIEMUND

Stephan

Engelhard Technologies G m b H

Germany

SKOGLUNDH

Magnus

ChalmersUniversity of Technology

Sweden

SUNG

Shiang

Engelhard Corporation

Iselin NJ, USA

TAHIR

Saad

King's College London

United Kingdom

TANAKA

Hirohisa

Daihatsu Motor Co. Ltd.

Japan

TERWAGNE

Albert


Belgium

THORM,i~HLEN

Peter


Sweden

TORNCRONA

Anders


Sweden

TOURET

Olivier

Rh6ne - Poulenc Chimie S.A.

France

VAN DEN TILLAART

Johan

Degussa AG

Germany

VAN GEMERT

R.


VAN KOOTEN

Wijnand

Delft University

The Netherlands

VAN SETTEN

Barry

Delft University of Technology

The Netherlands

692 VAN YPEREN

Rene

Degussa AG

Germany

VARGA

Judith

Joseph Attila University

Hungary

VISART

Thierry


Belgium

W.A. BAKAR

W. Azelee

Universityof Technology Malaysia

Malaysia

WAHLBERG

Annika


Sweden


WALKER WALSH

Michael

Arlington VA USA

WARREN

James

Johnson Matthey Ltd

WATLING

Timothy

University of Reading, Whiteknights UnitedKingdom

WEIBEL

Michel

Daimler Benz AG

Germany

WESTERBERG

Bjorn

Chemical Engineering II

Sweden

ZAKUMBAEVA

Gaoukhar

ZANDIRI

Stefania

Centre Ricerche Fiat

Republic of Kazakhstan Italy

ZENG

Hua Chun

NationalUniversity of Singapore

Singapore

United Kingdom

693 STUDIES IN SURFACE SCIENCE A N D CATALYSIS Advisory Editors: B. Delmon, Universit6 Catholique de Louvain, Louvain-la-Neuve, Belgium J.T. Yates, University of Pittsburgh, Pittsburgh, PA, U.S.A. Volume 1

Volume 2

Volume 3

Volume 4

Volume 5

Volume 6 Volume 7 Volume 8 Volume 9 Volume 10 Volume 11

Volume 12 Volume 13 Volume 14 Volume 15

Preparation of Catalysts I.Scientific Bases for the Preparation of Heterogeneous Catalysts. Proceedings of the First International Symposium, Brussels, October 14-17,1975 edited by B. Delmon, P.A.Jacobs and G. Poncelet The Control of the Reactivity of Solids. A Critical Survey of the Factors that Influence the Reactivity of Solids, with Special Emphasis on the Control ofthe Chemical Processes in Relation to Practical Applications by V.V. Boldyrev, M. Bulens and B. Delmon Preparation of Catalysts I1. Scientific Bases for the Preparation of Heterogeneous Catalysts. Proceedings of the Second International Symposium, Louvain-la-Neuve, September 4-7, 1978 edited by B. Delmon, P.Grange, P.Jacobs and G. Poncelet Growth and Properties of Metal Clusters. Applications to Catalysis and the Photographic Process. Proceedings of the 32nd International Meeting of the Soci~t~ de Chimie Physique, Villeurbanne, September 24-28, 1979 edited by J. Bourdon Catalysis by Zeolites. Proceedings of an International Symposium, Ecully (Lyon), September 9-11, 1980 edited by B. Imelik, C. Naccache, u Ben Taarit, J.C. Vedrine, G. Coudurier and H. Praliaud Catalyst Deactivation. Proceedings of an International Symposium, Antwerp, October 13-15,1980 edited by B. Delmon and G.E Froment New Horizons in Catalysis. Proceedings of the 7th International Congress on Catalysis, Tokyo, June 30-July4, 1980. Parts A and B edited by T. Seiyama and K. Tanabe Catalysis by Supported Complexes by u u B.N. Kuznetsov and V.A. Zakharov Physics of Solid Surfaces. Proceedings of a Symposium, Bechyhe, September 29-October 3,1980 edited by M. L&zni~:ka Adsorption at the Gas-Solid and Liquid-Solid Interface. Proceedings of an International Symposium, Aix-en-Provence, September 21-23, 1981 edited by J. Rouquerol and K.S.W. Sing Metal-Support and Metal-Additive Effects in Catalysis. Proceedings of an International Symposium, Ecully (Lyon), September 14-16, 1982 edited by B. Imelik, C. Naccache, G. Coudurier, H. Praliaud, P. Meriaudeau, P. Gallezot, G.A. Martin and J.C. Vedrine Metal Microstructures in Zeolites. Preparation - Properties - Applications. Proceedings of a Workshop, Bremen, September 22-24, 1982 edited by P.A. Jacobs, N.I. Jaeger, P.Jin3 and G. Schulz-Ekloff Adsorption on Metal Surfaces. An Integrated Approach edited by J. Bdnard Vibrations at Surfaces. Proceedings of the Third International Conference, Asilomar, CA, September 1-4, 1982 edited by C.R. Brundle and H. Morawitz Heterogeneous Catalytic Reactions Involving Molecular Oxygen by G.I. Golodets

694 Volume 16

Volume 17 Volume 18 Volume 19 Volume 20 Volume 21 Volume 22 Volume 23 Volume 24 Volume 25 Volume 26 Volume 27 Volume 28 Volume 29 Volume 30 Volume 31


Preparation of Catalysts i11.Scientific Bases for the Preparation of Heterogeneous Catalysts. Proceedings of the Third International Symposium, Louvain-la-Neuve, September 6-9, 1982 edited by G. Poncelet, P. Grange and P.A.Jacobs Spillover of Adsorbed Species. Proceedings of an International Symposium, Lyon-Villeurbanne, September 12-16, 1983 edited by G.M. Pajonk, S.J. Teichner and J.E. Germain Structure and Reactivity of Modified Zeolites. Proceedings of an International Conference, Prague, July 9-13, 1984 edited by P.A.Jacobs, N.I. Jaeger, P.JidJ, V.B. Kazansky and G. Schulz-Ekloff Catalysis on the Energy Scene. Proceedings of the 9th Canadian Symposium on Catalysis, Quebec, P.Q.,September 30-October 3, 1984 edited by S. Kaliaguine and A. Mahay Catalysis by Acids and Bases. Proceedings of an International Symposium, Villeurbanne (Lyon), September 25-27, 1984 edited by B. Imelik, C. Naccache, G. Coudurier, Y. Ben Taarit and J.C. Vedrine Adsorption and Catalysis on Oxide Surfaces. Proceedings of a Symposium, Uxbridge, June 28-29, 1984 edited by M. Che and G.C. Bond Unsteady Processes in Catalytic Reactors by Yu.Sh. Matros Physics of Solid Surfaces 1984 edited by J. Koukal Zeolites: Synthesis, Structure, Technology and Application. Proceedings of an International Symposium, Portoro~-Portorose, September 3-8, 1984 edited by B. Dr;~aj,S. HoEevar and S. Pejovnik Catalytic Polymerization of Olefins. Proceedings of the International Symposium on Future Aspects of Olefin Polymerization, Tokyo, July 4-6, 1985 edited by T. Keii and K. Soga Vibrations at Surfaces 1985. Proceedings of the Fourth International Conference, Bowness-on-Windermere, September 15-19, 1985 edited by D.A. King, N.V. Richardson and S. Holloway Catalytic Hydrogenation edited by L. Cerven~ New Developments in Zeolite Science and Technology. Proceedings of the 7th International Zeolite Conference, Tokyo, August 17-22, 1986 edited by Y. Murakami, A. lijima and J.W. Ward Metal Clusters in Catalysis edited by B.C. Gates, L. Guczi and H. Kn6zinger Catalysis and Automotive Pollution Control. Proceedings of the First International Symposium, Brussels, September 8-11, 1986 edited by A. Crucq and A. Frennet Preparation of Catalysts IV. Scientific Bases for the Preparation of Heterogeneous Catalysts. Proceedings of the Fourth International Symposium, Louvain-la-Neuve, September 1-4, 1986 edited by B. Delmon, P.Grange, P.A. Jacobs and G. Poncelet Thin Metal Films and Gas Chemisorption edited by R Wissmann Synthesis of High-silica Aluminosilicate Zeolites edited by P.A.Jacobs and J.A. Martens Catalyst Deactivation 1987. Proceedings of the 4th International Symposium, Antwerp, September 29-October 1, 1987 edited by B. Delmon and G.F. Froment Keynotes in Energy-Related Catalysis edited by S. Kaliaguine

695 Volume 36 Volume 37 Volume 38 Volume 39 Volume 40 Volume 41

volume 42 Volume 43 Volume 44

Volume 45 Volume 46



Methane Conversion. Proceedings of a Symposium on the Production of Fuels and Chemicals from Natural Gas, Auckland, April 27-30, 1987 edited by D.M. Bibby, C.D. Chang, R.F. Howe and S. Yurchak Innovation in Zeolite Materials Science. Proceedings of an International Symposium, Nieuwpoort, September 13-17, 1987 edited by P.J. Grobet, W.J. Mortier, E.F. Vansant and G. Schulz-Ekloff Catalysis 1987. Proceedings of the 10th North American Meeting of the Catalysis Society, San Diego, CA, May 17-22, 1987 edited by J.W. Ward Characterization of Porous Solids. Proceedings of the IUPAC Symposium (COPS I), Bad Soden a. Ts., April 26-29,1987 edited by K.K. Unger, J. Rouquerol, K.S.W. Sing and H. Kral Physics of Solid Surfaces 1987. Proceedings of the Fourth Symposium on Surface Physics, Bechyne Castle, September 7-11, 1987 edited by J. Koukal Heterogeneous Catalysis and Fine Chemicals. Proceedings of an International Symposium, Poitiers, March 15-17, 1988 edited by M. Guisnet, J. Barrault, C. Bouchoule, D. Duprez, C. Montassier and G. Perot Laboratory Studies of Heterogeneous Catalytic Processes by E.G. Christoffel, revised and edited by Z. Pa&l Catalytic Processes under Unsteady-State Conditions by Yu. Sh. Matros Successful Design of Catalysts. Future Requirements and Development. Proceedings ofthe Worldwide Catalysis Seminars, July, 1988, on the Occasion of the 30th Anniversary of the Catalysis Society of Japan edited by T. Inui Transition Metal Oxides. Surface Chemistry and Catalysis by H.H. Kung Zeolites as Catalysts, Sorbents and Detergent Builders. Applications and Innovations. Proceedings of an International Symposium, W~rzburg, September 4-8,1988 edited by H.G. Karge and J. Weitkamp Photochemistry on Solid Surfaces edited by M. Anpo and T. Matsuura Structure and Reactivity of Surfaces. Proceedings of a European Conference, Trieste, September 13-16, 1988 edited by C. Morterra, A. Zecchina and G. Costa Zeolites: Facts, Figures, Future. Proceedings of the 8th International Zeolite Conference, Amsterdam, July 10-14, 1989. Parts A and B edited by P.A. Jacobs and R.A. van Santen Hydrotreating Catalysts. Preparation, Characterization and Performance. Proceedings of the Annual International AIChE Meeting, Washington, DC, November 27-December 2, 1988 edited by M.L. Occelli and R.G. Anthony New Solid Acids and Bases. Their Catalytic Properties by K. Tanabe, M. Misono, Y. Ono and H. Hattori Recent Advances in Zeolite Science. Proceedings of the 1989 Meeting of the British Zeolite Association, Cambridge, April 17-19, 1989 edited by J. Klinowsky and P.J. Barrie Catalyst in Petroleum Refining 1989. Proceedings of the First International Conference on Catalysts in Petroleum Refining, Kuwait, March 5-8, 1989 edited by D.L. Trimm, S. Akashah, M. Absi-Halabi and A. Bishara Future Opportunities in Catalytic and Separation Technology edited by M. Misono, Y. Moro-oka and S. Kimura

696 Volume 55

New Developments in Selective Oxidation. Proceedings of an International Symposium, Rimini, Italy, September 18-22, 1989 edited by G. Centi and F.Trifiro Volume 56 Olefin Polymerization Catalysts. Proceedings of the International Symposium on Recent Developments in Olefin Polymerization Catalysts, Tokyo, October 23-25,1989 edited by T. Keii and K. Soga Volume 57A Spectroscopic Analysis of Heterogeneous Catalysts. Part A: Methods of Surface Analysis edited by J.L.G. Fierro Volume 57B Spectroscopic Analysis of Heterogeneous Catalysts. Part B: Chemisorption of Probe Molecules edited by J.L.G. Fierro Volume 58 Introduction to Zeolite Science and Practice edited by H. van Bekkum, E.M. Flanigen and J.C. Jansen Volume 59 Heterogeneous Catalysis and Fine Chemicals II. Proceedings of the 2nd International Symposium, Poitiers, October 2-6, 1990 edited by M. Guisnet, J. Barrault, C. Bouchoule, D. Duprez, G. Pdrot, R. Maurel and C. Montassier Volume 60 Chemistry of Microporous Crystals. Proceedings of the International Symposium on Chemistry of Microporous Crystals, Tokyo, June 26-29, 1990 edited by T. Inui, S. Namba and T. Tatsumi Volume 61 Natural Gas Conversion. Proceedings of the Symposium on Natural Gas Conversion, Oslo, August 12-17, 1990 edited by A. Holmen, K.-J. Jens and S. Kolboe Volume 62 Characterization of Porous Solids II. Proceedings of the IUPAC Symposium (COPS II), Alicante, May 6-9, 1990 edited by E Rodriguez-Reinoso, J. Rouquerol, K.S.W. Sing and K.K. Unger Preparation of Catalysts V. Scientific Bases for the Preparation of Heterogeneous Volume 63 Catalysts. Proceedings of the Fifth International Symposium, Louvain-la-Neuve, September 3-6, 1990 edited by G. Poncelet, P.A.Jacobs, P.Grange and B. Delmon New Trends in CO Activation Volume 64 edited by L. Guczi Catalysis and Adsorption by Zeolites. Proceedings of ZEOCAT90, Leipzig, Volume 65 August 20-23, 1990 edited by G. (~hlmann, H. Pfeifer and R. Fricke Dioxygen Activation and Homogeneous Catalytic Oxidation. Proceedings of the Volume 66 Fourth International Symposium on Dioxygen Activation and Homogeneous Catalytic Oxidation, Balatonffired, September 10-14, 1990 edited by L.I. Simfindi Structure-Activity and Selectivity Relationships in Heterogeneous Catalysis. Volume 67 Proceedings ofthe ACS Symposium on Structu re-Activity Relationships in Heterogeneous Catalysis, Boston, MA, April 22-27, 1990 edited by R.K. Grasselli and A.W. Sleight Catalyst Deactivation 1991. Proceedings of the Fifth International Symposium, Volume 68 Evanston, IL, June 24-26, 1991 edited by C.H. Bartholomew and J.B. Butt Zeolite Chemistry and Catalysis. Proceedings of an International Symposium, Volume 69 Prague, Czechoslovakia, September 8-13, 1991 edited by P.A.Jacobs, N.I. Jaeger, L. Kubelkovfi and B. Wichterlovfi Poisoning and Promotion in Catalysis based on Surface Science Concepts and Volume 70 Experiments by M. Kiskinova

697 Volume 71 Volume 72

Volume 73 Volume 74 Volume 75 Volume 76 Volume 77

Volume78

Volume79 Volume 80 Volume 81 Volume82

Volume 83 Volume84

Volume 85 Volume 86 Volume 87

Catalysis and Automotive Pollution Control II. Proceedings of the 2nd International Symposium (CAPoC 2), Brussels, Belgium, September 10-13, 1990 edited by A. Crucq New Developments in Selective Oxidation by Heterogeneous Catalysis. Proceedings of the 3rd European Workshop Meeting on New Developments in Selective Oxidation by Heterogeneous Catalysis, Louvain-la-Neuve, Belgium, April 8-10, 1991 edited by R Ruiz and B. Delmon Progress in Catalysis. Proceedings of the 12th Canadian Symposium on Catalysis, Banff, Alberta, Canada, May 25-28, 1992 edited by K.J. Smith and E.C. Sanford Angle-Resolved Photoemission. Theory and Current Applications edited by S.D. Kevan New Frontiers in Catalysis, Parts A-C. Proceedings of the 10th International Congress on Catalysis, Budapest, Hungary, 19-24 July, 1992 edited by L. Guczi, F. Solymosi and R T6t6nyi Fluid Catalytic Cracking: Science and Technology edited by J.S. Magee and M.M. Mitchell, Jr. New Aspects of Spillover Effect in Catalysis. For Development of Highly Active Catalysts. Proceedings of the Third International Conference on Spillover, Kyoto, Japan, August 17-20, 1993 edited by T. Inui, K. Fujimoto, T. Uchijima and M. Masai Heterogeneous Catalysis and Fine Chemicals III. Proceedings of the 3rd International Symposium, Poitiers, April 5 - 8, 1993 edited by M. Guisnet, J. Barbier, J. Barrault, C. Bouchoule, D. Duprez, G. P6rot and C. Montassier Catalysis: An Integrated Approach to Homogeneous, Heterogeneous and Industrial Catalysis edited by J.A. Moulijn, RW.N.Mo van Leeuwen and R.A. van Santen Fundamentals of Adsorption. Proceedings of the Fourth International Conference on Fundamentals of Adsorption, Kyoto, Japan, May 17-22, 1992 edited by M. Suzuki Natural Gas Conversion II. Proceedings ofthe Third Natural Gas Conversion Symposium, Sydney, July 4-9, 1993 edited by H.E. Curry-Hyde and R.F. Howe New Developments in Selective Oxidation II. Proceedings of the Second World Congress and Fourth European Workshop Meeting, Benalm~dena, Spain, September 20-24, 1993 edited by V. Cortes Corberdn and S. Vic Bell6n Zeolites and Microporous Crystals. Proceedings of the International Symposium on Zeolites and Microporous Crystals, Nagoya, Japan, August 22-25, 1993 edited by T. Hattori and T. Yashima Zeolites and Related Microporous Materials: State of the Art 1994. Proceedings ofthe 10th International Zeolite Conference, Garmisch-Partenkirchen, Germany, July 17-22, 1994 edited by J. Weitkamp, H.G. Karge, H. Pfeifer and W. H61derich Advanced Zeolite Science and Applications edited by J.C. Jansen, M. St6cker, H.G. Karge and J.Weitkamp Oscillating Heterogeneous Catalytic Systems by M.M. Slin'ko and N.I. Jaeger Characterization of Porous Solids III. Proceedings of the IUPAC Symposium (COPS III), Marseille, France, May 9-12, 1993 edited by J.Rouquerol, F. Rodriguez-Reinoso, K.S.W. Sing and K.K. Unger

698 Volume 88 Volume89

Volume90 Volume91

Volume 92

Volume 93 Volume94 Volume 95 Volume 96

Volume 97 Volume98

Volume 99 Volume 100

Volume 101 Volume 102 Volume 103 Volume 104 Volume 105

Catalyst Deactivation 1994. Proceedings ofthe 6th International Symposium, Ostend, Belgium, October 3-5, 1994 edited by B. Delmon and G.F. Froment Catalyst Design for Tailor-made Polyolefins. Proceedings of the International Symposium on Catalyst Design for Tailor-made Polyolefins, Kanazawa, Japan, March 10-12, 1994 edited by K. Soga and M. Terano Acid-Base Catalysis II. Proceedings of the International Symposium on Acid-Base Catalysis II, Sapporo, Japan, December 2-4, 1993 edited by H. Hattori, M. Misono and Y. Ono Preparation of Catalysts VI. Scientific Bases for the Preparation of Heterogeneous Catalysts. Proceedings of the Sixth International Symposium, Louvain-La-Neuve, September 5-8, 1994 edited by G. Poncelet, J. Martens, B. Delmon, P.A. Jacobs and P.Grange Science and Technology in Catalysis 1994. Proceedings of the Second Tokyo Conference on Advanced Catalytic Science and Technology, Tokyo, August 21-26, 1994 edited by Y. Izumi, H. Arai and M. Iwamoto Characterization and Chemical Modification of the Silica Surface by E.F.Vansant, P.Van Der Voort and K.C. Vrancken Catalysis by Microporous Materials. Proceedings of ZEOCAT'95, Szombathely, Hungary, July 9-13,1995 edited by H.K. Beyer, H.G.Karge, I. Kiricsi and J.B. Nagy Catalysis by Metals and Alloys by V. Ponec and G.C. Bond Catalysis and Automotive Pollution Control II1. Proceedings of the Third International Symposium (CAPoC3), Brussels, Belgium, April 20-22, 1994 edited by A. Frennet and J.-M. Bastin Zeolites: A Refined Tool for Designing Catalytic Sites. Proceedings of the International Symposium, Qu6bec, Canada, October 15-20, 1995 edited by L. Bonneviot and S. Kaliaguine Zeolite Science 1994: Recent Progress and Discussions. Supplementary Materials to the 10th International Zeolite Conference, Garmisch-Partenkirchen, Germany, July 17-22, 1994 edited by H.G. Karge and J. Weitkamp Adsorption on New and Modified Inorganic Sorbents edited by A. Dqbrowski and V.A. Tertykh Catalysts in Petroleum Refining and Petrochemical Industries 1995. Proceedings of the 2nd International Conference on Catalysts in Petroleum Refining and Petrochemical Industries, Kuwait, April 22-26, 1995 edited by M. Absi-Halabi, J. Beshara, H. Qabazard and A. Stanislaus 1lth International Congress on Catalysis - 40th Anniversary. Proceedings ofthe 1lth ICC, Baltimore, MD, USA, June 30-July 5, 1996 edited by J. W. Hightower, W.N. Delgass, E. Iglesia and A.T. Bell Recent Advances and New Horizons in Zeolite Science and Technology edited by H. Chon, S.I. Woo and S. -E. Park Semiconductor Nanoclusters - Physical, Chemical, and Catalytic Aspects edited by P.V. Kamat and D. Meisel Equilibria and Dynamics of Gas Adsorption on Heterogeneous Solid Surfaces edited by W. Rudzifiski, W.A. Steele and G. Zgrablich Progress in Zeolite and Microporous Materials Proceedings ofthe 1lth International Zeolite Conference, Seoul, Korea, August 12-17,1996 edited by H. Chon, S.-K. Ihm and Y.S. Uh

699 Volume 106

Volume 107

Volume 108



Volume 113

Volume 114


Hydrotreatment and Hydrocracking of Oil Fractions Proceedings ofthe 1st International Symposium / 6th European Workshop, Oostende, Belgium, February 17-19, 1997 edited by G.F. Froment, B. Delmon and R Grange Natural Gas Conversion IV Proceedings ofthe 4th International Natural Gas Conversion Symposium, Kruger Park, South Africa, November 19-23, 1995 edited by M. de Pontes, R.L. Espinoza, C.P. Nicolaides, J.H. Scholtz and M.S. Scurrell Heterogeneous Catalysis and Fine Chemicals IV Proceedings of the 4th International Symposium on Heterogeneous Catalysis and Fine Chemicals, Basel, Switzerland, September 8-12, 1996 edited by H.U. Blaser, A. Baiker and R. Prins Dynamics of Surfaces and Reaction Kinetics in Heterogeneous Catalysis. Proceedings ofthe International Symposium, Antwerp, Belgium, September 15-17,1997 edited by G.F. Froment and K.C. Waugh Third World Congress on Oxidation Catalysis. Proceedings of the Third World Congress on Oxidation Catalysis, San Diego, CA, U.S.A., 21-26 September 1997 edited by R.K. Grasselli, S.T. Oyama, A.M. Gaffney and J.E. Lyons Catalyst Deactivation 1997. Proceedings of the 7th International Symposium, Cancun, Mexico, October 5-8, 1997 edited by C.H. Bartholomew and G.A. Fuentes Spillover and Migration of Surface Species on Catalysts. Proceedings ofthe 4th International Conference on Spillover, Dalian, China, September 15-18, 1997 edited by Can Li and Qin Xin Recent Advances in Basic and Applied Aspects of Industrial Catalysis. Proceedings ofthe 13th National Symposium and Silver Jubilee Symposium of Catalysis of India, Dehradun, India, April 2-4,1997 edited by T.S.R. Prasada Rao and G. Murali Dhar Advances in Chemical Conversions for Mitigating Carbon Dioxide. Proceedings of the 4th International Conference on Carbon Dioxide Utilization, Kyoto, Japan, September 7-11, 1997 edited by T. Inui, M. Anpo, K. Izui, S. Yanagida and T. Yamaguchi Methods for Monitoring and Diagnosing the Efficiency of Catalytic Converters. A patent-oriented survey by M. Sideris Catalysis and Automotive Pollution Control IV. Proceedings ofthe 4th International Symposium (CAPoC4), Brussels, Belgium, April 9-11, 1997 edited by N. Kruse, A. Frennet and J.-M. Bastin


Catalysis and automotive pollution control IV: proceedings of the Fourth International Symposium (CAPoC4), Brussels, Belgium, April 9-11, 1997

Catalysis and Automotive Pollution Control II

Dynamics of surfaces and reaction kinetics in heterogeneous catalysis: proceedings of the international symposium, Antwerp, Belgium, September 15-17, 1997

Catalysis and Automotive Pollution Control (Studies in Surface Science and Catalysis)

Zeolite Chemistry and Catalysis Proceedings of an International Symposium Prague

Catalysis and Automotive Pollution Control III (Studies in Surface Science and Catalysis, Volume 96)

Environmental Pollution and Control, Fourth Edition

Environmental Pollution and Control, Fourth Edition

System Configuration Management: ECOOP'98 SCM-8 Symposium, Brussels, Belgium, July 20-21, 1998, Proceedings

Catalysis by Zeolites: International Symposium Proceedings (Studies in surface science and catalysis)

System Configuration Management: ECOOP'98 SCM-8 Symposium, Brussels, Belgium, July 20-21, 1998, Proceedings

Circuit Cellar (April 1997)

Environmental Pollution and Control

Pollution Engineering April 2011

Perception and Evaluation of Cultural Landscapes: Proceedings of an International symposium, Zakynthos, December 1997

Quantum theory and symmetries. Proceedings of the 3rd international symposium

Computer mathematics : proceedings of the Fourth Asian Symposium (ASCM 2000)

Sustainable Automotive Technologies 2011: Proceedings of the 3rd International Conference

Heterogeneous Catalysis and Fine Chemicals III: Proceedings of the 3rd International Symposium, Poitiers, April 5-8, 1993 (Studies in Surface Science and Catalysis)

Lepton and photon'01: proceedings of the XX International Symposium

Preparation of Catalysts: 2nd: International Symposium Proceedings (Studies in Surface Science and Catalysis)

Coordination Models and Languages: 7th International Conference, COORDINATION 2005, Namur, Belgium, April 20-23, 2005, Proceedings

Industrial Combustion Pollution and Control

Integrated Pollution Prevention and Control

Frontiers of Fundamental Physics : Proceedings of the Sixth International Symposium

Industrial Combustion Pollution and Control

Pollution: Causes, Effects and Control

Pollution Control and Energy Needs

Costs of Air Pollution Control

Frontiers of Applied Mathematics: Proceedings of the 2nd International Symposium

Monitoring, Control and Effects of Air Pollution

Catalysis and automotive pollution control IV: proceedings of the Fourth International Symposium (CAPoC4), Brussels, Belgium, April 9-11, 1997

Catalysis and Automotive Pollution Control II

Dynamics of surfaces and reaction kinetics in heterogeneous catalysis: proceedings of the international symposium, Antwerp, Belgium, September 15-17, 1997

Catalysis and Automotive Pollution Control (Studies in Surface Science and Catalysis)

Zeolite Chemistry and Catalysis Proceedings of an International Symposium Prague

Catalysis and Automotive Pollution Control III (Studies in Surface Science and Catalysis, Volume 96)

Environmental Pollution and Control, Fourth Edition

Environmental Pollution and Control, Fourth Edition

System Configuration Management: ECOOP'98 SCM-8 Symposium, Brussels, Belgium, July 20-21, 1998, Proceedings

Catalysis by Zeolites: International Symposium Proceedings (Studies in surface science and catalysis)

System Configuration Management: ECOOP'98 SCM-8 Symposium, Brussels, Belgium, July 20-21, 1998, Proceedings

Circuit Cellar (April 1997)

Environmental Pollution and Control

Pollution Engineering April 2011

Perception and Evaluation of Cultural Landscapes: Proceedings of an International symposium, Zakynthos, December 1997

Quantum theory and symmetries. Proceedings of the 3rd international symposium

Computer mathematics : proceedings of the Fourth Asian Symposium (ASCM 2000)

Sustainable Automotive Technologies 2011: Proceedings of the 3rd International Conference

Heterogeneous Catalysis and Fine Chemicals III: Proceedings of the 3rd International Symposium, Poitiers, April 5-8, 1993 (Studies in Surface Science and Catalysis)

Lepton and photon'01: proceedings of the XX International Symposium

Preparation of Catalysts: 2nd: International Symposium Proceedings (Studies in Surface Science and Catalysis)

Coordination Models and Languages: 7th International Conference, COORDINATION 2005, Namur, Belgium, April 20-23, 2005, Proceedings

Industrial Combustion Pollution and Control

Integrated Pollution Prevention and Control

Frontiers of Fundamental Physics : Proceedings of the Sixth International Symposium

Industrial Combustion Pollution and Control

Pollution: Causes, Effects and Control

Pollution Control and Energy Needs

Costs of Air Pollution Control

Frontiers of Applied Mathematics: Proceedings of the 2nd International Symposium

Monitoring, Control and Effects of Air Pollution

Recommend Documents